EDITORIAL
Postoperative Sore Throat: More Answers
Than Questions
Phillip E. Scuderi, MD
P ostoperative sore throat (POST) is a common adverse
event after general anesthesia. Typically, the incidence
of POST is highest in patients who are tracheally
intubated; however, POST also occurs when a laryngeal mask
airway (LMA) is used.1 Even patients who are managed with
a facemask are not immune.1 Most of the measures that have
been recommended for reducing this complication have been
directed at limiting the physical trauma that might result from
airway instrumentation and manipulation. Surprisingly few
investigations have evaluated pharmacologic interventions
as a means of reducing POST. Furthermore, no single drug
has achieved widespread acceptance in the clinical commu-
nity. In this issue of Anesthesia & Analgesia, 4 articles
describe simple prophylactic measures that seem to signifi-
cantly reduce the incidence of POST.2–5 Two of these
articles evaluate the effectiveness of topical benzydamine
hydrochloride applied to the cuff of the endotracheal tube,
directly to the pharyngeal mucosa, or both.2,3 A third article
evaluates the efficacy of inhaled fluticasone propionate.4
The fourth article evaluates Strepsils威, a nonprescription
lozenge that contains 2 active ingredients, amylmetacresol
and 2,4-dichlorobenzyl alcohol.5
When considered in aggregate, these 4 articles raise a
number of interesting questions. The expression “sore
throat” is obviously common to the vernacular of many
different cultures, yet it provides at best a parsimonious
description of the actual phenomena. Consequently, the
expression “postoperative sore throat” likely represents a
broad constellation of signs and symptoms. For instance, in
its simplest form, sore throat is a lay description of phar-
yngitis, which in itself can have a variety of causes.
However, sore throat may also include a variety of symp-
toms including laryngitis, tracheitis, hoarseness, cough, or
dysphagia. Postoperatively, it seems most plausible that the
symptoms are the result of mucosal injury with resulting
inflammation caused by the process of airway instrumen-
tation (i.e., laryngoscopy and suctioning) or the irritating
effects of a foreign object (i.e., endotracheal tube, LMA, or
From the Department of Anesthesiology, Wake Forest University School of
Medicine, Winston-Salem, North Carolina.
Accepted for publication June 14, 2010.
Disclosure: The author reports no conflicts of interest.
Address correspondence and reprint requests to Phillip E. Scuderi, MD,
Department of Anesthesiology, Wake Forest University School of Medicine,
Medical Center Blvd., Winston-Salem, NC 27157-1009. Address e-mail to
pscuderi@wfubmc.edu.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181ee85c7
October 2010 • Volume 111 • Number 4
oral airway). The site or sites of mucosal injury would
obviously vary depending on the airway device. For in-
stance, endotracheal intubation can result in injury to any
portion of the pharynx as well as injury to the larynx and
trachea. Placement of an LMA can reasonably be expected
to cause injury to pharyngeal mucosa in the supraglottic
regions only, whereas the use of a facemask with an oral
airway should result in injury to only the oropharynx,
assuming that no other injuries occurred because of suc-
tioning or other airway maneuvers. It is therefore some-
what surprising to note that the reported incidence of POST
after LMA insertion is, at least in some studies, remarkably
similar to that seen with endotracheal intubation.6,7 Al-
though this might lead one to infer that the mechanism and
location of injury must also be similar, a number of facts
seem to contradict this assumption. For instance, reducing
the size of endotracheal tubes results in a significant
decrease in the incidence of POST.8 The design of tube cuffs
has also been an area of intense research. The size,
pressure/volume characteristics, and shape of cuff have all
been implicated in tracheal mucosal injury and resultant
POST.9 –12 Conversely, it has been suggested that cuff
inflation pressure has less of a role in POST when an LMA
is used.6 Both airway devices are clearly capable of induc-
ing mucosal irritation and both can cause POST in patients
at rates that are not strikingly different. Yet, anatomically,
the site or sites of injury cannot be the same.
There are several interesting observations that arise
when one examines the data presented in the 4 articles
published in this issue of Anesthesia & Analgesia. The data
from Tazeh-kand et al.4 demonstrate that inhalation of
fluticasone propionate before the induction of anesthesia
significantly reduces the incidence of POST at 1 hour and
24 hours after surgery compared with a placebo control.
This is not necessarily an unexpected result. Topical13,14
and systemic steroids15 have been demonstrated to reduce
the incidence of POST presumably because of their sys-
temic antiinflammatory effects. More puzzling are the data
presented by Ebneshahidi and Mohseni.5 Patients who
received a Strepsils lozenge before the induction of anes-
thesia had a significantly lower incidence of POST and
hoarseness both in the postanesthesia care unit and at 24
hours after surgery than did the placebo control group that
received a flavored lozenge without the active ingredient.
Unless one postulates a systemic effect from the active
ingredients in the Strepsils lozenge (i.e., amylmetacresol
and 2,4-dichlorobenzyl alcohol), the effect site must be the
www.anesthesia-analgesia.org 831
 EDITORIAL
pharyngeal mucosa. Whereas it is plausible to postulate a
reduction on pharyngeal irritation due to the lozenge, it is
harder to postulate a mechanism for the reduction in
“hoarseness” that was reported in this study. It is also
difficult to understand how a preoperative lozenge could
reduce a reaction to injury to the larynx and trachea and the
resultant laryngotracheitis that must have a role in POST
that occurs after intubation.
Conversely, 2 of the articles2,3 describe a reduction in
POST with the application of benzydamine hydrochloride
to the endotracheal tube cuff alone compared with a
placebo control (normal saline and distilled water, respec-
tively). It seems unlikely that the small dose of benzydam-
ine hydrochloride (1.5 and 0.75 mg, respectively) used
would have resulted in a systemic effect. Therefore, the
reduction in POST that was observed in both of these
studies must be assumed to have resulted from a localized
decrease in mucosal injury and/or inflammatory response.
The incidence of and reduction in POST is strikingly similar
when Strepsils lozenges were used when compared with the
application of benzydamine hydrochloride to the endotra-
cheal tube cuff. There are unavoidable questions that must be
asked: How can a lozenge that is administered orally result in
a similar reduction in POST when compared with the topical
application of benzydamine hydrochloride to the endotra-
cheal tube cuff? In addition, How can either an oral lozenge or
topical antiinflammatory agent applied to an endotracheal
tube cuff yield reductions in POST that compare favorably
with a more widespread application of topical steroids to both
the pharyngeal and laryngotracheal mucosa?
POST is unquestionably a common adverse event after
general anesthesia. A number of physical factors have been
implicated as noted above. Most notable would seem to be
endotracheal tube and cuff design and the approach to airway
management (i.e., endotracheal tube, LMA, or mask anesthe-
sia). In addition, female gender, younger patients, gynecologic
surgery, and the use of succinylcholine also seem to increase
the incidence.16 Of particular note, the use of topical lidocaine
appears to confer no benefit and may in fact make POST
worse,17,18 a fact that seems to have been confirmed by Hung
et al.2 However, what actually causes POST remains some-
thing of a mystery. Hoarseness is a physical sign and can
certainly be evaluated objectively by a careful observer. Dys-
phagia, although a symptom, is perhaps less likely to be
influenced by intersubject variation in reporting, particularly
if questioning is done by a trained observer and is consistently
applied. Sore throat is more problematic. As noted above, the
location of mucosal injury can vary widely but still result in a
subject complaining of a “sore throat.” The 4 articles pre-
sented in this issue of Anesthesia & Analgesia describe very
different strategies yet all achieved positive results. However,
we are no closer to understanding the actual etiology of POST
than we were before. A reasonable question to ask is “Does it
matter?” I believe that the answer is “yes.” If the precise
etiology (or etiologies) of pain after airway management can
be determined, it increases the likelihood that a specific
therapy or therapies could be recommended that would
decrease symptoms and improve outcomes. POST is not the
most important adverse event to avoid, at least from a
patient’s perspective.19 Nevertheless, it is an adverse event
that could easily be significantly decreased or even potentially
832 www.anesthesia-analgesia.org
eliminated. The 4 studies presented here may provide the
impetus for a more careful evaluation of POST resulting in
more precisely targeted therapies.
AUTHOR CONTRIBUTIONS
PES designed and conducted the study, analyzed the data, and
wrote the manuscript. This author approved the final manuscript.
REFERENCES
1. Higgins PP, Chung F, Mezei G. Postoperative sore throat after
ambulatory surgery. Br J Anaesth 2002;88:582– 4
2. Hung NK, Wu CT, Chan SM, Lu CH, Hang YS, Yeh CC, Lee MS,
Cherng CH. The effect on postoperative sore throat of spraying
the endotracheal tube cuff with benzydamine hydrochloride, 10%
lidocaine, and 2% lidocaine. Anesth Analg 2010;111:882– 6
3. Huang YS, Hung NK, Lee MS, Kuo CP, Yu JC, Huang GS,
Cherng CH, Wong CS, Chu CH, Wu CT. The effectiveness of
benzydamine hydrochloride spray on the endotracheal tube
cuff or oral mucosa for postoperative sore throat. Anesthesia
Analg 2010;111:887–91
4. Tazeh-kand NF, Eslami B, Mohammadian K. Inhaled flutica-
sone propionate reduces postoperative sore throat, cough and
hoarseness. Anesth Analg 2010;111:895– 8
5. Ebneshahidi A, Mohseni M. Strepsils威 tablets reduce sore throat and
hoarseness after tracheal intubation. Anesth Analg 2010;111:892–4
6. Wakeling HG, Butler PJ, Baxter PJ. The laryngeal mask airway:
a comparison between two insertion techniques. Anesth Analg
1997;85:687–90
7. Joshi GP, Inagaki Y, White PF, Taylor-Kennedy L, Wat LI,
Gevirtz C, McCraney JM, McCulloch DA. Use of the laryngeal
mask airway as an alternative to the tracheal tube during
ambulatory anesthesia. Anesth Analg 1997;85:573–7
8. Stout DM, Bishop MJ, Dwersteg JF, Cullen BF. Correlation of
endotracheal tube size with sore throat and hoarseness follow-
ing general anesthesia. Anesthesiology 1987;67:419 –21
9. Loeser EA, Kaminsky A, Diaz A, Stanley TH, Pace NL. The
influence of endotracheal tube cuff design and cuff lubrication
on postoperative sore throat. Anesthesiology 1983;58:376 –9
10. Loeser EA, Bennett GM, Orr DL, Stanley TH. Reduction of
postoperative sore throat with new endotracheal tube cuffs.
Anesthesiology 1980;52:257–9
11. Loeser EA, Hodges M, Gliedman J, Stanley TH, Johansen RK,
Yonetani D. Tracheal pathology following short-term intuba-
tion with low- and high-pressure endotracheal tube cuffs.
Anesth Analg 1978;57:577–9
12. Loeser EA, Orr DL II, Bennett GM, Stanley TH. Endotracheal
tube cuff design and postoperative sore throat. Anesthesiology
1976;45:684 –7
13. Ayoub CM, Ghobashy A, Koch ME, McGrimley L, Pascale V,
Qadir S, Ferneini EM, Silverman DG. Widespread application
of topical steroids to decrease sore throat, hoarseness, and
cough after tracheal intubation. Anesth Analg 1998;87:714 – 6
14. Sumathi PA, Shenoy T, Ambareesha M, Krishna HM. Con-
trolled comparison between betamethasone gel and lidocaine
jelly applied over tracheal tube to reduce postoperative sore
throat, cough, and hoarseness of voice. Br J Anaesth 2008;
100:215– 8
15. Park SH, Han SH, Do SH, Kim JW, Rhee KY, Kim JH.
Prophylactic dexamethasone decreases the incidence of sore
throat and hoarseness after tracheal extubation with a double-
lumen endobronchial tube. Anesth Analg 2008;107:1814 – 8
16. McHardy FE, Chung F. Postoperative sore throat: cause, pre-
vention and treatment. Anaesthesia 1999;54:444 –53
17. Herlevsen P, Bredahl C, Hindsholm K, Kruhøffer PK. Prophy-
lactic laryngo-tracheal aerosolized lidocaine against postopera-
tive sore throat. Acta Anaesthesiol Scand 1992;36:505–7
18. Loeser EA, Stanley TH, Jordan W, Machin R. Postoperative
sore throat: influence of tracheal tube lubrication versus cuff
design. Can Anaesth Soc J 1980;27:156 – 8
19. Macario A, Weinger M, Carney S, Kim A. Which clinical
anesthesia outcomes are important to avoid? The perspective
of patients. Anesth Analg 1999;89:652– 8
ANESTHESIA & ANALGESIA

Activated Clotting Times, Heparin Responses, and
Antithrombin: Have We Been Wrong All These Years?
Jerrold H. Levy, MD, FAHA, and Roman M. Sniecinski, MD
I t was recognized early in the development of modern
cardiopulmonary bypass (CPB) that the fluidity of
blood needed to be maintained for extracorporeal cir-
culation, especially with the evolution of early oxygen-
ators.1 It was not until the anticoagulant properties of
heparin were discovered that blood could circulate over
nonendothelial surfaces without activating the coagulation
cascade. Today, almost a century after its isolation by
Howell and McLean in 1916, heparin remains the mainstay
agent for cardiac surgery because of several advantages.2
Profound levels of anticoagulation can be quickly obtained
(AntiXa levels of 3–5 U/mL) and an antidote, protamine, is
readily available. In addition, heparin has a relatively short
context-sensitive half-life, and can be used in patients with
renal dysfunction. However, an important limiting aspect
of heparin is that it requires a cofactor, antithrombin (also
referred to as antithrombin III or ATIII), for its anticoagu-
lation effect. Thus, heparin is still used extensively as an
anticoagulant despite the development of newer agents
that have been the focus of significant clinical interest and
consternation.3
The most common method for determining the adequacy
of anticoagulation for cardiac surgery is the activated
clotting time (ACT), a modified whole blood Lee-White
clotting assay that can be affected by multiple factors.4
Although there is not universal agreement on the ideal
ACT for CPB, when heparin does not produce the
desired ACT increase, “heparin resistance” is said to
have occurred.5 The term is misleading, however, be-
cause it is really an alteration in the heparin dose
response, possibly because of decreased antithrombin.
“Heparin resistance” is perhaps more aptly termed “al-
tered heparin responsiveness.”6
Antithrombin levels markedly decrease after CPB.7
These decreased levels may lead to a dangerous increase in
thrombin activity.8 It would seem logical that this could
lead to postoperative thrombotic complications, suggesting
the need for antithrombin not only before CPB, but after it as
From the Department of Anesthesiology, Emory University School of
Medicine, Cardiothoracic Anesthesiology and Critical Care, Emory Health-
care, Atlanta, Georgia.
Accepted for publication June 22, 2010.
Disclosure: The authors report no conflicts of interest.
Address correspondence and reprint requests to Jerrold H. Levy, MD, FAHA,
Department of Anesthesiology, Emory University Hospital, 1364 Clifton Rd., NE,
Atlanta, GA 30322. Address e-mail to jlevy01@emory.edu.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181f08a80
October 2010 • Volume 111 • Number 4
EDITORIAL
well. In this issue of the journal, 3 investigations from the
same group add further information and pose new questions
about the interrelationship of antithrombin and heparin.
In the first study, Garvin et al.9 retrospectively evaluated
the Hepcon HMS PLUS System (Medtronic Inc., Minneapo-
lis, MN) for its accuracy in predicting heparin dose re-
sponses for cardiac surgery with CPB. ACT-measured
heparin dose response and heparin concentrations were
evaluated in 3880 patients after a heparin dose calculated to
achieve a target ACT. The result was wide variability in
measurements. A target ACT of 300 seconds was not
obtained in 7% of patients, and a target ACT of 350 seconds
was not obtained in 17% of patients. The investigators also
found that calculated and measured heparin dose re-
sponses were not related at any heparin dose.
In the second study, the authors examined whether
there was a direct association among preoperative anti-
thrombin activity, heparin dose responses, and the heparin
sensitivity index from 304 patients after CPB using Hepcon
HMS PLUS Systems.10 The authors used multivariate linear
regression to identify independent predictors of heparin
dose response. The baseline antithrombin activity was
normal in this study and was not associated with either
baseline or postheparin ACT, heparin dose response, or
heparin sensitivity index. Of note, only 16% of patients (49
of 304) in the study presented with low baseline antithrom-
bin levels as defined as ⬍80%.
In the third study, the authors prospectively evaluated
whether low levels of antithrombin were associated with
postoperative major adverse cardiac events in 1403 patients
undergoing coronary artery bypass graft.11 Major adverse
cardiac events were defined as postoperative death, reop-
eration for coronary graft occlusion, myocardial infarction,
stroke, pulmonary embolism, or cardiac arrest until first
hospital discharge. Antithrombin activity levels were mea-
sured preoperatively, post-CPB, and on postoperative days
1 to 5. Major adverse cardiac events occurred in 146
patients (10.4%) and were independently associated with
postoperative antithrombin but not preoperative anti-
thrombin levels.
What do these 3 studies tell us? For starters, they prove
that an altered heparin response cannot be predicted by
preoperative antithrombin alone. Heparin’s effect on co-
agulation is rather unpredictable, at least as measured by
ACT, an important perspective but not a new finding.12,13
For example, Metz and Keats14 gave 193 patients a single
dose of heparin (300 U/kg). In 51 patients (26.4%), ACT
www.anesthesia-analgesia.org 833
 EDITORIAL
values were ⬍400 seconds, including 4 patients ⬍300
seconds.
The target ACT values of 300 or 350 seconds for the 3
studies may seem relatively low for many clinical practices
and the rate of failure to meet those targets may seem
unusually high. Previous reports suggest there is a 5% to
10% chance of a patient developing altered heparin respon-
siveness. The current studies targeted ACT values ⬎350,
which may account for the authors finding a 17% incidence
of altered heparin responsiveness.14,15
Realistically, we still do not know the ideal ACT to
initiate CPB, which is reflected in the widespread variabil-
ity in clinical practice.4,16 Several studies have demon-
strated that thrombin is still activated at ACT values of 400
to 480 seconds during CPB.17,18 As shown by the response
of patients with factor XII deficiency, a prolonged ACT
does not guarantee anticoagulation.19
All of this calls into question the validity of using ACTs
to assess adequate anticoagulation. Clinicians need to un-
derstand the factors that influence both anticoagulation and
its clinical measurement. The ACT is a relatively primordial
tool, consisting only of a tube, some dirt or glass for
activation, and some heat and agitation to speed the
reaction. As shown in Figure 1, there are a host of other
factors besides heparin concentration and antithrombin
levels that affect ACT values. Unfortunately, there is no
“gold standard” measurement to validate ACTs, complicat-
ing assessment of the relationship between ACT and al-
tered heparin responsiveness.
Based on the current studies, is there a role for
antithrombin administration to improve heparin respon-
siveness? Multiple studies report that antithrombin
supplementation improves intraoperative anticoagulation,
increases ACT, and reduces biochemical markers of hemo-
static activation.13,20 –22 Although these 3 studies did not
investigate antithrombin administration, the authors did
find a relationship between postoperative antithrombin
levels and major adverse cardiac events. This corroborates
the observation by Ranucci et al.,23 that low levels of
antithrombin activity in the intensive care unit are associ-
ated with a poor outcome in cardiac surgery. However,
834 www.anesthesia-analgesia.org
Figure 1. The activated clotting time (ACT) is a
whole blood point of care coagulation test used
extensively in cardiac surgery and in the cardiac
catheterization laboratory to monitor the anticoagu-
lant effect of different agents including unfraction-
ated heparin. In the ACT, blood is added to a
cartridge or tube that contains an activator, usually
celite or kaolin, to speed the process by increasing
contact activation by the intrinsic coagulation cas-
cade. Clot formation in the ACT represents the
interaction of plasma coagulation components
(e.g., factors and fibrinogen), platelets, and red
blood cells as this is a whole blood clotting assay.
However, clot formation in the ACT is influenced by
multiple factors that include platelet count and
platelet function, factor deficiencies, fibrinogen
levels, pharmacologic agents (anticoagulants in
platelet inhibitors), temperature(especially hypo-
thermia), and contact activation inhibitors (e.g.,
aprotinin).
there are still only a few supporting studies demonstrating
that antithrombin supplementation improves clinical out-
comes, and these studies were not conducted in patients
undergoing cardiac surgery.24,25
In summary, the current 3 studies provide additional
data about the complex issues of anticoagulation, heparin
responsiveness, and outcomes, including the role of anti-
thrombin. Have we been wrong all these years regarding
ACTs, heparin responsiveness, and the role of antithrom-
bin? More than 50 years after the development of CPB, we
are still asking many of the questions that the early
pioneers confronted. We need better monitors and better
understanding of anticoagulation adequacy to treat alter-
ations in heparin response and assess therapeutic efficacy.
We hope it will not take another 50 years to find these
answers.
REFERENCES
1. Galletti PM. Cardiopulmonary bypass: a historical perspective.
Artif Organs 1993;17:675– 86
2. Wardrop D, Keeling D. The story of the discovery of heparin
and warfarin. Br J Haematol 2008;141:757– 63
3. Spyropoulos AC. Brave new world: the current and future use
of novel anticoagulants. Thromb Res 2008;123:S29 –35
4. Despotis GJ, Gravlee G, Filos K, Levy J. Anticoagulation
monitoring during cardiac surgery: a review of current and
emerging techniques. Anesthesiology 1999;91:1122–51
5. Staples MH, Dunton RF, Karlson KJ, Leonardi HK, Berger RL.
Heparin resistance after preoperative heparin therapy or in-
traaortic balloon pumping. Ann Thorac Surg 1994;57:1211– 6
6. Levy JH. Heparin resistance and antithrombin: should it still be
called heparin resistance? J Cardiothorac Vasc Anesth 2004;
18:129 –30
7. Zaidan JR, Johnson S, Brynes R, Monroe S, Guffin AV. Rate of
protamine administration: its effect on heparin reversal and
antithrombin recovery after coronary artery surgery. Anesth
Analg 1986;65:377– 80
8. Sniecinski R, Szlam F, Chen EP, Bader SO, Levy JH, Tanaka
KA. Antithrombin deficiency increases thrombin activity
after prolonged cardiopulmonary bypass. Anesth Analg
2008;106:713– 8
9. Garvin S. Heparin concentration-based anticoagulation for
cardiac surgery fails to reliably predict heparin bolus dose
requirement. Anesth Analg 2010;111:849 –55
ANESTHESIA & ANALGESIA
Activated Clotting Times, Heparin Responses, and Antithrombin
10. Garvin S. Heparin dose response is independent of perioper-
ative antithrombin activity in patients undergoing coronary
artery bypass graft surgery using low heparin concentrations.
Anesth Analg 2010;111:856 – 61
11. Garvin S. Postoperative activity, but not preoperative activity,
of antithrombin is associated with major adverse cardiac
events after coronary artery bypass graft surgery. Anesth
Analg 2010;111:862–9
12. Gravlee GP, Case LD, Angert KC, Rogers AT, Miller GS.
Variability of the activated coagulation time. Anesth Analg
1988;67:469 –72
13. Levy JH, Montes F, Szlam F, Hillyer CD. The in vitro effects of
antithrombin III on the activated coagulation time in patients
on heparin therapy. Anesth Analg 2000;90:1076 –9
14. Metz S, Keats AS. Low activated coagulation time during
cardiopulmonary bypass does not increase postoperative
bleeding. Ann Thorac Surg 1990;49:440 – 4
15. Ranucci M, Isgro G, Cazzaniga A, Ditta A, Boncilli A, Cotza M,
Carboni G, Brozzi S. Different patterns of heparin resistance:
therapeutic implications. Perfusion 2002;17:199 –204
16. Lobato RL, Despotis GJ, Levy JH, Shore-Lesserson LJ, Carlson
MO, Bennett-Guerrero E. Anticoagulation management during
cardiopulmonary bypass: a survey of 54 North American
institutions. J Thorac Cardiovasc Surg 2010;139:1665– 6
17. Brister SJ, Ofosu FA, Buchanan MR. Thrombin generation
during cardiac surgery: is heparin the ideal anticoagulant?
Thromb Haemost 1993;70:259 – 62
18. Slaughter TF, LeBleu TH, Douglas JM Jr, Leslie JB, Parker JK,
Greenberg CS. Characterization of prothrombin activation
during cardiac surgery by hemostatic molecular markers.
Anesthesiology 1994;80:520 – 6
October 2010 • Volume 111 • Number 4
19. Salmenpera M, Rasi V, Mattila S. Cardiopulmonary bypass
in a patient with factor XII deficiency. Anesthesiology 1991;
75:539 – 41
20. Avidan MS, Levy JH, van Aken H, Feneck RO, Latimer RD, Ott
E, Martin E, Birnbaum DE, Bonfiglio LJ, Kajdasz DK, Despotis
GJ. Recombinant human antithrombin III restores heparin
responsiveness and decreases activation of coagulation in
heparin-resistant patients during cardiopulmonary bypass.
J Thorac Cardiovasc Surg 2005;130:107–13
21. Levy JH, Despotis GJ, Szlam F, Olson P, Meeker D, Weis-
inger A. Recombinant human transgenic antithrombin in
cardiac surgery: a dose-finding study. Anesthesiology 2002;
96:1095–102
22. Williams MR, D’Ambra AB, Beck JR, Spanier TB, Morales DL,
Helman DN, Oz MC. A randomized trial of antithrombin
concentrate for treatment of heparin resistance. Ann Thorac
Surg 2000;70:873–7
23. Ranucci M, Frigiola A, Menicanti L, Ditta A, Boncilli A, Brozzi
S. Postoperative antithrombin levels and outcome in cardiac
operations. Crit Care Med 2005;33:355– 60
24. Fourrier F, Chopin C, Goudemand J, Hendrycx S, Caron C,
Rime A, Marey A, Lestavel P. Septic shock, multiple organ
failure, and disseminated intravascular coagulation: compared
patterns of antithrombin III, protein C, and protein S deficien-
cies. Chest 1992;101:816 –23
25. Eid A, Wiedermann CJ, Kinasewitz GT. Early administration of
high-dose antithrombin in severe sepsis: single center results
from the KyberSept-trial. Anesth Analg 2008;107:1633– 8
www.anesthesia-analgesia.org 835
 EDITORIAL
Bridging the Gap to Reduce Surgical Site Infections
Frank J. Overdyk, MSEE, MD
S urgical site infections (SSI) remain one of the most
persistent and costly preventable complications in
health care. Although as anesthesiologists, we may
perceive our responsibility limited to the timely administra-
tion of prophylactic antibiotics and sterile placement of inva-
sive catheters, we feel the pain of our surgical colleagues
when their service suffers a rash of SSI, and they resort to
strategies ranging from ones with sound scientific basis to
mere superstition. Over the years, I have seen operating
rooms undergo lockdowns worthy of a maximum-security
prison, bathed in dance club–like ultraviolet light, or trans-
formed to resemble a spacewalk rehearsal. Yet the problem
of SSI persists, second only to urinary tract infections
among the most common types of nosocomial infections in
hospitals.1 Approximately 1 in 50 surgical procedures are
complicated by SSI, which translates to an estimated
290,485 cases annually, of which 13,088 result in death. It is
no surprise that SSI have made the 2010 Joint Commission
National Patient Safety Goals for the second consecutive
year.*
The importance of high tissue oxygen levels in im-
proved surgical outcomes and reduced surgical infection
rates has been understood for many years. High tissue
oxygen concentrations enhance the effects of leukocytes
and antibiotics on microbes and are indicative of adequate
tissue perfusion. Oxygen delivery to the tissues, a function
of both bloodflow and oxygen-carrying capacity, has been
the underpinning of many of the nonpharmacologic strat-
egies to prevent surgical wound infection.2 Traditionally,
liberal intraoperative fluid management was thought to
improve outcomes by improving perfusion, yet more re-
cent data suggest that tissue edema from overly aggressive
hydration may be detrimental. We remain without a con-
sensus on the optimal volume replacement strategy for
optimal surgical outcomes.3 Similarly, although the sym-
pathectomy accompanying regional anesthesia improves
tissue perfusion and certain outcomes (such as preserved
graft patency in vascular surgery and decreased incidence
of deep vein thrombosis), its impact on SSI is unclear.
Avoiding intraoperative hypothermia helps maintain tissue
*From http://www.jointcommission.org/PatientSafety/NationalPatient
SafetyGoals/.
From Medical University of South Carolina, Charleston, South Carolina.
Accepted for publication June 14, 2010.
Address correspondence to Frank J. Overdyk, MSEE, MD, Medical Univer-
sity of South Carolina, 167 Ashley Avenue, Suite 301, MSC 912 Charleston
SC 29425-9120. Address e-mail to overdykf@musc.edu.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181ee85b1
836 www.anesthesia-analgesia.org
perfusion, reduces SSI, and shortens hospitalization.4 Mild
intraoperative hypoventilation has been shown to increase
tissue oxygen levels due to the vasodilating effect of carbon
dioxide in obese patients.5 More clarity may be brought to
the optimal intraoperative fluid, temperature, ventilation,
and oxygenation strategies that minimize SSI if we could
easily measure subcutaneous tissue oxygen partial pressure
(Sto2). Arguably, Sto2 represents the common pathway of
cellular “contentment” during the “decisive period,” which
is defined as the hours surrounding surgical incision in
which prophylactic antibiotics have been shown to have the
greatest impact on reducing SSI.6
In this issue of Anesthesia & Analgesia, Govinda et al. use
a noninvasive infrared spectroscope (NIRS) to measure
Sto2 as a predictor of SSI that manifests as late as 30 days
postoperatively.7 The authors used a receiver operator
curve to optimize the sensitivity and specificity of upper-
arm Sto2 measurements to differentiate patients who de-
velop an SSI from those who do not. They found Sto2 below
66% measured 75 minutes after colonic surgery on the
upper arm to be predictive for SSI diagnosed, on average, 9
days after surgery. This NIRS measurement was a better
predictor for SSI than were 2 clinically accepted SSI risk
assessment metrics, the National Nosocomial Infections
Surveillance System and Study on the Efficacy of Nosoco-
mial Infection Control.
As an introductory study on this technology, its limitations
are several and well delineated in the article. However, if this
result is reproducible upon further investigation, we may
have an important “early warning system” for SSI and a tool
with which to dissect the problem of SSI into 2 components.
First, this noninvasive technique may allow us to investigate
the different intraoperative anesthetic strategies involving
fluid management, ventilation, oxygenation, hemodynamic
indices, and temperature that optimize Sto2 during the imme-
diate postoperative period. Second, once we optimize and
control for Sto2, we can explore additional postoperative
strategies to minimize SSI. The current debate on the
efficacy of high intraoperative and postoperative Fio2 in
reducing SSI demonstrates in which cases this Sto2 mea-
surement may be helpful. A recent meta-analysis of 5
randomized, double-blind clinical trials found a relative
risk reduction for SSI of 33% when hyperoxic gas mixtures
(Fio2 ⬇ 80%) were used intraoperatively and immediately
postoperatively for 2 to 6 hours versus a normoxic (Fio2
⬇30%) mixture.8 Yet, few anesthesiologists and surgeons
have adopted this approach, perhaps because a benefit
realized only once in every 33rd patient (number needed to
treat) may not justify the inconvenience and risks of
October 2010 • Volume 111 • Number 4
Noninvasive Infrared Spectroscope and Surgical Site Infection
patients wearing nonrebreather oxygen masks up to 6
hours postoperatively. Atelectasis and increased pulmo-
nary complications in the hyperoxic group, particularly in
obese patients, remain a concern.9 Furthermore, these re-
sults were contradicted by a randomized control trial
powered to detect an identical 33% relative risk reduction
between an Fio2 of 80% and one of 30%, continued for only
2 hours postoperatively.10 That study found no decreased
risk of SSI with the higher inspired oxygen concentration.
Was the shorter duration of postoperative hyperoxia deci-
sive? Or were intraoperative fluid and temperature man-
agement suboptimal for preventing SSI, as is suggested by
Hunt and Hopf in an accompanying editorial?11 Perhaps
immediate postoperative Sto2 measurements will give us a
“halftime score,” which we can use to alter the game plan,
and concentrate on postoperative interventions to reduce
SSI, such as high supplemental oxygen concentrations,
effective pain management, postoperative warming, and
appropriate discontinuation of prophylactic antibiotics
within 24 hours of surgery end.12 However, if Sto2 mea-
surements in the recovery room suggest that the “SSI die is
cast,” it is clear that our efforts to reduce SSI should be
focused on intraoperative strategies, including those not
associated with oxygenation and perfusion, such as timely
antibiotic prophylaxis within 1 hour before surgical inci-
sion and tight control of blood glucose levels.
Additional work with Sto2 is warranted. Although dif-
ferences in dependent variables such as intraoperative
temperature, volume replacement, hemodynamics, surgical
duration, and surgical technique between the patients with
and without SSI were nonsignificant in this study, the
patients with SSI weighed more and had more pain, and
Fio2 during NIRS measurements was not controlled.
Clearly, the impact of these confounding variables on the
results must be clarified. This study also contradicted
previous work in terms of the anatomic location of the Sto2
measurement, suggesting that measurement location is not
merely nuance, but may have important ramifications in
terms of the reproducibility of these results.
Lastly, the trade-off between sensitivity and specificity
in determining a cutoff point (Sto2 ⬍66%) needs to be
considered in the context of the outcome. This threshold for
Sto2 was chosen by the authors from the receiver operator
curve to optimize both sensitivity and specificity. In the
case of SSI, the annual cost to hospitals is approximately
$7.5 billion, and the toll on patients and their families
incalculable.13 It may be wise to sacrifice specificity for
†From http://www.aqihq.org/. Accessed May 25, 2010.
October 2010 • Volume 111 • Number 4
better sensitivity, if most of the intra- and postoperative
risk mitigation strategies can be implemented cost-
effectively. NIRS should help answer these questions.
As the above discussion illustrates, SSI is not merely a
surgical issue, but one that anesthesiologists influence
through intraoperative decisions and postoperative pain
management strategies. This may be reflected at some point
by the inclusion of SSI in the National Anesthesia Clinical
Outcomes Registry.† Our involvement in investigating and
reducing SSI will be appreciated by our administrators,
surgeons, and our patients, and we will continue to be
recognized as critical and invaluable members of the peri-
operative team.
REFERENCES
1. Klevens RM, Edwards J, Richards C, Horan T, Gaynes R,
Pollock D, Cardo D. Estimating health care–associated infec-
tions and deaths in U.S. hospitals. Public Health Rep 2007;
122:160 – 66
2. Sessler D. Non-pharmacologic prevention of surgical wound
infection. Anesthesiol Clin 2006;24:279 –97
3. Chappel D, Jacob M, Hofmann-Kiefer K, Conzen P, Rehm M.
Rational approach to perioperative fluid management. Anes-
thesiology 2008;109:723– 40
4. Kurz A, Sessler D, Lenhardt R. Perioperative normothermia to
reduce the incidence of surgical wound infection and shorten
hospitalization. NEJM 1996;334:1209 –15
5. Hager H, Reddy D, Mandadi G, Pulley D, Eagon JC, Sessler D,
Kurz A. Hypercapnia improves tissue oxygenation in mor-
bidly obese surgical patients. Anesth Analg 2006;103:677– 81
6. Classen DC, Evans RS, Pestotnik SL, Horn SD, Menlove RL,
Burke JP. The timing of prophylactic administration of antibi-
otics and the risk of surgical-wound infection. N Engl J Med
1992;326(5):281– 6
7. Govinda R, Kasuya Y, Bala E, Mahboobi R, Devarajan J, Sessler
D, Akca O. Early postoperative subcutaneous tissue oxygen
predicts surgical site infection. Anesth Analg 2010;111:946 –52
8. Qadan M, Akca O, Mahid S, Hornung C, Polk H. Perioperative
supplemental oxygen therapy and surgical site infection: a
meta-analysis of randomized controlled trials. Arch Surg
2009;144(4):359 – 66
9. Zoremba M, Dette F, Hunecke T, Braunecker S, Wulf H. The
influence of perioperative oxygen concentration on postopera-
tive lung function in moderately obese adults. Eur J Anaesthe-
siol 2010;27:501–7
10. Meyhoff C, Wetterslev J, Jorgensen L. For the PROXI random-
ized clinical trial. JAMA 2009;302(14):1543–50
11. Hunt T, Hopf H. High inspired oxygen fraction and surgical
site infection. JAMA 2009;302:1588 –9
12. Bratzler DW, Houck PM. Antimicrobial prophylaxis for sur-
gery: an advisory statement from the National Surgical Infec-
tion Prevention Project. Amer J Surg 2005;189:395– 404
13. Stone PW, Braccia D, Larson E. Systematic review of economic
analyses of health care–associated infections. Am J Infect
Control 2005;33:501–9
www.anesthesia-analgesia.org 837
 EDITORIAL
Labor Pain, Analgesia, and Chronobiology: What
Factor Matters?
Yvan Touitou, PhD,* Garance Dispersyn, PhD,† and Laure Pain, MD†
A lleviation and control of pain are of major impor-
tance in medicine. Depending on the causes and on
the clinical situations, the patterns of pain may
vary during the day, with peak and trough times reported
at different times of day, although sometimes with
contradictory results.1 The therapeutic effects of local
anesthetics and opioid analgesics also demonstrate cir-
cadian variations.1
Two articles in this issue of Anesthesia & Analgesia report
investigations in parturients into the relationship between
time of administration of intrathecal bupivacaine,2 or intra-
thecal fentanyl versus systemic hydromorphone,3 and the
duration of analgesia. In both studies, the duration of
analgesia was defined as the time from the first adminis-
tration of analgesia during labor until the second request
for analgesia. The 2 groups of investigators approached the
analysis of the chronobiology of analgesic requirements
during labor from somewhat different perspectives. Scav-
one et al.3 addressed the problem from a clinical perspec-
tive, with their main objective being to determine whether
time-related effects might have confounded the results of a
previous study by the group.4 Shafer et al.,2 on the other
hand, used their clinical data to examine the problems that
arise when developing models to analyze the often com-
plex periodic data obtained from clinical situations.
Scavone et al.3 compared neuraxial versus systemic
opioid analgesia in 692 healthy parturients early in labor.
At the first request for analgesia, patients in the neuraxial
group were given intrathecal fentanyl 25 ␮g, whereas
patients in the systemic analgesia group received 1 mg IV
and 1 mg IM hydromorphone. Subjects given their initial
analgesia between 0701 hours and 2300 hours were consid-
ered as the daytime group; those given their first analgesic
between 2301 hours and 0700 hours, the nighttime group.
The authors found no difference in the median duration of
either neuraxial or systemic analgesia, irrespective of the
time of administration. A major advantage claimed by the
authors for their study was the large number of patients
From *Unité de Chronobiologie, Fondation Ophtalmologique A. de Roths-
child, Paris, France; and †INSERM U666 (GRERCA), CHU de Strasbourg,
Faculté de médecine, Strasbourg, France.
Accepted for publication June 6, 2010.
Address correspondence to Yvan Touitou, Chronobiology Unit, Fondation
Ophtalmologique A. de Rothschild, 29 rue Manin, 75019 Paris, France.
Address e-mail to yvan.touitou@upmc.fr.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181ee85d9
838 www.anesthesia-analgesia.org
investigated, considerably greater than that of previous
studies of this nature.
The paper by Shafer et al.2 represents the final product
of a manuscript initially submitted to Anesthesia & Analge-
sia, reporting the chronobiology of the duration of analgesia
following intrathecal bupivacaine. After peer review and
extensive reanalysis of the original data, the manuscript has
evolved into a much more interesting exploration of the
methods used in chronobiological analysis to detect the
influence of external factors and in particular how one can
use these methods to detect possible artifacts in the data.
This is an important illustration of the value of a well-
conducted peer review process.
Shafer et al.2 make the important point in their paper
that very different conclusions can be drawn, depending on
the statistical methods used to analyze the data. They used
3 different smoothing functions to explore circadian
rhythms graphically. These revealed a bimodal pattern in
the duration of analgesia, with 1 peak at around 0630 hours
and a subsequent peak in the afternoon or evening. In
contrast, an analysis of variance (ANOVA) did not show
any significant difference in the duration of analgesia,
irrespective of the timing of the intrathecal injection of
bupivacaine. Fitting the data to a simple cosine function of
analgesia duration versus time demonstrated a periodic
waveform with a period of 8 hours, but with peaks that
corresponded with only 1 of the 3 peaks identified by the
smoother functions. The authors then used a bootstrap
analysis to show that 2 individual points were responsible
for the statistical significance in their cosine fit. Removing
these points from the dataset also removed the previously
observed rhythmic effect of intrathecal bupivacaine on
analgesia. The authors concluded that these 2 points were
likely to be artifacts, though they acknowledge this might
be uncertain, corresponding to the change in nursing and
anesthesia shifts.
The literature contains conflicting reports on the time-
dependent effects of neuraxial local anesthetics and opi-
oids.1 These differences may be a consequence of the wide
variety of factors that need to be considered when inter-
preting the results from chronobiology experiments. They
also emphasize the importance of a very stringent experi-
mental protocol, including methodology and statistical
methods used to analyze the potential rhythmic effects of a
drug. Many factors, both internal and external, can influ-
ence the biological rhythms of a drug’s action. These
include temporal variations in light– dark, rest–activity,
October 2010 • Volume 111 • Number 4
Labor Pain, Analgesia, and Chronobiology
fasting– eating cycles, and other environmental factors such
as stress, alcohol consumption, tobacco use, and caffeine
intake. All of these are able to alter the parameters charac-
terizing a biological rhythm and should be considered as
cofactors that can mask or unmask any circadian effects of
drugs.5 When studying labor pain, specific obstetrical fac-
tors should be controlled for, because they may have
considerable influence on pain intensity, e.g., parity, spon-
taneous or pharmacologically induced labor, stage of cer-
vical dilation, rupture of the membrane, labor duration,
and pharmacological induction of uterine contractions
(oxytocin administration).6 – 8
Ideally, although not always feasible, studies should con-
sider subpopulations of patients, considering all of these
different clinical situations. As an example, one may analyze
the duration of intrathecal analgesia in a population of partu-
rients, nulliparous, with cervical dilation at 2 cm, with no
administration of oxytocin at all, and find circadian variations
that could be masked in another study examining a different
subgroup of population. This might explain the discrepancies
between the published studies. However, other factors can
bias results. Although factors such as the timing of the shift
changes in the work team can bias results, one needs to be
cautious about this type of explanation. For example, the time
when the morning nursing shift begins could coincide with
the peak in cortisol, which is lowest in the evening and
reaches its highest concentration during the early waking
hours. Which factor is then influencing the duration of anal-
gesia, the change of nursing team or the endogenous circadian
rhythm of cortisol?
Although biological rhythms are genetically deter-
mined, they are continuously modulated (adjusted in time)
by periodic events in the environment called synchroniz-
ers.5 In mammals, the main synchronizer is light. Rhythm
synchronization of subjects is the “gold standard” of any
chronobiological study protocol. This means that, in addi-
tion to the factors discussed above, the subjects should keep
as close as possible to their usual lifestyle, maintaining the
environmental factors (synchronizers) that impose their
period on biological rhythms, such as sleep–wake time,
rest–activity cycle, light– dark cycle, fasting– eating, etc.5
Thus, it is advisable to verify the synchronization of the
subjects’ circadian time organization by checking rhythm
markers (body temperature, cortisol, melatonin, etc.) as an
initial step or as an integral aspect of the investigative
protocol.9 The importance of the synchronizers that influ-
ence endogenous rhythms varies according to the variable
or the function under investigation. Some synchronizers
are more predominant than others, e.g., the light– dark
cycle and the sleep–wake activity in humans. In some
experimental situations, however, synchronizers that oth-
erwise might be considered minor can assume predomi-
nance. Changes in any one of these synchronizers may lead
to changes in the temporal relationship between biological
rhythms.5
Circadian rhythms can also be subject to seasonal modu-
lation, and different circadian rhythm patterns have been
described in the same individuals but at different times of
the year.10 –12 Failure to consider this could explain some
of the apparently discrepant results in the literature. All of
those factors need be considered and standardized in any
October 2010 • Volume 111 • Number 4
study dealing with biological rhythms to avoid potential
biases.
Many exogenous factors (masking agents) can influence
circadian patterns by masking the endogenous rhythms of
the biological clock. A circadian rhythm can be masked by
any environmental signal to which the organism is sensi-
tive.13 To unmask the endogenous circadian structure, an
experimental protocol called “constant routine” can be
used. This involves subjects staying awake but lying down
for 24 to 36 hours, in an environment of constant tempera-
ture, humidity, and lighting, with identical and regularly
spaced meals. Though this protocol is considered by many
as the gold standard, it obviously presents several limita-
tions,14 and for research in fields such as obstetrics is
seldom practicable.
Subtracting the estimated exogenous or “masking” com-
ponent of a rhythm from the observed rhythm may reveal
the underlying endogenous component. Regression models
have been developed to identify masking effects, but sepa-
ration of the exogenous from the endogenous component is
difficult.5 When the data can be approximated by a sinu-
soidal model, the cosinor function allows calculation of
rhythm parameters such as the mesor (the mean level that
is equal to the 24-hour average), amplitude (half of the
peak-to-trough difference of the fitted cosine function), and
acrophase (the peak time of the rhythm given in degrees or
hours and minutes). When the number of subjects is small
or the density of samplings low, the interindividual vari-
ability should be smoothed. Nonparametric tests are useful
because they ensure that the absolute values do not influ-
ence the results. Whatever methods are used, they should
be critically evaluated for their ability to test the hypotheses
under question and to assess the time series data of the
variable(s) being studied.
Finally, why are the studies by Scavone et al.3and Shafer
et al.2 important? Both studies found that time of day did
not appear to influence the duration of analgesia produced
by intrathecal local anesthetics or opioids. Knowledge of
the time dependency of drugs is important because drug
effect can be optimized and toxicity minimized by basing
drug administration on the circadian patterns of drug
activity. Several studies have demonstrated circadian time–
dependent changes in the toxicity and the pharmacokinetic
disposition of local anesthetics.15 Shafer et al., in addition to
demonstrating the importance of the peer review process,
provide a set of clear guidelines that future investigators in
this field should find valuable, and hopefully help them
avoid many of the potential pitfalls in chronobiological
research.
ACKNOWLEDGMENTS
This editorial was solicited and handled by Dr. James Bovill,
Guest Editor-in-Chief for Anesthesia & Analgesia.
REFERENCES
1. Bruguerolle B, Labrecque G. Rhythmic pattern in pain and
their chronotherapy. Adv Drug Del Rev 2007;59:883–95
2. Shafer SL, Lemmer B, Boselli E, Boiste F, Bouvet L, Allaouch-
iche B, Chassard D. Pittfalls in chronobiology: a suggested
analysis using intrathecal bupivacaine analgesia as an ex-
ample. Anesth Analg 2010;111:980 –5
www.anesthesia-analgesia.org 839
 EDITORIAL
3. Scavone BM, McCarthy RJ, Wong CA, Sullivan JT. The influ-
ence of time of day of administration on duration of opioid
labor analgesia. Anesth Analg 2010;111:986 –91
4. Wong CA, Scavone BM, Peaceman AM, McCarthy RJ, Sullivan
JT, Diaz NT, Yaghmour E, Marcus RJ, Sherwani SS, Sproviero
MT, Yilmaz M, Patel R, Robles C, Grouper S. The risk of
cesarean delivery with neuraxial analgesia given early versus
late in labor. N Engl J Med 2005;352:655– 65
5. Touitou Y, Haus E, eds. Biological Rhythms in Clinical and
Laboratory Medicine. Berlin: Springer, 1992
6. Sheiner E, Sheiner EK, Shoham-Vardi I. The relationship between
parity and labor pain. Int J Gynecol Obstet 1998;63:287– 8
7. Aya AGM, Vialles N, Mangin R, Robert C, Ferrer JM, Ripart J,
de la Courssaye JE. Chronobiology of labour pain perception:
an observational study. Br J Anaesth 2004;93:451–3
8. Veira WS, Hidalgo MPL, da Silva LT, Caumo W. Biological
rhythms of spinal-epidural labor analgesia. Chronobiol Int
2010;27:865–78
9. Selmaoui B, Touitou Y. Reproducibility of the circadian
rhythms of serum cortisol and melatonin in healthy subjects. A
study of three different 24-h cycles over six weeks. Life Sci
2003;73:3339 – 49
840 www.anesthesia-analgesia.org
10. Touitou Y, Lagoguey M, Bogdan A, Reinberg A, Beck H.
Seasonal rhythms of plasma gonadotrophins: their persistence
in elderly men and women. J Endocrinol 1983;96:15–21
11. Levi F, Canon C, Touitou Y, Reinberg A, Mathé G. Seasonal
modulation of the circadian time structure of circulating T and
natural killer lymphocyte subsets from healthy subjects. J Clin
Invest 1988;81:407–13
12. Reinberg A, Touitou Y, Levi F, Nicolai A. Circadian and
seasonal changes in ACTH-induced effects in healthy young
men. Eur J Clin Pharmacol 1983;25:657– 65
13. Mrosovsky N. Masking: history, definitions, and measure-
ment. Chronobiol Int 1999;16:415–29
14. Waterhouse JM, Redfern P, Minors DS. An introduction to
circadian rhythms and their measurement in humans. In:
Redfern P, ed. Chronotherapeutics. London: Pharmaceutical
Press, 2003:1–30
15. Chassard D, Bruguerolle B. Chronobiology and anesthesia.
Anesthesiology 2004;100:413–27
ANESTHESIA & ANALGESIA

Can an Acute Pain Service Be Cost-Effective?
Eric Sun, MD, PhD,*† Franklin Dexter, MD, PhD,‡ and Alex Macario, MD, MBA§
I n many countries around the world, the anesthesiolo-
gist is the primary physician responsible for pain con-
trol in the first 24 hours after surgery. However, in the
United States of America (USA), postoperative analgesia is
typically managed by the surgeon. This is because their
professional fee includes this responsibility while the pa-
tient remains in the hospital and when the patient returns
home. At the other end of the spectrum is a dedicated
Acute Pain Service team with expertise and authority for
managing a patient’s surgical pain.
The well-done randomized clinical trial conducted by
Lee et al.1 and described in this issue of the journal
estimates the cost-effectiveness of an anesthesiologist-led,
nurse-based Acute Pain Service, mainly charged with man-
aging IV patient-controlled analgesia. The control group
consisted of patients receiving IM or IV boluses of opioids
as needed by nurses on the ward.1 It is not known how
frequently this technique is used in other parts of the
world, including the USA.
Our objectives for this editorial were to:
• Estimate the total annual number of surgical inpa-
tients in the USA that might benefit from care pro-
vided by an Acute Pain Service.
• Define possible structures and functions of an Acute
Pain Service because these may vary among institutions,
and affect the economic viability of such a service.
• Outline research questions that deserve further study
to help hospitals make the needed investment in an
Acute Pain Service.
HOW MANY PATIENTS ARE AT RISK FOR PAIN
AFTER INPATIENT SURGERY IN THE USA?
To obtain an estimate of the number of inpatients having
surgery, the Nationwide Inpatient Sample from the Health-
care Cost and Utilization Project was used. It samples 20%
of inpatient discharges from 1056 academic, community,
and acute care hospitals in 42 states, and is intended to be
From the *RAND Corporation, Santa Monica; †the Department of Health
Services, University of California, Los Angeles, Los Angeles, California;
‡University of Iowa, Iowa City, Iowa; and §Department of Anesthesia,
Stanford University School of Medicine, Stanford, California.
Accepted for publication July 9, 2010.
Dr. Sun is funded by the Agency for Healthcare Research and Quality
through the UCLA/RAND training grant T32 HS 000046.
Disclosure: The authors report no conflicts of interest.
Address correspondence and reprint requests to Alex Macario, MD, MBA,
Department of Anesthesia H3580, Stanford University School of Medicine,
Stanford, CA 94305-5640. Address e-mail to amaca@stanford.edu.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181f33533
October 2010 • Volume 111 • Number 4
EDITORIAL
representative of 90% of all hospital discharges in the USA.
For 2008, the year with the most recently available data,
there were approximately 9.1 million inpatients having
surgery in the USA. This constitutes approximately 23% of
all inpatient visits, and a 4% annual growth rate since 1998
(Table 1). These data exclude federal and prison hospitals.
The data in Table 1 provide insight into the size and
scope of the demand for pain management in the postop-
erative setting. From a national health policy perspective,
there are many different surgical procedures.2 However,
cesarean deliveries, orthopedics, and general surgery pro-
cedures should be a priority for optimizing clinical acute
postoperative pain care because they account for a large
fraction of all inpatient surgery.
With public and private payers continuing to face tight-
ening budgets, and with single bundled episode of care
payments for hospital and physician care looming, it is an
opportunity for anesthesiologists to consider additional
venues where the specialty can provide value. The dedi-
cated Acute Pain Service seems to be a logical choice, given
that it is a straightforward extension of the care anesthesi-
ologists already provide in the operating room and chronic
pain settings. However, given the expenses of such a
service, what are the incentives for hospitals and anesthe-
siologists to participate? On the clinical side, controlling
pain is of importance to patients, a goal made more difficult
by the increasing number of patients who are admitted
with chronic pain syndromes including opioid prescrip-
tions superimposed on their acute surgical pain. The po-
tential clinical benefits of a dedicated Acute Pain Service
have resulted in support from a variety of organizations
such as the American Society of Anesthesiologists,3 the
Royal College of Surgeons, and the College of Anesthetists.
In addition to these clinical benefits, an Acute Pain Service
provides important visibility to the anesthesia group within a
hospital. Patient satisfaction with pain control comprises an
important component of many measures of hospital quality,
such as Press Ganey scores. Many facilities are providing
resources to an Acute Pain Service to increase these scores. In
fact, a patient’s Press Ganey response to, “How well your pain
was controlled” was the second most important variable
correlated to whether a patient recommended the hospital to
someone else. By the way, the number one item was “Staff
worked together to care for you.”4
POSSIBLE STRUCTURES AND FUNCTIONS OF AN
ACUTE PAIN SERVICE
Despite these benefits, the use of a dedicated Acute Pain
Service to manage postoperative pain did not gain wide-
spread adoption in the 1990s in the USA5–7 or abroad.8,9
www.anesthesia-analgesia.org 841
 EDITORIAL

Table 1. Thirty Most Common Diagnosis Related Groups for Inpatients Having Surgery Requiring Anesthesia
Discharges coded without Discharges coded with
complications and complications and Total in the
DRG name comorbidities comorbidities USA in 2008
Cesarean 904,000 472,000 1,376,000
Major joint replacement or reattachment of lower extremity 912,000 56,000 968,000
Uterine and adnexa procedure for nonmalignancy 430,000 97,000 527,000
Major small and large bowel procedures 95,000 275,000 370,000
Laparoscopic cholecystectomy without common duct exploration 191,000 137,000 328,000
Hip and femur procedures except major joint (e.g., hip fracture) 85,000 169,000 254,000
Appendectomy 196,000 26,000 222,000
Back and neck procedures except spinal fusion 157,000 51,000 208,000
Spinal fusion except cervical 198,000 9100 207,100
Lower extremity and humerus procedures except hip, foot, femur 149,000 54,000 203,000
(e.g., tibia fracture)
Cervical spinal fusion 123,000 26,000 149,000
Major cardiovascular procedures 84,000 53,000 137,000
OR procedures for obesity 109,000 20,000 129,000
Craniotomy and endovascular intracranial procedures 58,000 66,000 124,000
Extensive OR procedure unrelated to principal diagnosis 24,000 99,000 123,000
Major chest procedures 37,000 86,000 123,000
Extracranial procedures 85,000 25,000 110,000
Coronary bypass with cardiac catheterization 68,000 38,000 106,000
Other respiratory system OR procedures 14,000 92,000 106,000
Vaginal delivery with sterilization and/or D&C 101,000 None 101,000
Major male pelvic procedures 79,000 17,000 96,000
Infectious and parasitic diseases with OR procedure 1000 91,000 92,000
Transurethral prostatectomy 44,000 45,000 89,000
Peritoneal adhesiolysis 35,000 53,000 88,000
Coronary bypass without cardiac catheterization 63,000 24,000 87,000
Revision of hip or knee replacement 41,000 43,000 84,000
Hernia procedures except inguinal and femoral 50,000 33,000 83,000
Appendectomy with complicated principal diagnosis 48,000 24,000 72,000
Cardiac valve and other major cardiothoracic procedures without 12,000 59,000 71,000
cardiac catheterization
Total 4,393,000 2,240,100 6,633,100
Note: Because these are hospital discharges, a patient could have had more than 1 visit to the operating room for a surgical procedure during 1 hospitalization.
Thus, the actual number of operations is likely higher than the numbers reported here.
DRG ⫽ diagnosis related group; USA ⫽ United States of America; OR ⫽ operating room; D&C ⫽ dilatation and curettage.

Sometimes Acute Pain Services are services in name only.
For example, 36% of 446 hospitals in Germany operated
an Acute Pain Service but half did not have specific
personnel assigned to the service, policies for nights and
weekends, written protocols for pain management, or
regular assessment and documentation of pain scores at
least once a day.10 The challenges are in part financial,
such as how coding/billing affects profitability, but also
include organizational change barriers such as compet-
ing managerial and clinical agendas, staff shortages,
local politics, professional hierarchies, and workload
changes.11
In Italy, although 42% of 163 public hospitals had an
organized Acute Pain Service, continuous IV analgesia
using elastomeric infusion systems was the most frequently
used analgesic technique, and IV patient-controlled analge-
sia and epidural techniques were used in fewer than 14% of
patients.12 In Ireland, a 2007 study showed that 71%
of teaching hospitals had a formalized Acute Pain Service,
of which 85% were established after 1990.13
Moreover, where a dedicated Acute Pain Service exists,
there is variability in its structure and the types of services
it provides. To illustrate the variability in Acute Pain
Service staffing and services provided at several USA
hospitals, we created Table 2, which is based on informal
communication with colleagues at those facilities.
The heterogeneity in structure and function of pain
services across facilities makes it difficult to state unequivo-
cally whether an Acute Pain Service is or is not cost-
effective. To define the exact nature of an Acute Pain
Service is a crucial component of its economic study.
Anesthesia groups in the USA currently are likely to
provide an Acute Pain Service for a variety of reasons
such as:
• An anesthesia residency training program exists at the
hospital. As of 2008, a month-long acute pain experi-
ence for anesthesia trainees is required.
• A hospital’s practice is to place a large number of
peripheral nerve catheters, which then extends the
anesthesiologist’s responsibility longer and leads to an
Acute Pain Service.
• The anesthesia group is paid for its service by third
party payers such as the federal government or pri-
vate insurance, and the marginal revenue exceeds the
marginal costs.
• The anesthesia group’s contract and professional
service agreement with the hospital stipulates its
existence. The hospital may subsidize the Acute
Pain Service budget if there is a perceived competi-
tive advantage to the hospital to attract more
cases.

842 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

Acute Pain Service Economics

Table 2. Staffing and Care Provided by Acute Pain Services

Hospital
A B C D E F G
Anesthesiologist (FTEs/d) 0.8 1 1 1 1 1 X
Housestaff (FTEs/d) 1 1 1 No 1 1 No
RN (FTEs/d) No 1 0 No 1 No No
NP (FTEs/d) No No 1 No No No No
Other staff Fellow Pharmacist No No No No No
No. of patients/d 12 18 12 (Monday)–30 4 10–20 8–20 10–15
(Friday)
Services 0 ⫽ never provided; 10 ⫽ most frequently provided service
Manage routine IV PCA 0 0 0 2 1 0 10
Manage complex IV PCA (e.g., patients 10 8 3 7 3 8 2
on chronic pain medications)
Manage peripheral nerve catheter/ 10 10 10 10 10 3 0
infusions
Manage intraoperative spinal opioids 10 10 0 2 2 1 3
Consult on nonsurgical acute pain (e.g., 10 3 5 2 4 3 1
lumbar puncture, epidural catheters
for rib fractures)
Manage epidural catheter 10 10 10 0 9 10 1
Outpatient nerve block catheters No Yes No No Yes No No
24-h coverage Yes Yes Yes Yes Yes Yes Yes
Note: Hospitals listed in this table are academic medical centers except hospitals D and G, which are community hospitals. Hospital D has no formal acute pain
service in place but it is being proposed for funding. Hospital G has the post 2nd call person round on the patients in the morning, and the 1st call person answers
the calls at all other times.
FTE ⫽ full time equivalent; RN ⫽ registered nurse; NP ⫽ nurse practitioner; PCA ⫽ patient-controlled analgesia.

• The surgeons’ desire to deflect the night phone calls
and pain control questions to the anesthesia group
especially if the anesthesia group is being paid for
taking calls (depending on the contract) whereas the
surgeons are not.
Financial considerations are often largely responsible for
the slow adoption of a dedicated Acute Pain Service.
Although the study by Lee et al. in this issue of the journal,
along with the studies that preceded it,14 represent impor-
tant first steps, more research is needed to find ways to
overcome organizational constraints and deliver the spe-
cialized care of an Acute Pain Service in an economically
viable manner.
RESEARCH QUESTIONS THAT DESERVE
FURTHER STUDY
The literature leaves unanswered the question of what ser-
vices a dedicated Acute Pain Service should provide and the
most cost-effective method in which to provide them. Studies
to date have compared a wide variety of staffing models and
services to more conventional methods of postoperative pain
management. However, no studies have explicitly compared
various Acute Pain Service models with each other. In all
likelihood, the most cost-effective way to implement an Acute
Pain Service will vary among hospitals because of local
variations in culture and personnel, for example. In addition,
smaller hospitals that perform a range of surgical cases may
face more variability in the demand for expert postoperative
pain management. This observation is based on the law of
large numbers, which states that there is more variability in
smaller samples. Such low volume hospitals will need to find
ways to manage this variability, such as pooling staff and
resources across hospitals.
More work also needs to be done to quantify the efficacy
of a dedicated Acute Pain Service. A typical outcome
considered is some measure of pain scores. The Quality-of-
Recovery instrument, for instance, has 9 questions, at least
4 of which pertain to pain or side effects from analgesics.15
However, a challenge for the health economist is that these
measures need to be translated into units more easily
understood by decision makers.
For example, efforts could be made to translate these
pain scores to quality-adjusted life years, the measure of
efficacy used by many public payers.16 However, the time
horizon in acute surgical care studies is typically measured
in days and weeks and not months or years as for treat-
ments aimed at relieving chronic conditions. As a result, a
dedicated Acute Pain Service may have little effect on the
conventional metric of quality-adjusted life years. There-
fore, additional efforts should be made to define what
measures payers should consider, and present economic
analyses in a way that practicing clinicians and payers can
understand and translate to their practice.
Future studies could also report alternative outcomes of
relevance to hospital administrators. For example, a group of
pain service interventions increased Press Ganey measure-
ments of patient satisfaction from the 87th to the 99th percen-
tile.17 More understanding is needed regarding which of the
several reported interventions had the largest impact.
Efforts to date to estimate the efficacy of a dedicated Acute
Pain Service have included consideration of the average effect
in a study population. However, this approach may obscure
the utility of an Acute Pain Service for 2 reasons.18 The first
reason concerns differences in patient responses. If a dedicated
Acute Pain Service provides a large benefit for a small portion
of the population, the average effect will be small, but clearly
the Acute Pain Service still has an important clinical role.
The second reason concerns dependence in responses
across different alternatives to pain management. Put sim-
ply, the issue here is whether patients who fail to respond

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 843

 EDITORIAL
to basic methods of pain management are likely to see large
improvement with the expertise of an Acute Pain Service. If
an Acute Pain Service is likely to provide large benefits for
patients who are not responding to other modalities, the
service also has an important clinical role, even if its
average effect (measured across a larger population) is
small. This latter possibility is particularly relevant in some
hospitals where the anesthesiologist-staffed Acute Pain
Service typically focuses on particularly difficult cases that
have been refractory to more conventional methods. Ulti-
mately, these factors suggest that a dedicated Acute Pain
Service can improve its cost-effectiveness by targeting its
efforts toward patients most likely to benefit. Unfortu-
nately, our ability to identify these patients remains limited
and is another area for further study.19
Finally, in calculating the costs of an Acute Pain Service,
the present study focuses on the costs of the service itself,
measured as staff time and the costs of administered
medications. Personnel costs will differ among countries
both in the absolute cost of an anesthesiologist and nurse,
and in the relative cost of one to the other. This markedly
affects the optimal mix of providers for an Acute Pain
Service.20 In addition, these measures of costs are incom-
plete, because they do not account for the net costs, taking
into account the possibility that improved pain manage-
ment may reduce costs by shortening hospital stay or
decreasing the probability of complications. As an example,
total knee arthroplasty patients discharged home with a
continuous femoral nerve block had reduced hospital
length of stay and associated costs and charges, not includ-
ing postdischarge resource utilization.21 Because stake-
holders are likely to be interested in the net costs of running
an Acute Pain Service, future studies should investigate
how pain management affects total hospitalization and
downstream costs.
Providing an Acute Pain Service is not solely an anes-
thesia issue. The hospital facility needs to invest in training,
supplies, and policies. In fact, a dedicated Acute Pain
Service provides the opportunity for anesthesiologists to
extend their influence in perioperative medicine for the
millions of inpatients having surgery. The tasks involved
align well with the anesthesiologist’s existing skill set.
However, the wide variety of practice models in the USA
and abroad suggests there has not been agreement on how
to provide these services in a financially viable way. Given
the large number of surgical inpatients, optimizing pain
management may yield large savings.
The truth is that whether a dedicated Acute Pain Service
can be a cost-effective way of managing postoperative pain
depends on multiple conditions, such as work culture,
surgical case mix, and reimbursement systems.
AUTHOR CONTRIBUTIONS
All authors helped design and conduct the study, analyze the
data, and write the manuscript. All authors approved the final
manuscript.
RECUSE NOTE
Franklin Dexter is the Section Editor of Economics, Education,
and Policy for the Journal. The manuscript was handled by
Spencer Liu, Section Editor for pain Medicine. Dr. Dexter was
not involved in any way with the editorial process or decision.
844 www.anesthesia-analgesia.org
REFERENCES
1. Lee A, Chang SKC, Chen PP, Gin T, Lau ASC, Chiu CH. The
costs and benefits of extending the role of the acute pain
service on clinical outcomes after major elective surgery.
Anesth Analg 2010;111:1042–50
2. Macario A. Truth in scheduling: is it possible to accurately
predict how long a surgical case will last? Anesth Analg
2009;108:681–5
3. American Society of Anesthesiologists Task Force on Acute
Pain Management. Practice guidelines for acute pain manage-
ment in the perioperative setting: an updated report by the
American Society of Anesthesiologists Task Force on Acute
Pain Management. Anesthesiology 2004;100:1573– 81
4. Hospital Pulse Report 2009. Available at: http://www.pressganey.
com/galleries/default-file/Hospital_Pulse_Report_2009.pdf. Ac-
cessed June 15, 2010
5. Carr DB, Miaskowski C, Dedrick SC, Williams GR. Manage-
ment of perioperative pain in hospitalized patients: a national
survey. J Clin Anesth 1998;10:77– 85
6. Ready LB. How many acute pain services are there in the
United States, and who is managing patient-controlled analge-
sia? Anesthesiology 1995;82:322
7. Warfield CA, Kahn CH. Acute pain management: programs in
U.S. hospitals and experiences and attitudes among U.S.
adults. Anesthesiology 1995;83:1090 – 4
8. Rawal N, Allvin R. Acute pain services in Europe: a 17-nation
survey of 105 hospitals. The EuroPain Acute Pain Working
Party. Eur J Anaesthesiol 1998;15:354 – 63
9. Powell AE, Davies HT, Bannister J, Macrae WA. Rhetoric and
reality on acute pain services in the UK: a national postal
questionnaire survey. Br J Anaesth 2004;92:689 –93
10. Stamer UM, Mpasios N, Stüber F, Maier C. A survey of acute
pain services in Germany and a discussion of international
survey data. Reg Anesth Pain Med 2002;27:125–31
11. Powell AE, Davies HT, Bannister J, Macrae WA. Challenge of
improving postoperative pain management: case studies of
three acute pain services in the UK National Health Service.
Br J Anaesth 2009;102:824 –31
12. Coluzzi F, Savoia G, Paoletti F, Costantini A, Mattia C.
Postoperative pain survey in Italy (POPSI): a snapshot of
current national practices. Minerva Anestesiol 2009;75:622–31
13. Hu P, Owens T, Harmon D. A survey of acute pain services in
teaching hospitals in the Republic of Ireland. Ir J Med Sci
2007;176:225– 8
14. Lee A, Chan S, Chen PP, Gin T, Lau AS. Economic evaluations
of acute pain service programs: a systematic review. Clin J Pain
2007;23:726 –33
15. Myles PS, Hunt JO, Nightingale CE, Fletcher H, Beh T, Tanil D,
Nagy A, Rubinstein A, Ponsford JL. Development and psycho-
metric testing of a quality of recovery score after general
anesthesia and surgery in adults. Anesth Analg 1999;88:83–90
16. Tan JM, Macario A. How to evaluate whether a new technol-
ogy in the operating room is cost-effective from society’s
viewpoint. Anesthesiol Clin 2008;26:745– 64
17. Philips BD, Liu SS, Wukovits B, Boettner F, Waldman S,
Liguori G, Goldberg S, Goldstein L, Melia J, Hare M, Jasphey L,
Tondel S. Creation of a novel recuperative pain medicine
service to optimize postoperative analgesia and enhance pa-
tient satisfaction. HSS J 2010;6:61–5
18. Basu A, Philipson TJ. The impact of comparative effectiveness
research on health and health care spending. NBER Working
Paper No. 15633, 2010
19. Raja SN, Jensen TS. Predicting postoperative pain based on
preoperative pain perception: are we doing better than the
weatherman? Anesthesiology 2010;112:1311–2
20. Macario A. What does one minute of operating room time
cost? J Clin Anesth 2010;22:233– 6
21. Ilfeld BM, Mariano ER, Williams BA, Woodard JN, Macario A.
Hospitalization costs of total knee arthroplasty with a continu-
ous femoral nerve block provided only in the hospital versus
on an ambulatory basis: a retrospective, case-control, cost-
minimization analysis. Reg Anesth Pain Med 2007;32:46 –54
ANESTHESIA & ANALGESIA
 Society of Cardiovascular Anesthesiologists
Cardiovascular Anesthesiology Section Editor: Charles W. Hogue, Jr.
Perioperative Echocardiography and Cardiovascular Education Section Editor: Martin J. London
Hemostasis and Transfusion Medicine Section Editor: Jerrold H. Levy

Heparin Concentration–Based Anticoagulation for

Cardiac Surgery Fails to Reliably Predict Heparin
Bolus Dose Requirements
Sean Garvin, MD,* Daniel C. FitzGerald, CCP,† George Despotis, MD,‡§储 Prem Shekar, MD,†
and Simon C. Body, MBChB, MPH*

BACKGROUND: Hemostasis management has evolved to include sophisticated point-of-care

systems that provide individualized dosing through heparin concentration–based anticoagulation. The
Hepcon HMS Plus system (Medtronic, Minneapolis, MN) estimates heparin dose, activated
clotting time (ACT), and heparin dose response (HDR). However, the accuracy of this test has not
been systematically evaluated in large cohorts.
METHODS: We examined institutional databases for all patients who underwent cardiac surgery
with cardiopulmonary bypass (CPB) at our institution from February 2005 to July 2008. During
this period, the Hepcon HMS Plus was used exclusively for assessment of heparin dosing and
coagulation monitoring. Detailed demographic, surgical, laboratory, and heparin dosing data
were recorded. ACT, calculated and measured HDR, and heparin concentrations were recorded.
Performance of the Hepcon HMS Plus was assessed by comparison of actual and target ACT
values and calculated and measured HDR.
RESULTS: In 3880 patients undergoing cardiac surgery, heparin bolus dosing to a target ACT
resulted in wide variation in the postheparin ACT (r2 ⫽ 0.03). The postheparin ACT did not reach
the target ACT threshold in 7.4% (i.e., when target ACT was 300 s) and 16.9% (i.e., when target
ACT was 350 s) of patients. Similarly, the target heparin level calculated from the HDR did not
correlate with the postbolus heparin level, with 18.5% of samples differing by more than 2 levels
of the assay. Calculated and measured HDR were not linearly related at any heparin level.
CONCLUSIONS: The Hepcon HMS Plus system poorly estimates heparin bolus requirements in
the pre-CPB period. Further prospective studies are needed to elucidate what constitutes
adequate anticoagulation for CPB and how clinicians can reliably and practically assess
anticoagulation in the operating room. (Anesth Analg 2010;111:849 –55)

A dministration of heparin, and monitoring of whole

blood activated clotting time (ACT), is still the
mainstay of anticoagulation for cardiopulmonary
bypass (CPB).1,2 Hemostasis management has evolved to
include more sophisticated point-of-care systems that pro-
vide individualized dosing of heparin and protamine,
through heparin concentration– based estimation of hepa-
rin dosage, and measurement of ACT.3,4 Individualized
heparin concentration– based anticoagulation management
has been reported to cause less activation of thrombin,
decreased fibrinolysis, reduced blood loss, fewer transfusions,
and a decreased risk of reexploration for hemorrhage,5– 8
From the *Department of Anesthesiology, Perioperative and Pain Medicine,
and †Division of Cardiac Surgery, Brigham and Women’s Hospital, Harvard
Medical School, Boston, Massachusetts; and ‡Departments of Pathology,
§Immunology, and 储Anesthesiology, Washington University School of
Medicine, St. Louis, Missouri.
Accepted for publication June 22, 2009.
Supported by departmental funds.
Address correspondence and reprint requests to Simon C. Body, MBChB,
MPH, Brigham and Women’s Hospital, Harvard Medical School, 75 Francis
St., Boston, MA 02215. Address e-mail to body@zeus.bwh.harvard.edu.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181b79d09
although there is still a lack of consensus in this area.9,10
Previous studies of heparin-level monitoring devices, includ-
ing the Hepcon HMS Plus, had varied results, with some
studies showing good agreement between whole blood hep-
arin concentration and either ACT before CPB11 or plasma
anti-Xa levels before or during CPB,12 whereas others showed
a lack of significant relationship between ACT and heparin
concentration13 or whole blood versus anti-Xa heparin con-
centration.14 All of these studies were limited by their small
sample sizes. Importantly, there is previously observed inter-
individual variation in measured heparin levels for a target
ACT required for CPB.
Therefore, we examined the relationship between indi-
vidualized in vitro estimation of the heparin level for a
specific target ACT, and the subsequently observed ACT
and heparin level measured after bolus administration of
the heparin dose response (HDR)-estimated heparin
dose. We hypothesized that: 1) there would be strong
correlation between calculated and measured HDR for
individual patients; 2) residual variation in the predicted
HDR would be due to patient factors that are accounted for
in the calculation of the HDR; and 3) the HDR-estimated
heparin dose would achieve the desired ACT for CPB in all
patients.

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 849

Anticoagulation for Cardiac Surgery
METHODS
The Hepcon HMS Plus (Medtronic, Minneapolis, MN) was
introduced in our institution in February 2004, in concert
with a change in heparin coagulation management. Use of
this system was based on a belief that adequate anticoagula-
tion for CPB using a heparin-coated circuit could be achieved
with lower ACTs as long as reasonable, albeit lower heparin
levels than had been historically used were maintained.
Furthermore, there was a perception that reduced heparin
administration may result in reduced postoperative bleeding.
After IRB approval, we examined institutional databases
for all patients who underwent CPB for cardiac surgery at
our institution, from February 2005 to July 2008 (n ⫽ 4667).
Patients with incomplete perioperative anticoagulation
data were excluded from analysis (n ⫽ 562). The majority of
patients were excluded for having a target ACT of ⬎350 s
or subsequent administration of aprotinin (n ⫽ 157), no
record of postheparinization heparin level (n ⫽ 144), no
record of baseline ACT (n ⫽ 91), or no record of height or
weight (n ⫽ 67). Furthermore, we excluded an additional
225 patients who were not anticoagulated using the insti-
tutional protocol described below, yielding 3880 analyzed
patients.
Anticoagulation Management
The Hepcon HMS Plus was used for anticoagulation man-
agement according to the manufacturer’s recommenda-
tions.15 The estimated blood volume for each patient was
calculated using the manufacturer’s instructions,15 accord-
ing to the method described by Allen et al.14 After induc-
tion of anesthesia, baseline kaolin ACT, predicted HDR,
predicted heparin concentration, and heparin bolus calcu-
lations were performed according to the manufacturer’s
instructions, using HDR cartridges encompassing whole
blood heparin concentration ranging from 0.4 to 3.4 U/mL.
The Hepcon HMS Plus recommended CPB prime heparin
dose based on a 750- or 1000-mL prime volume was added
to the calculated heparin bolus and administered via a
central venous catheter. Heparin was not added to the CPB
pump prime. Three minutes after United States Pharma-
copeia porcine heparin administration and 10 min after
onset of CPB, heparin concentration and ACT were remea-
sured. All patients received an ⑀-aminocaproic acid load of
7.5–10 g over 1 h, after the initial blood draw for baseline ACT,
but before heparin administration and blood sampling for
measurement of postheparin ACT. ⑀-Aminocaproic acid was
subsequently infused at a rate of 1.25–1.5 g/h.
Throughout the entire study period, an ACT of more
than 350 s and a minimum heparin concentration of 2
U/mL were used in patients undergoing non– coronary
artery bypass graft (CABG) procedures or CABG with the
use of cardiotomy suction. Patients undergoing primary
CABG surgery without the use of cardiotomy suction
before March 2007 were anticoagulated using a protocol
that proscribed a minimum ACT of 300 s before the institu-
tion of CPB. After February 2007, to standardize anticoagula-
tion management, the minimum ACT before CPB was in-
creased to 350 s with a minimum heparin concentration of 2.0
U/mL for CABG surgery. These protocols were driven by a
desire to reduce heparin and protamine doses to reduce blood
850 www.anesthesia-analgesia.org
loss but were not supported by prior clinical evidence of
efficacy.
We also observed variability in anticoagulation practice
that was not proscribed by a protocol. Some perfusionists
administered additional heparin when the measured hep-
arin level was ⬍1.4 or 2.0 U/mL, according to personal
preference. These patients were excluded from analysis, as
previously described.
Data Collection and Statistical Analysis
Detailed demographic data, surgical indications, preopera-
tive laboratory values, operative data, heparin bolus dose,
ACT values, and measured heparin concentrations were
routinely collected into a centralized database. The calcu-
lated HDR was calculated as the target ACT ⫺ baseline
ACT divided by the target heparin level and reported as
s 䡠 U⫺1 䡠 mL⫺1. The calculated HDR differed from the
method used by the Hepcon HMS Plus,15 but prospective
comparison of the Hepcon HMS Plus and the above
calculation in 15 patients showed excellent correlation (r2 ⫽
0.991). This calculation uses the target heparin concentra-
tion identified from the HDR slope generated by the
Hepcon HMS Plus, not the protocol-driven heparin level.
Thus, the calculation of the HDR slope is independent of
the clinical protocol and unaffected by the heparin level
desired or obtained after heparin dosing.
The measured HDR was calculated as postheparin
ACT ⫺ baseline ACT divided by the measured heparin
level and reported as s 䡠 U⫺1 䡠 mL⫺1. By using the measured
postheparinization ACT and measured heparin level, the
measured HDR is independent of the clinical protocol
because the HDR slope has been demonstrated to be linear
over the clinical range of heparin concentrations.11
The Hepcon HMS Plus heparin assay has limited fidel-
ity, reporting only 6 categorical heparin levels, with inter-
mediate values of heparin reported to the closest level.
Accordingly, we could not directly compare the calculated
and measured HDR. Therefore, we report the calculated
HDR for each measured postbolus heparin level. Pearson
product-moment correlation was performed between cal-
culated and measured values of HDR, and heparin concen-
tration and ACT data. ACT data at each heparin level were
compared with Wilcoxon’s ranked sum test. Statistical
analysis was performed with JMP version 7.02 (SAS, Cary,
NC). All data are presented as mean ⫾ sd or median and
interquartile range, as appropriate. A 2-sided P ⬍ 0.05
was considered as showing statistical significance.
RESULTS
A total of 3880 patients were analyzed. Baseline demo-
graphics and perioperative data are summarized in Table 1.
Target ACTs of 300 and 350 s were used in 23.4% and 76.6%
of patients, respectively. Heparin dose, ACT, and heparin
level after the heparin bolus for each target ACT are
detailed in Table 2. The mean heparin dose calculated to
achieve a target ACT of 300 s was 152 U/kg, and for a
target ACT of 350 s was 179 U/kg. After administration of
the HDR-estimated heparin dose, the target ACT was not
achieved in 7.4% of patients with a target ACT of 300 s, and
it was not reached in 16.9% of patients with a target ACT of
350 s (Fig. 1). Additional heparin was administered either
ANESTHESIA & ANALGESIA
Table 1. Demographics and Perioperative Data
Study cohort
(N ⴝ 3880)
Age 65.2 ⫾ 13.4
Male gender 2538 (65.4%)
Height (cm) 171 ⫾ 10
Weight (kg) 82.2 ⫾ 19.0
BSA (m2) 1.94 ⫾ 0.24
Indication for surgery
Coronary artery disease 2394 (63.3%)
Valve disease 1932 (53.1%)
Aortic dissection 41 (1.2%)
Medical history
Previous myocardial infarction 989 (26.2%)
Hypertension 2550 (67.5%)
Insulin-dependent diabetes 343 (9.1%)
Renal insufficiency (creatinine 145 (4.0%)
ⱖ2.0 mg/dL) Figure 1. Observed activated clotting time (ACT) at each target ACT
Preoperative medications level. Box plot represents the 5th, 25th, median, 75th, and 95th
Aspirin 2387 (63.1%) percentiles of data.
Adenosine diphosphate inhibitors 182 (4.8%)
Intravenous heparin 249 (6.6%)
Warfarin 34 (0.9%)
Preoperative laboratory values before or immediately after initiation of CPB to 91% of
Hematocrit (%) 38.8 ⫾ 4.9 patients who failed to achieve their target ACT of 300 s and
Platelet count (⫻109/L) 247 ⫾ 81 to 77% of patients who failed to achieve their target ACT of
INR 1.14 ⫾ 0.30 350 s.
INR ⬎1.4 280 (7.8%)
There was wide variation among patients for the ACT
Partial thromboplastin time (s) 36.2 ⫾ 15.9
Partial thromboplastin time ⬎37 s 733 (20.9%) seen at each heparin level (Fig. 2). Measured HDR varied
substantially among patients (mean ⫾ sd, 104.7 ⫾ 24.3
Data are presented as mean ⫾ 1SD for demographics.
The upper limit of the laboratory reference range for partial thromboplastin s 䡠 U⫺1 䡠 mL⫺1; median [10%–90% interquartile range], 102
time is 37 s. [77–137]). Correlation (r2) between the calculated and mea-
BSA ⫽ body surface area; CABG ⫽ coronary artery bypass graft; INR ⫽ sured HDR was poor at all heparin levels (r2 ⬍ 0.02; Fig. 3).
International normalized ratio.

Table 2. Intraoperative Anticoagulation Management

Mean ⴞ SD Median (10th–90th percentile)
Baseline ACT (s) 138 ⫾ 17 137 (118–159)
Target ACT of 300 s (N ⫽ 908)
Initial heparin dose (U) 12,994 ⫾ 3909 12,000 (9000–18,000)
Heparin dose (U/kg) 152 ⫾ 38 149 (108–203)
Postbolus ACT (s) 387 ⫾ 73 377 (309–466)
Postbolus ACT ⬍300 s 67 (7.4%)
CPB ⫹ 10 min ACT (s) 340 ⫾ 54 335 (274–409)
Postheparin bolus CPB ⴙ 10 min
Heparin level (U/mL)
⬍1.4 16 (1.8%) 250 (27.9%)
1.4 232 (25.6%) 498 (55.5%)
2.0 460 (50.7%) 118 (13.2%)
2.7 137 (15.1%) 21 (2.3%)
3.4 60 (6.6%) 10 (1.1%)
⬎3.4 3 (0.3%) 0 (0%)
Mean ⴞ SD Median (10th–90th percentile)
Target ACT of 350 s (N ⫽ 2972)
Initial heparin dose (U) 14,313 ⫾ 4013 14,000 (10,000–20,000)
Heparin dose (U/kg) 179 ⫾ 42 174 (132–232)
Postbolus ACT (s) 418 ⫾ 83 406 (332–509)
Postbolus ACT ⬍350 s 501 (16.9%)
CPB ⫹ 10 min ACT (s) 386 ⫾ 73 377 (313–466)
Postheparin bolus CPB ⴙ 10 min
Heparin level (U/mL)
⬍1.4 20 (0.7%) 278 (9.4%)
1.4 438 (14.7%) 1716 (58.2%)
2.0 1449 (48.8%) 789 (26.8%)
2.7 731 (24.6%) 125 (4.2%)
3.4 315 (10.6%) 41 (1.4%)
⬎3.4 19 (0.6%) 0 (0%)
ACT ⫽ activated clotting time; CPB ⫽ cardiopulmonary bypass.

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 851

Anticoagulation for Cardiac Surgery

Figure 2. Observed activated clotting time at each measured post-

bolus heparin level. Box plot represents the 5th, 25th, median, 75th,
and 95th percentiles of data.

To exclude an outlier effect, the analysis was repeated in a

subgroup of patients within 1 sd of the mean postheparin
ACT, with similar results.
The predicted heparin level expected after heparin bolus
administration was compared with the heparin level mea-
sured after heparin bolus. Levels that differed by more than
0.3 U/mL differed by more than the sensitivity (1 channel)
of the assay and were observed in 51.0% of samples (Fig. 4).
Levels that differed by more than 0.7 U/mL differed by
more than 2 channels in the assay and were observed in
18.5% of samples.

DISCUSSION
We examined the performance of the Hepcon HMS Plus to
guide anticoagulation in 3880 patients undergoing cardiac
surgical procedures. Data provided by the Hepcon HMS
Plus were used to calculate heparin dosing throughout CPB
and to direct the administration of protamine after CPB. We
observed wide variability in heparin dose requirements to
achieve an adequate ACT for CPB, as others have previ-
ously described.1 We also observed poor correlation of the
calculated HDR with the measured HDR. This led to ACT
values that were less than the target ACT values when
heparin dosing was guided by estimation of HDR, in
7.4%–16.9% of patients. The imprecision for the calculated
HDR may explain the high frequency of failure to reach the
Figure 3. Correlation between calculated and measured heparin
target ACT. dose response. Each panel contains data for individuals with a
Interpatient variability in HDR is well described, and specific measured heparin level after heparinization. Panel A repre-
use of a dose-response plot recommended by Bull et al.3 sents a measured heparin level of 1.4 U/mL (r2 ⬍ 0.01). Panel B
provides the basis of Hepcon HMS Plus calculations. Our represents a measured heparin level of 2.0 U/mL (r2 ⫽ 0.02). Panel
C represents a measured heparin level of 2.7 U/mL (r2 ⫽ 0.01).
initial hypothesis that the HDR slope calculated by the
Panel D represents a measured heparin level of 3.4 U/mL (r2 ⫽
Hepcon HMS Plus would correlate with the measured 0.03).
HDR slope is not supported by the data. The interindividual
variability in heparin response could be attributable to a

852 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA


Figure 4. Target and measured heparin level observed after bolus
heparin administration. Box plot represents the 5th, 25th, median,
75th, and 95th percentiles of data.
number of in vivo factors, including error in the estimation of
blood volume, effects of release of tissue factor pathway
inhibitor (TFPI) by heparin in vivo but not ex vivo, extravascu-
lar sequestration of heparin, plasma protein binding, circulat-
ing antithrombin, and platelet activation.16
Estimation of blood volume is a potential source of error
in heparin dose calculation. The methodology used by
several previous studies5,6,12 is predominantly based on the
method described by Allen et al.14 in 1956. The magnitude
of error in estimation of blood volume in individual
patients undergoing cardiac surgery has not been system-
atically evaluated. It is probable that the severity and
nature of cardiac disease may be a substantial source of
variation in blood volume estimates, thereby contributing
to imprecision of point-of-care anticoagulation management
instruments. Further investigation aimed at improving estima-
tion of blood volume in patients undergoing cardiac surgery
may result in better performance of point-of-care heparin
concentration– based anticoagulation systems.
Potentially contributing to the imprecision of the Hep-
con HMS Plus are the effects of release of TFPI by heparin
in vivo but not ex vivo. TFPI in vivo is bound to endothelial
cell surfaces with small amounts circulating in plasma and
bound to platelets.17 TFPI is an endogenous serine protease
inhibitor that exerts its action primarily through neutraliz-
ing factor Xa by complexing with factor Xa and forming a
TFPI-FXa complex, while also providing feedback inhibi-
tion of the factor VIIa–tissue factor complex.18 Plasma TFPI
is released from endothelial surface stores by administra-
tion of unfractionated and low-molecular-weight hepa-
rins,19 increasing plasma levels during CPB.20 Brodin et
al.21 recently demonstrated significant synergy between
antithrombin and TFPI in postheparin plasma with an
almost equal contribution of each to the prolongation of
anticoagulation in a laboratory setting. In the absence of an
endothelial source of TFPI, the contribution of TFPI to
October 2010 • Volume 111 • Number 4
heparin responsiveness may be underestimated by the
initial HDR calculation performed by the Hepcon HMS
Plus. This source of error in the calculated HDR would
seemingly lead to increased heparin dosages and higher
postheparin ACT, which is incongruous with our finding of
7.4%–16.9% of patients failing to achieve their target ACT,
but congruous with the mean ACT being higher than the
target ACT in our overall population. Although the exact
contribution of TFPI to point-of-care measurements of
heparin responsiveness has not been established, strong
correlation between the TFPI-responsive Heptest (Ameri-
can Diagnostica, Stamford, CT) measurements of heparin
level and ACT in the pre-CPB period22 has been observed,
and release of TFPI correlates with the Heptest measure-
ment of heparin response in normal volunteers receiving IV
heparin.23 These observations reinforce the importance of
TFPI in heparin responsiveness and point to a likely impact
on point-of-care assessments of HDR.
Earlier studies of these heparin-level monitoring de-
vices, including the Hepcon HMS Plus, had varied results
with some studies showing good agreement between
whole blood heparin concentration and either ACT before
CPB11 or plasma anti-Xa levels,12 whereas others showed a
lack of significant relationship between ACT and heparin
concentration24 or whole blood versus anti-Xa heparin
concentration during CPB.13,14 One study12 revealed sub-
stantial interpatient variability in HDR; however, the linear
relationship between heparin concentration and kaolin
ACT within individual patients was generally exceptional.
Our retrospective analysis demonstrated significant vari-
ability in the responsiveness of the kaolin ACT and heparin
in vivo versus kaolin ACT and heparin ex vivo (i.e., after
IV administration of heparin), and calculated versus
measured HDR, respectively (Fig. 3). This was paralleled
by the finding that the Hepcon HMS Plus– guided hep-
arin administration failed to result in adequate anti-
coagulation for CPB in 16% of patients in this study
when the target ACT was 350 s, even with the CPB prime
heparin dose being administered as part of the initial IV
bolus dose.
Early work performed by Bull et al.25 and Young
et al.26 continues to form the basis for the ACT target used
to safely institute CPB, with many centers using an ACT of
more than 480 s as a target. In 1981, Jobes et al.27 deter-
mined that a heparin level of 2 U/mL was associated with
an ACT ⬎300 s in ⬎95% of patients. These data were not
the basis for our use of a heparin level of 2 U/mL in the
latter portion of the study period, but are in line with our
observation of wide variation in ACT values at specific
heparin levels. The theoretical benefit of using heparin
concentration– based anticoagulation is that a stable hepa-
rin level may be achieved, thereby minimizing hemostatic
activation7,28 and keeping the patient fully anticoagulated
throughout the CPB period, which would lead to reduced
blood loss and transfusion.5,6 We observed wide variability
in ACT for any given heparin level (Fig. 2). Almost all
studies using heparin concentration have used this histori-
cal reference as the target for defining adequate anticoagu-
lation, but establishment of a “safe heparin level” has not
been systematically studied.
www.anesthesia-analgesia.org 853
Anticoagulation for Cardiac Surgery
This study is limited by its retrospective nature. We
were not able to capture all of the perioperative factors that
may influence anticoagulation and an individual’s re-
sponse to heparin administration. In addition, important
clinical outcomes are not part of this data set, including
perioperative complications that may relate to anticoagula-
tion management, such as bleeding, transfusion, reexplora-
tion, myocardial infarction, new or worsening renal insuf-
ficiency, and cerebral vascular accident. Furthermore, our
comparison of the calculated and measured HDR is limited
by the imprecision of the automated protamine titration
method, which measures whole blood heparin concentration
using discrete cutoffs using only the 6 channels of the car-
tridge. Thus, we are left with 4 likely sources of disparity
between calculated and measured HDR: 1) inaccurate esti-
mate of the patient’s blood volume; 2) lack of fidelity of
measured heparin concentration because only 6 categories
of heparin level describe the linear range of heparin con-
centration; 3) inherent inaccuracy of the device; and 4)
differences between the anticoagulant activity of heparin ex
vivo compared with its several actions in vivo, notably upon
the release of TFPI. We were unable to subset these possible
causes further, but they seem to be significant.
We conclude that the Hepcon HMS Plus fails to consis-
tently provide the therapeutic heparin bolus dose uniformly
in all patients based on the wide discrepancy in calculated
versus measured HDR. This can lead to inadequate heparin
doses to achieve a target ACT for CPB in as much as 16.9% of
patients. However, the Hepcon HMS Plus was able to identify
an adequate heparin dose for the majority of the patients.
Because this study, did not compare the Hepcon HMS Plus
with empiric dosing regimens, we are uncertain whether
empiric regimens can either under- or overperform when
compared to this system. Further prospective studies are
needed to elucidate what constitutes adequate anticoagula-
tion for CPB and how clinicians can reliably and practically
assess anticoagulation in the operating room.
REFERENCES
1. Akl BF, Vargas GM, Neal J, Robillard J, Kelly P. Clinical
experience with the activated clotting time for the control of
heparin and protamine therapy during cardiopulmonary by-
pass. J Thorac Cardiovasc Surg 1980;79:97–102
2. Hattersley PG. Activated coagulation time of whole blood.
JAMA 1966;196:436 – 40
3. Bull BS, Huse WM, Brauer FS, Korpman RA. Heparin therapy
during extracorporeal circulation. II. The use of a dose-
response curve to individualize heparin and protamine dos-
age. J Thorac Cardiovasc Surg 1975;69:685–9
4. Raymond PD, Ray MJ, Callen SN, Marsh NA. Heparin moni-
toring during cardiac surgery. Part 1: Validation of whole-
blood heparin concentration and activated clotting time. Per-
fusion 2003;18:269 –76
5. Despotis GJ, Joist JH, Hogue CW Jr, Alsoufiev A, Kater K,
Goodnough LT, Santoro SA, Spitznagel E, Rosenblum M,
Lappas DG. The impact of heparin concentration and activated
clotting time monitoring on blood conservation. A prospective,
randomized evaluation in patients undergoing cardiac opera-
tion. J Thorac Cardiovasc Surg 1995;110:46 –54
6. Jobes DR, Aitken GL, Shaffer GW. Increased accuracy and
precision of heparin and protamine dosing reduces blood loss
and transfusion in patients undergoing primary cardiac opera-
tions. J Thorac Cardiovasc Surg 1995;110:36 – 45
854 www.anesthesia-analgesia.org
7. Despotis GJ, Joist JH, Hogue CW Jr, Alsoufiev A, Joiner-Maier
D, Santoro SA, Spitznagel E, Weitz JI, Goodnough LT. More
effective suppression of hemostatic system activation in pa-
tients undergoing cardiac surgery by heparin dosing based on
heparin blood concentrations rather than ACT. Thromb Hae-
most 1996;76:902– 8
8. Bowie J, Kemma G. Automated management of heparin anti-
coagulation in cardiovascular surgery. Proc Am Acad Cardio-
vasc Perf 1985;6:1–10
9. Shore-Lesserson L, Reich DL, DePerio M. Heparin and prota-
mine titration do not improve haemostasis in cardiac surgical
patients. Can J Anaesth 1998;45:10 – 8
10. Slight RD, Buell R, Nzewi OC, McClelland DB, Mankad PS. A
comparison of activated coagulation time-based techniques for
anticoagulation during cardiac surgery with cardiopulmonary
bypass. J Cardiothorac Vasc Anesth 2008;22:47–52
11. Despotis GJ, Alsoufiev AL, Spitznagel E, Goodnough LT,
Lappas DG. Response of kaolin ACT to heparin: evaluation
with an automated assay and higher heparin doses. Ann
Thorac Surg 1996;61:795–9
12. Despotis GJ, Summerfield AL, Joist JH, Goodnough LT, San-
toro SA, Spitznagel E, Cox JL, Lappas DG. Comparison of
activated coagulation time and whole blood heparin measure-
ments with laboratory plasma anti-Xa heparin concentration in
patients having cardiac operations. J Thorac Cardiovasc Surg
1994;108:1076 – 82
13. Hardy JF, Belisle S, Robitaille D, Perrault J, Roy M, Gagnon L.
Measurement of heparin concentration in whole blood with
the Hepcon/HMS device does not agree with laboratory
determination of plasma heparin concentration using a chro-
mogenic substrate for activated factor X. J Thorac Cardiovasc
Surg 1996;112:154 – 61
14. Allen TH, Peng MT, Chen KP, Huang TF, Chang C, Fang HS.
Prediction of blood volume and adiposity in man from body
weight and cube of height. Metabolism 1956;5:328 – 45
15. Medtronic Perfusion Systems. Hepcon HMS Plus operator’s
manual. Minneapolis, MN, 2001
16. Ranucci M, Isgro G, Cazzaniga A, Ditta A, Boncilli A, Cotza M,
Carboni G, Brozzi S. Different patterns of heparin resistance:
therapeutic implications. Perfusion 2002;17:199 –204
17. Bajaj MS, Kuppuswamy MN, Saito H, Spitzer SG, Bajaj SP.
Cultured normal human hepatocytes do not synthesize
lipoprotein-associated coagulation inhibitor: evidence that en-
dothelium is the principal site of its synthesis. Proc Natl Acad
Sci USA 1990;87:8869 –73
18. Broze GJ Jr, Miletich JP. Characterization of the inhibition of
tissue factor in serum. Blood 1987;69:150 –5
19. Hoppensteadt DA, Walenga JM, Fasanella A, Jeske W, Fareed
J. TFPI antigen levels in normal human volunteers after
intravenous and subcutaneous administration of unfraction-
ated heparin and a low molecular weight heparin. Thromb Res
1995;77:175– 85
20. Adams MJ, Cardigan RA, Marchant WA, Grocott MP, Mythen
MG, Mutch M, Purdy G, Mackie IJ, Machin SJ. Tissue factor
pathway inhibitor antigen and activity in 96 patients receiving
heparin for cardiopulmonary bypass. J Cardiothorac Vasc
Anesth 2002;16:59 – 63
21. Brodin E, Appelbom H, Osterud B, Hilden I, Petersen LC,
Hansen JB. Regulation of thrombin generation by TFPI in
plasma without and with heparin. Transl Res 2009;153:124 –31
22. Hellstern P, Bach J, Simon M, Saggau W. Heparin monitoring
during cardiopulmonary bypass surgery using the one-step
point-of-care whole blood anti-factor-Xa clotting assay heptest-
POC-Hi. J Extra Corpor Technol 2007;39:81– 6
23. Hoffmann U, Harenberg J, Bauer K, Huhle G, Tolle AR,
Feuring M, Christ M. Bioequivalence of subcutaneous and
intravenous body-weight-independent high-dose low-
molecular-weight heparin Certoparin on anti-Xa, Heptest, and
tissue factor pathway inhibitor activity in volunteers. Blood
Coagul Fibrinolysis 2002;13:289 –96
24. Culliford AT, Gitel SN, Starr N, Thomas ST, Baumann FG,
Wessler S, Spencer FC. Lack of correlation between activated
clotting time and plasma heparin during cardiopulmonary
bypass. Ann Surg 1981;193:105–11
ANESTHESIA & ANALGESIA
25. Bull BS, Korpman RA, Huse WM, Briggs BD. Heparin therapy
during extracorporeal circulation. I. Problems inherent in
existing heparin protocols. J Thorac Cardiovasc Surg
1975;69:674 – 84
26. Young JA, Kisker CT, Doty DB. Adequate anticoagulation
during cardiopulmonary bypass determined by activated clot-
ting time and the appearance of fibrin monomer. Ann Thorac
Surg 1978;26:231– 40
October 2010 • Volume 111 • Number 4
27. Jobes DR, Schwartz AJ, Ellison N, Andrews R, Ruffini RA,
Ruffini JJ. Monitoring heparin anticoagulation and its neutral-
ization. Ann Thorac Surg 1981;31:161– 6
28. Koster A, Fischer T, Praus M, Haberzettl H, Kuebler WM,
Hetzer R, Kuppe H. Hemostatic activation and inflammatory
response during cardiopulmonary bypass: impact of heparin
management. Anesthesiology 2002;97:837– 41
www.anesthesia-analgesia.org 855
Heparin Dose Response Is Independent of
Preoperative Antithrombin Activity in Patients
Undergoing Coronary Artery Bypass Graft Surgery
Using Low Heparin Concentrations
Sean Garvin, MD,* Daniel Fitzgerald, CCP,† Jochen D. Muehlschlegel, MD,* Tjörvi E. Perry, MD,*
Amanda A. Fox, MD,* Stanton K. Shernan, MD,* Charles D. Collard, MD,‡ Sary Aranki, MD,†
and Simon C. Body, MBChB, MPH*

BACKGROUND: Unfractionated heparin’s primary mechanism of action is to enhance the

enzymatic activity of antithrombin (AT). We hypothesized that there would be a direct association
between preoperative AT activity and both heparin dose response (HDR) and heparin sensitivity
index (HSI) in patients undergoing coronary artery bypass graft surgery.
METHODS: Demographic and perioperative data were collected from 304 patients undergoing
primary coronary artery bypass graft surgery. AT activity was measured after induction of general
anesthesia using a colorimetric method (Siemens Healthcare Diagnostics, Tarrytown, NY).
Activated coagulation time (ACT), HDR, and HSI were measured using the Hepcon HMS Plus
system (Medtronic, Minneapolis, MN). Heparin dose was calculated for a target ACT using
measured HDR by the same system. Multivariate linear regression was performed to identify
independent predictors of HDR. Subgroup analysis of patients with low AT activity (⬍80% normal;
⬍0.813 U/mL) who may be at risk for heparin resistance was also performed.
RESULTS: Mean baseline ACT was 135 ⫾ 18 seconds. Mean calculated HDR was 98 ⫾ 21
s/U/mL. Mean baseline AT activity was 0.93 ⫾ 0.13 U/mL. Baseline AT activity was not
significantly associated with baseline or postheparin ACT, HDR, or HSI. Addition of AT activity to
multivariable linear regression models of both HDR and HSI did not significantly improve model
performance. Subgroup analysis of 49 patients with baseline AT ⬍80% of normal levels did not
reveal a relationship between low AT activity and HDR or HSI. Preoperative AT activity, HDR, and
HSI were not associated with cardiac troponin I levels on the first postoperative day, intensive
care unit duration, or hospital length of stay.
CONCLUSION: Although enhancing AT activity is the primary mechanism by which heparin
facilitates cardiopulmonary bypass anticoagulation, low preoperative AT activity is not associated
with impaired response to heparin or to clinical outcomes when using target ACTs of 300 to 350
seconds. (Anesth Analg 2010;111:856 –61)

M ore than 50 years after heparin’s discovery,1

Rosenberg and Damus2,3 identified its primary
mechanism of action as enhancement of the en-
zymatic activity of antithrombin (AT). AT binds and inac-
tivates serine proteases’ contact activation and common
pathways, principally factors IIa, Xa, IXa, and VIIa, by
decreasing their binding efficiency for substrate. AT’s inhi-
bition of serine protease activity is increased several orders
of magnitude by heparin as a result of a conformational
change induced by binding of a specific pentasaccharide
unit of heparin with high affinity for AT.4
Heparin resistance has been reported in 4% to 22% of
patients undergoing cardiopulmonary bypass (CPB).5–10
Potential risk factors include age older than 65 years,
From the *Department of Anesthesiology, Perioperative and Pain Medicine,
and †Division of Cardiac Surgery, Brigham and Women’s Hospital, Harvard
Medical School, Boston, Massachusetts; and ‡Division of Cardiovascular
Anesthesia at the Texas Heart Institute, Baylor College of Medicine, Saint
Luke’s Episcopal Hospital, Houston, Texas.
Accepted for publication November 18, 2009.
Address correspondence and reprint requests to Simon C. Body, MBChB,
MPH, Brigham and Women’s Hospital, Harvard Medical School, 75 Francis
St., Boston, MA 02215. Address e-mail to body@zeus.bwh.harvard.edu.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181ce1ffa
platelet count ⬎300,000 cells/mm3, recent heparin expo-
sure, and AT deficiency.9 Preoperative AT activity of ⬍80%
of normal has been associated with reduced heparin re-
sponse in adults undergoing cardiac surgery.11 Postopera-
tive AT deficiency has also been associated with worsened
clinical outcomes, including increased intensive care unit
(ICU) length of stay, risk of reexploration for bleeding, and
thromboembolic events.12
Because the primary mechanism of action of heparin is
to facilitate the enzymatic activity of AT, we surmised that
reduced AT activity would be associated with heparin
response. Little evidence is available to support this hy-
pothesis, with most data regarding AT activity and heparin
dose response (HDR) reported from studies examining resto-
ration of heparin responsiveness in cardiac surgical patients
by administration of recombinant AT concentrate, usually
without measurement of the patient’s AT level.5– 8,13,14 We
aimed to assess the relationship between preoperative AT
activity and heparin sensitivity in a cohort of primary
coronary artery bypass graft (CABG) patients and addi-
tionally assess this association in a subgroup of subjects
with low preoperative AT activity. We further examined
whether AT activity or measures of heparin sensitivity
were associated with severity of myocardial injury or

856 www.anesthesia-analgesia.org October 2010 • Volume 111 • Number 4

duration of ICU and hospital stays, as surrogates of sever-
ity of illness.
METHODS
With IRB approval and individual patient consent, 346
patients undergoing primary CABG using CPB from Feb-
ruary 2005 to December 2006 were enrolled into a parent
study called the CABG Genomics Program with the aim of
determining genetic risks for adverse perioperative out-
comes. Patients were excluded from the parent study if
they were younger than 20 years of age; underwent repeat
or off-pump CABG, planned concomitant valve, or other
cardiac surgery; had a preoperative hematocrit ⬍25%; or if
they had received leukocyte-rich blood products within 30
days before surgery. For this secondary analysis, detailed
demographic data, preoperative laboratory values, opera-
tive data, heparin bolus dose, anticoagulation data includ-
ing activated coagulation time (ACT) values, and measured
heparin concentrations were collected from hospital records.
Patients with missing laboratory or demographic data (n ⫽
27) and those receiving warfarin (n ⫽ 15) were further
excluded from analysis, yielding 304 analyzable patients.
ICU length of stay until first discharge was recorded in
hours. Hospital length of stay was recorded in days, with
the surgical day and last hospital day included as whole
days.
Baseline blood samples for ACT, HDR, AT, and other
assays were obtained after induction of anesthesia but
before surgical incision (described as preoperative hereaf-
ter). Free unbound AT activity was measured with a
colorimetric method (Modular Analytics biochemistry ana-
lyzer, Siemens Healthcare Diagnostics, Tarrytown, NY)
performed by Charles River Laboratory (Montreal, Can-
ada). The assay limit of quantitation was 21.6%. To report
human plasma AT activity results in IU/mL, the National
Institute for Biological Standards and Control Second In-
ternational Reference Standard was used to determine a
conversion factor of activity in IU/mL ⫽ AT activity in % ⫻
0.0102. AT content was measured using an immunoneph-
elometric method using a BN-100 Prospec nephelometer
(Dade Behring Diagnostics, Marburg, Germany). Coeffi-
cient of variation for AT measurement was ⬍5% within
assay and ⬍10% across assays. The assay limit of quanti-
tation was 0.00672 mg/mL. To report human plasma AT
content results in IU/mL, National Institute for Biological
Standards and Control Second International Reference
Standard was used to determine a conversion factor of
content in IU/mL ⫽ content in g/L ⫻ 3.64. Both assays
measure free AT, rather than AT complexed with heparin.
Plasminogen-activator inhibitor-1, tissue factor, d-dimer,
protein C, and cardiac troponin I (cTnI) were measured
using a sandwich immunoassay on a triage platform using
monoclonal and polyclonal antibodies (Biosite, San Diego,
CA). Hemoglobin, platelet, and white blood counts, along
with prothrombin time, partial thromboplastin time, and
international normalized ratio were measured by a central
hematology laboratory according to institutional protocols.
Complete blood counts were performed using the Advia
2120 Hematology System (Siemens Healthcare Diagnostics,
Deerfield, IL), and coagulation studies were performed on
STA-R Evolution (Diagnostica Stago, Parsippany, NJ).
October 2010 • Volume 111 • Number 4
For anticoagulation management, the Hepcon HMS Plus
system (Medtronic, Minneapolis, MN) was used according
to the manufacturer’s recommendations.15 The estimated
blood volume for each patient was calculated using the
manufacturer’s instructions,15 according to the method
described by Allen et al.16 After induction of anesthesia,
baseline kaolin ACT, predicted HDR, predicted heparin
concentration, and heparin bolus calculations were per-
formed according to the manufacturer’s instructions, using
heparin-protamine titration cartridges encompassing whole
blood heparin concentration ranging from 0.7 to 3.4 U/mL
and kaolin as the activator. The recommended Hepcon
HMS Plus CPB prime heparin dose based on a 750- or
1000-mL prime volume was added to the calculated hepa-
rin bolus and administered via a central venous catheter.
Heparin was not added to the CPB pump prime. Through-
out the entire study period, an ACT of ⬎350 seconds was
used in patients undergoing surgery where cardiotomy
suction was to be used. Patients undergoing primary CABG
surgery without the use of cardiotomy suction were anti-
coagulated using a protocol that prescribed a minimal ACT
of 300 seconds before the institution of CPB. Three minutes
after USP porcine heparin (APP Pharmaceuticals, Schaum-
burg, IL) administration, heparin concentration and ACT
were remeasured. This heparin management protocol
including use of heparin-coated circuits was adopted in
a comprehensive institutional program to reduce the rate
of reoperation for bleeding. All patients received an
⑀-aminocaproic acid initial loading dose of 7.5 to 10 g over
1 hour, after the initial blood draw for AT level and baseline
ACT, but before heparin administration and blood sam-
pling for measurement of postheparin ACT.
HDR was measured as the difference in ACT between
target and baseline ACT measurements, divided by target
heparin level estimated from the Hepcon HMS Plus system.
Because the Hepcon HMS Plus system has a limited fidelity
in reporting whole blood heparin concentrations, in that it
provides values with discrete categories (i.e., 0.7, 1.4, 2.0,
2.7, and 3.4 U/mL) rather than a continuous variable, the
heparin sensitivity index (HSI) was also calculated from
change in ACT between before and after heparin adminis-
tration, divided by heparin dose, per kilogram of body
weight.11 The same calculation was performed using hep-
arin dose per liter of estimated blood volume without
significantly changing the results; therefore, body weight
was used.
We estimated the accessible effect size for HDR, using
the available sample size (n ⫽ 319), a 20% rate of heparin
resistance based on prior studies, a mean HDR of 99
s/U/mL and an sd of 22 s/U/mL, a type I error rate of 5%
and a type II error rate of 20%. We estimated that we would
be able to observe differences in HDR of 9 s/U/mL, which
we thought would be more sensitive than a clinically
important difference.
Data are presented as mean ⫾ sd when normally
distributed, or median and 5th and 95th percentiles when
not normally distributed. Data that were nonnormally
distributed were compared using the Wilcoxon ranked sum
test. The Student t test was used to compare means of
normally distributed data. Subgroup analysis of patients
with low AT activity (⬍80% of laboratory normal; AT
www.anesthesia-analgesia.org 857
Antithrombin and Heparin Response

Table 1. Demographics and Perioperative Data of Patients with Low Antithrombin (AT) Activity (<80% of
Normal; Defined as AT Activity <0.813 U/mL) and Normal AT Activity
Study cohort AT activity <80% of normala AT activity >80% of normala
(N ⴝ 304) (N ⴝ 49) (N ⴝ 255) P
Age 65 ⫾ 9 70 ⫾ 9 64 ⫾ 9 ⬍0.0001
Male gender 247 (81.3%) 42 (85.7%) 205 (80.4%) 0.43
Race (Caucasian) 272 (89.5%) 44 (89.8%) 228 (89.4%) 1.0
Height (cm) 173 ⫾ 9 174 ⫾ 9 173 ⫾ 9 0.56
Weight (kg) 88.3 ⫾ 18.6 87.0 ⫾ 16.6 88.4 ⫾ 19.0 0.62
Medical history
Previous myocardial infarction 130 (42.7%) 27 (55.1%) 103 (40.4%) 0.06
Myocardial infarction within prior 2 wk 56 (18.4%) 16 (32.7%) 40 (15.7%) 0.008
Hypertension 224 (73.7%) 30 (61.2%) 194 (76.1%) 0.03
Hypercholesterolemia 236 (77.6%) 33 (67.4%) 203 (79.6%) 0.06
Insulin-dependent diabetes 31 (10.2%) 9 (18.4%) 22 (8.6%) 0.07
Renal insufficiency (creatinine ⱖ2.0 mg/dL) 8 (2.6%) 2 (4.1%) 6 (2.4%) 0.62
Liver failure 0 (0%) 0 (0%) 0 (0%) —
Preoperative medications
Aspirin 256 (84.2%) 38 (77.6%) 218 (85.4%) 0.20
Platelet inhibitor (excluding aspirin) 71 (23.3%) 14 (28.6%) 57 (22.4%) 0.36
Intravenous heparin in current admission 88 (28.9%) 33 (67.4%) 55 (21.6%) ⬍0.0001
Any statin 252 (82.8%) 41 (83.7%) 211 (82.8%) 1.0
Preoperative laboratory values
Hematocrit (%) 39.9 ⫾ 4.5 39.5 ⫾ 5.3 40.0 ⫾ 4.3 0.52
Platelet count (109/mL) 237 (169–322) 245 (170–339) 236 (167–317) 0.55
White cell count (103/mL) 8.0 (5.5–11.2) 8.6 (5.4–11.8) 7.8 (5.5–11.1) 0.27
INR 1.0 (1.0–1.1) 1.0 (1.0–1.2) 1.0 (1.0–1.1) 0.02
INR ⬎1.4 0 (0%) 0 (0%) 0 (0%) —
Partial thromboplastin time (s) 30.2 (26.5–63.2) 39 (28–94) 30 (26–40) ⬍0.0001
Antithrombin content (U/mL) 0.862 ⫾ 0.123 0.703 (0.582–0.790) 0.881 (0.772–1.04)
Antithrombin content (mg/mL) 0.23 (0.196–0.282) 0.193 (0.160–0.217) 0.242 (0.212–0.286)
Antithrombin activity (U/mL) 0.933 ⫾ 0.126 0.761 (0.649–0.811) 0.961 (0.855–1.11)
Antithrombin activity (%) 91.1 (76.3–107.2) 74.8 (63.6–79.5) 93.8 (83.7–108.7)
Protein C (␮g/mL) 5.01 (2.27–6.98) 4.75 (3.32–6.63) 5.07 (3.53–6.96) 0.10
Plasminogen activator inhibitor-1 (ng/mL) 1.53 (0.0–20.2) 1.18 (0.23–3.88) 1.58 (0.06–6.92) 0.16
D-dimer (ng/mL) 183.3 (0.0–2569) 142.8 (6.8–1084) 186.6 (7.2–1381) 0.52
Tissue factor (pg/mL) 16.4 (0.03–495.6) 12.6 (0.04–73.3) 16.5 (0.04–118.8) 0.49
Heparin administration
Baseline ACT (s) 135 ⫾ 18 138 (114–163) 136 (114–156) 0.12
Target ACT
300 s 257 (84.5%) 41 (16.0%) 216 (84.0%)
350 s 47 (15.5%) 8 (17.0%) 42 (83.0%) 0.86
Heparin dose (U/kg) 150 (86–246) 154 (112–198) 150 (111–211) 0.75
Heparin dose ⬎200 U/kg 59 (19.4%) 8 (16.3%) 51 (20.0%) 0.69
ACT after heparin administration
300 s 366 (300–456) 357 (300–469) 366 (300–456) 0.53
350 s 389 (324–460) 395 (366–425) 389 (319–470) 0.81
Calculated HDR slope (s/U/mL) 98 ⫾ 21 96 ⫾ 20 98 ⫾ 21 0.50
Heparin sensitivity index (s/U/kg) 1.59 ⫾ 0.41 1.58 ⫾ 0.40 1.60 ⫾ 0.41 0.68
Clinical outcomes
Cardiac troponin I on POD 1 (␮g/L) 0.99 (0.13–7.72) 1.17 (0.31–3.21) 0.97 (0.32–3.89) 0.48
ICU duration (h) 45 (21–94) 48 (24–116) 44 (21–90) 0.13
Postsurgical hospital stay (d) 6 (4–10) 6 (4–11) 6 (4–10) 0.17
Data presented as mean ⫾ SD or median (10th–90th percent confidence intervals).
INR ⫽ International Normalized Ratio; ACT ⫽ activated clotting time; POD 1 ⫽ postoperative day 1; HDR ⫽ heparin dose response; ICU ⫽ intensive care unit.
a
Defined as 100% of the laboratory normal for AT activity, not corrected for the National Institute for Biological Standards and Control Second International
Reference Standard (see Methods).

activity ⬍0.813 U/mL) who might be at risk for heparin RESULTS

resistance was performed. Multivariable linear regres-
sion modeling was performed to examine for clinical and
laboratory predictors of HDR and HSI; variables with
univariate P values ⬍0.2 were entered into a combined
forward/backward stepwise linear regression model, with
an exit P value of 0.05. A 2-sided P ⬍ 0.05 was considered
as showing statistical significance. Statistical analyses were
performed using SAS version 9.1.3 and JMP 7.03 (SAS
Institute, Cary, NC).
Baseline demographics and perioperative data for 304
patients are summarized in Table 1. Thirty-two patients
(10.5%) required additional heparin administration before
the institution of CPB for failing to achieve their respective
target ACT with administration of the heparin dose esti-
mated by the HepCon HMS Plus system. No patient
received ⬎300 U/kg heparin. No patients were given fresh
frozen plasma or AT supplementation to reach target ACT
values.

858 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

Figure 1. Preoperative antithrombin (AT)
activity versus heparin dose response
(HDR) (r2 ⬍ 0.001; P ⫽ 0.59). Confi-
dence interval of HDR for a single mea-
surement of AT activity is shaded. HDR
slope (s/U/mL) ⫽ 94.3 ⫹ 5.31 ⫻ AT
activity (U/mL).

Table 2. Multivariable Predictors of Heparin-Dose Response (HDR)

Without AT activity With AT activity
r2 ⴝ 0.138 r2 ⴝ 0.140
Heparin dose response (s/U/mL) Estimate P Estimate P
Female gender ⫺1.82 ⫾ 1.66 0.27 ⫺2.02 ⫾ 1.67 0.23
Age (y) 0.16 ⫾ 0.13 0.23 0.176 ⫾ 0.131 0.18
Caucasian race ⫺3.81 ⫾ 1.87 0.042 ⫺3.69 ⫾ 1.87 0.049
Hypercholesterolemia 5.61 ⫾ 1.39 ⬍0.0001 5.52 ⫾ 1.40 ⬍0.0001
Hematocrit 0.71 ⫾ 0.29 0.016 0.70 ⫾ 0.29 0.016
Prothrombin time (s) 5.57 ⫾ 1.82 0.002 5.83 ⫾ 1.84 0.002
White cell count (⫻106/mL) ⫺1.52 ⫾ 0.51 0.003 ⫺1.50 ⫾ 0.51 0.004
AT activity (U/mL) 8.32 ⫾ 9.64 0.39
AT ⫽ antithrombin.

AT activity and content were highly correlated (r2 ⫽
0.801), therefore AT activity is reported. Higher AT levels
(P ⬍ 0.05) were observed in females, younger individuals,
current smokers, hypercholesterolemia, and individuals
who had not had a myocardial infarction within the previ-
ous 2 weeks. No association was observed between base-
line AT activity and baseline ACT (r2 ⬍ 0.001; P ⫽ 0.10) or
with HDR (r2 ⬍ 0.001; P ⫽ 0.59) (Fig. 1), HSI (r2 ⬍ 0.001; P ⫽
0.73), or platelet count (r2 ⫽ 0.004; P ⫽ 0.24). Those patients
with recent preoperative heparin exposure had signifi-
cantly lower AT activity (0.87 vs 0.96 U/mL; P ⬍ 0.001) but
did not show a significant difference in heparin require-
ments or heparin responsiveness by any measure including
HDR or HSI.
Individuals with AT activity ⬍80% normal (⬍0.813 U/mL)
had higher partial thromboplastin time and were more likely
to have received heparin during the current admission before
surgery, but otherwise showed no differences in coagulation
testing or management (Table 1). No relationship between AT
activity and HDR was observed in these patients (r2 ⬍ 0.001).
Furthermore, there was no evidence of diminished heparin
responsiveness in this group, because 94.1% (48 of 51)
achieved the target ACT after administration of the calculated
heparin bolus dose. Neither HDR nor HSI was significantly
related to AT activity in univariate relationship (Fig. 1), or
after accounting for other covariates using multivariable lin-
ear regression (Tables 2 and 3).
Preoperative AT activity, HDR, and HSI were not
associated with cTnI levels on the first postoperative day
(all r2 ⬍ 0.002; P ⬎ 0.5), ICU duration, or hospital length
of stay.
DISCUSSION
We observed no relationship between preoperative AT
activity and response to heparin, measured using either
HDR or HSI, in these 304 patients undergoing primary

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 859

Antithrombin and Heparin Response
Table 3. Multivariable Predictors of Heparin Sensitivity Index (HSI)
Without AT activity
r ⴝ 0.136
2
Heparin sensitivity index (s/U/kg) Estimate
Female gender 0.070 ⫾ 0.032
Age (y) 0.001 ⫾ 0.003
Caucasian race 0.019 ⫾ 0.037
Platelet count (⫻109/mL) ⫺0.0007 ⫾ 0.0003
PT (s) 0.10 ⫾ 0.04
White cell count (⫻106/mL) ⫺0.022 ⫾ 0.011
PAI-1 level (ng/mL) 0.032 ⫾ 0.009
Tissue factor level (pg/mL) ⫺0.0008 ⫾ 0.0003
AT activity (U/mL)
PAI-1 ⫽ plasminogen activator inhibitor level-1; AT ⫽ antithrombin; PT ⫽ prothrombin time.
CABG surgery. Additional subgroup analysis in patients
with baseline AT activity ⬍80% failed to identify AT
activity as a predictor of HDR or HSI. However, no patient
exhibited a requirement for a heparin dose ⬎300 U/kg to
achieve target ACTs of either 300 or 350 seconds.
Acquired AT deficiency is common in patients with
previous heparin administration,17 critical illness, severe
hepatic dysfunction, and after major cardiovascular sur-
gery.18,19 After cardiac surgery, lower levels of AT have
been independently associated with prolonged ICU stay
and a higher incidence of neurologic and thromboembolic
events.12 In this cohort of patients, we observed associa-
tions between clinical markers of anticoagulation (pro-
thrombin time and ACT) and factors that may relate to the
severity of illness including white blood cell count and
hematocrit, and HDR. However, neither AT activity nor
heparin sensitivity was associated with severity of myocar-
dial injury, ICU length of stay, or hospital length of stay, as
surrogates of severity of illness.
Administration of unfractionated heparin remains the
mainstay of anticoagulation management for patients un-
dergoing cardiac surgery requiring CPB, with the goal of
maintaining therapeutic anticoagulation, thereby prevent-
ing thromboembolic complications. Its unique properties of
rapidly providing systemic anticoagulation that can be
maintained for the duration of CPB and rapid complete
reversal with subsequent administration of protamine
make heparin the anticoagulant of choice for CPB. Interin-
dividual variability in heparin response is well described,20
and contributing to this variation may be previously ob-
served risk factors for diminished heparin responsiveness
that include low AT levels, platelet count ⬎300,000
cells/mm3, and heparin pretreatment.9 Although we ob-
served substantial variability in heparin responsiveness,
our data failed to show any relationship between heparin
response and AT activity or heparin pretreatment.
AT is critical for maintenance of anticoagulation during
CPB and is consumed during the process.21 Despotis et
al.14 demonstrated a strong association between in vitro AT
and heparin responsiveness measured by ACT slope over
the range of AT levels of 0.2 to 1 U/mL. At AT levels above
1 U/mL, there was no further increase in heparin respon-
siveness. However, there was much weaker association in
31 patients undergoing cardiac surgery with CPB.14 Simi-
larly, Dietrich et al.11 observed a diminished HSI in adult
860 www.anesthesia-analgesia.org
With AT activity
r2 ⴝ 0.137
P Estimate P
0.029 0.068 ⫾ 0.032 0.037
0.79 0.001 ⫾ 0.003 0.72
0.61 0.021 ⫾ 0.037 0.58
0.013 ⫺0.0007 ⫾ 0.0003 0.013
0.006 0.10 ⫾ 0.04 0.005
0.048 ⫺0.022 ⫾ 0.011 0.048
0.0005 0.031 ⫾ 0.009 0.0007
0.015 ⫺0.0008 ⫾ 0.0003 0.016
0.121 ⫾ 0.198 0.54
patients with preoperative AT activity ⬍80% of normal,
whereas no relationship was found in patients with normal
AT levels. Our data contrast with this, in that we observed
no relationship between AT activity and heparin respon-
siveness in either group.
The ACT response to heparin is complex and affected by
different factors as previously noted. Although AT en-
hancement is a primary mechanism of heparin’s action,
other factors may influence HDR including tissue factor
pathway inhibitor levels in vivo but not in vitro, extravas-
cular sequestration of heparin, plasma protein binding,
leukocyte lactoferrin, and activated platelets.7,14,22 Tissue
factor pathway inhibitor is an endothelium-derived endog-
enous serine protease inhibitor with enhanced expression
after heparin administration23 that may enhance anticoagu-
lation in vivo.24 Other endogenous mechanisms may
further modify the activity of heparin in vivo. Higher
molecular weight heparin (⬎13 kDa) is sequestered and
deactivated by endothelial endocytosis and depolymeriza-
tion.25 In addition, neutrophil-derived lactoferrin can neu-
tralize heparin by ionic binding.22 Heparin has shown to be
similarly neutralized by histidine-rich glycoproteins such
as vitronectin, fibronectin, and kininogen.7 Platelet factor 4
is a potent heparin binding agent released from activated
platelets that reverses the ACT,26,27 potentially contributing
to heparin resistance. These endogenous mechanisms may
be important in modifying heparin’s anticoagulant activity
in vivo, but difficult to quantify in vitro.
This study has important limitations. Our study did not
identify patients with severe heparin resistance, with only
10.5% of patients failing to reach the target ACT, and none
required fresh frozen plasma or AT to initiate CPB. Perhaps
the relatively low ACT target contributed to the lack of
identification of heparin-resistant patients. Historically,
CPB has been initiated with higher ACTs to give a margin
of safety28 and in many studies, a large number of patients
diagnosed as heparin resistant had target ACTs ⬎400
seconds.5– 8,13,14 Definitions of heparin resistance based on
a dose of heparin to achieve a specific target ACT such as
requiring ⬎500 IU/kg to achieve a target ACT of 480
seconds assume linearity of the HDR.10 Although Despotis
et al.29 observed a strong linear relationship between ACT
and heparin concentration observed over a range of hepa-
rin concentrations, to assume that our results could be
ANESTHESIA & ANALGESIA
extrapolated to higher target ACTs would not be appropri-
ate. Furthermore, we did not measure circulating concen-
trations of several important members of the coagulation
pathway that may have affected these results.
In conclusion, we found that heparin responses were
independent of preoperative plasma AT activity in patients
undergoing primary CABG using target ACTs of 300 to 350
seconds. Preoperative heparin exposure was associated
with diminished AT activity, but no change in HDR was
observed. Perioperative AT activity, HDR, or HSI were not
associated with cTnI levels on the first postoperative day,
ICU duration, or hospital length of stay.
DISCLOSURE
Dr. Garvin was the recipient of a research fellowship funded
by Talecris Biotherapeutics, which allowed time for generation
of this and other articles. Dr. Body has received consulting fees
from Talecris Biotherapeutics for another study.
Talecris Biotherapeutics paid for the costs of antithrombin
analyses performed by Charles River Laboratories in Ontario,
Canada, but had no input into these analyses.
REFERENCES
1. McLean J. The thromboplastin action of cephalin. Am J Physiol
1916;41:250 –7
2. Rosenberg RD, Damus PS. The purification and mechanism of
action of human antithrombin-heparin cofactor. J Biol Chem
1973;248:6490 –505
3. Rosenberg RD. Actions and interactions of antithrombin and
heparin. N Engl J Med 1975;292:146 –51
4. Langdown J, Johnson DJ, Baglin TP, Huntington JA. Allosteric
activation of antithrombin critically depends upon hinge re-
gion extension. J Biol Chem 2004;279:47288 –97
5. Avidan MS, Levy JH, Scholz J, Delphin E, Rosseel PM, Howie
MB, Gratz I, Bush CR, Skubas N, Aldea GS, Licina M, Bonfiglio
LJ, Kajdasz DK, Ott E, Despotis GJ. A phase III, double-blind,
placebo-controlled, multicenter study on the efficacy of recom-
binant human antithrombin in heparin-resistant patients
scheduled to undergo cardiac surgery necessitating cardiopul-
monary bypass. Anesthesiology 2005;102:276 – 84
6. Avidan MS, Levy JH, van Aken H, Feneck RO, Latimer RD, Ott
E, Martin E, Birnbaum DE, Bonfiglio LJ, Kajdasz DK, Despotis
GJ. Recombinant human antithrombin III restores heparin
responsiveness and decreases activation of coagulation in
heparin-resistant patients during cardiopulmonary bypass.
J Thorac Cardiovasc Surg 2005;130:107–13
7. Ranucci M, Isgro G, Cazzaniga A, Ditta A, Boncilli A, Cotza M,
Carboni G, Brozzi S. Different patterns of heparin resistance:
therapeutic implications. Perfusion 2002;17:199 –204
8. Williams MR, D’Ambra AB, Beck JR, Spanier TB, Morales DL,
Helman DN, Oz MC. A randomized trial of antithrombin
concentrate for treatment of heparin resistance. Ann Thorac
Surg 2000;70:873–7
9. Ranucci M, Isgro G, Cazzaniga A, Soro G, Menicanti L, Frigiola
A. Predictors for heparin resistance in patients undergoing
coronary artery bypass grafting. Perfusion 1999;14:437– 42
10. Staples M, Dunton D, Karlson K, Leonardi H, Berger R.
Heparin resistance after preoperative heparin therapy or in-
traaortic ballon pumping. Ann Thorac Surg 1994;57:1211– 6
October 2010 • Volume 111 • Number 4
11. Dietrich W, Braun S, Spannagl M, Richter JA. Low preopera-
tive antithrombin activity causes reduced response to heparin
in adult but not in infant cardiac-surgical patients. Anesth
Analg 2001;92:66 –71
12. Ranucci M, Frigiola A, Menicanti L, Ditta A, Boncilli A, Brozzi
S. Postoperative antithrombin levels and outcome in cardiac
operations. Crit Care Med 2005;33:355– 60
13. Lemmer JH Jr, Despotis GJ. Antithrombin III concentrate to
treat heparin resistance in patients undergoing cardiac sur-
gery. J Thorac Cardiovasc Surg 2002;123:213–7
14. Despotis GJ, Levine V, Joist JH, Joiner-Maier D, Spitznagel E.
Antithrombin III during cardiac surgery: effect on response of
activated clotting time to heparin and relationship to markers
of hemostatic activation. Anesth Analg 1997;85:498 –506
15. Medtronic Perfusion Systems. Hepcon HMS Plus Operator’s
Manual. Minneapolis, MN: Medtronic Perfusion Systems, 2001
16. Allen TH, Peng MT, Chen KP, Huang TF, Chang C, Fang HS.
Prediction of blood volume and adiposity in man from body
weight and cube of height. Metabolism 1956;5:328 – 45
17. Dietrich W, Spannagl M, Schramm W, Vogt W, Barankay A,
Richter JA. The influence of preoperative anticoagulation on
heparin response during cardiopulmonary bypass. J Thorac
Cardiovasc Surg 1991;102:505–14
18. Maclean PS, Tait RC. Hereditary and acquired antithrombin
deficiency: epidemiology, pathogenesis and treatment options.
Drugs 2007;67:1429 – 40
19. Levy JH, Despotis GJ, Szlam F, Olson P, Meeker D, Weisinger A.
Recombinant human transgenic antithrombin in cardiac surgery:
a dose-finding study. Anesthesiology 2002;96:1095–102
20. Bull BS, Korpman RA, Huse WM, Briggs BD. Heparin therapy
during extracorporeal circulation. I. Problems inherent in existing
heparin protocols. J Thorac Cardiovasc Surg 1975;69:674 – 84
21. Hashimoto K, Yamagishi M, Sasaki T, Nakano M, Kurosawa H.
Heparin and antithrombin III levels during cardiopulmonary
bypass: correlation with subclinical plasma coagulation. Ann
Thorac Surg 1994;58:799 – 804; discussion 804 –5
22. Wu HF, Lundblad RL, Church FC. Neutralization of heparin
activity by neutrophil lactoferrin. Blood 1995;85:421– 8
23. Bajaj MS, Kuppuswamy MN, Saito H, Spitzer SG, Bajaj SP.
Cultured normal human hepatocytes do not synthesize
lipoprotein-associated coagulation inhibitor: evidence that en-
dothelium is the principal site of its synthesis. Proc Natl Acad
Sci USA 1990;87:8869 –73
24. Brodin E, Appelbom H, Osterud B, Hilden I, Petersen LC,
Hansen JB. Regulation of thrombin generation by TFPI in
plasma without and with heparin. Transl Res 2009;153:124 –31
25. Barzu T, van Rijn JL, Petitou M, Tobelem G, Caen JP. Heparin
degradation in the endothelial cells. Thromb Res 1987;47:601–9
26. Lane DA, Pejler G, Flynn AM, Thompson EA, Lindahl U.
Neutralization of heparin-related saccharides by histidine-rich
glycoprotein and platelet factor 4. J Biol Chem 1986;261:3980 – 6
27. Levy JH, Cormack JG, Morales A. Heparin neutralization by
recombinant platelet factor 4 and protamine. Anesth Analg
1995;81:35–7
28. Young JA, Kisker CT, Doty DB. Adequate anticoagulation
during cardiopulmonary bypass determined by activated clot-
ting time and the appearance of fibrin monomer. Ann Thorac
Surg 1978;26:231– 40
29. Despotis GJ, Summerfield AL, Joist JH, Goodnough LT, San-
toro SA, Spitznagel E, Cox JL, Lappas DG. Comparison of
activated coagulation time and whole blood heparin measure-
ments with laboratory plasma anti-Xa heparin concentration in
patients having cardiac operations. J Thorac Cardiovasc Surg
1994;108:1076 – 82
www.anesthesia-analgesia.org 861
Postoperative Activity, but Not Preoperative Activity,
of Antithrombin Is Associated with Major Adverse
Cardiac Events After Coronary Artery Bypass
Graft Surgery
Sean Garvin, MD,* Jochen D. Muehlschlegel, MD,* Tjörvi E. Perry, MD,* Junliang Chen, PhD,†
Kuang-Yu Liu, PhD,* Amanda A. Fox, MD,* Charles D. Collard, MD,‡ Sary F. Aranki, MD,§
Stanton K. Shernan, MD,* and Simon C. Body, MB, ChB, MPH*

BACKGROUND: Low levels of antithrombin (AT) have been independently associated with prolonged
intensive care unit stay and an increased incidence of neurologic and thromboembolic events after
cardiac surgery. We hypothesized that perioperative AT activity is independently associated with
postoperative major adverse cardiac events (MACEs) in patients undergoing coronary artery bypass
graft (CABG) surgery.
METHODS: We prospectively studied 1403 patients undergoing primary CABG surgery with cardio-
pulmonary bypass (CPB) (http://clinicaltrials.gov/show/NCT00281164). The primary clinical end
point was occurrence of MACE, defined as a composite outcome of any one or more of the following:
postoperative death, reoperation for coronary graft occlusion, myocardial infarction, stroke, pulmo-
nary embolism, or cardiac arrest until first hospital discharge. Plasma AT activity was measured
before surgery, after post-CPB protamine, and on postoperative days (PODs) 1–5. Multivariate logistic
regression modeling was performed to estimate the independent effect of perioperative AT activity
upon MACE.
RESULTS: MACE occurred in 146 patients (10.4%), consisting of postoperative mortality (n ⫽ 12),
myocardial infarction (n ⫽ 108), stroke (n ⫽ 17), pulmonary embolism (n ⫽ 8), cardiac arrest (n ⫽
16), or a subsequent postoperative or catheter-based treatment for graft occlusion (n ⫽ 6). AT activity
at baseline did not differ between patients with (0.91 ⫾ 0.13 IU/mL; n ⫽ 146) and without (0.92 ⫾
0.13 IU/mL; n ⫽ 1257) (P ⫽ 0.18) MACE. AT activity in both groups was markedly reduced
immediately after CPB and recovered to baseline values over the ensuing 5 PODs. Postoperative AT
activity was significantly lower in patients with MACE than those without MACE. After adjustment for
clinical predictors of MACE, AT activity on PODs 2 and 3 was associated with MACE.
CONCLUSIONS: Preoperative AT activity is not associated with MACE after CABG surgery. MACE
is independently associated with postoperative AT activity but only at time points occurring
predominantly after the MACE. (Anesth Analg 2010;111:862–9)

A ntithrombin (AT) is a serine protease inhibitor

(serpin) and the principal inhibitor of the final
common pathway of the coagulation system by
inactivation of circulating thrombin (factor IIa) and factor
Xa, among other serine proteases. Heparin increases AT
activity 2000- to 4000-fold due to a conformational change
in the quaternary structure of AT by heparin binding and
through formation of a ternary complex of thrombin, AT,
and heparin. Heparin-augmented AT activity is still the
principal mechanism of anticoagulation for cardiopulmo-
nary bypass (CPB).
From the *Department of Anesthesiology, Perioperative and Pain Medicine,
Brigham and Women’s Hospital, Harvard Medical School, Boston, Massa-
chusetts; †Talecris Biotherapeutics, Research Triangle Park, Durham, North
Carolina; ‡Baylor College of Medicine Division of Cardiovascular Anesthe-
sia at the Texas Heart Institute, Saint Luke’s Episcopal Hospital, Houston,
Texas; and §Division of Cardiac Surgery, Brigham and Women’s Hospital,
Harvard Medical School, Boston, Massachusetts.
Accepted for publication June 22, 2009.
The authors had full access to the data and take responsibility for its
integrity. All authors have read and agree to the manuscript as written.
Address correspondence and reprint requests to Simon C. Body, MB, ChB,
MPH, Department of Anesthesiology, Perioperative and Pain Medicine,
Brigham and Women’s Hospital, 75 Francis St., Boston, MA 02115. Address
e-mail to body@zeus.bwh.harvard.edu.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181b7908c
AT levels are decreased after administration of heparin
due to degradation of the ternary complex. Additionally,
acquired AT deficiency is common in patients with critical
illness, severe hepatic dysfunction, and after major cardio-
vascular surgery.1,2 The magnitude of reduction in AT after
cardiac surgery is similar to that in patients with heterozy-
gous AT deficiency, which is associated with increased risk
of thromboembolic events.3,4 After cardiac surgery, lower
levels of AT have been independently associated with
prolonged intensive care unit (ICU) stay and a higher
incidence of neurologic and thromboembolic events.5 We
therefore examined for independent association between
perioperative AT activity and the frequency of postopera-
tive major adverse cardiac events (MACEs) in patients
undergoing coronary artery bypass graft (CABG) surgery.
METHODS
The study cohort was obtained from a continuing prospective
longitudinal parent study of 1447 patients undergoing primary
CABG surgery with CPB between August 2001 and May 2006 at
2 United States academic medical centers (CABG Genomics
Program; http://clinicaltrials.gov/show/NCT00281164). With
IRB approval, written informed consent was obtained from each
patient. Patients were excluded from the parent study if they
were younger than 20 yr old, underwent repeat or off-pump

862 www.anesthesia-analgesia.org October 2010 • Volume 111 • Number 4

CABG, had a preoperative hematocrit ⬍25%, or if they had
received leukocyte-rich blood products within 30 days before
surgery. All patients enrolled in the parent study were included.
Patients without postoperative cardiac troponin I (cTnI) levels
(n ⫽ 43) were excluded from further analysis. Demographic
data, medical and surgical history, medications, and out-
comes were recorded by trained research staff using defined
protocols in a purpose-built case report form. The examina-
tion of the relationship between AT levels and MACE was not
prespecified in the original parent study.
Perioperative anticoagulation protocols differed be-
tween institutions. At Brigham and Women’s Hospital,
patients received 300 U per kg body weight of porcine
heparin to achieve an activated clotting time (ACT) of ⬎400
s, until February 2004. From February 2004, patients re-
ceived a Hepcon HMS Plus (Medtronic, Minneapolis, MN)
calculated dose of porcine heparin to achieve an ACT of
either 300 or 350 s. At Texas Heart Institute, 300 U per kg
body weight of either bovine or porcine heparin was given
to achieve an ACT of ⬎400 s.
Clinical End Points for the Test Cohort
The primary clinical end point was prespecified as the
occurrence of a MACE, defined as a composite outcome of
any one or more of the following: postoperative mortality
(defined as all deaths occurring within 30 days of the
operation or occurring during the primary hospitalization),
reoperation for coronary graft occlusion, myocardial infarc-
tion (MI) (predefined as peak postoperative cTnI concen-
tration ⬎12 ng/mL, being the upper 8th percentile), cardiac
arrest (defined as a postoperative event requiring cardio-
pulmonary resuscitation) until first hospital discharge,
thromboembolic event consisting of stroke (defined as a
clinical diagnosis of focal or global neurological deficit), or
pulmonary embolism (diagnosed by ventilation perfusion
scan of moderate to high probability or by a positive
pulmonary angiogram).
Cardiac Biomarker Assay
Blood samples were obtained before surgery, 5 min after
administration of post-CPB protamine, and on the mornings
of postoperative days (PODs) 1–5. Citrated plasma was stored
in vapor-phase liquid nitrogen until analysis for cTnI with a
sandwich immunoassay on a Triage® platform using mono-
clonal and polyclonal antibodies (Biosite, San Diego, CA) at
a single core facility. Patient caregivers were not aware of
the results of the assays as they were performed after
patient discharge.
Antithrombin Assays
AT activity was measured with a colorimetric method using a
Modular Analytics biochemistry analyzer (Siemens Health-
care Diagnostics, Tarrytown, NY). The assay limit of quanti-
tation was 21.6%. To report human plasma AT activity results
in IU/mL, the National Institute for Biological Standards and
Control Second International Reference Standard was used to
determine a conversion factor of activity in IU/mL ⫽ activity
in % ⫻ 0.0102. AT content was measured using an immuno-
ephelometric method using a BN-100 ProSpec nephelometer
(Dade Behring Diagnostics, Marburg, Germany). The assay
limit of quantitation was 0.00672 mg/mL. To report human
October 2010 • Volume 111 • Number 4
plasma AT content results in IU/mL, the National Institute
for Biological Standards and Control Second International
Reference Standard was used to determine a conversion
factor of content in IU/mL ⫽ content in g/L ⫻ 3.64. Both
assays measure free AT rather than AT complexed with
heparin. Assays were performed by Charles River Labora-
tory in Montreal, Canada by personnel blinded to outcome
status. Subsequent comparison of paired AT activity and
content data revealed high correlation (r2 ⫽ 0.878), so only
the activity is reported.
Statistical Methods
Statistical analyses were performed using SAS, version
9.1.3, and JMP 7.0 (SAS Institute, Cary, NC). AT activity
was normally distributed at all time points, so was not
transformed. Data are presented as mean (sd) and median
with 10%–90% interquantile range, unless otherwise stated.
Continuous variables were compared using analysis of vari-
ance or Wilcoxon Mann-Whitney ranked sum test when
appropriate. Categorical variables were compared with ␹2 or
Fisher’s exact test.
Multivariate logistic regression modeling was per-
formed to identify and account for MACE risk factors that
might confound any association between low AT activity
and MACE. The multivariate analysis used a forward
stepwise technique to identify independent risk factors for
MACE, whereby clinically relevant demographic variables
and variables with a two-tailed univariate P ⱕ 0.2 were
entered into the model and P ⱕ 0.2 was necessary to remain
in the model. Age, gender, race, body mass index, and institu-
tion were forced into the model. Nagelkerke generalized r2 and
likelihood ratio test were used to determine the additional
predictive value of AT upon MACE. F tests were used to
compare generalized r2. Odds ratios and 95% confidence
intervals for a 0.1 IU/mL decrease in AT activity were
estimated. A two-sided P ⬍ 0.05 was considered significant.
RESULTS
The cohort comprised 1403 patients undergoing CABG
surgery whose characteristics are described in Table 1.
MACE occurred in 146 patients (10.4%), consisting of
postoperative mortality (n ⫽ 12), MI (n ⫽ 108), stroke (n ⫽
17), pulmonary embolism (n ⫽ 8), cardiac arrest (n ⫽ 16), or
a subsequent postoperative or catheter-based treatment for
graft occlusion (n ⫽ 6). Nineteen patients had 2 or 3 events,
usually MI, with either subsequent death or stroke. Most
adverse events occurred before or on POD 2. Of 12 patients
with operative mortality, 2 patients died at POD 0 and 10
patients died on or after POD 5. Of 17 patients with stroke,
6 patients had a stroke on or before POD 2. Of 108 patients
with MI, 89 patients had a diagnosis of MI first occurring
on POD 1. MACE frequency did not differ between insti-
tutions (Table 1).
AT activity and content were measured at the 7 time
points (Table 2). AT activity at baseline did not differ
between patients with MACE (0.91 ⫾ 0.13 IU/mL; n ⫽ 146)
and those without MACE (0.92 ⫾ 0.13 IU/mL; n ⫽ 1257)
(P ⫽ 0.18). AT activity was significantly reduced at the
post-CPB measurement compared with the preoperative
time point (P ⬍ 0.0001) and returned to baseline levels over
the ensuing 5-day period in patients with and without MACE.
www.anesthesia-analgesia.org 863
Antithrombin and Adverse Events

Table 1. Demographic and Clinical Characteristics of the Cohort as a Whole and Stratified by the
Occurrence of Major Adverse Cardiac Events (MACE)
Entire cohort Patients with MACE Patients without MACE
(n ⴝ 1403) (n ⴝ 146) (n ⴝ 1257) P
Age
⬍55 yr 240 (17.1) 22 (15.1) 218 (17.3)
55 to ⬍65 yr 494 (35.2) 51 (34.9) 443 (35.2)
65 to ⬍75 yr 422 (30.1) 42 (28.8) 380 (30.2)
75 to ⬍85 yr 231 (16.5) 29 (19.9) 202 (16.1)
At least 85 yr 16 (1.1) 2 (1.4) 14 (1.1) 0.75
Male 1136 (81.0) 109 (74.7) 1027 (81.7) 0.045
Caucasian race 1189 (84.6) 116 (79.5) 1073 (85.4) 0.068
Body mass index (kg/m2) 29.4 (5.5) 30.2 (5.5) 29.3 (5.4) 0.063
Institution
BWH 1061 (75.6) 109 (10.3) 952 (89.7) 0.76
THI 342 (24.4) 37 (10.8) 305 (89.2)
Medical history
Diabetes (drug treated; %) 466 (33.2) 52 (35.6) 414 (32.9) 0.52
Pulmonary disease (%) 224 (16.0) 17 (11.6) 207 (16.5) 0.13
Creatinine (mg/dL) 1.10 (0.334) 1.13 (0.322) 1.10 (0.335) 0.27
Hematocrit (%) 40.11 (4.724) 39.48 (4.825) 40.2 (4.708) 0.099
Hypertension (%) 1051 (74.9) 117 (80.1) 934 (74.3) 0.12
Hypercholesterolemia (%) 1045 (74.5) 106 (72.6) 939 (74.7) 0.58
Previous MI 620 (44.2) 83 (56.9) 537 (42.7) 0.002
Time since last MI
⬍2 wk 256 (18.3) 44 (30.1) 212 (16.9)
2–13 wk 47 (3.4) 6 (4.1) 41 (3.3)
⬎13 wk 266 (19.0) 31 (21.2) 235 (18.7)
Never 834 (59.4) 65 (44.5) 769 (61.2) 0.0002
Previous thrombolysis 71 (5.1) 13 (8.9) 58 (4.6) 0.025
IABP placed preoperatively 38 (2.7) 11 (7.5) 27 (2.2) 0.001
Arrhythmia requiring therapy 148 (10.6) 19 (13.0) 129 (10.3) 0.31
Peripheral vascular disease 130 (9.3) 20 (13.7) 110 (8.8) 0.051
Prior PVD procedure 39 (2.8) 10 (6.9) 29 (2.3) 0.0049
Prior stroke 63 (4.5) 10 (6.9) 53 (4.2) 0.15
LVEF preoperative ⬍40% 182 (13.0) 29 (19.9) 153 (12.2) 0.013
Medications—preoperative
ACE inhibitor 648 (46.2) 66 (45.2) 582 (46.3) 0.86
Beta-blocker 1094 (78.0) 116 (79.5) 978 (77.8) 0.75
Ca⫹⫹ antagonist 195 (13.9) 24 (16.4) 171 (13.6) 0.35
Aspirin 1072 (76.4) 113 (77.4) 959 (76.3) 0.84
HMG CoA reductase inhibitor 1082 (77.1) 107 (73.3) 975 (77.6) 0.24
Heparin (intravenous) 351 (25.0) 47 (32.2) 304 (24.2) 0.035
Platelet inhibitor (not aspirin) 304 (21.7) 32 (21.9) 272 (21.6) 0.94
Preoperative laboratory data
Hemoglobin (g/dL) 13.7 (1.7) 13.5 (1.7) 13.7 (1.6) 0.12
Creatinine (mg/dL) 1.10 (0.33) 1.13 (0.32) 1.10 (0.34) 0.27
Platelet count (109/mL) 240 (72) 234 (70) 241 (72) 0.32
cTnI (ng/mL) 0.4 (2.5) 1.9 (7.4) 0.2 (0.7) 0.006
Intraoperative management
Number of grafts
1 28 (2.0) 2 (1.4) 26 (2.1)
2 188 (13.4) 27 (18.5) 161 (12.8)
3 627 (44.8) 63 (43.2) 564 (44.9)
ⱖ4 558 (39.8) 54 (37.0) 504 (40.2) 0.28
CPB duration (min) 99.2 (42.21) 119.63 (57.821) 96.78 (39.353) ⬍0.0001
Aortic cross-clamp duration (min) 71.5 (35.03) 81.92 (45.125) 70.27 (33.471) 0.003
IABP placed intraoperatively 63 (4.5) 23 (15.8) 40 (3.2) ⬍0.0001
Heparin administration (mg)
BWH before February 2004 206 (83.7) 208 (76.0) 206 (85.0) 0.8319
BWH after February 2004 194 (61.0) 187 (54.7) 195 (61.5) 0.4608
THI 306 (92.0) 300 (80.6) 306 (93.4) 0.6998
Other surgical procedure
Concurrent mitral valve 35 (2.5) 10 (6.9) 25 (2.0) 0.0020
Concurrent aortic valve 22 (1.6) 3 (2.1) 19 (1.5) 0.49
Concurrent other valve 2 (0.1) 2 (1.4) 0 0.01
Other cardiac surgery 108 (7.7) 24 (16.4) 84 (6.7) ⬍0.0001
Other noncardiac surgery 17 (1.2) 1 (0.7) 16 (1.3) 1.0
(Continued)

864 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

 Table 1. Continued
Entire cohort Patients with MACE Patients without MACE
(n ⴝ 1403) (n ⴝ 146) (n ⴝ 1257) P
Intraoperative inotropes
Epinephrine 345 (24.6) 59 (40.4) 286 (22.8) ⬍0.0001
Norepinephrine 169 (12.1) 20 (13.7) 149 (11.9) 0.52
Phenylephrine 668 (47.6) 63 (43.2) 605 (48.1) 0.25
Dopamine ⬎5␮g 䡠 kg⫺1 䡠 min⫺1 38 (2.7) 3 (2.1) 35 (2.8) 0.79
Dobutamine 18 (1.3) 1 (0.7) 17 (1.4) 1.0
Milrinone 29 (2.1) 6 (4.1) 23 (1.8) 0.11
Vasopressin 17 (1.2) 3 (2.1) 14 (1.1) 0.41
Postoperative inotropes
Epinephrine 230 (16.4) 51 (34.9) 179 (14.2) ⬍0.0001
Norepinephrine 143 (10.2) 31 (21.2) 112 (8.9) ⬍0.0001
Phenylephrine 113 (8.1) 11 (7.5) 102 (8.1) 0.81
Dopamine ⬎5␮g 䡠 kg⫺1 䡠 min⫺1 30 (2.1) 8 (5.5) 22 (1.8) 0.009
Dobutamine 12 (0.9) 3 (2.1) 9 (0.7) 0.12
Milrinone 32 (2.3) 9 (6.2) 23 (1.8) 0.004
Vasopressin 91 (6.5) 26 (17.8) 65 (5.2) ⬍0.0001
Intraoperative and postoperative
transfusiona (median, 10–90
percentile)
Red blood cell transfusion (units) 1 (0–5) 2 (0–8) 1 (0–5) ⬍0.0001
Coagulation factor transfusion (units) 0 (0–2) 0 (0–4) 0 (0–2) 0.0055
Postoperative events
HLOS (d) ⬎12 d 130 (9.3) 43 (29.5) 87 (6.9) ⬍0.0001
ICU LOS (d) ⬎4 d 136 (9.7) 44 (30.1) 92 (7.3) ⬍0.0001
Peak postoperative cTnI ⬎12 ng/mL 108 (8.0) 108 (76.6) 0 ⬍0.0001
Peak postoperative cTnI 4.06 (8.6) 21.9 (17.8) 2.0 (2.1) ⬍0.0001
a
All blood products administered during the hospital stay. Red blood cell transfusion includes both packed red blood cells and whole blood. Coagulation factor
transfusion includes fresh frozen plasma, cryoprecipitate, and platelet transfusion.
BWH ⫽ Brigham and Women’s Hospital; THI ⫽ Texas Heart Institute; MI ⫽ myocardial infarction; cTnI ⫽ cardiac troponin I; IABP ⫽ intraaortic balloon pump;
LVEF ⫽ left ventricular ejection fraction; ACE ⫽ angiotensin converting enzyme; CPB ⫽ cardiopulmonary bypass; HLOS ⫽ hospital length of stay; ICU LOS ⫽ intensive
care unit length of stay; PVD ⫽ peripheral vascular disease.

Table 2. Postoperative Changes in Antithrombin Activity in Patients with Major Adverse Cardiac Events
(MACE) (n ⴝ 146) and Without MACE (n ⴝ 1257): Results of Univariate Analysis at Each Time Point
Postoperative day
Baseline Post-CPB 1 2 3 4 5
Antithrombin activity
(IU/mL)
Patients with MACE 0.91 ⫾ 0.13 0.59* ⫾ 0.12 0.66* ⫾ 0.12 0.72* ⫾ 0.12 0.78* ⫾ 0.13 0.85* ⫾ 0.15 0.90† ⫾ 0.16
Patients without 0.92 ⫾ 0.13 0.62 ⫾ 0.11 0.71 ⫾ 0.13 0.77 ⫾ 0.12 0.83 ⫾ 0.12 0.90 ⫾ 0.13 0.95 ⫾ 0.14
MACE
Change in
antithrombin
activity from
baseline (IU/mL)
Patients with MACE — ⫺0.32 ⫾ 0.13 ⫺0.24† ⫾ 0.13 ⫺0.19† ⫾ 0.13 ⫺0.13† ⫾ 0.14 ⫺0.06* ⫾ 0.15 ⫺0.00† ⫾ 0.16
Patients without — ⫺0.30 ⫾ 0.12 ⫺0.22 ⫾ 0.13 ⫺0.16 ⫾ 0.13 ⫺0.09 ⫾ 0.13 ⫺0.02 ⫾ 0.13 0.03 ⫾ 0.13
MACE
Percentage change in
antithrombin
activity from
baseline (%)
Patients with MACE — ⫺34.7† ⫾ 12.8 ⫺26.2† ⫾ 12.3 ⫺20.4* ⫾ 12.7 ⫺13.5† ⫾ 15.0 ⫺5.9† ⫾ 16.4 0.7† ⫾ 17.9
Patients without — ⫺32.0 ⫾ 11.2 ⫺22.7 ⫾ 13.5 ⫺16.1 ⫾ 13.3 ⫺9.2 ⫾ 14.7 ⫺1.1 ⫾ 14.5 4.6 ⫾ 15.5
MACE
Data are reported as mean ⫾ standard deviation.
CPB ⫽ cardiopulmonary bypass.
Significance is reported between MACE groups at each time point: *P ⱕ 0.001 and †P ⬍ 0.05 by Student’s t-test.

Postoperative AT activity was significantly lower in patients
with MACE than those without MACE (Table 2).
Decreased preoperative AT activity was independently
predicted by older age, male gender, and prior heparin use
within the same hospitalization (Table 3). Other clinical
variables that may possibly be indicative of recent heparin
use at prior recent hospitalization, such as recent MI, were
also independently predictive. Decreased preoperative AT
activity was also independently associated with lower
platelet count and increased partial thromboplastin time,

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 865

Antithrombin and Adverse Events

Table 3. Multivariable Predictors of Preoperative Antithrombin Activity

Predictors of preoperative AT activity (U/mL)
(n ⴝ 1194a; r2 ⴝ 0.168)
Univariate Multivariate Standard error of
Predictor estimate estimate multivariate estimate P
Age (1 yr increment) ⫺0.0020 ⫺0.0022 0.0004 ⬍0.0001
Gender (female) 0.0288 0.0177 0.0059 0.0028
Race (Caucasian) ⫺0.0078 ⫺0.0026 0.0054 0.624
Weight (1 kg increment) ⫺0.0003 ⫺0.0003 0.0002 0.233
Height (1 cm increment) ⫺0.0007 ⫺0.0031 0.0005 0.568
Institution 0.0183 0.0081 0.0053 0.129
Previous MI (Y) ⫺0.0273 ⫺0.0104 0.0036 0.004
Preoperative platelet count (10 ⫻ 0.0023 0.0013 0.0005 0.006
109/mL increment)
Preoperative PTT (1 s increment) ⫺0.0014 ⫺0.0008 0.0002 ⬍0.0001
Preoperative heparin use ⫺0.0863 ⫺0.0352 0.0042 ⬍0.0001
Preoperative diuretic use 0.0275 0.0147 0.0042 0.0005
a
209 subjects were missing one or more of the model’s predictor variables and are not included in this analysis.
AT ⫽ antithrombin; MI ⫽ myocardial infarction; PTT ⫽ partial thromboplastin time.

Table 4. Multivariable Predictors of Post-CPB Antithrombin (AT) Activity

Predictors of post-CPB AT activity (U/mL)
(n ⴝ 1205a; r2 ⴝ 0.504)
Univariate Multivariate Standard error of
Predictor estimate estimate multivariate estimate P
Age (1 yr increment) ⫺0.0030 ⫺0.0006 0.0002 0.0110
Gender (female) 0.0356 0.0023 0.0039 0.5581
Race (Caucasian) ⫺0.0068 ⫺0.0022 0.0033 0.4979
Height (1 cm increment) 0.0023 0.0010 0.0003 0.0045
Weight (1 kg increment) 0.0015 0.0006 0.0001 ⬍0.0001
Institution ⫺0.0492 ⫺0.0236 0.0038 ⬍0.0001
Preoperative AT activity (U/mL) 0.4870 0.4698 0.0174 ⬍0.0001
CPB duration (10 min increment) ⫺0.006 ⫺0.004 0.001 ⬍0.0001
Lowest venous temperature during CPB 0.0030 0.0044 0.0011 ⬍0.0001
(1°C increment)
Intraoperative packed red blood cell ⫺0.0095 ⫺0.0085 0.0021 ⬍0.0001
transfusion (1 unit increment)
Intraoperative cryoprecipitate transfusion 0.0497 0.0491 0.0150 0.0011
(1 pooled unit increment)
Post-CPB hemoglobin (1 g/dL increment) 0.0177 0.0184 0.0018 ⬍0.0001
a
198 subjects were missing one or more of the model’s predictor variables and are not included in this analysis.
CPB ⫽ cardiopulmonary bypass.

independent of recent heparin use, perhaps indicating a
dose effect of heparin administration upon decreased AT
activity. In the 1205 patients who had complete data for all
variables in the model, decreased post-CPB AT activity was
independently predicted by lower preoperative AT activity
and clinical variables that indicate greater hemodilution,
such as lower body weight and height and increased
transfusion incidence or a prolonged procedure (Table 4).
A clinical model predicting MACE was developed
(Table 5; adjusted r2 ⫽ 0.156) for 1403 patients who had
complete data for all variables in the clinical model. Vari-
ables that have been associated with MACE in prior
studies, notably recent MI, longer perfusion time, a require-
ment for intraaortic counterpulsation, and red blood cell
transfusion were also associated with MACE in this study.
AT activities at baseline, post-CPB, PODs 1–5, and the
change from baseline at these time points for each patient
were added to the clinical model, one time point at a time.
Preoperative, post-CPB, and POD 1 AT activity were not
independently predictive of MACE, whereas AT activity on
PODs 2 and 3 was independently predictive of MACE. MACE
was independently associated with change in AT activity
from baseline on PODs 2–5 (Table 5). The receiver operat-
ing characteristic of the model was significantly improved
by the addition of AT activity on PODs 2– 4 to the model
(Fig. 1).
DISCUSSION
This prospective, observational, cohort study confirms earlier
findings that preoperative AT activity is reduced in patients
with recent exposure to heparin or recent MI.5,6 Furthermore,
AT activity is reduced after CPB, likely due to consumption
by heparin administration6 and dilution. AT activity remained
significantly reduced from baseline levels over the postoperative
period, recovering to baseline levels within 5 days, on average.
In this population of patients, MACE was associated
with factors that have been previously associated with MI
or mortality including recent MI, peripheral vascular dis-
ease, longer perfusion time, use of intraaortic counterpul-
sation, and red blood cell transfusion.7–10 The clinical

866 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

 Table 5. Multivariable Predictors of Major Adverse Cardiac Events (MACE)
Without AT activity information in model
(n ⴝ 1403; r2 ⴝ 0.156)
95% confidence
Predictor Odds ratio interval P
Age
⬍55 yr 1.00 —
55–64 yr 1.26 0.72–2.21
65–74 yr 1.26 0.70–2.28
75–84 tears 1.66 0.86–3.20
ⱖ85 yr 0.70 0.11–4.36 0.58
Gender (female) 1.13 0.72–1.78 0.60
Race (Caucasian) 0.59 0.36–0.95 0.03
Body mass index
⬍20 0.67 0.15–3.04
20–24.9 0.52 0.28–0.96
25–34.9 0.48 0.30–0.77
ⱖ35 1.00 — 0.03
Institution 0.73 0.44–1.21 0.22
Myocardial infarction ⬍2 wk prior 1.70 1.11–2.59 0.014
Prior peripheral vascular procedure 3.38 1.53–7.46 0.003
Perfusion time (10 min increment) 1.09 1.05–1.14 ⬍0.0001
Concurrent other cardiac procedure 1.47 0.82–2.67 0.20
Preoperative or intraoperative IABP 4.02 2.18–7.42 ⬍0.0001
RBC transfusion during hospital stay (per unit) 1.10 1.04–1.16 0.0004

Additional Predictive Value of AT Activity at Each Time Point Added to the Clinical Model
AT activity (0.1 IU/mL Odds 95% Confidence P value of AT P value of improvement
decrease) Ratio interval variable in overall model
Preoperative AT activity 0.98 0.85–1.13 0.7820 0.7822
Postoperative AT activity 1.10 0.91–1.31 0.3193 0.3179
POD 1 AT activity 1.13 0.96–1.33 0.1565 0.1543
POD 2 AT activity 1.26 1.07–1.48 0.0056 0.0053
POD 3 AT activity 1.19 1.01–1.41 0.0412 0.0399
POD 4 AT activity 1.17 1.00–1.37 0.0506 0.0493
POD 5 AT activity 1.17 1.00–1.37 0.0553 0.0546
Change in AT activity Post-CPBa 1.15 0.93–1.41 0.1888 0.4058
Change in AT activity Day 1a 1.20 1.00–1.45 0.0558 0.1394
Change in AT activity Day 2a 1.36 1.13–1.62 0.0009 0.0035
Change in AT activity Day 3a 1.26 1.04–1.52 0.0163 0.0525
Change in AT activity Day 4a 1.26 1.06–1.51 0.0104 0.0294
Change in AT activity Day 5a 1.20 1.01–1.44 0.0407 0.1201
a
Refers to change in AT activity from the baseline level.
AT ⫽ antithrombin; RBC ⫽ red blood cell; POD ⫽ postoperative day.

model of MACE generated from this cohort had modest
predictive value, similar to prior clinical models.9 The
addition of AT activity on POD 2 and beyond to the model
modestly improved model performance. However, the
clinical value of AT activity as a predictor of MACE is
limited by the majority of adverse events that comprised
MACE occurring before POD 2. Specifically, the majority of
MACE events were MI, which was typically manifested as
the peak cTnI level occurring on POD 1. Rather, it may be
that lower postoperative AT activity may be a consequence
of more extensive surgery and other factors that are asso-
ciated with increased incidence of MACE, although such
assertion cannot be proven in this observational cohort. We
cannot exclude the possibility that lower AT levels may
have enhanced postoperative MACE.
Increased risk of adverse cardiovascular events has been
associated with lower levels of AT in other critically ill
populations, such as patients with severe sepsis.11–13 These
observations have prompted clinical trials of supplemental
AT administration.14 –17 Although several trials have
shown survival benefit, a meta-analysis of 20 trials encom-
passing 3458 patients failed to show a survival benefit of
administration of AT to critically ill patients.18
There are limited data regarding AT activity and ad-
verse outcomes after cardiac surgery. The majority of
studies describe use of AT for “heparin resistance” during
CPB and lack postoperative clinical outcome data.19 –25 A
single well-conducted observational study of 647 patients
evaluated the association between preoperative and imme-
diate postoperative AT levels and outcomes in cardiac
surgery patients. Low levels of AT activity upon ICU
arrival were associated with prolonged ICU stay, higher
rate of reexploration for bleeding, thromboembolism, and
adverse neurologic sequelae.5 Our study examined a longer
time period that encompassed the period of MACE occur-
rence and the recovery of AT levels, thus providing addi-
tional insights. Importantly, we replicated the finding that
AT level after CPB is associated with clinical factors indica-
tive of longer, more extensive procedures, perhaps with
more hemodilution.

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 867

Antithrombin and Adverse Events
Figure 1. Receiver operating characteristics of the addition of
antithrombin (AT) activity on postoperative days (PODs) 2– 4 to the
clinical model.
Although our study contributes to our understanding of
the role AT plays in the perioperative period, we are
concerned by the absence of a temporal correlation between
lower AT activity and MACE in our cohort, likely indicat-
ing that AT level is not an important determinant of
thrombotic complications such as MI, stroke, and graft
occlusion in the cardiac surgical setting. Thus, our obser-
vational cohort cannot directly address any putative bio-
logical mechanism for the role of AT in MACE generation.
CONCLUSION
In a large cohort of patients undergoing CABG surgery,
postoperative AT activity was independently associated with
MACE. Because this occurs at time points predominantly after
the MACE event, the clinical utility of AT as a biomarker of
risk remains unknown.
DISCLOSURE
There are four potential conflicts of interest. Dr. Garvin was the
recipient of a Research Fellowship funded by Talecris Biothera-
peutics, which allowed time for generation of this and other
articles. Dr. Chen is an employee of Talecris Biotherapeutics
and performed the majority of the analyses. To prevent the
appearance or substance of conflict, Simon Body personally
directed the conduct of all analyses and confirmed the conduct
of the analyses by reviewing the code and output of all
analyses. In addition, Simon Body personally reran the impor-
tant components of the analyses to confirm the findings and
can attest that there was no potential for conflict in the analysis
phase. Dr. Body received a total of ⬍$5000 to allow his time for
travel to Talecris in North Carolina to coordinate the analyses.
Talecris also paid for the costs of antithrombin analyses
performed by Charles River Laboratories in Ontario, Canada,
but had no opportunity to intervene in these analyses. In brief,
868 www.anesthesia-analgesia.org
we believe there is no conflict of interest in the conduct of this
study.
REFERENCES
1. Maclean PS, Tait RC. Hereditary and acquired antithrombin
deficiency: epidemiology, pathogenesis and treatment options.
Drugs 2007;67:1429 – 40
2. Levy JH, Despotis GJ, Szlam F, Olson P, Meeker D, Weisinger A.
Recombinant human transgenic antithrombin in cardiac surgery:
a dose-finding study. Anesthesiology 2002;96:1095–102
3. Sanson BJ, Simioni P, Tormene D, Moia M, Friederich PW,
Huisman MV, Prandoni P, Bura A, Rejto L, Wells P, Mannucci
PM, Girolami A, Buller HR, Prins MH. The incidence of venous
thromboembolism in asymptomatic carriers of a deficiency of
antithrombin, protein C, or protein S: a prospective cohort
study. Blood 1999;94:3702– 6
4. Simioni P, Sanson BJ, Prandoni P, Tormene D, Friederich PW,
Girolami B, Gavasso S, Huisman MV, Buller HR, Wouter ten
Cate J, Girolami A, Prins MH. Incidence of venous thrombo-
embolism in families with inherited thrombophilia. Thromb
Haemost 1999;81:198 –202
5. Ranucci M, Frigiola A, Menicanti L, Ditta A, Boncilli A, Brozzi
S. Postoperative antithrombin levels and outcome in cardiac
operations. Crit Care Med 2005;33:355– 60
6. Ranucci M, Ditta A, Boncilli A, Cotza M, Carboni G, Brozzi S,
Bonifazi C, Tiezzi A. Determinants of antithrombin consump-
tion in cardiac operations requiring cardiopulmonary bypass.
Perfusion 2004;19:47–52
7. Bradshaw PJ, Jamrozik K, Le M, Gilfillan I, Thompson PL.
Mortality and recurrent cardiac events after coronary artery
bypass graft: long term outcomes in a population study. Heart
2002;88:488 –94
8. Charlesworth DC, Likosky DS, Marrin CA, Maloney CT,
Quinton HB, Morton JR, Leavitt BJ, Clough RA, O’Connor GT.
Development and validation of a prediction model for strokes
after coronary artery bypass grafting. Ann Thorac Surg
2003;76:436 – 43
9. Koch CG, Weng YS, Zhou SX, Savino JS, Mathew JP, Hsu PH,
Saidman LJ, Mangano DT. Prevalence of risk factors, and not
gender per se, determines short- and long-term survival after
coronary artery bypass surgery. J Cardiothorac Vasc Anesth
2003;17:585–93
10. van Brussel BL, Plokker HW, Voors AA, Ernst JM, Ernst NM,
Knaepen PJ, Koomen EM, Tijssen JG, Vermeulen FE. Multivar-
iate risk factor analysis of clinical outcome 15 years after
venous coronary artery bypass graft surgery. Eur Heart J 1995;
16:1200 – 6
11. Martinez MA, Pena JM, Fernandez A, Jimenez M, Juarez S,
Madero R, Vazquez JJ. Time course and prognostic significance
of hemostatic changes in sepsis: relation to tumor necrosis
factor-alpha. Crit Care Med 1999;27:1303– 8
12. Sakr Y, Reinhart K, Hagel S, Kientopf M, Brunkhorst F.
Antithrombin levels, morbidity, and mortality in a surgical
intensive care unit. Anesth Analg 2007;105:715–23
13. Wilson RF, Mammen EF, Tyburski JG, Warsow KM, Kubinec
SM. Antithrombin levels related to infections and outcome.
J Trauma 1996;40:384 –7
14. Baudo F, Caimi TM, de Cataldo F, Ravizza A, Arlati S, Casella
G, Carugo D, Palareti G, Legnani C, Ridolfi L, Rossi R,
D’Angelo A, Crippa L, Giudici D, Gallioli G, Wolfler A, Calori
G. Antithrombin III (ATIII) replacement therapy in patients
with sepsis and/or postsurgical complications: a controlled
double-blind, randomized, multicenter study. Intensive Care
Med 1998;24:336 – 42
15. du Cheyron D, Bouchet B, Bruel C, Daubin C, Ramakers M,
Charbonneau P. Antithrombin supplementation for anticoagu-
lation during continuous hemofiltration in critically ill patients
with septic shock: a case-control study. Crit Care 2006;10:R45
16. Eid A, Wiedermann CJ, Kinasewitz GT. Early administration of
high-dose antithrombin in severe sepsis: single center results
from the KyberSept-trial. Anesth Analg 2008;107:1633– 8
ANESTHESIA & ANALGESIA
17. Kienast J, Juers M, Wiedermann CJ, Hoffmann JN, Ostermann
H, Strauss R, Keinecke HO, Warren BL, Opal SM. Treatment
effects of high-dose antithrombin without concomitant heparin
in patients with severe sepsis with or without disseminated
intravascular coagulation. J Thromb Haemost 2006;4:90 –7
18. Afshari A, Wetterslev J, Brok J, Moller A. Antithrombin III in
critically ill patients: systematic review with meta-analysis and
trial sequential analysis. BMJ 2007;335:1248 –51
19. Avidan MS, Levy JH, van Aken H, Feneck RO, Latimer RD, Ott
E, Martin E, Birnbaum DE, Bonfiglio LJ, Kajdasz DK, Despotis
GJ. Recombinant human antithrombin III restores heparin
responsiveness and decreases activation of coagulation in
heparin-resistant patients during cardiopulmonary bypass.
J Thorac Cardiovasc Surg 2005;130:107–13
20. Conley JC, Plunkett PF. Antithrombin III in cardiac surgery: an
outcome study. J Extra Corpor Technol 1998;30:178 – 83
October 2010 • Volume 111 • Number 4
21. Despotis GJ, Levine V, Joist JH, Joiner-Maier D, Spitznagel E.
Antithrombin III during cardiac surgery: effect on response of
activated clotting time to heparin and relationship to markers
of hemostatic activation. Anesth Analg 1997;85:498 –506
22. Kanbak M. The treatment of heparin resistance with anti-
thrombin III in cardiac surgery. Can J Anaesth 1999;46:581–5
23. Lemmer JH Jr, Despotis GJ. Antithrombin III concentrate to
treat heparin resistance in patients undergoing cardiac sur-
gery. J Thorac Cardiovasc Surg 2002;123:213–7
24. Levy JH, Montes F, Szlam F, Hillyer CD. The in vitro effects of
antithrombin III on the activated coagulation time in patients
on heparin therapy. Anesth Analg 2000;90:1076 –9
25. Williams MR, D’Ambra AB, Beck JR, Spanier TB, Morales DL,
Helman DN, Oz MC. A randomized trial of antithrombin
concentrate for treatment of heparin resistance. Ann Thorac
Surg 2000;70:873–7
www.anesthesia-analgesia.org 869
␤2-Adrenergic Receptor-Coupled Phosphoinositide
3-Kinase Constrains cAMP-Dependent Increases in
Cardiac Inotropy Through Phosphodiesterase
4 Activation
Christopher J. Gregg,* Jochen Steppan, MD,* Daniel R. Gonzalez, PhD,† Hunter C. Champion, MD, PhD,‡
Alexander C. Phan,* Daniel Nyhan, MD,* Artin A. Shoukas, PhD,* Joshua M. Hare, MD,†
Lili A. Barouch, MD,* and Dan E. Berkowitz, MD*

BACKGROUND: Emerging evidence suggests that phosphoinositide 3-kinase (PI3K) may modu-
late cardiac inotropy; however, the underlying mechanism remains elusive. We hypothesized that
␤2-adrenergic receptor (AR)-coupled PI3K constrains increases in cardiac inotropy through cyclic
adenosine monophosphate (cAMP)-dependent phosphodiesterase (PDE) activation.
METHODS: We tested the effects of PI3K and PDE4 inhibition on myocardial contractility by using
isolated murine cardiac myocytes to study physiologic functions (sarcomere shortening [SS] and
intracellular Ca⫹ transients), as well as cAMP and PDE activity.
RESULTS: PI3K inhibition with the reversible inhibitor LY294002 (LY) resulted in a significant
increase in SS and Ca2⫹ handling, indicating enhanced contractility. This response depended on
Gi␣ protein activity, because incubation with pertussis toxin (an irreversible Gi␣ inhibitor)
abolished the LY-induced hypercontractility. In addition, PI3K inhibition had no greater effect on
SS than both a PDE3,4 inhibitor (milrinone) and LY combined. Furthermore, LY decreased PDE4
activity in a concentration-dependent manner (58.0% of PDE4 activity at LY concentrations of 10
␮M). Notably, PI3K␥ coimmunoprecipitated with PDE4D. The ␤2-AR inverse agonist, ICI 118,551
(ICI), abolished induced increases in contractility.
CONCLUSIONS: PI3K modulates myocardial contractility by a cAMP-dependent mechanism
through the regulation of the catalytic activity of PDE4. Furthermore, basal agonist-independent
activity of the ␤2-AR and its resultant cAMP production and enhancement of the catalytic activity
of PDE4 through PI3K represents an example of integrative cellular signaling, which controls
cAMP dynamics and thereby contractility in the cardiac myocyte. These results help to explain the
mechanism by which milrinone is able to increase myocardial contractility in the absence of
direct ␤-adrenergic stimulation and why it can further augment contractility in the presence of
maximal ␤-adrenergic stimulation. (Anesth Analg 2010;111:870 –7)

T he ␤-adrenergic receptor (␤-AR) isoforms character-

ized in the heart have distinct but overlapping signal
transduction mechanisms that regulate many aspects
of myocardial pump function. Augmentation of cardiac
inotropy through ␤-ARs is mediated through a well-
studied system that increases intracellular cyclic adenosine
monophosphate (cAMP) leading to increases in cAMP-
dependent protein kinase A activity.1 The ␤1-AR is coupled
to the heterotrimeric Gs protein and is considered the
primary means for catecholamine-induced increases in
cAMP concentrations and thus myocardial contractility.
The role of the ␤2-AR in modulation of contractility is more
From the *Johns Hopkins Medical Institutions (current affiliation: University
of California, San Diego), †Johns Hopkins Medical Institutions (current
affiliation: University of Miami), ‡Johns Hopkins Medical Institutions,
Baltimore, Maryland (current affiliation: University of Pittsburgh).
Accepted for publication June 1, 2010.
This work was supported in part by a grant from the National Space
Biomedical Research Institute (CA00405) through National Aeronautics &
Space Administration (AS) and a National Institutes of Health grant (R01
AG 021,523) (DEB).
Disclosure: The authors report no conflict of interest.
Address correspondence to Dan E. Berkowitz, MD, Johns Hopkins Medical
Institutions, 600 N Wolfe Street, CCM Tower 711, Baltimore, MD 21287.
Address e-mail to dberkowi@bme.jhu.edu.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181ee8312
complex, as this receptor is coupled to both Gs and Gi, the
latter of which is recognized to attenuate inotropy through
the cGMP dependent effects of nitric oxide produced from
endothelial nitric oxide synthase 3. ␤2-AR signaling is likely
compartmentalized in light of the dual and seemingly
opposing effects of these signaling cascades.2 Those cas-
cades are critically dependent on cAMP concentrations and
spatial localization.3 Thus, changes in the balance between
cAMP production through adenylyl cyclase activation and
cAMP breakdown through cAMP-dependent phosphodies-
terases (PDE) will modulate protein kinase A activity and
myocardial contractility.
The roles of phosphoinositide 3-kinase (PI3K), a fam-
ily of lipid kinases whose downstream targets include
bioactive lipids and proteins, in signaling has emerged
over the past few years.4 Specifically, it has been dem-
onstrated that the ␤-AR kinase–1 (␤-ARK1) and PI3K
interact and that ␤-ARK1 recruitment of PI3K is crucial
for mediating ␤2-AR internalization as a component of
the receptor/internalization scaffold complex.1 In addi-
tion to its role in receptor internalization, the direct role
of PI3K in modulating myocardial contractility has been
investigated in an elegant study observing the role of
PI3K in progression of myocardial hypertrophy. Specifi-
cally, mice deficient in the catalytic subunit of the PI3K␥

870 www.anesthesia-analgesia.org October 2010 • Volume 111 • Number 4

Figure 1. Phosphoinositide 3-kinase constrains contractility in cardiac myocytes. (A,B) Effects of the reversible phosphoinositide 3-kinase
(PI3K) inhibitor LY (10 ␮M) on sarcomere shortening (SS) (A, n ⫽ 14, *** P ⬍ 0.001 vs baseline) and systolic [Ca2⫹]i (B, n ⫽ 12, *** P ⬍
0.001 vs baseline). Washout confirms the reversibility for both SS (n ⫽ 9, n.s. vs baseline) and [Ca2⫹]i (n ⫽ 8, n.s. vs baseline). Sample
transients are shown above the graphic panels. (C) LY (0.1 to 100 ␮M) dose response showing a concentration-dependent increase in SS with
an EC50 of 2.2 ␮M. (D) Effects of wortmannin (5 nM) on SS (n ⫽ 4, ** P ⬍ 0.01 vs baseline). (E) Monoexponential time constant, ␶ (index
of lusitropy), fit to diastolic portion of averaged Ca2⫹ fluorescence transients (n ⫽ 15, * P ⬍ 0.05 vs baseline fit) and washout.

isoform (p110␥⫺/⫺) demonstrate increased basal myo-
cardial contractility in a cAMP-dependent manner, and
inhibition of PI3K augments isoproterenol-induced con-
tractile responses.5 The cellular mechanism of this effect,
however, remains incompletely understood.
Thus, we hypothesized that PI3K is a negative upstream
regulator of myocardial contractility by modulating cAMP-
dependent PDE activity.
METHODS
Animals
Animal treatment and care was approved and provided in
accordance with the institutional animal care and use
committee of the Johns Hopkins University School of
Medicine, as well as with the National Institutes of Health
and American Physiological Society guidelines. Three
5-month-old C57/BL6J mice were purchased from Jackson
Laboratories (Bar Harbor, ME).
Cardiac Myocyte Isolation
Myocytes were isolated by enzymatic digestion with colla-
genase type 2 (1 g/L; Worthington Biochemical, Lakewood,
NJ) and protease type XIV (0.1 g/L), as previously de-
scribed.6,7 Cell suspension was obtained by mechanically
disrupting digested ventricles, filtering, centrifugation, and

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 871

PI3K Constrains Cardiac Contractility

resuspension in Tyrode solution (1.0 mM Ca2⫹). The myo-

cyte cell suspension was incubated with 5 ␮mol/L fura-
2/am (Molecular Probes, Eugene, OR), then transferred to
an inverted microscope (TE 200; Nikon), continuously
superfused with Tyrode solution and stimulated at 1 Hz.
Change in average sarcomere length was determined by
fast Fourier transform of the Z-line density trace to the
frequency domain. Calcium concentrations were measured
using a dual-excitation spectrofluorometer (IonOptix, Mil-
ton, MA), as previously described.6,7 The experiments were
conducted in the presence or absence of the reversible PI3K
inhibitor LY294002 (LY, 10 ␮M), the irreversible PI3K
inhibitor wortmanin (5 nM), the Gi␣ protein inhibitor
pertussis toxin (PTX, 1.5 ␮g/mL), the ␤1-and ␤2-
adrenoreceptor agonist isoproterenol (50 nM), the PDE3,4
inhibitor milrinone (0.3, 3, 30, 300 ␮M), the PDE4 inhibitor
rolipram (1, 10, 100 ␮M), and ICI 118,551 (ICI, 100 nM).

Phosphodiesterase Activity Analysis

Total low Km cAMP-dependent PDE activity was assayed
by fluorescence polarization (Molecular Devices, Sunny-
vale, CA) and read on a microplate reader (Perkin Elmer,
Wellesley, MA), as described by the manufacturer in the
presence or absence of LY (1, 3, 10 ␮mol/L), isoproterenol
(100 nmol/L), milrinone (30 ␮mol/L), and ICI (100
nmol/L).
Cyclic Nucleotide Assay
Myocytes were used in a cAMP enzyme immunoassay
(Amersham Pharmacia Biotech, Piscataway, NJ), as de-
scribed by the manufacturer, in the presence or absence of
isoproterenol (50 nM), LY (10 ␮M), milrinone (Mil) (30 ␮M),
and ICI (100 nM).
Figure 2. Enhancement of myocardial contractility by PI3K inhibition
is sensitive to Gi activity. Effects of the irreversible Gi␣ inhibitor
pertussis toxin (PTX) on basal and phosphoinositide-3 kinase (PI3K)
inhibited myocardial contractility. LY significantly increased contrac-
tility in non-PTX treated myocytes (n ⫽ 6, ** P ⬍ 0.01 vs baseline).
In PTX-treated myocytes, LY response was significantly blunted,
compared with LY only group (n ⫽ 6, ** P ⬍ 0.01, LY vs LY ⫹ PTX)
and was not significantly different than baseline (n ⫽ 6, P ⬎ 0.05 vs
baseline).
value ⬍0.05 after t tests for pairs or nonparametric ranking
and one-way ANOVA.

Coimmunoprecipitation
PI3K (p110␥) or control (bovine serum albumin) antibodies
(both: Cell Signaling, Danvers, MA) were cross-linked to
protein A/G beads (Pierce Biotechnology, Rockford, IL)
and used to immunoprecipitate proteins from mouse heart
cell lysates. Western blots were then performed with a
PDE4D antibody.
Western Blots
The entire coimmunoprecipitation was denatured, re-
duced (150 mmol/L dithiothreitol), and separated by
Tris-Glycine/sodium dodecyl sulfate-polyacrylamide gel
electrophoresis (Invitrogen, Carlsbad, CA). The proteins
were transferred to polyvinylidene fluoride membranes
(Billerica, MA) with Towbin transfer buffer, and the
membranes were blocked with 5% nonfat dry milk/tris
buffered saline (NFDM/TBS). Primary antibodies (PI3-
kinase, Cell Signaling; PDE4D, Fabgennix, Frisco, TX)
were diluted in 5% NFDM/TBS. The immunoblots were
washed and incubated with the appropriate secondary
antibodies (Santa Cruz Biotechnology, Santa Cruz CA).
The membranes were washed again and developed with
SuperSignal Pico and Femto substrate (Pierce Biotechnol-
ogy, Rockford, IL).
RESULTS
PI3K Constrains Contractility in Isolated
Cardiac Myocytes
Figure 1 illustrates that the dose-dependent inhibition of
PI3K results in an increased sarcomere shortening (SS) and
systolic [Ca2⫹]i. In detail, inhibition of PI3K with the
specific reversible inhibitor LY (10 ␮M) resulted in a 2.21 ⫾
0.12-fold increase in SS (Fig. 1A) and a 1.68 ⫾ 0.10-fold
increase in systolic [Ca2⫹]i (Fig. 1B). This increase in
contractility was reversible and concentration dependent
(Fig. 1C). Similar results were observed with the irrevers-
ible PI3K inhibitor wortmannin, which produced a 2.31 ⫾
0.04-fold increase in SS (Fig. 1D). Furthermore, the time
constant of Ca2⫹ fluorescence decay decreased from 0.16 ⫾
0.01 to 0.11 ⫾ 0.01 s upon PI3K inhibition (Fig. 1E).
PI3K Is Regulated Through Gi␣ Signaling
Figure 2 shows the relation between PI3K and Gi␣ signal-
ing. The enhanced contractility mediated by PI3K inhibi-
tion was abolished by preincubation with the Gi␣ inhibitor,
PTX. LY increased SS 2.52 ⫾ 0.75-fold, while in the presence
of PTX, SS did not change significantly (Fig. 2).

Data Analysis and Statistical Procedures PI3K Inhibition Enhances ␤-AR-Stimulated

Data are presented as mean ⫾ SEM, with the exception of Contractility in Myocytes
physiologic data that are presented as change from base- Figure 3 demonstrates the additive effect of PI3K inhibition
line ⫾ SEM. Statistical significance was considered for P and ␤-AR stimulation. The combination of PI3K inhibition

872 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

Figure 3. Synergistic effect of isoproterenol and LY on myocyte contractility. Cumulative data showing significant increases in sarcomere
shortening (SS) (A) with adrenergic stimulation alone (isoproterenol: n ⫽ 9, * P ⬍ 0.05 vs baseline) and the addition of phosphoinositide-3
kinase (PI3K) inhibition (isoproterenol ⫹ LY: n ⫽ 9, ** P ⬍ 0.01 vs baseline, # P ⬍ 0.01 isoproterenol versus isoproterenol ⫹ LY). Systolic
[Ca2⫹]i (B) also showed significant enhancement with adrenergic stimulation alone (isoproterenol: n ⫽ 9, * P ⬍ 0.05 vs baseline) and the
addition of PI3K inhibition (isoproterenol ⫹ LY: n ⫽ 9, ** P ⬍ 0.01 vs baseline, # P ⬍ 0.01 isoproterenol versus isoproterenol ⫹ LY). Sample
transients from these experiments are shown above the graphic panels.

decreased PDE4 activity in a concentration-dependent

Figure 4. PI3K inhibition increases myocyte cAMP production and
augments adrenergically stimulated cAMP. Effect of isoproterenol
(50 nM), LY (10 ␮M) and their combination on myocyte cAMP
levels (fmol/␮g protein, n ⫽ 6, * P ⬍ 0.05). Both LY and
isoproterenol increased cAMP levels independently, however, the
combination of isoproterenol and LY increased cAMP levels above
levels seen with isoproterenol and LY alone, (n ⫽ 8, *** P ⬍
0.001, baseline versus isoproterenol ⫹ LY, # P ⬍ 0.01 isopro-
terenol versus isoproterenol ⫹ LY).
with LY and ␤-AR stimulation with isoproterenol aug-
mented SS 1.97 ⫾ 0.27-fold and systolic [Ca2⫹]i 1.50 ⫾
0.13-fold over isoproterenol alone (Fig. 3).
PI3K Inhibition Increases Myocyte cAMP Levels
As shown in Figure 4, increasing cAMP levels are linked to
PI3K inhibition and ␤-AR stimulation. Isoproterenol in-
creased cAMP levels from 5.20 ⫾ 1.20 to 18.48 ⫾ 2.21
fmol/␮g protein, and LY increased cAMP levels to 24.32 ⫾
2.73 fmol/␮g protein. The combination of LY and isopro-
terenol had a synergistic effect and augmented cAMP levels
further to 37.44 ⫾ 3.78 fmol/␮g protein (Fig. 4).
PI3K Is Coupled to cAMP-Dependent
PDE Activity
Figure 5 and 6 demonstrate the mechanism by which PI3K
regulates contractility in a cAMP-dependent manner. LY
manner, which paralleled the results from the PDE4 spe-
cific inhibitor CB524717 (Fig. 5A). The PDE3,4 inhibitor Mil
(30 ␮M) and LY increased cAMP levels from 4.70 ⫾ 0.77 to
14.59 ⫾ 2.12 and 19.85 ⫾ 3.42 fmol/␮g protein, respectively
(Fig. 5B). There was no difference when incubated with LY
and Mil alone or a combination of both (18.28 ⫾ 4.31
fmol/␮g protein).
Mil and the PDE4 specific inhibitor rolipram resulted in
a concentration-dependent increase in SS and systolic
[Ca2⫹]i (Fig. 6A, B). At EC50 concentrations, Mil (10 ␮M)
produced a 1.35 ⫾ 0.08-fold increase in SS, and LY pro-
duced a 2.27 ⫾ 0.25-fold increase. The combination of Mil
and LY yielded no synergistic effect (2.25 ⫾ 0.18-fold
increase), while the addition of isoproterenol (100 nM)
yielded an additional enhancement in contractility (3.41 ⫾
0.32-fold change in SS, Fig. 6C).
Agonist Independent ␤2-AR Coupling Through
PI3K Maintains Contractile Tone
The results depicted in Figure 7 illustrate that decreasing
basal agonist-independent cAMP production (with ICI), the
catalytic activity of PDE isoforms (regulated by PI3K) cease
to have an effect on cardiac contractility. Myocytes incu-
bated with the ␤2-AR inverse agonist ICI demonstrated a
time and concentration-dependent decrease in SS (data not
shown) and exhibited a decrease in SS from baseline (Fig.
7A). Furthermore, ICI completely reversed LY and Mil-
induced contractile responses, cAMP levels, and systolic
[Ca2⫹]i (Fig. 7, B–D).
Interaction of PDE4 and PI3K␥
PDE4D was detected in HeLA and U937 cells (control) and
in homogenates immunoprecipitated with PI3K␥ (p110␥)
but not in homogenates immunoprecipitated with bovine
serum albumin. As expected, PI3K␥ (p110␥) was detected

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 873

PI3K Constrains Cardiac Contractility

Figure 5. PI3K inhibition decreases the activity of cAMP-dependent PDE4 to increase myocardial cAMP levels. Concentration-dependent
depression of LY (A) and CB524717 (panel inlay) on phosphodiesterase (PDE) 4 activity (** P ⬍ 0.01, n ⫽ 5 vs baseline activity). (B) Effect
of LY (10␮61,517), milrinone (30 ␮M) alone, and in combination on myocyte cAMP levels (fmol/␮g protein). LY and milrinone increased
myocyte cAMP levels above baseline (* P ⬍ 0.05, n ⫽ 4 vs baseline). There is no significant increase in cAMP levels if the drugs are combined
(n ⫽ 4, p ⫽ n.s. LY versus milrinone, milrinone versus LY ⫹ milrinone, LY versus LY ⫹ milrinone).

Figure 6. PI3K regulates myocardial contractility by cAMP-dependent PDE activation. Concentration-dependent enhancement of myocyte
sarcomere shortening (SS) by milrinone (A, n ⫽ 4) and rolipram (B, n ⫽ 10). (C) Effect of milrinone (30 ␮M), LY (5 ␮M), and isoproterenol (10
nM) alone and in combination on myocyte SS and systolic [Ca2⫹]i. Milrinone and LY alone produced significant increases in SS and systolic
[Ca2⫹]i. However, administration of a combination of both LY and milrinone produced no added effect over LY alone (n ⫽ 13, p ⫽ n.s. LY versus
LY, milrinone). Addition of isoproterenol to milrinone and LY produced a further increase in SS (n ⫽ 13, ** P ⬍ 0.01, milrinone ⫹ LY ⫹
isoproterenol versus milrinone ⫹ LY).

in lysates immunoprecipitated with the same antibodies
(Fig. 8).
DISCUSSION
In the present study, we have demonstrated that PI3K
negatively modulates myocardial contractility by a Gi␣-
dependent mechanism. Furthermore, PI3K negatively regu-
lates adrenergic stimulation of myocardial contractility
since inhibition significantly enhances ␤-AR–mediated con-
tractility. ␤2-AR– coupled PI3K decreases cAMP by a
mechanism that involves activation of PDE4, as PI3K
inhibition decreases PDE4 activity and increases cAMP
levels. Finally, this mechanism is constitutively activated
given that the inverse ␤2-AR agonist ICI completely re-
verses both PI3K and PDE-mediated increases in contrac-
tility and cAMP levels (Fig. 9).
We demonstrated that inhibition of PI3K enhances ad-
renergic stimulation of cardiac myocytes, which is consis-
tent with the findings of Jo et al.8 and Crackower et al.5
However, it does differ significantly from Jo et al.8 with
regard to the effect of LY on basal contractility. While they
demonstrated no effect of PI3K inhibition with LY on basal

874 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

Figure 7. Agonist independent coupling of the ␤2-AR to cAMP production and the role of PI3K. (A) Depression of myocyte sarcomere shortening
(SS) by the ␤2-AR inverse agonist ICI 118,551 (100 nM) (n ⫽ 6, * P ⬍ 0.05 vs baseline). (B) ICI abolishes LY and milrinone-induced increases
in SS (n ⫽ 10, ** P ⬍ 0.01 LY or milrinone versus baseline; p ⫽ n.s. LY or milrinone ⫹ ICI versus baseline) and systolic [Ca2⫹]i (D, n ⫽ 11,
* P ⬍ 0.05 LY versus baseline; ** P ⬍ 0.01 milrinone versus baseline; p ⫽ n.s. LY or milrinone ⫹ ICI versus baseline). (C) Both LY (n ⫽ 6,
*** P ⬍ 0.001 vs baseline) and milrinone (n ⫽ 6, ** P ⬍ 0.01 vs baseline) increased cAMP levels that were completely abolished by ICI
coincubation. Sample transients are shown above their respective graphic panels.

Figure 8. Molecular interaction between PI3K and PDE4D. Phospho-

diesterase (PDE) 4D was detected in lysates from HeLa and U937
cells (control). In mouse heart homogenates immunoprecipitated
with phosphoinositide 3-kinase (PI3K) (p110␥), PDE4D was detected
but not in homogenates immunoprecipitated with A/G-linked bovine
serum albumin Ab.
cellular contractility,8 we demonstrated a concentration-
dependent increase in both SS and systolic [Ca2⫹]i. This is
consistent with the findings of Crackower et al.5 who
demonstrated significantly enhanced basal contractility in
isolated myocytes from mice in which the catalytic subunit
of the ␥ isoform of PI3K (p110␥) has been knocked out. We
agree with Crackower et al.5 that myocyte cAMP levels are
modulated by the activity of PI3K. This is in contrast to the
conclusion made by Jo et al. 8 that the ␤2-AR–mediated
enhancement in contractility is not mediated by an increase
in cAMP, because they demonstrate no difference in myo-
cyte cAMP levels in response to a ␤2-AR selective agonist in
the presence or absence of LY. However, these experiments
Figure 9. Schematic of the proposed PI3K interaction with the ␤-AR.
␤2-AR ⫽ beta-2 adrenoreceptor; AC ⫽ adenylate cyclase; Gs ⫽
G-protein S; Gi ⫽ G-protein i; PI3K ⫽ phosphoinositide 3-kinase;
PDE ⫽ phosphodiesterase; cAMP ⫽ cyclic adenosine monophos-
phate; PKA ⫽ protein kinase A.
were performed in the presence of the nonspecific PDE
inhibitor isobutylmethylxanthine. This difference in the
assays performed could explain their different conclusion,
while indirectly supporting our hypothesis that inhibition
of PI3K enhances contractility through a pathway involving
PDE inhibition.
We show that PI3K is coupled to cAMP-dependent
PDE4 activity, which is in contrast to Patrucco et al.9 who
demonstrated that mice expressing a PI3Kg kinase dead
mutant (PI3KgKD) have no overt alteration in myocardial

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 875

PI3K Constrains Cardiac Contractility
contractility. They interpreted their results to indicate that
the kinase domain of the PI3K was not necessary for
modulating PDE and constraining cAMP-dependent alter-
ations in myocardial contractility. Therefore, Patrucco et
al.9 suggested that protein-protein interaction between
PI3Kg and the PDE3B isoform is sufficient for PDE3B
activation. Our data, on the other hand, suggest that the
PI3K catalytic domain may indeed be important in activa-
tion of a downstream PDE, in this case PDE4D. LY294002,
a specific reversible inhibitor of PI3K, does so by competing
with adenosine triphosphate for the active site of the
catalytic subunit for PI3K␥, p110␥.10 Our results with
LY294002 and wortmannin support the notion that kinase
activity is critically important for the modulation of cAMP-
dependent inotropic changes. We cannot refute the idea
that LY294002, in binding and inhibiting the kinase activity
of PI3K, may, in addition, alter the binding of a PDE to
p110␥. An alternative explanation is that PI3K may act
through an intermediate to result in regulation of a PDE.
This is hypothetically supported by the observation that
PDE has regulatory phosphorylation sites for Akt, the
kinase downstream of PI3K.11,12 Thus, PI3K may not only
act as a scaffold for PDE9 but may also be phosphorylated
by the enzyme or its downstream effector Akt.
PDE3 and PDE4 are the major isoforms that degrade
cAMP in cardiac myocytes.13 Recent data suggest that these
isoforms localize to distinct compartments in the cell and
have different functional roles.3,13–15 Specifically, PDE4
appears to be functionally coupled to the pool of cAMP
stimulated by ␤-ARs as a result of spatial confinement by a
localization signal, while PDE3 appears to regulate the
general pool of cAMP stimulated by forskolin.13 This is
consistent with the observations of Xiang et al.16 who
demonstrated that PDE4 rather than PDE3 is required for
␤2-AR signaling in neonatal mouse myocytes. They demon-
strated an enhanced response to isoproterenol in the presence
of selective PDE4 inhibitors and not in cells treated with a
selective PDE3 inhibitor. Furthermore, PDE4 knockout mice
have augmented responses to isoproterenol, while myocytes
from ␤2-AR knockout mice have no augmented response to
PDE4 inhibition.16 These findings are consistent with our
findings and support the idea that PDE4 is essential for ␤2-AR
signaling. Moreover, we have extended the observation that
this phenomenon is dependent on PI3K.
There are several shortcomings of the present study that
need to be considered. First, all physiologic experiments
were done in a rodent model of isolated myocyte and not
verified in vivo or in human tissues. There are considerable
differences between species, and it is not possible to
generalize all results and translate them to a different
species. Also, there could be a direct biochemical effect of
PI3K inhibitors on PDE activity. None, however, have yet
been described in the literature. Moreover, even though we
showed that PI3K acts through PDE4, the cAMP-dependent
PDE isoform remains controversial.
Milrinone, a well-known PDE 3,4 inhibitor, is frequently
used to treat advanced heart failure, especially in the context
of cardiac surgery. Our novel finding that inhibition of PI3K
through Gi␣ enhances myocardial contractility by a PDE4-
dependent mechanism that is independent of direct ␤-AR
agonist activation further demonstrates the exquisite balance
876 www.anesthesia-analgesia.org
of cAMP production and cAMP breakdown in the cardiac
myocyte and sheds a light on Milrinone’s mechanism of
action (Fig. 9). In another way, cAMP is a convergence point
of two opposing signaling pathways. More specifically, cAMP
levels are the integrated output of two temporally and spa-
tially sensitive signaling inputs that serve to regulate myocar-
dial inotropic responses. This signaling convergence of cAMP
production and breakdown at cAMP levels indicates tighter
regulation of cAMP availability than has been previously
identified. This could potentially prevent a destructive, run-
away cAMP event on the second timescale or could control
the gain of the ␤2-AR over the lifetime of the organism by
relatively regulating transcription of both Gs and Gi proteins.
Furthermore, by indirectly inhibiting the breakdown of
cAMP, Milrinone is able to increase myocardial contrac-
tility in the absence of direct ␤-adrenergic stimulation and
further augments contractility even in the presence of
maximal ␤-adrenergic stimulation, as seen clinically. Thus,
an understanding of the importance of this physiologic
pathway in the regulation of myocyte contractility is im-
portant in defining the molecular mechanism underlying
Milrinone’s action as a critical inotrope in our limited
armamentarium.
REFERENCES
1. Rockman HA, Koch WJ, Lefkowitz RJ. Seven-transmembrane-
spanning receptors and heart function. Nature 2002;415:206 –12
2. Devic E, Xiang Y, Gould D, Kobilka B. Beta-adrenergic receptor
subtype-specific signaling in cardiac myocytes from beta(1)
and beta(2) adrenoceptor knockout mice. Mol Pharmacol
2001;60:577– 83
3. Baillie GS, Houslay MD. Arrestin times for compartmentalised
cAMP signalling and phosphodiesterase-4 enzymes. Curr
Opin Cell Biol 2005;17:129 –34
4. Oudit GY, Sun H, Kerfant BG, Crackower MA, Penninger JM,
Backx PH. The role of phosphoinositide-3 kinase and PTEN in
cardiovascular physiology and disease. J Mol Cell Cardiol
2004;37:449 –71
5. Crackower MA, Oudit GY, Kozieradzki I, Sarao R, Sun H, Sasaki
T, Hirsch E, Suzuki A, Shioi T, Irie-Sasaki J, Sah R, Cheng HY,
Rybin VO, Lembo G, Fratta L, Oliveira-dos-Santos AJ, Benovic JL,
Kahn CR, Izumo S, Steinberg SF, Wymann MP, Backx PH,
Penninger JM. Regulation of myocardial contractility and cell size
by distinct PI3K-PTEN signaling pathways. Cell 2002;110:737– 49
6. Barouch LA, Harrison RW, Skaf MW, Rosas GO, Cappola TP,
Kobeissi ZA, Hobai IA, Lemmon CA, Burnett AL, O’Rourke B,
Rodriguez ER, Huang PL, Lima JA, Berkowitz DE, Hare JM.
Nitric oxide regulates the heart by spatial confinement of nitric
oxide synthase isoforms. Nature 2002;416:337–9
7. Khan SA, Lee K, Minhas KM, Gonzalez DR, Raju SV, Tejani
AD, Li D, Berkowitz DE, Hare JM. Neuronal nitric oxide
synthase negatively regulates xanthine oxidoreductase inhibi-
tion of cardiac excitation-contraction coupling. Proc Natl Acad
Sci U S A 2004;101:15944 – 8
8. Jo SH, Leblais V, Wang PH, Crow MT, Xiao RP. Phosphatidyl-
inositol 3-kinase functionally compartmentalizes the concur-
rent G(s) signaling during beta2-adrenergic stimulation. Circ
Res 2002;91:46 –53
9. Patrucco E, Notte A, Barberis L, Selvetella G, Maffei A,
Brancaccio M, Marengo S, Russo G, Azzolino O, Rybalkin SD,
Silengo L, Altruda F, Wetzker R, Wymann MP, Lembo G,
Hirsch E. PI3Kgamma modulates the cardiac response to
chronic pressure overload by distinct kinase-dependent and
-independent effects. Cell 2004;118:375– 87
10. Walker EH, Pacold ME, Perisic O, Stephens L, Hawkins PT,
Wymann MP, Williams RL. Structural determinants of phos-
phoinositide 3-kinase inhibition by wortmannin, LY294002,
quercetin, myricetin, and staurosporine. Mol Cell 2000;6:
909 –19
ANESTHESIA & ANALGESIA
11. Kenan Y, Murata T, Shakur Y, Degerman E, Manganiello
VC. Functions of the N-terminal region of cyclic nucleotide
phosphodiesterase 3 (PDE 3) isoforms. J Biol Chem 2000;
275:12331– 8
12. Wijkander J, Landstrom TR, Manganiello V, Belfrage P, Deger-
man E. Insulin-induced phosphorylation and activation of
phosphodiesterase 3B in rat adipocytes: possible role for
protein kinase B but not mitogen-activated protein kinase or
p70 S6 kinase. Endocrinology 1998;139:219 –27
13. Mongillo M, McSorley T, Evellin S, Sood A, Lissandron V,
Terrin A, Huston E, Hannawacker A, Lohse MJ, Pozzan T,
Houslay MD, Zaccolo M. Fluorescence resonance energy
transfer-based analysis of cAMP dynamics in live neonatal rat
cardiac myocytes reveals distinct functions of compartmental-
ized phosphodiesterases. Circ Res 2004;95:67–75
October 2010 • Volume 111 • Number 4
14. Baillie GS, Sood A, McPhee I, Gall I, Perry SJ, Lefkowitz RJ,
Houslay MD. beta-Arrestin-mediated PDE4 cAMP phosphodi-
esterase recruitment regulates beta-adrenoceptor switching
from Gs to Gi. Proc Natl Acad Sci U S A 2003;100:940 –5
15. Perry SJ, Baillie GS, Kohout TA, McPhee I, Magiera MM, Ang
KL, Miller WE, McLean AJ, Conti M, Houslay MD, Lefkowitz
RJ. Targeting of cyclic AMP degradation to beta 2-adrenergic
receptors by beta-arrestins. Science 2002;298:834 – 6
16. Xiang Y, Naro F, Zoudilova M, Jin SL, Conti M, Kobilka B.
Phosphodiesterase 4D is required for beta2 adrenoceptor
subtype-specific signaling in cardiac myocytes. Proc Natl Acad
Sci U S A 2005;102:909 –14
www.anesthesia-analgesia.org 877
 ECHO ROUNDS
Intraoperative Transesophageal Echocardiography-
Guided Patent Ductus Arteriosus Ligation in an
Asymptomatic Nonbacterial Endocarditis Patient
Haibo Song, MD, Fei Liu, MD, Ke Dian, MD, and Jin Liu, MD
O n routine examination an asymptomatic 16-year-
old girl was found to have a continuous (machin-
ery like) murmur in the second intercostal space. A
transthoracic echocardiogram (TTE) demonstrated a patent
ductus arteriosus (PDA) with continuous left-to-right
shunting and a mobile hyperechogenic vegetation (40
mm ⫻ 4 mm) attached to the wall of the pulmonary artery,
with normal-sized right heart and pulmonary artery pres-
sure. Three consecutive blood cultures were negative for
bacterial growth. A chest radiograph was normal. The
patient was diagnosed with PDA and noninfective endo-
carditis. Consent for publication of this case was obtained
from the patient and her parents.
Intraoperative transesophageal echocardiography (TEE)
(Philip IE33 ⫻ 7 to 3) during PDA ligation showed a
slightly enlarged left atrium and ventricle, mild mitral
valve regurgitation, and a normal triscuspid valve. Mod-
erate aortic regurgitation and a 6-mm vegetation (Fig. 1
A) (Video 1; see Supplemental Digital Content 1,
http://links.lww.com/AA/A177; see the Appendix for
video legends) was noted in the midesophageal aortic valve
long-axis view. Using live three-dimensional TEE imaging,
we noted the vegetation to be present on the right coronary
cusp (Figure 1B) (Video 2; see Supplemental Digital Con-
tent 2, http://links.lww.com/AA/A178; see the Appendix
for video legends). Imaging in the midesophageal right
ventricular inflow– outflow view revealed a dilated main
pulmonary artery (MPA) without pulmonic valve regurgi-
tation. A PDA that measured 18 mm long and 8 mm wide
was detected in the modified upper esophageal aortic arch
short-axis view. A mobile vegetation was seen in the
pulmonary artery on the downstream side of the PDA (Fig.
2A). Color Doppler imaging demonstrated flow into the
MPA from descending aorta (Figure 2B). Continuous wave
Doppler showed a continuous spectrum; the peak velocity
of the systolic shunt flow was 4.5 m/s, and diastolic flow
was 3 m/s.
After institution of cardiopulmonary bypass, TEE-
guided ligation of the PDA was first performed (Video 2,
clip 2). During the procedure, TEE monitoring of the
anterior wall of the pulmonary artery documented the
presence of the vegetation. Intraoperative surgical findings
included a dilated and atherosclerotic PDA with small
From the West China Hospital, Sichuan University, Chengdu, Sichuan,
China.
Accepted for publication May 27, 2010.
Supplemental digital content is available for this article. Direct URL citations
appear in the printed text and are provided in the HTML and PDF versions
of this article on the journal’s Web site (www.anesthesia-analgesia.org).
Address correspondence to Fei Liu, MD, West China Hospital, Sichuan
University, Chengdu, Sichuan 610041, China. Address e-mail to
liufeicd@yahoo.cn.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181effffb
878 www.anesthesia-analgesia.org
calcified vegetations across the lumen and at the pulmo-
nary artery end. Vegetations were removed from the pul-
monic and aortic valves. Finally, aortic valvuloplasty was
performed. TEE confirmed the absence of flow through the
PDA, and the absence of vegetations and residual regurgi-
tation through the aortic valve when the operation was
finished. Pathologic examination of the vegetations proved
to be thrombi, and cultures were negative.
TEE has been shown to be more sensitive than TTE for
identifying small (⬍1 cm) valvular vegetations.1 In this case
it provided confirmatory information as well as detecting
an aortic valve vegetation and valvular insufficiency not
appreciated on preoperative TTE, which prevents infective
endocarditis (IE) and other complications after surgery.
We believe that this is the first case report of an
asymptomatic patient with nonbacterial thrombotic endo-
carditis in a patient with PDA. The fetal ductus arteriosus
arises from the aorta opposite the origin of the left subcla-
vian artery to the bifurcation of the MPA at the origin of the
left pulmonary artery. PDA normally closes in the weeks
after birth, but persists in 10% of cases of adult congenital
heart disease.2 PDA-related IE is a rare but very dangerous
event, which could cause septic shock or even pulmonary
thrombolization.3,4 Turbulent bloodflow produced by PDA
is likely a causative factor in the development of nonbac-
terial endocarditis and IE. Continuous turbulent flow
causes microscopic trauma to the endothelium of the
pulmonary artery and aortic valve, which provides an
environment conducive to platelet and fibrin deposition
and progression to IE if septicemia sets in. Before the
introduction of antibiotic therapy and surgical closure,
PDA-related IE had a 45% mortality rate. With recent
advancement in diagnostic tools, especially echocardiogra-
phy, preoperative antimicrobial therapies, and surgery, the
mortality rate of patients with PDA-related IE seems to
have declined.3
PDA is usually demonstrated in the upper esophageal
ascending aortic short-axis view with color Doppler flow
mapping.2 However, we made a small change to enhance
visualization of the PDA. We first get the upper esophageal
aortic arch short-axis view. We advanced the probe 1 to 2
cm from this view and rotated it from 0° to 60° to obtain a
view of the descending aorta and MPA (Figure 2). Color-
flow Doppler was used to detect flow through the PDA
from 0° to 60°, from which we could visualize left-to-right
continuous shunt flow originating from the descending
aorta. The systolic velocity was 4.5 m/s, yielding a peak
gradient between the descending aortic artery and pulmo-
nary artery of 81 mm Hg, and a diastolic velocity (3 m/s)
yielding the lowest gradient of 49 mm Hg.
TEE visualization of the PDA is potentially problematic
because of interposition of the trachea between the esoph-
agus and aortic arch (Figure 3 A). We used the technique
described by Li et al.,5 in which a saline-filled balloon is
October 2010 • Volume 111 • Number 4
Comprehensive TEE for PDA Ligation

Figure 1. A, Two-dimensional transesophageal

echocardiographic examination of the aortic valve
long axis view. Arrow indicates a vegetation overly-
ing the aortic valve. B, Live 3-dimensional view
shows the vegetation overlying the right coronary
artery cusp. DAO ⫽ descending aorta; AO ⫽ aorta;
RV ⫽ right ventricle; LV ⫽ left ventricle; AVV ⫽
aortic valve vegetation; RCS ⫽ right coronary cusp;
LCS ⫽ left coronary cusp; NCS ⫽ non-coronary
cusp.

Figure 2. A, Two-dimensional transesophageal

echocardiographic examination of the upper esoph-
ageal aortic arch short-axis sectional view at 30°.
Arrows indicate the vegetation in the pulmonary
artery. B, Color doppler image of shunted flow
through the patent ductus arteriosus in the upper
esophageal aortic arch short-axis view. MPA ⫽
main pulmonary artery; PV ⫽ pulmonary vegetation;
DAO ⫽ descending aorta; PDA ⫽ patent ductus
arteriosus.

Figure 3. A, Desending aorta and a blurred image of

the main pulmonary artery. B, Relationship between
the esophagus, left main bronchus, and patent
ductus arteriosus (PDA). We used a saline balloon
in the left main bronchus to improve the image of
the PDA. C, Clear pulmonary artery after we used a
saline balloon. DAO ⫽ descending aorta, MPA ⫽
main pulmonary artery, AAO ⫽; AO arch ⫽ aortic
arch, LPA ⫽ left pulmonary artery, LB ⫽ left main
bronchus; TEE ⫽ transesophageal echocardiogra-
phy; RPA ⫽ right pulmonary artery; LPA ⫽ left
pulmonary artery.

advanced through the endotracheal tube after initiation of
cardiopulmonary bypass but before aortic cross-clamping
to attenuate the ultrasound scattering by the tracheal air
column (Fig. 3 B). This resulted in substantial enhancement
of the PDA in this case (Fig. 3C).
In summary, in this case, TEE monitoring played an
important role in PDA ligation, facilitating assessment of
residual bloodflow. It also allowed monitoring of the status
of the vegetations in the pulmonary artery and on the aortic
valve.

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 879

 ECHO ROUNDS

APPENDIX: VIDEO LEGENDS
Video 1. Real-time monitoring of persistent patent ductus arteriosus
(PDA) during surgical ligation. Initial bloodflow from the descending
aorta to the pulmonary artery ceases by the end of the video
recording. White arrow shows colorful bloodflow ceases after PDA
ligation. MPA ⫽ main pulmonary artery; DAO ⫽ descending aorta.
Video 2. In the first part of the video is the live 2-dimensional
transesophageal echocardiography (TEE) view of aortic valve vegeta-
tion in the aortic valve long-axis view. White arrow indicates a
vegetation overlying the aortic valve. In the second part of the video
is the live 3-dimensional TEE view of aortic valve vegetation. White
arrow indicates a vegetation overlying the aortic valve. LV ⫽ left
ventricle; LA ⫽ left atrium; AO ⫽ aorta.
REFERENCES
1. Shapiro SM, Young E, De Guzman S, Ward J, Chiu CY, Ginzton
LE, Bayer AS. Transesophageal echocardiography in diagnosis
of infective endocarditis. Chest 1994;105:377– 82
2. Neuman MD, Fox JD, Muehlschlegel JD. Incidental discovery of
a large patent ductus arteriosus in an adult during aortic
reconstruction: echocardiographic findings and diagnostic di-
lemmas. Anesth Analg 2007;105:1227– 8
3. Ozkokeli M, Ates M, Uslu N, Akcar M. Pulmonary and aortic
valve endocarditis in an adult patient with silent patent ductus
arteriosus. Jpn Heart J 2004;45:1057– 61
4. Kouris NT, Sifaki MD, Kontogianni DD, Zaharos I, Kalkandi
EM, Grassos HE, Babalis DK. Patent ductus arteriosus endarte-
ritis in a 40-year-old woman, diagnosed with transesophageal
echocardiography. A case report and a brief review of the
literature. Cardiovasc Ultrasound 2003;1:2
5. Li YL, Wong DT, Wei W, Liu J. A new method for detecting the
proximal aortic arch and innominate artery by transesophageal
echocardiography. Anesthesiology 2006;105:226 –7

Clinician’s Key Teaching Points By Martin M. Stechert, MD, Roman M. Sniecinski, MD,
and Martin J. London, MD
• Patent ductus arteriosus (PDA) is the persistence of a normal fetal connection between the left pulmonary artery and
the descending aorta. Clinical manifestations are determined by the size of the PDA and the degree of left-to-right
shunting. Although large PDAs cause left ventricular overload and are usually diagnosed in the newborn, a small lesion
may remain undetected into adulthood. Turbulent flow through the PDA can lead to endothelial injury, complicated by
nonbacterial thrombotic endocarditis, bacterial seeding, and infective endocarditis.
• Transesophageal echocardiography (TEE) can demonstrate a PDA with the upper esophageal aortic arch views, which
show a dilated main pulmonary artery and, when color-flow Doppler is applied, turbulent flow into the left pulmonary
artery from the descending aorta. However, because of the interposition of the air-filled trachea over the aortic arch,
the left pulmonary artery is not always well visualized. Therefore, transthoracic echocardiography using the
suprasternal and parasternal windows is the preferred method of diagnosis.
• In this case, the authors initially obtained the upper esophageal aortic short-axis view, advanced the TEE probe 1 to
2 cm, then adjusted the omniplane angle to 60° to successfully visualize the PDA. The image was further improved
by inserting a saline-filled balloon through the endotracheal tube while on bypass, which decreased scattering of the
ultrasound by endotracheal air.
• TEE may be useful for perioperative PDA assessment, provided an appropriate echo window can be found. In select
patients, visualization might be improved by filling the endotracheal cuff with normal saline to provide such a window.

880 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

Society for Ambulatory Anesthesiology
Section Editor: Peter S. A. Glass

Effect on Postoperative Sore Throat of Spraying the

Endotracheal Tube Cuff with Benzydamine
Hydrochloride, 10% Lidocaine, and 2% Lidocaine
Nan-Kai Hung, MD,* Ching-Tang Wu, MD,* Shun-Ming Chan, MD,* Chueng-He Lu, MD,*
Yuan-Shiou Huang, MD,* Chun-Chang Yeh, MD,* Meei-Shyuan Lee, DPH,†
and Chen-Hwan Cherng, MD, DMSc*

BACKGROUND: Postoperative sore throat (POST) is a common complication after endotracheal

intubation. We compared the effectiveness on POST of spraying the endotracheal tube (ETT) cuff
with benzydamine hydrochloride, 10% lidocaine, and 2% lidocaine.
METHODS: Three hundred seventy-two patients were randomly allocated into 4 groups. The ETT
cuffs in each group were sprayed with benzydamine hydrochloride, 10% lidocaine hydrochloride,
2% lidocaine hydrochloride, or normal saline before endotracheal intubation. After insertion, the
cuffs were inflated to an airway leak pressure of 20 cm H2O. Anesthesia was maintained with
propofol. The patients were examined for sore throat (none, mild, moderate, or severe) at 1, 6,
12, and 24 hours after extubation.
RESULTS: The highest incidence of POST occurred at 6 hours after extubation in all groups.
There was a significantly lower incidence of POST in the benzydamine group than 10% lidocaine,
2% lidocaine, and normal saline groups (P ⬍ 0.05) at each observation time point. At 6 hours
after extubation, the incidence of POST was significantly lower in the benzydamine group (17.0%)
compared with 10% lidocaine (53.7%), 2% lidocaine (37.0%), and normal saline (40.8%) groups
(P ⬍ 0.05). The benzydamine group had significantly decreased severity of POST compared with
the 10% lidocaine, 2% lidocaine, and normal saline groups (P ⬍ 0.05) at each observation time
point. Compared with the 2% lidocaine and normal saline groups, the 10% lidocaine group had
significantly increased severity of POST at 1, 6, and 12 hours after extubation. There were no
significant differences among groups in local or systemic side effects.
CONCLUSIONS: Spraying benzydamine hydrochloride on the ETT cuff is a simple and effective
method to reduce the incidence and severity of POST. (Anesth Analg 2010;111:882–6)

P ostoperative sore throat (POST) after general anes-

thesia with an endotracheal tube (ETT) is an unde-
sirable outcome1 with an incidence varying from
40% to 100%.2–5 Although the symptoms resolve spontane-
ously without any treatment, prophylactic management for
decreasing its frequency and severity is still recommended
to improve the quality of postanesthesia care.6 – 8
Several pharmacological methods have been suggested
to reduce POST including inhaling beclomethasone; apply-
ing lidocaine spray or lidocaine to the ETT; administering
aspirin, ketamine, or benzydamine hydrochloride; or gar-
gling with azulene sulfonate.6,9 –12 Benzydamine hydro-
chloride is a topical nonsteroidal antiinflammatory drug
that also has analgesic, antipyretic, and antimicrobial
From the *Department of Anesthesiology, Tri-Service General Hospital and
National Defense Medical Center; and †School of Public Health, National
Defense Medical Center, Taipei, Taiwan, Republic of China.
Accepted for publication January 11, 2010.
Supported by grants from Tri-Service General Hospital (TSGH-C98-125) of
Taiwan, Republic of China.
Address correspondence and reprint requests to Dr. Chen-Hwan Cherng,
Department of Anesthesiology, Tri-Service General Hospital and National
Defense Medical Center, #325, Section 2, Chenggung Rd., Neihu 114, Taipei,
Taiwan, ROC. Address e-mail to cherng1018@yahoo.com.tw.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181d4854e
properties.13–16 It has been reported that moderate to
severe sore throat may be resolved with gargling benzy-
damine hydrochloride.17 In addition, it is widely used in
radiation-induced oral mucositis,18 for arthritis as a gel
ointment preparation applied to the skin,13 and for symp-
tomatic treatment of acute sore throat.17,19 Preventive topical
benzydamine hydrochloride applied to the oropharyngeal
cavity before endotracheal intubation or before endotracheal
intubation and continuously for 48 hours postoperatively has
been reported to decrease the incidence and severity of POST
after ETT insertion and laryngeal mask airway inser-
tion.19 –21 However, patients may sustain numbness or the
sensation of tingling in the tissues in the oral cavity, dry
mouth, thirst, and nausea because of application of benzy-
damine hydrochloride by gargling or oropharyngeal
spray.21 Combes et al.22 demonstrated that mucosal dam-
age occurring at the cuff level is thought to be an important
causative factor for tracheal morbidity. Therefore, applica-
tion of benzydamine hydrochloride to the ETT rather than
gargling may provide an alternative and effective method
to reduce the incidence and severity of POST.
The hypothesis of this study was that simply spraying
benzydamine hydrochloride on the endotracheal cuff will
provide better prevention of POST after extubation than
10% lidocaine hydrochloride, 2% lidocaine hydrochloride,
or normal saline sprayed over the cuff of the ETT.

882 www.anesthesia-analgesia.org October 2010 • Volume 111 • Number 4

METHODS
After obtaining hospital ethics committee approval and
written informed consent, 420 patients (n ⫽ 105 per group),
ASA physical status I to III, aged 18 to 84 years, were
included in this prospective, randomized, and double-blind
study. All patients were scheduled for surgical procedures
under general anesthesia with orotracheal intubation in the
supine position. The exclusion criteria included patients
undergoing oral cavity surgery, cervical spine surgery,
thyroid surgery, patients with a difficult airway (after ⬎1
orotracheal intubation attempt), requiring a nasogastric
tube, or a history of perioperative sore throat.
Patients were randomly allocated into 4 groups by
choosing blinded envelopes. All study solutions and nor-
mal saline were contained in similarly configured bottles.
The grouping and medication were as follows:
1. The benzydamine (Comfflam, United Biomedical, Asia)
(containing ethanol, glycerin, and menthol as additives)
group: 10 puffs of benzydamine hydrochloride were
sprayed on the ETT cuffs, which contain approximately
1.5 mg benzydamine hydrochloride.
2. The 10% lidocaine (Xylocaine Spray 10%, AstraZeneca,
Sweden) (containing ethanol, polyethylene glycol 400,
menthol, saccharin, and essence of banana as addi-
tives) group: 10 puffs of 10% lidocaine hydrochloride
were sprayed on the ETT cuffs, which contain ap-
proximately 100 mg lidocaine hydrochloride.
3. The 2% lidocaine (Xylocaine 2%, AstraZeneca) (con-
taining sodium chloride as an additive) group: 10
puffs of 2% lidocaine hydrochloride were sprayed on
the ETT cuffs, which contain approximately 20 mg
lidocaine hydrochloride.
4. The normal saline group: 10 puffs of normal saline
were sprayed on the ETT cuffs, which contain ap-
proximately 1 mL normal saline.
Sterile ETTs (ILM Endotracheal Tube, Euromedical, Ke-
dah, Malaysia) with high-volume, low-pressure cuffs were
used. Application of substances to the ETT cuff was per-
formed 5 minutes before the induction of anesthesia. Male
and female patients received 7.5- and 7.0-mm inner diam-
eter ETTs, respectively. Before intubation, oxygen was
administered, adequate IV access was established, and
standard American Society of Anesthesiologists clinical
monitoring was applied. After administration of 2 to 3
␮g/kg fentanyl, induction was accomplished with 2 to 2.5
mg/kg propofol and 0.6 mg/kg rocuronium. Endotracheal
intubation was performed by residents with at least 2 years
of experience and attending physicians, who were blinded
to group allocation. The ETT cuffs were inflated with room
air to achieve a seal at 20 cm H2O of peak airway pressure.
Anesthesia was maintained by a target-controlled infusion
system (Fresenius Base Primea威, Brezins, France) with
propofol, intermittent fentanyl, and cisatracurium/rocuronium
administered as required. Nitrous oxide was not used. At
the end of surgery, neuromuscular blockade was antago-
nized by neostigmine and atropine. After full recovery and
awakening, the ETT was removed after gentle suctioning of
oral secretions. Patients were then transferred to the post-
anesthesia care unit.
October 2010 • Volume 111 • Number 4
At 1, 6, 12, and 24 hours after extubation, the patients
were asked about sore throat and hoarseness by a single
investigator who was blinded to the group allocation. POST
was graded by a modified 4-point scale (1– 4): 1, no sore
throat; 2, mild sore throat (complains of sore throat only on
asking); 3, moderate sore throat (complains of sore throat
on his or her own); and 4, severe sore throat (change of
voice or hoarseness, associated with throat pain).10,11
We recorded the patients’ age, sex, weight, height,
duration of surgery, total fentanyl consumption, type of
surgery, and postoperative analgesia methods. Potential
side effects associated with tracheal intubation or the
study drugs, such as nausea, vomiting, cough, throat
numbness or stinging, dry mouth, and hoarseness, were
also recorded.
Before initiation of the study, power analysis was per-
formed. A minimum of 91 patients was required in each
group to detect a decrease of the incidence of POST from
40% to 20% with a power of 80% and a significance level of
95%. Statistical analysis was performed using the SPSS for
Windows version 15 (SPSS, Chicago, IL). Results are ex-
pressed as mean (SD), median with range, or percentage.
Patient age, height, weight, and duration of surgery were
compared among groups and tested statistically by analysis
of variance. The incidence of POST and side effects among
groups were tested by ␹2 tests. To avoid a type I error, for
those significant variables in the ␹2 test, we recalculated all
possible six 2 ⫻ 2 ␹2 tests by applying the Bonferroni
inequality to adjust the ␣ level (i.e., P[␹2 ⬎ 6.97] ⫽ 0.05/6 ⫽
0.0083 for 1 degree of freedom). Differences in the severity
of symptoms among groups were evaluated by Kruskal-
Wallis tests. Furthermore, the Dunn procedure was applied
to compare the difference among groups. The area under
the curve (AUC) by adding 3 trapezoid areas was gener-
ated from the level of POST and time (within 24 hours).
Both differences in the severity of symptoms and AUC
were tested by Kruskal-Wallis analysis of variance and
followed by the Dunn procedure. Values of P ⬍ 0.05 were
considered statistically significant.
RESULTS
Characteristics of the study groups are shown in Table 1.
There were 11, 12, 13, and 12 patients withdrawn from the
benzydamine, 10% lidocaine, 2% lidocaine, and normal
saline groups, respectively, because of ⬎1 attempt at oro-
tracheal intubation or nasogastric tube insertion during the
operation. Therefore, there were 372 patients enrolled in
this study.
In the 24-hour evaluation period, the highest incidences
of POST occurred at the sixth hour time interval after
extubation, with incidences (95% confidence interval)
of 17.0% (9.4%–24.6%), 53.7% (43.6%– 63.9%), 37.0%
(27.1%– 46.8%), and 40.8% (30.7%–50.9%) in the benzydam-
ine, 10% lidocaine, 2% lidocaine, and normal saline groups,
respectively. There was a significantly lower incidence of
POST in the benzydamine group than the 10% lidocaine,
2% lidocaine, and normal saline groups (P ⬍ 0.05) at each
observation time point (Table 2). Similarly, the benzydam-
ine group had significantly decreased severity of POST
www.anesthesia-analgesia.org 883
Benzydamine Hydrochloride Reduces Postoperative Sore Throat

Table 1. Demographic Data of the Patients and Data Related to the Surgery
Benzydamine 10% lidocaine 2% lidocaine
hydrochloride hydrochloride hydrochloride Normal saline
Group (n ⴝ 94) (n ⴝ 93) (n ⴝ 92) (n ⴝ 93)
Gender (M/F) 45/49 48/45 46/46 48/45
Age (y) 48.5 (16) 47.8 (15) 45.3 (17) 46.3 (17)
Height (cm) 162.9 (9.2) 162.5 (8.8) 162.7 (9.3) 161.6 (9.5)
Weight (kg) 62.2 (11.4) 63.8 (13.4) 64.3 (12.5) 62.3 (12.1)
Duration of surgery (min) 183.7 (110) 181.6 (91) 183.8 (114) 176.8 (100)
Total fentanyl consumption (␮g) 173 (73) 173 (59) 173 (61) 171 (61)
Postoperative analgesia method (PCA/meperidine IM) 78/16 76/17 75/17 78/15
Types of surgery
Colon rectal surgery 6 4 6 4
General surgery 24 18 20 16
Genitourinary surgery 3 5 5 4
Gynecologic surgery 20 16 18 24
Ophthalmologic surgery 12 17 16 15
Orthopedic surgery 23 18 16 21
Plastic surgery 6 15 11 9
Values are presented as mean (SD) or number of patients.
PCA ⫽ patient-controlled analgesia.

Table 2. Incidence (n, %) and Severity (Median, Range) of Postoperative Sore Throat
Benzydamine 10% lidocaine 2% lidocaine Normal saline
Evaluation time (n ⴝ 94) (n ⴝ 93) (n ⴝ 92) (n ⴝ 93) P*
1 h after extubation
Incidence 10 (10.6%)* 30 (33.2%) 21 (22.8%) 22 (23.7%) 0.005
Severity 1 (1–4)* 1 (1–4)†‡ 1 (1–4) 1 (1–4) 0.004
6 h after extubation
Incidence 16 (17.0%)* 50 (53.7%) 34 (37.0%) 38 (40.8%) ⬍0.001
Severity 1 (1–4)* 2 (1–4)†‡ 1 (1–4) 1 (1–4) ⬍0.001
12 h after extubation
Incidence 5 (5.3%)* 37 (39.8%) 22 (23.9%) 30 (32.2%) ⬍0.001
Severity 1 (1–4)* 1 (1–4)†‡ 1 (1–4) 1 (1–4) ⬍0.001
24 h after extubation
Incidence 2 (2.1%)* 25 (26.9%) 16 (17.4%) 19 (20.4%) ⬍0.001
Severity 1 (1–4)* 1 (1–4)† 1 (1–4) 1 (1–4) ⬍0.001
Values are presented as number of patients (%) or median (range).
* P ⬍ 0.05, benzydamine group versus 10% lidocaine, 2% lidocaine, and normal saline groups.
† P ⬍ 0.05, 10% lidocaine group versus 2% lidocaine group.
‡ P ⬍ 0.05, 10% lidocaine group versus normal saline.

compared with the 10% lidocaine, 2% lidocaine, and nor-
mal saline groups (P ⬍ 0.05) at each observation time point
(Table 2). Compared with 2% lidocaine and normal saline
groups, the 10% lidocaine group had significantly in-
creased severity of POST at 1, 6, and 12 hours after
extubation (Table 2). Of the mean AUCs generated from the
level of POST and time (within 24 hours), the benzydamine
group (25.6) had a significantly smaller area than the 10%
lidocaine (41.4) and normal saline groups (33.0) (P ⬍ 0.05),
but not the 2% lidocaine group (37.5). There were no
significant differences in AUC among any other groups
(data not shown).
There were no significant differences among groups for
the potential side effects relevant to tracheal intubation or
study drugs.
DISCUSSION
This study demonstrated that spraying benzydamine hy-
drochloride on an ETT cuff may reduce the incidence and
severity of POST compared with applying 10% lidocaine,
2% lidocaine, and normal saline. An unexpected finding
was that applying 10% lidocaine on the ETT cuff increased
the severity of POST.
POST is one of the common side effects associated
with endotracheal intubation.2–5 Tracheal mucosa lesions
after intubation and an overinflated ETT cuff have been
proposed to be a possible cause of POST.22 These com-
plications can occur after even a “smooth” intubation.
Immediate POST may be primarily due to the action of
extubation, and late POST may be related to tracheal
mucosa trauma.23 According to the results of this study,
the highest incidence of POST occurred at the sixth hour
after extubation, but not the first hour. Sore throat at the
first hour after extubation might be masked by residual
analgesic effects after general anesthesia.
Benzydamine hydrochloride is indicated for the relief of
painful conditions of the mouth and throat such as tonsil-
litis, sore throat, radiation mucositis, and postorosurgical
and periodontal procedures. Previous studies demon-
strated that topical application of benzydamine hydrochlo-
ride to the pharynx before laryngeal mask airway or ETT
insertion decreased the incidence of POST.19 –21 The side

884 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

effects of topical use of benzydamine hydrochloride in-
clude local numbness or burning, stinging sensation, nau-
sea or vomiting, cough, dry mouth, throat discomfort,
drowsiness, and headache,24 which may be evident before
induction of anesthesia. To avoid these adverse effects,
we applied benzydamine hydrochloride on the ETT cuff
instead of perioperative topical application to the oral
pharyngeal cavity. We found that this maneuver pro-
vided excellent prevention of POST and reduced its
incidence from placebo or 2% lidocaine spray by ⬎50%.
Therefore, the application of benzydamine hydrochlo-
ride on the ETT cuffs may provide a simple and effective
method to attenuate the incidence and severity of POST
after tracheal intubation.
Application of lidocaine spray to the oral pharyngeal
cavity before intubation seems to increase the incidence of
sore throat.5,25,26 In this study, we also found that spraying
10% lidocaine on the ETT cuff also increased the severity of
POST compared with 2% lidocaine or placebo. Ten percent
lidocaine solution contains ethanol, polyethylene glycol
400, menthol, saccharin, and macrogolum as additives in
the solvent, whereas the 2% lidocaine solution we used
contained sodium chloride as an additive. In fact, both
menthol and ethanol can irritate tracheal mucosa, poten-
tially causing tracheal mucosa damage, thus leading to
increased severity of POST. However, Soltani and Agha-
davoudi27 reported that using intracuff lidocaine (ETT cuffs
prefilled with 7 to 8 mL of 2% lidocaine for 90 minutes
before intubation and refilled with enough 2% lidocaine
after intubation) was superior to spraying topical 10%
lidocaine on laryngopharyngeal structures or on the distal
end of the ETT for decreasing the incidence of POST.
Theoretically, chemical irritation from the additives may be
avoided by using intracuff lidocaine. We also found that 2%
lidocaine spray did not attenuate the incidence and severity
of POST compared with normal saline. The duration of the
analgesic effect of lidocaine spray applied to oral mucosa is
⬍15 minutes.28 In this study, at the end of surgery (aver-
aging 180 minutes after tracheal intubation), the analgesic
effect of lidocaine spray might have already disappeared.
This probably explains why we found the incidence of
POST to be no different between the 2% lidocaine and
normal saline groups.
One limitation of our study is that there was no record
of coughing or bucking at the time of extubation. Although
the extubation protocol was the same in all groups, we did
not evaluate the correlation between the frequency of
coughing or bucking at the time of extubation and the
incidence of POST. The second limitation is that benzy-
damine hydrochloride is available under different trade
names in different countries, its formulations are quite
different in each country, and the additives might also vary.
The drug is not available in the United States, Canada, and
most of Western Europe (the United Kingdom being an
exception). For this reason, the results of this study might
not be widely applicable. Moreover, the safety and dosage
of benzydamine hydrochloride applied to the trachea need
further investigation, even though we did not find any
adverse effects in our patients. The third limitation is that
the additives to 2% and 10% lidocaine solution are differ-
ent, which may have influenced the result.
October 2010 • Volume 111 • Number 4
In conclusion, application of benzydamine hydrochlo-
ride on the ETT cuff effectively attenuates the incidence
and severity of POST. Application of 10% lidocaine spray
should be avoided because of worsening of POST, and
spraying 2% lidocaine on the ETT cuff does not prevent
POST.
REFERENCES
1. Macario A, Weinger M, Carney S, Kim A. Which clinical
anesthesia outcomes are important to avoid? The perspective
of patients. Anesth Analg 1999;89:652– 8
2. Biro P, Seifert B, Pasch T. Complaints of sore throat after
tracheal intubation: a prospective evaluation. Eur J Anaesthe-
siol 2005;22:307–11
3. Al-Qahtani AS, Messahel FM. Quality improvement in anes-
thetic practice—incidence of sore throat after using small
tracheal tube. Middle East J Anesthesiol 2005;18:179 – 83
4. Monroe MC, Gravenstein N, Saga-Rumley S. Postoperative
sore throat: effect of oropharyngeal airway in orotracheally
intubated patients. Anesth Analg 1990;70:512– 6
5. McHardy FE, Chung F. Postoperative sore throat: cause, pre-
vention and treatment. Anaesthesia 1999;54:444 –53
6. el Hakim M. Beclomethasone prevents postoperative sore
throat. Acta Anaesthesiol Scand 1993;37:250 –2
7. Monem A, Kamal RS. Postoperative sore throat. J Coll Physi-
cians Surg Pak 2007;17:509 –14
8. Park SH, Han SH, Do SH, Kim JW, Rhee KY, Kim JH.
Prophylactic dexamethasone decreases the incidence of sore
throat and hoarseness after tracheal extubation with a double-
lumen endobronchial tube. Anesth Analg 2008;107:1814 – 8
9. Sumathi PA, Shenoy T, Ambareesha M, Krishna HM. Con-
trolled comparison between betamethasone gel and lidocaine
jelly applied over tracheal tube to reduce postoperative sore
throat, cough, and hoarseness of voice. Br J Anaesth 2008;
100:215– 8
10. Canbay O, Celebi N, Sahin A, Celiker V, Ozgen S, Aypar U.
Ketamine gargle for attenuating postoperative sore throat. Br J
Anaesth 2008;100:490 –3
11. Agarwal A, Nath SS, Goswami D, Gupta D, Dhiraaj S, Singh
PK. An evaluation of the efficacy of aspirin and benzydamine
hydrochloride gargle for attenuating postoperative sore throat:
a prospective, randomized, single-blind study. Anesth Analg
2006;103:1001–3
12. Ogata J, Minami K, Horishita T, Shiraishi M, Okamoto T,
Terada T, Sata T. Gargling with sodium azulene sulfonate
reduces the postoperative sore throat after intubation of the
trachea. Anesth Analg 2005;101:290 –3
13. Quane PA, Graham GG, Ziegler JB. Pharmacology of benzy-
damine. Inflammopharmacology 1998;6:95–107
14. Baldock GA, Brodie RR, Chasseaud LF, Taylor T, Walmsley
LM, Catanese B. Pharmacokinetics of benzydamine after intra-
venous, oral, and topical doses to human subjects. Biopharm
Drug Dispos 1991;12:481–92
15. Guglielmotti A, Aquilini L, Rosignoli MT, Landolfi C, Soldo L,
Coletta I, Pinza M. Benzydamine protection in a mouse model
of endotoxemia. Inflamm Res 1997;46:332–5
16. Modeer T, Yucel-Lindberg T. Benzydamine reduces prosta-
glandin production in human gingival fibroblasts challenged
with interleukin-1 beta or tumor necrosis factor alpha. Acta
Odontol Scand 1999;57:40 –5
17. Turnbull RS. Benzydamine Hydrochloride (Tantum) in the
management of oral inflammatory conditions. J Can Dent
Assoc 1995;61:127–34
18. Epstein JB, Silverman S Jr, Paggiarino DA, Crockett S, Schubert
MM, Senzer NN, Lockhart PB, Gallagher MJ, Peterson DE,
Leveque FG. Benzydamine HCl for prophylaxis of radiation-
induced oral mucositis: results from a multicenter, random-
ized, double-blind, placebo-controlled clinical trial. Cancer
2001;92:875– 85
19. Gulhas N, Canpolat H, Cicek M, Yologlu S, Togal T, Durmus
M, Ozcan Ersoy M. Dexpanthenol pastille and benzydamine
hydrochloride spray for the prevention of post-operative sore
throat. Acta Anaesthesiol Scand 2007;51:239 – 43
www.anesthesia-analgesia.org 885
Benzydamine Hydrochloride Reduces Postoperative Sore Throat
20. Mazzarella B, Macarone Palmieri A, Mastronardi P, Spatola R,
Lamarca S, De Rosa G, Mastella A. Benzydamine for the
prevention of pharyngo-laryngeal pathology following tra-
cheal intubation. Int J Tissue React 1987;9:121–9
21. Kati I, Tekin M, Silay E, Huseyinoglu UA, Yildiz H. Does
benzydamine hydrochloride applied preemptively reduce sore
throat due to laryngeal mask airway? Anesth Analg 2004;
99:710 –2
22. Combes X, Schauvliege F, Peyrouset O, Motamed C, Kirov K,
Dhonneur G, Duvaldestin P. Intracuff pressure and tracheal
morbidity: influence of filling with saline during nitrous oxide
anesthesia. Anesthesiology 2001;95:1120 – 4
23. Keane WM, Denneny JC, Rowe LD, Atkins JP Jr. Complications
of intubation. Ann Otol Rhinol Laryngol 1982;91:584 –7
24. Passali D, Volonte M, Passali GC, Damiani V, Bellussi L.
Efficacy and safety of ketoprofen lysine salt mouthwash versus
benzydamine hydrochloride mouthwash in acute pharyngeal
inflammation: a randomized, single-blind study. Clin Ther
2001;23:1508 – 8
886 www.anesthesia-analgesia.org
25. Maruyama K, Sakai H, Miyazawa H, Iijima K, Toda N,
Kawahara S, Hara K. Laryngotracheal application of lidocaine
spray increases the incidence of postoperative sore throat after
total intravenous anesthesia. J Anesth 2004;18:237– 40
26. Hara K, Maruyama K. Effect of additives in lidocaine spray on
postoperative sore throat, hoarseness and dysphagia after total
intravenous anaesthesia. Acta Anaesthesiol Scand 2005;49:
463–7
27. Soltani HA, Aghadavoudi O. The effect of different lidocaine
application methods on postoperative cough and sore throat.
J Clin Anesth 2002;14:15– 8
28. Schonemann NK, van der Burght M, Arendt-Nielsen L, Bjer-
ring P. Onset and duration of hypoalgesia of lidocaine spray
applied to oral mucosa—a dose response study. Acta Anaes-
thesiol Scand 1992;36:733–5
ANESTHESIA & ANALGESIA
The Effectiveness of Benzydamine Hydrochloride
Spraying on the Endotracheal Tube Cuff or Oral
Mucosa for Postoperative Sore Throat
Yuan-Shiou Huang, MD,* Nan-Kai Hung, MD,* Meei-Shyuan Lee, MPH,† Chang-Po Kuo, MD,*
Jyh-Cherng Yu, MD,‡ Go-Shine Huang, MD,* Chen-Hwan Cherng, MD, DMSC,*
Chih-Shung Wong, MD, PhD,* Chi-Hong Chu, MD, PhD,‡ and Ching-Tang Wu, MD*

BACKGROUND: The etiology of postoperative sore throat (POST) is considered to be the result
of laryngoscopy, intubation damage, or inflated cuff compression of the tracheal mucosa. In this
study, we compared the effectiveness in alleviating POST using different approaches to
benzydamine hydrochloride (BH) administration by spraying the endotracheal tube (ET) cuff or the
oropharyngeal cavity, or both.
METHODS: Three hundred eighty patients were included in this prospective and double-blind
study, which was randomized into 4 groups: group A, oropharyngeal cavity spray of BH, and
distilled water on the ET cuff; group B, both the oropharyngeal cavity and the ET cuff received BH
spray; group C, the ET cuff received BH spray, and the oropharyngeal cavity received distilled
water; and group D, distilled water sprayed on both the ET tube and into the oropharyngeal cavity.
The patients were examined for sore throat (none, mild, moderate, severe) at 0, 2, 4, and 24
hours postextubation.
RESULTS: The incidence of POST was 23.2%, 13.8%, 14.7%, and 40.4% in groups A, B, C, and
D, respectively. POST occurred significantly less frequently in groups B and C compared with
group D (odds ratio: 0.36; 95% confidence interval: 0.21– 0.60; P ⬍ 0.05). However, there was
no significant difference between groups A and D (odds ratio: 0.62; 95% confidence interval:
0.38 –1.01). Moreover, there was no significant interaction between spraying BH over the
oropharyngeal cavity and the ET cuff on the incidence of POST (P ⫽ 0.088). The severity of POST
was significantly more intense in group D compared with groups B and C (P ⬍ 0.001). Group B
had a significantly higher incidence of local numbness, burning, and/or stinging sensation
compared with patients in group D (P ⬍ 0.05).
CONCLUSIONS: This study indicates that spraying BH on the ET cuff decreases the incidence
and severity of POST without increased BH-related adverse effects. (Anesth Analg 2010;
111:887–91)

P ostoperative sore throat (POST) after intubated gen-

eral anesthesia (GA) is a troublesome complication
and is recognized as one of the undesirable outcomes
in the postoperative period.1 The incidence of POST ranges
from 21% to 66%2,3 in accordance with different surgical
and anesthetic manipulations.4,5 Therapeutic management
for decreasing its frequency and severity is still advised to
improve the quality of postanesthesia care even though it
will resolve without treatment.4,6 Various methods have
been reported to alleviate POST, such as using a smaller
sized endotracheal tube (ET), lubricating the ET cuff, and
avoiding excessive intracuff pressure.2 In addition, several
analgesic and antiinflammatory drugs have also been used,
including IV dexamethasone,4 inhaled beclomethasone,
and gargling with azulene sulfonate,6 aspirin,7 or benzy-
damine hydrochloride (BH).8,9 BH is a topical nonsteroidal
From the *Department of Anesthesiology, †School of Public Health, and
‡Division of General Surgery, Tri-Service General Hospital and National
Defense Medical Center, Taipei, Taiwan, Republic of China.
Accepted for publication April 25, 2010.
Disclosure: The authors report no conflicts of interest.
Address correspondence and reprint requests to Ching-Tang Wu, MD,
Department of Anesthesiology, Tri-Service General Hospital and Na-
tional Defense Medical Center, #325, Section 2, Chenggung Rd., Neihu
114, Taipei, Taiwan. Address e-mail to hong18002@hotmail.com or
wuchingtang@msn.com.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181e6d82a
antiinflammatory drug and has analgesic, antipyretic, an-
timicrobial, and antiinflammatory effects.10 It has been
reported that moderate to severe oral inflammatory condi-
tions may be resolved by gargling BH.11 In addition, it is
widely used in radiation-induced oral mucositis,12 for
arthritis as a gel ointment preparation applied to the skin,10
and for symptomatic treatment of acute sore throat.3,11 The
aim of this study was to compare the effectiveness of 0.15%
BH in alleviating POST using different approaches: spray-
ing on the ET cuff, into the oropharyngeal cavity, or both.
METHODS
This study was approved by our institutional ethics com-
mittee, and written informed consent was obtained from all
participants. In this prospective, randomized, and double-
blind study, we enrolled 380 patients. All patients were
scheduled for elective surgery under GA with ET intuba-
tion. Patients who had head and neck surgery, cervical
spine surgery, or thyroid surgery, and had a history of
preoperative sore throat were excluded. Patients with first
attempt failed laryngoscopy, postoperative endotracheal
intubation or reintubation, and nasogastric tube insertion
were also excluded from further analysis.
All patients were randomly assigned into 4 groups (95
patients in each group). The grouping and medication were
as follows: group A, 5 puffs of 0.15% BH (total 0.75 mg,
Comfflam; United Biomedical Asia, Taiwan) were sprayed

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 887

Topical Benzydamine Hydrochloride Attenuates Postoperative Sore Throat
into the oropharyngeal cavity and 5 puffs of distilled water
were sprayed on the ET cuff (5 puffs containing approxi-
mately 0.5 mL distilled water); group B, the oropharyngeal
cavity and the ET cuff were both sprayed with 5 puffs of
BH; group C, the ET cuff was sprayed with 5 puffs of BH
and the oropharyngeal cavity was sprayed with 5 puffs of
distilled water; and group D, the oropharyngeal cavity and
the ET cuff were both sprayed with 5 puffs of distilled
water. All medications were sprayed 5 minutes before
induction of anesthesia by a nurse anesthetist blinded to the
treatment. The treatments were prepared by a pharmacist
in our pharmacy department blinded to the medication so
that the different treatments had the same external appear-
ance. No patient received premedication, and standard
monitors were applied in the operating room. GA was
induced with 2 to 3 ␮g/kg fentanyl, 1 to 1.5 mg/kg
lidocaine, and 2 to 2.5 mg/kg propofol. Laryngoscopy and
ET intubation were facilitated by administration of 0.6
mg/kg rocuronium. Patients were tracheally intubated
with a 7.0- and 6.5-mm inner diameter ET with a high-
volume and low-pressure cuff (ILM Endotracheal Tube;
Euromedical, Kedah, Malaysia), respectively. Intubations
were performed by residents with at least 2 years of
experience or attending physicians who were blinded to the
treatment. The ET cuffs were inflated with room air to a
cuff pressure of 20 to 25 cm H2O. We measured the cuff
pressure immediately after ET intubation using a manom-
eter (VBM, Sulz, Germany) that was connected to the pilot
balloon of the ET cuff via a 3-way stopcock, and the cuff
pressure was measured once in each patient at 60 minutes
after intubation. Anesthesia was maintained with 8% to
12% desflurane in a total flow of 300 mL/min oxygen
under a closed system without nitrous oxide. Fentanyl was
administered depending on the surgical stimulus and the
hemodynamic response; the dosage was approximately 1
␮g/kg/h after intubation. Neuromuscular transmission
was monitored using train-of-four supramaximal stimula-
tions (2 Hz, 50 mA) (TOF Watch SX; Organon, Dublin,
Ireland). Intravenous boluses of cisatracurium 2 mg were
administered when more than 2 responses were detected in
train-of-four stimulation until closure of the peritoneum
was commenced. At the end of surgery, residual neuro-
muscular blockade was antagonized by neostigmine and
atropine. When the T4/T1 ratio reached ⱖ90% and patients
could follow simple commands, the patients were trache-
ally extubated after gentle oropharyngeal suction and then
transferred to the postanesthesia care unit.
When arriving at the postanesthesia care unit (0 hour)
and thereafter at 2, 4, and 24 hours, patients were assessed
for the incidence and severity of sore throat by another
anesthesiologist blinded to the treatment. POST was
graded on a 4-point scale (0 –3): 0, no sore throat; 1, mild
sore throat (complaints of sore throat only when
prompted); 2, moderate sore throat (complaints of sore
throat volunteered by the patient without prompting); and
3, severe sore throat (change of voice or hoarseness, asso-
ciated with throat pain).11 Total POST cases were those
patients who reported any degree of sore throat over the
24-hour evaluation period. We recorded the patient’s age,
sex, smoking history, weight, height, duration of surgery,
888 www.anesthesia-analgesia.org
hoarseness, nausea and vomiting, and the type of postop-
erative analgesia. Potential side effects associated with the
study drugs, such as local numbness, burning, and/or
stinging sensation, cough, and dry mouth, were also re-
corded during spraying and postoperatively by partici-
pants’ self report.
Before initiation of the study, a power calculation was
performed to determine the required sample size based on
our institute’s previous data. A minimum of 91 patients
was required in each group to detect a decreased incidence
of POST from 40% to 20% with a power of 80% and a
significance level of 0.05. To compensate for potential
dropouts, we enrolled 95 patients in each group. Statistical
analysis was performed using SPSS for Windows version
14 (SPSS, Chicago, IL). Data are expressed as mean (SD),
and median with range or percentage. We used a 1-way
analysis of variance to compare patients’ ages, heights,
weights, and durations of surgery among groups. The
overall incidence of POST and side effects among groups
were tested using ␹2 tests. For those significant variables in
the ␹2 test, we recalculated all possible six 2 ⫻ 2 ␹2 tests by
applying the Bonferroni inequality to adjust ␣ level [i.e.,
P(␹2 ⬎ 6.97) ⫽ 0.05/6 ⫽ 0.0083 for 1 degree of freedom] to
avoid type I error. A Kruskal-Wallis test followed by the
Dunn procedure was applied to compare the differences in
the severity of POST among groups. We used univariable
and multivariable logistic regressions for evaluating the
interaction between 2 treatments (mucosa and cuff) and the
relative risk, as odds ratio (OR) and 95% confidence inter-
val (CI) of BH on POST. Probability values ⬍0.05 were
considered statistically significant.
RESULTS
The characteristics of the study groups are shown in Table
1; there were no significant differences among the 4 groups
in age, sex, height, weight, body mass index, and the
duration of anesthesia. Two patients in groups B and D
received unanticipated nasogastric tubes, and the statistical
analysis was performed without those patients.
Table 2 lists the incidence and severity of POST for the
study groups at 0, 2, 4, and 24 hours postoperatively.
Within the 24-hour period of evaluation, the overall inci-
dence of POST (patients with any POST during 24-hour
evaluation/patient numbers) in groups A, B, C, and D was
23.2%, 13.8%, 14.7%, and 40.4%, respectively, mostly at the
second hour after extubation. There were significantly
fewer incidences in groups B and C, but not in group A
compared with group D, for the total incidence of POST.
The severity of POST was significantly higher in group D
compared with groups B and C (P ⬍ 0.001).
No statistically significant differences of POST incidence
were found among patients who received BH spray only on
the ET cuff, only in the oropharyngeal cavity, or both,
which indicated no additive or synergistic effects of appli-
cation of BH (P ⫽ 0.088). Patients who received BH on the
ET cuff had a significantly lower risk of POST (OR: 0.36;
95% CI: 0.21– 0.60) compared with the placebo group.
Those who received BH in the oropharyngeal cavity also
had less risk of POST (OR: 0.62; 95% CI: 0.38 –1.01) com-
pared with the placebo group, although not significant.
After adjusting for potential confounders (age, gender,
ANESTHESIA & ANALGESIA
Table 1. Demographic Data and Duration of Surgery
Group

A (n ⴝ 95) B (n ⴝ 94) C (n ⴝ 95) D (n ⴝ 94) P value

Gender (male/female) 49/46 44/50 45/50 44/50 0.919
Age (y) 49.2 (18) 47.1 (15) 48.7 (16) 45.5 (18) 0.446
Height (cm) 163.7 (8.0) 162.1 (8.8) 162.9 (9.4) 162.4 (9.6) 0.631
Weight (kg) 63.2 (12.1) 62.9 (10.8) 62.0 (11.5) 62.7 (11.7) 0.897
BMI (kg/m2) 23.5 (3.5) 23.9 (3.6) 23.3 (3.4) 23.7 (3.9) 0.655
Duration of anesthesia (min) 175 (99) 186 (91) 176 (84) 175 (83) 0.917
Intraoperative fentanyl use 318 (65) 312 (57) 310 (60) 306 (60) 0.706
Smoking habit 35 40 38 42 0.964
Postoperative analgesia (PCA with 80/15 78/16 77/18 75/19 0.862
fentanyl/demerol, prn)
Type of surgery
Colon rectal surgery 8 6 5 6 0.850
General surgery 21 20 21 18 0.954
Genitourinary surgery 4 5 7 5 0.815
Gynecologic surgery 22 16 18 19 0.758
Ophthalmologic surgery 12 16 16 18 0.673
Orthopedic surgery 22 20 20 18 0.928
Plastic surgery 6 11 8 10 0.585
BMI ⫽ body mass index; PCA ⫽ patient-controlled analgesia; prn ⫽ pro re nata (when necessary).
Values are presented as mean (SD) or number.

Table 2. The Incidence and Severity of Postoperative Sore Throat for the Study Groups
Evaluation time Group A (n ⴝ 95) Group B (n ⴝ 94) Group C (n ⴝ 95) Group D (n ⴝ 94) P value
0h 6 (6.3%)* 4 (4.2%)* 5 (5.2%)* 20 (20.6%) ⬍0.001
Grading of discomfort 0.001
Mild 4 2 3 12
Moderate 2 2 2 8
Severe 0 0 0 0
2h 22 (23.2%) 13 (13.8%)* 14 (14.7%)* 38 (40.4%) ⬍0.001
Grading of discomfort ⬍0.001
Mild 13 5 6 9
Moderate 5 4 5 15
Severe 4 4 3 14
4h 12 (12.6%)* 8 (8.3%)* 10 (10.4%)* 30 (30.9%) ⬍0.001
Grading of discomfort 0.001
Mild 6 3 5 8
Moderate 2 1 2 8
Severe 4 4 3 14
24 h 10 (10.5%) 7 (7.3%)* 8 (8.3%)* 22 (22.7%) 0.003
Grading of discomfort 0.039
Mild 4 2 4 5
Moderate 2 1 1 3
Severe 4 4 3 14
Total incidence of POST 22 (23.2%) 13 (13.8%)* 14 (14.7%)* 38 (40.4%) ⬍0.001
Severity of POST 0 (0–3) 0 (0–3)* 0 (0–3)* 0 (0–3) ⬍0.001
POST ⫽ postoperative sore throat.
Values are presented as number of subjects (%) or median (range).
* P ⬍ 0.05, compared with group D. The incidence of POST was done by ␹2 test by applying the Bonferroni inequality to adjust ␣ level 关i.e., P(␹2 ⬎ 6.97) ⫽
0.05/6 ⫽ 0.0083 for 1 degree of freedom兴 to avoid type I error. The severity of POST was determined by Kruskal-Wallis test and pairwise posteriori comparisons
were done by Dunn procedure.

smoking, and duration of anesthesia), the association be-
tween treatments for the risk of developing POST did not
change substantially (Table 3).
The incidence of local numbness, burning, or stinging
sensation in groups A, B, C, and D was 8.42%, 10.6%, 3.19%,
and 1.06%, respectively. Only group B had a significantly
higher incidence of local numbness, burning, or stinging
sensation than group D (P ⬍ 0.05). There were 4, 4, 3, and
5 patients who had hoarseness in groups A, B, C, and D,
respectively. There was no significant difference in the
potential side effects relevant to BH, such as nausea or
vomiting, cough, and mouth dryness (Table 4).
DISCUSSION
Our major finding was that spraying BH on the ET cuff may
reduce the incidence and severity of POST up to 24 hours
postoperatively compared with the application of distilled
water. Moreover, spraying BH in the oropharyngeal cavity
did not reduce the incidence and severity of POST. Indeed,
there was no additional advantage from spraying BH on the

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 889

Topical Benzydamine Hydrochloride Attenuates Postoperative Sore Throat

into the oropharyngeal cavity before induction of anesthe-

Table 3. Odds Ratios and 95% Confidence sia may ameliorate the local irritation of BH.
Intervals of Benzydamine Hydrochloride on
Postoperative Sore Throat at 2 Hours The first limitation in our study is the pharmacokinetics
After Surgery and safety of BH on the tracheal mucosa. BH is well absorbed
No. of sore
through the oral mucosa, and its effects last for 1.84 hours.15 In
throats/no. Univariable Multivariablea our study, BH may be well absorbed after application to the
of patients OR (95% CI) OR (95% CI) trachea because of its thinner epithelial layer. However,
BH on mucosa concern for damage to the tracheal structure should be
Yes 36/189 0.63 (0.39–1.03) 0.65 (0.40–1.06)
considered, even though we did not find any adverse event in
No 51/189 1.00 (reference) 1.00 (reference)
BH on cuff our limited data. Silvestrini et al.16 had reported that gross
Yes 27/189 0.36 (0.22–0.60)* 0.36 (0.22–0.60)* and histological structure of the major organs such as lung,
No 60/189 1.00 (reference) 1.00 (reference) liver, kidney, and spleen were not altered by benzydamine in
OR ⫽ odds ratio; CI ⫽ confidence interval; BH ⫽ benzydamine hydrochloride. an animal study. However, no relevant human study regard-
a
Adjusted for gender, age (in years), smoking, and anesthetic time. ing a histological examination of the trachea has been re-
* P ⬍ 0.05 compared with the reference group.
ported. A second limitation is that we did not ensure that each
group did in fact have an equal amount of BH, although we
Table 4. The Local and Systemic Side Effects of limited the dose to 5 sprays each on either the ET or into the
Benzydamine Hydrochloride in the Study Groups oropharyngeal cavity, and an equal dosage was prescribed. A
Group A Group B Group C Group D third limitation of our randomized study is that postoperative
(n ⴝ 95) (n ⴝ 94) (n ⴝ 95) (n ⴝ 94) pain management can influence POST, although there was no
Nausea 20 25 22 25 significant difference in postoperative pain management. At
Vomiting 18 16 24 20 the same time, dry oxygen and IV lidocaine would affect
Cough 19 22 20 18
Local stinging or 8 10* 3 1
POST; however, we used these routinely and standardized
numbness of the the dosage for each patient.
throat and mouth In conclusion, BH topical spray on the ET cuff before
Dry mouth 42 46 45 48 intubation and GA can reduce the incidence and severity of
Hoarseness 4 4 3 5
POST without increased BH-related adverse effects.
Values are presented as number.
* P ⬍ 0.05 compared with group D. ␹2 test and applying the Bonferroni
inequality to adjust ␣ level 关i.e., P(␹2 ⬎ 6.97) ⫽ 0.05/6 ⫽ 0.0083 for 1 REFERENCES
degree of freedom兴 to avoid type I error. 1. Macario A, Weinger M, Carney S, Kim A. Which clinical
anesthesia outcomes are important to avoid? The perspective
of patients. Anesth Analg 1999;89:652– 8
cuff and into the oropharyngeal cavity simultaneously for 2. Al-Qahtani AS, Messahel FM. Quality improvement in anes-
POST prevention, even though there was a decreased trend of thetic practice: incidence of sore throat after using small
POST incidence when spraying BH on the ET cuff compared tracheal tube. Middle East J Anesthesiol 2005;18:179 – 83
with into the oropharyngeal cavity. Our results indirectly 3. Gulhas N, Canpolat H, Cicek M, Yologlu S, Togal T, Durmus
M, Ozcan Ersoy M. Dexpanthenol pastille and benzydamine
demonstrated that mucosal irritation occurring at the level of hydrochloride spray for the prevention of post-operative sore
the ET cuff is the most important causative factor for tracheal throat. Acta Anaesthesiol Scand 2007;51:239 – 43
morbidity, and this outcome was consistent with previous 4. Park SH, Han SH, Do SH, Kim JW, Rhee KY, Kim JH.
studies.13,14 Prophylactic dexamethasone decreases the incidence of sore
Local spraying of BH can resolve moderate to severe sore throat and hoarseness after tracheal extubation with a double-
lumen endobronchial tube. Anesth Analg 2008;107:1814 – 8
throat after ET intubation.11,15 Preventive topical application 5. Sumathi PA, Shenoy T, Ambareesha M, Krishna HM. Controlled
of BH into the oropharyngeal cavity before intubation and comparison between betamethasone gel and lidocaine jelly ap-
continuous use for 48 hours after the operation effectively plied over tracheal tube to reduce postoperative sore throat,
decreased the incidence and severity of POST after ET and cough, and hoarseness of voice. Br J Anaesth 2008;100:215– 8
also laryngeal mask airway insertion.3,8 In this study, we 6. Ogata J, Minami K, Horishita T, Shiraishi M, Okamoto T,
Terada T, Sata T. Gargling with sodium azulene sulfonate
showed that spraying BH into the oropharyngeal cavity did
reduces the postoperative sore throat after intubation of the
not reduce the incidence and severity of POST. This might be trachea. Anesth Analg 2005;101:290 –3
because of the small dosage of BH used in the oropharyngeal 7. Agarwal A, Nath SS, Goswami D, Gupta D, Dhiraaj S, Singh
cavity (0.75 mg) in our study compared with previous studies PK. An evaluation of the efficacy of aspirin and benzydamine
(which ranged from 2.16 to 22.5 mg).3,7 hydrochloride gargle for attenuating postoperative sore throat:
There are many side effects (64%) of topical use of BH, a prospective, randomized, single-blind study. Anesth Analg
2006;103:1001–3
including local numbness, burning or stinging sensation, 8. Kati I, Tekin M, Silay E, Huseyinoglu UA, Yildiz H. Does benzy-
nausea or vomiting, cough, dry mouth, thirst, throat dis- damine hydrochloride applied preemptively reduce sore throat
comfort, drowsiness, and headache.15 Kati et al.8 and due to laryngeal mask airway? Anesth Analg 2004;99:710 –2
Agarwal et al.7 reported local numbness, dysgeusia, and 9. Hung NK, Wu CT, Chan SM, Lu CH, Huang YS, Yeh CC, Lee
throat irritation in BH-pretreated patients. In our current MS, Cherng CH. Effect on postoperative sore throat of spray-
ing the endotracheal tube cuff with benzydamine hydrochlo-
study, we found that 8% to 10% of patients experienced
ride, 10% lidocaine, and 2% lidocaine. Anesth Analg 2010 Mar
these side effects after topical application of BH before 19 [Epub ahead of print]
anesthesia induction. To avoid these drawbacks, we sug- 10. Quane PA, Graham GG, Ziegler JB. Pharmacology of benzy-
gest that the application of BH onto the ET cuff instead of damine. Inflammopharmacology 1998;6:95–107

890 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

11. Turnbull RS. Benzydamine hydrochloride (Tantum) in the
management of oral inflammatory conditions. J Can Dent
Assoc 1995;61:127–34
12. Epstein JB, Silverman S Jr, Paggiarino DA, Crockett S, Schubert
MM, Senzer NN, Lockhart PB, Gallagher MJ, Peterson DE,
Leveque FG. Benzydamine HCl for prophylaxis of radiation-
induced oral mucositis: results from a multicenter, random-
ized, double-blind, placebo-controlled clinical trial. Cancer
2001;92:875– 85
13. Combes X, Schauvliege F, Peyrouset O, Motamed C, Kirov K,
Dhonneur G, Duvaldestin P. Intracuff pressure and tracheal
morbidity: influence of filling with saline during nitrous oxide
anesthesia. Anesthesiology 2001;95:1120 – 4
October 2010 • Volume 111 • Number 4
14. Peppard SB, Dickens JH. Laryngeal injury following
short-term intubation. Ann Otol Rhinol Laryngol 1983;92:
327–30
15. Passali D, Volonte M, Passali GC, Damiani V, Bellussi L.
Efficacy and safety of ketoprofen lysine salt mouthwash versus
benzydamine hydrochloride mouthwash in acute pharyngeal
inflammation: a randomized, single-blind study. Clin Ther
2001;23:1508 –18
16. Silvestrini B, Barcellona PS, Garau A, Catanese B. Toxicology of
benzydamine. Toxicol Appl Pharmacol 1967;10:148 –59
www.anesthesia-analgesia.org 891
Strepsils姞 Tablets Reduce Sore Throat and
Hoarseness After Tracheal Intubation
Amin Ebneshahidi, MD, and Masood Mohseni, MD

BACKGROUND: Amyl-m-cresol (Strepsils姞) has been successfully used in the prophylaxis and
treatment of oral inflammations, but its effects on postintubation sore throat and hoarseness are
unknown. We conducted this study to evaluate the effects of Strepsils in reducing postintubation
sore throat and hoarseness.
METHODS: One hundred fifty patients, ASA physical status I to II, scheduled to undergo general
anesthesia and elective orthopedic or gynecologic surgery were enrolled. Participants were
randomly allocated to receive either Strepsils or identical-looking placebo tablets immediately
before arrival to the operating room. The incidence and severity of postoperative sore throat and
hoarseness were evaluated immediately and 24 hours after surgery.
RESULTS: The incidence of early postoperative sore throat was 13.7% and 33.3% and
hoarseness was 12.3% and 26.4% in the Strepsils and placebo groups, respectively (P ⬍ 0.05).
One day after surgery, the incidence of sore throat decreased to 6.8% and 18.1% in the Strepsils
and control groups, respectively. The incidence of hoarseness 1 day after the operation
decreased to 8.2% in the Strepsils group and 19.4% in the placebo group, but the difference
remained statistically significant (P ⬍ 0.05).
CONCLUSION: Perioperative use of Strepsils tablets reduces postoperative sore throat and
hoarseness of voice. (Anesth Analg 2010;111:892–4)

P ostoperative sore throat and hoarseness are minor but

frequent complications of endotracheal intubation that
occur in up to 50% of patients.1,2 These complications
negatively influence patient satisfaction and occasionally re-
quire treatment with supplementary analgesics. Several drugs
such as clonidine,3 betamethasone gel,4 and chamomile-
extract spray5 have been used to reduce postoperative sore
throat and hoarseness with varying efficacy. A simple, safe,
and inexpensive perioperative intervention to prevent post-
operative sore throat and hoarseness would be useful.
Amyl-m-cresol (Strepsils威) has been successfully used in the
treatment of oral inflammation and for the prevention of pain
and inflammation after oral surgery,6,7 but the effects on postin-
tubation sore throat and hoarseness are unknown. This study
was conducted to evaluate whether the perioperative use of
Strepsils lozenges would reduce the incidence of postoperative
sore throat and hoarseness compared with placebo.
METHODS
Patients
This prospective study was approved by the ethics com-
mittee of Sadi Hospital, and written informed consent was
obtained from all patients. One hundred fifty ASA physical
status I to II patients, aged 19 to 63 years, undergoing
elective orthopedic or gynecologic surgery under general
anesthesia who were expected to remain in the hospital for
⬎24 hours were enrolled. Patients with significant sore
throat for any reason or obvious hoarseness were not
included. Participants were randomly allocated to receive
From the Department of Anesthesiology, Sadi Hospital, Isfahan, Iran.
Accepted for publication November 29, 2009.
Supported by departmental funds.
Address correspondence and reprint requests to Amin Ebneshahidi, MD,
Department of Anesthesiology, Sadi Hospital, Khodaverdi Alley, Sadi Blv,
Isfahan, Iran. Address e-mail to amin.ebneshahidi@gmail.com.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181d00c60
either Strepsils or identical-looking placebo tablets imme-
diately before arrival to the operating room (45 minutes
before induction of anesthesia on average). A blinded nurse
administered the tablets to the patients.
Strepsils honey and lemon lozenge (Boots, Nottingham, UK)
contained the active ingredients 2,4-dichlorobenzyl alcohol 1.2
mg, amylmetacresol 0.6 mg, sucrose, glucose syrup, honey,
tartaric acid, peppermint oil, terpeneless lemon oil, and quinoline
yellow (E104). The placebo tablets (Anata, Tabriz, Iran) contained
sugar, glucose, citric acid, lemon flavor, and color (E104). The
lozenges were not easily distinguishable from placebo tablets
with regard to color, taste, and smell. The randomization was
performed by the hospital pharmacy using a table of random
numbers, and the patients, nurses, and anesthetic technician who
were involved in the patients’ care and data recording were
blinded to the nature of the assignment.
Method of Anesthesia
All patients were premedicated with oral oxazepam 10 mg
and ranitidine 150 mg 2 hours before surgery. Fentanyl 3 to
4 ␮g 䡠 kg⫺1 and IV lidocaine 1.5 mg 䡠 kg⫺1 were adminis-
tered 3 to 5 minutes before tracheal intubation. After the
administration of 100% oxygen at 5 L min⫺1 for several
minutes, anesthesia was induced with propofol 1.5 to
2 mg 䡠 kg⫺1 and atracurium (0.5 mg 䡠 kg⫺1). Laryngoscopy
was performed 3 to 5 minutes after atracurium injection
when adequate neuromuscular blockade was confirmed
using train-of-four nerve stimulation. A train-of-four ⬍10%
was used as an indication of satisfactory neuromuscular
blockade for intubation. Difficulty in laryngoscopy was
graded as 1, no difficulty; 2, only posterior extremity of
glottis visible; 3, only epiglottis seen; and 4, no recognizable
structures. Endotracheal intubation was performed using a
polyvinyl chloride endotracheal tube (Supa, Tehran, Iran)
with 7- to 8-mm internal diameter for women and 7.5- to
8.5-mm internal diameter for men. The high-volume, low-
pressure cuff was inflated until no air leak could be heard
with peak airway pressure at 20 cm H2O. Cricoid pressure

892 www.anesthesia-analgesia.org October 2010 • Volume 111 • Number 4

was not applied for intubation. Patients requiring nasogas-
tric tube placement or those in whom laryngoscopy was Table 1. Group Baseline Characteristics
attempted more than once were withdrawn.
and Covariates
Strepsils Placebo
General anesthesia was maintained with propofol, fentanyl, Variables (n ⴝ 73) (n ⴝ 72)
and atracurium. Controlled mechanical ventilation with an initial Age (y)a 29 (9) 31 (10)
tidal volume of 10 mL 䡠 kg⫺1 and respiratory frequency of 10 Male sexb 43 (58.9) 40 (55.5)
breaths 䡠 min⫺1 was adjusted to maintain normocapnia. Weight (kg)a 58 (9) 57 (11)
Duration of laryngoscopy (s)a 13.7 (3.1) 14.3 (3.6)
Neuromuscular blockade was reversed with neostig- Duration of intubation (min)a 149 (44) 158 (51)
mine 0.05 mg 䡠 kg⫺1 and atropine 0.02 mg 䡠 kg⫺1. Grade of laryngoscopic viewc 2 (1,2) 2 (1,2)
When spontaneous ventilation was adequate and the Type of surgeryb
patients were able to follow verbal commands, tracheal Myomectomy/hysterectomy 9 (12.3) 7 (9.7)
Ovarectomy 4 (5.5) 5 (6.9)
extubation was performed immediately after suctioning of Diagnostic laparoscopy 10 (13.7) 12 (16.6)
the oropharynx at the discretion of the responsible anesthe- Bone tumor resection 7 (9.5) 8 (11.1)
siologist. An oral airway was inserted before extubation Total hip replacement 4 (5.5) 3 (4.2)
Traumatic fractures 31 (42.4) 29 (40.2)
and emergence in all patients. Shoulder surgery/arthroscopy 4 (5.5) 3 (4.2)
Bone grafting 4 (5.5) 5 (6.9)
Outcome Measures Positionb
Postoperative sore throat and hoarseness were evaluated Supine 55 (75.3) 52 (72.2)
within 20 minutes in the recovery room and 24 hours after Lateral decubitus 14 (19.2) 17 (23.6)
Sitting 4 (5.5) 3 (4.2)
surgery. At the time of the first evaluation, patients with a
a
Ramsay Sedation Score8 of 2 (cooperative, oriented, and b
Data are presented as mean (SD).
Data are presented as n (%).
tranquil) or 3 (responding to commands only) were in- c
Data are presented as median (25%, 75%).
cluded. The severity of sore throat was graded as follows: 0,
no sore throat; 1, minimal; 2, moderate; 3 severe; and for
hoarseness, the grading was as follows: 0, no hoarseness; 1,
slight hoarseness; 2, severe hoarseness; 3, cannot speak be-
cause of hoarseness. Additional analgesics were not adminis-
tered until completion of the first evaluation. All scales were
completed by a nurse blinded to the study groups.
Demographic data, perioperative steroid use, type of
surgery, the time needed for laryngoscopy (time from
opening the mouth to placement of the endotracheal tube),
intubation period (time from intubation to extubation), and
grades of laryngoscopic view as well as the incidence of
bucking were recorded by a blinded anesthetic technician. Figure 1. Incidence of sore throat and hoarseness at recovery from
Patients’ discomfort after lozenge administration including anesthesia and 24 hours postoperatively in the Strepsils (white) and
subjective symptoms of allergy to lozenges, bad taste or placebo (gray) groups. Data shown as percentages; P ⬍ 0.05 with ␹2 test.
smell, and nausea were also recorded by the same techni-
cian before induction of anesthesia.
room. The distribution of surgical procedures and positions
Statistical Analysis are shown in Table 1. Patient characteristics, grade of diffi-
Results are presented as either mean (SD) or percentages, as culty of laryngoscopy, and the duration of laryngoscopy
appropriate. The incidences of sore throat and hoarseness (13.7% vs 33.3%, P ⫽ 0.81) and intubation (12.3% vs 26.4%,
were compared between the 2 groups using the Fisher exact P ⫽ 0.14) were comparable between the 2 groups (Table 1).
test. The scores of sore throat and hoarseness as well as grade None of the patients complained of any kind of discomfort
of laryngoscopic view were compared between the 2 groups after using either Strepsils or placebo. The incidence of early
with the Mann-Whitney U test. The duration of laryngoscopy postoperative sore throat in patients who received Strepsils
and intubation period were compared between groups by an was about one-third compared with the placebo group (P ⫽
independent 2-sample t test. A nomogram based on estimated 0.003). Likewise, hoarseness was reported less frequently in
standardized difference was used to calculate the required the Strepsils than placebo group (P ⫽ 0.04). As shown in
sample size. With a power of 80% and ␣ level of 0.05 for Figure 1, the incidence of both sore throat and hoarseness
2-tailed statistical analysis, and estimated incidence of sore decreased at the 24-hour assessment after surgery, but the
throat of 0.351,9 and 0.10 in the 2 groups, a sample size of 70 differences between the 2 groups remained statistically sig-
patients for each group was calculated as being appropriate. nificant (P ⫽ 0.04). Further analysis showed that the mean ⫾
Correlation coefficient between grade of laryngoscopic view SD severity scores of early sore throat (0.52 ⫾ 0.85 vs 0.20 ⫾
and sore throat/hoarseness score in either group was assessed 0.57), late sore throat (0.22 ⫾ 0.50 vs 0.08 ⫾ 0.32), early
with Spearman analysis. Statistical analysis was performed hoarseness (0.36 ⫾ 0.65 vs 0.16 ⫾ 0.47), and late hoarseness
with SPSS version 11.0 software (SPSS, Chicago, IL). (0.22 ⫾ 0.45 vs 0.08 ⫾ 0.27) in the placebo group were
significantly higher than in the Strepsils group (P ⬍ 0.05).
RESULTS None of the patients complained of grade 3 hoarseness
Of 150 included participants, 5 patients were excluded be- (unable to speak because of hoarseness). Only 1 patient in
cause of unfavorable Ramsay Sedation Scores in the recovery the Strepsils group and 3 patients in the placebo group had

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 893

Strepsils威 Reduces Sore Throat and Hoarseness
severe sore throat. Pearson correlation coefficient analyses
between grade of laryngoscopic view and early sore throat
(r ⫽ 0.13, P ⬎ 0.05), 24-hour sore throat (r ⫽ 0.10, P ⬎ 0.05),
early hoarseness (r ⫽ 0.16, P ⬎ 0.05), and 24-hour hoarse-
ness (r ⫽ 0.10, P ⬎ 0.05) showed no statistically significant
results. Reanalysis of the data after the exclusion of 6
patients with perioperative steroid use yielded similar
results in all sets of analysis (data not shown).
DISCUSSION
Sore throat and hoarseness are minor but common postop-
erative complaints with an estimated incidence of 14.4% to
90% for sore throat1,9 and 10% to 50.1% for hoarseness.1,10
In this study, the overall incidence of sore throat and
hoarseness in the immediate postoperative period was
20.8% and 18.6%, respectively. Both complications were
shown to be significantly reduced with the use of Strepsils.
Likewise, 1-day follow-up of patients confirmed the effi-
cacy of Strepsils tablets for the prevention of postoperative
sore throat and hoarseness.
Several causal factors for sore throat and hoarseness
after intubation have been reported, including sex, large
tracheal tube size, cuff design,11,12 increased intracuff pres-
sure by nitrous oxide,13 use of succinylcholine, and pro-
longed laryngoscopy.14 In this study, patient characteristics
and duration of laryngoscopy and surgery were compa-
rable between the 2 groups. We excluded the potential
effects of difficult intubation on postoperative throat com-
plications in our series and had a standardized protocol for
induction, tracheal intubation, and extubation of patients.
Thus, it seems reasonable to conclude that the observed
difference in the outcome measures between the 2 groups
can be safely attributed to the favorable effects of Strepsils.
Prior studies have demonstrated that mucosal damage
can occur even after uneventful intubation for routine
surgery.12 Because of the potential role of inflammation in
the causation of tracheal morbidities, earlier investigators
have suggested the use of inhaled and topical steroids.15–17
Gáspár et al.6 used Strepsils for the treatment of 22 patients
with oral inflammatory diseases and preoperatively in 20
oral surgery cases. In this preliminary study, they reported
that Strepsils may be effective in the prophylaxis and
treatment of oral inflammation. In a complementary survey of
272 patients with either oral inflammatory diseases or treated
prophylactically for possible oral inflammation or infection,
Gáspár et al.7 found that healing was shortened by 30% and
pain and functional assessment were improved by 30% in
patients treated with Strepsils compared with control. Strep-
sils tablets were well tolerated by all patients. Taken together,
these data suggest that Strepsils is an effective antiinflamma-
tory choice for mucosal damage in the orotracheal cavity. In
this study, we demonstrated that postoperative sore throat
and hoarseness were reduced by using Strepsils lozenges.
Although we did not evaluate the effect of Strepsils on the
potential role of the inflammatory process in the generation of
these adverse effects, it is likely that this is its mechanism of
action based on the above literature.
Study Limitations
Responsiveness of patients in the immediate postoperative
period may be questioned. Although we excluded patients
894 www.anesthesia-analgesia.org
with inappropriate sedation scores, fentanyl or propofol still
circulating in the bloodstream immediately after surgery could
have affected their ability to accurately perceive their throat
symptoms. However, standardized protocols for intubation and
extubation of patients along with similar results in the later
(24-hour) evaluation of patients support our initial findings.
One limitation of this study is that we did not evaluate patient
satisfaction, which could be an important indicator of the efficacy
of our intervention. Finally, we did not evaluate the best timing
for the administration of Strepsils and the effects of repeated use,
especially during the day after surgery.
In conclusion, the perioperative use of Strepsils lozenges
may help eliminate sore throat and hoarseness after inpa-
tient surgery. The best timing of lozenge administration
and the efficacy of repeated use in patients with remaining
symptoms should be evaluated in further studies.
REFERENCES
1. Maruyama K, Sakai H, Miyazawa H, Toda N, Iinuma Y,
Mochizuki N, Hara K, Otagiri T. Sore throat and hoarseness
after total intravenous anaesthesia. Br J Anaesth 2004;92:541–3
2. Higgins PP, Chung F, Mezei G. Postoperative sore throat after
ambulatory surgery. Br J Anaesth 2002;58:582– 4
3. Maruyama K, Yamada T, Hara K. Effect of clonidine premedi-
cation on postoperative sore throat and hoarseness after total
intravenous anesthesia. J Anesth 2006;20:327–30
4. Kazemi A, Amini A. The effect of betamethasone gel in
reducing sore throat, cough, and hoarseness after laryngo-
tracheal intubation. Middle East J Anesthesiol 2007;19:197–204
5. Kyokong O, Charuluxananan S, Muangmingsuk V, Rodanant
O, Subornsug K, Punyasang W. Efficacy of chamomile-extract
spray for prevention of post-operative sore throat. J Med Assoc
Thai 2002;85(suppl 1):S180 –5
6. Gáspár L, Turi J, Toth BZ, Suri C, Vago P. [The use of benzyl alcohol
and amyl-m-cresol (Strepsils) in the oral cavity: review of the
literature and first clinical experiences]. Fogorv Sz 1998;91:143–50
7. Gáspár L, Szmrtyka A, Turi J, Tóth BZ, Suri C, Vágó P, Sefer A, al
Haj C. [Clinical experience with the use of benzylalcohol and
amyl-m-cresol (Strepsils) in stomatological diseases]. Fogorv Sz
2000;93:83–90
8. Ramsay MA, Savege TM, Simpson BR, Goodwin R. Controlled
sedation with alphaxalone-alphadolone. Br Med J 1974;22:656 – 65
9. Lev R, Rosen P. Prophylactic lidocaine use preintubation: a
review. J Emerg Med 1994;4:499 –506
10. Christensen AM, Willemoes-Larsen H, Lundby L, Jakobsen
KB. Postoperative throat complaints after tracheal intubation.
Br J Anaesth 1994;73:786 –7
11. Jensen PJ, Hommelgaard P, Sondergaard P, Eriksen S. Sore
throat after operation: influence of tracheal intubation, intra-
cuff pressure and type of cuff. Br J Anaesth 1982;54:453–7
12. McHardy FE, Chung F. Postoperative sore throat: cause, pre-
vention and treatment. Anaesthesia 1999;54:444 –53
13. Combes X, Schauvliege F, Peyrouset O, Motamed C, Kirov K,
Dhonneur G, Duvaldestin P. Intracuff pressure and tracheal
morbidity: influence of filling with saline during nitrous oxide
anesthesia. Anesthesiology 2001;95:1120 – 4
14. Higgins PP, Chung F, Mezei G. Postoperative sore throat after
ambulatory surgery. Br J Anaesth 2002;88:582– 4
15. Sumathi PA, Shenoy T, Ambareesha M, Krishna HM. Controlled
comparison between betamethasone gel and lidocaine jelly ap-
plied over tracheal tube to reduce postoperative sore throat,
cough, and hoarseness of voice. Br J Anaesth 2008;100:215– 8
16. Ayoub MC, Ghobashy A, McGrimley L, Koch ME, Qadir S,
Silverman DG. Wide spread application of topical steroids to
decrease sore throat, hoarseness and cough after tracheal
intubation. Anesth Analg 1998;87:714 – 6
17. el-Hakim M. Beclomethasone prevents postoperative sore
throat. Acta Anaesthesiol Scand 1993;37:250 –2
ANESTHESIA & ANALGESIA
Inhaled Fluticasone Propionate Reduces
Postoperative Sore Throat, Cough, and Hoarseness
Nasrin Faridi Tazeh-kand, MD, Bita Eslami, MPH, and Khadijeh Mohammadian, RN

BACKGROUND: Sore throat is a common complication after surgery. Postoperative cough and
hoarseness can also be distressing to patients. We sought to determine the effect of an inhaler
steroid on sore throat, cough, and hoarseness during the first 24 hours of the postoperative
period.
METHODS: We enrolled 120 women with ASA physical status I or II and term singleton pregnancy
who were scheduled for elective cesarean delivery under general anesthesia. Patients were
randomized into 2 groups: in the sitting position, group F patients received 500 ␮g inhaled
fluticasone propionate via a spacer device during 2 deep inspirations, after arrival in the operating
room, and group C had no treatment. The patients were interviewed by a blinded investigator for
postoperative sore throat, cough, and hoarseness at 1 and 24 hours after surgery.
RESULTS: There were no significant differences in age, height, weight, body mass index,
duration of surgery, intubation, and grade of laryngeal exposure between the 2 groups. The
incidence of sore throat, cough, and hoarseness was significantly lower in group F (3.33%,
3.33%, and 3.33%) compared with the control group (36.67%, 18.33%, and 35%) (P ⬍ 0.05 for
all comparisons), not only in the first postoperative hour but also 24 hours after surgery (13.33%,
13.33%, and 25% in group F vs 40%, 41.67%, and 50% in the control group). The incidence of
moderate and severe hoarseness in group F at the first hour was significantly less than the
control group (P ⬍ 0.05).
CONCLUSIONS: Inhaled fluticasone propionate decreases the incidence and severity of postop-
erative sore throat, cough, and hoarseness in patients undergoing cesarean delivery under
general anesthesia. (Anesth Analg 2010;111:895–8)

S ore throat is a common complication after surgery.

Postoperative cough and hoarseness also can be dis-
tressing to patients and affect patient satisfaction.
Many factors can contribute to postoperative sore throat,
and the incidence has been found to vary with the method
of airway management.1 The incidence is the highest after
tracheal intubation (45.4%), whereas after laryngeal mask
airway use the incidence is lower (17.5%) and much less
(3.3%) when a facemask is used for the maintenance of
anesthesia.1–3 Female sex increases the incidence and sever-
ity of postoperative sore throat.4
We sought to determine the effect of an inhaler steroid
on sore throat, cough, and hoarseness during the first 24
postoperative hours.
METHODS
After receiving hospital ethics committee approval and
informed consent from subjects, we enrolled 120 women
with ASA physical status I or II and term singleton preg-
nancies who were scheduled for elective cesarean delivery
under general anesthesia. The study was conducted in a
prospective, randomized, and single-blinded manner, from
August to December 2008. Patients with a history of
perioperative sore throat and asthma, Mallampati grade
⬎2, recent nonsteroidal antiinflammatory drug use, weight
From the Department of Anesthesiology, Roointan-Arash Hospital, Tehran
University of Medical Sciences, Tehran, Iran.
Accepted for publication October 26, 2009.
Address correspondence and reprint requests to Nasrin Faridi Tazeh-kand,
MD, Roointan-Arash Hospital, Rashid Ave., Tehranpars, Tehran, Iran.
Address e-mail to nfaridi@sina.tums.ac.ir.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181c8a5a2
⬎115 kg, diabetes mellitus, pregnancy-induced hyperten-
sion, ⬎2 attempts at intubation, and patients receiving steroid
therapy were excluded from the study. Patients were random-
ized into the 2 groups with the help of a computer-generated
table of random numbers.
In the sitting position, group F received 500 ␮g inhaled
fluticasone propionate (Flixotide™ Evohaler™ 250 ␮g, Glaxo
Wellcome Production, Evreux, France) via a spacer device
during 2 deep inspirations, after their arrival in the oper-
ating room, and group control (C) received no treatment. In
group F, the inhaler was given by an anesthesia nurse. All
patients were given the impression they would receive
medication to reduce the incidence of sore throat. They
were told that some of them would be randomized to
receive the medication by inhalation (F group) and some by
injection (C group). Those randomized to group C did not
receive any active drug.
Standard monitoring included noninvasive arterial
blood pressure, pulse oximetry, electrocardiogram, and
end-tidal carbon dioxide. Uterine displacement was
achieved by tilting the operating table to the left. We
inserted a wide-bore IV catheter into a forearm vein and
started a slow infusion of Ringer solution. In all patients,
after 4 minutes of administration of oxygen, rapid sequence
induction with cricoid pressure was achieved using thio-
pental 5 mg/kg and succinylcholine 1.5 mg/kg. The tra-
chea was intubated with a soft seal cuffed sterile polyvinyl
chloride endotracheal tube with a standard cuff (Supa
Medical Devices, Tehran, Iran) and an internal diameter of
7 mm. The tracheal tube cuff was inflated until no air
leakage could be heard with a peak airway pressure at 20
cm H2O. The degree of laryngeal exposure was ranked
from complete visualization of the vocal cords (grade I) to

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 895

Postoperative Sore Throat and Fluticasone

Table 1. Scoring System for Sore Throat, Cough, Table 2. Characteristics of Study Population
and Hoarseness Fluticasone
Score propionate Control
Sore throat group group
0 No sore throat (n ⴝ 60) (n ⴝ 60) P
1 Mild (less than a common cold) Age (y) 26.32 ⫾ 5.18 26.70 ⫾ 5.26 NS
2 Moderate (similar to a common cold) Height (cm) 162.03 ⫾ 4.90 161.39 ⫾ 4.99 NS
3 Severe (more than a common cold) Weight (kg) 79.18 ⫾ 10.91 76.79 ⫾ 12.16 NS
Cough Body mass index 30.09 ⫾ 3.41 29.42 ⫾ 4.13 NS
0 No cough (kg/m2)
1 Mild (less than a common cold) Duration of surgery 39.92 ⫾ 8.36 40.67 ⫾ 11.55 NS
2 Moderate (similar to a common cold) (min)
3 Severe (more than a common cold) Duration of tracheal 53.83 ⫾ 8.65 54 ⫾ 11.96 NS
Hoarseness intubation (min)
0 No hoarseness Grade of laryngeal
1 Mild (no hoarseness at the time of interview but had exposure
it previously) I 43 (71.67) 43 (71.67)
2 Moderate (is only perceived by the patient) II 15 (25) 17 (28.33)
3 Severe (recognizable at the time of interview) III 2 (3.33) 0 (0)
V 0 (0) 0 (0) NS
NS ⫽ not significant.
a partial view of the vocal cords (grade II), epiglottis only
(grade III), and the inability to view even the epiglottis in group F had moderate or severe (grades 2 and 3) sore
(grade IV). throat, cough, or hoarseness. However, in group C, mod-
Anesthesia was maintained with 50% nitrous oxide in erate and severe symptoms were reported in 3 patients for
oxygen and 0.5 minimum alveolar concentration of halo- sore throat, 3 for cough, and 5 for hoarseness. The incidence
thane. Atracurium was given as required for further muscle of combined moderate and severe hoarseness in group F in
relaxation. After delivery and clamping of the umbilical the first hour was significantly lower than in group C (P ⬍
cord, we administered 2 ␮g/kg fentanyl and 0.02 mg/kg 0.05). Meanwhile, the evaluation of complications 24 hours
midazolam and 10 IU oxytocin IV. At the end of surgery, after surgery showed that in group F, only 2 patients had
oxygen 100% was administered, and residual neuromuscu- moderate or severe sore throat, and 2 patients had moder-
lar block was antagonized using neostigmine and atropine. ate or severe hoarseness. None of the patients in group F
Oral suctioning was done just before extubation only. The had moderate or severe cough after 24 hours. However,
trachea was extubated after deflating the cuff when the moderate and severe complications were common in group
patient was fully awake and was following commands. All C and when combined, moderate and severe complications
patients received oxygen by a facemask after surgery. The were significantly more frequent than in group F (P ⬍ 0.05).
anesthesiologist intubating and providing care did not
know whether a patient had been allocated to either the F DISCUSSION
or C group. The patients were interviewed by a blinded We found that the incidence of postoperative sore throat,
investigator for postoperative sore throat, cough, and cough, and hoarseness was significantly less when inhaled
hoarseness at 1 and 24 hours after surgery, using the fluticasone (500 ␮g) was administered compared with no
questionnaire based on the scoring system in Table 1. treatment. Some studies found the incidence of postopera-
Based on the results of a pilot study with 30 patients in tive sore throat, cough, and hoarseness to be as high as 6.6%
each group that showed an incidence of problems (sore to 90%. The results of our study showed that the incidence
throat, cough, and hoarseness) of 56% in group F and 83% of these problems in group C was higher than in group F
in group C, we calculated that 60 patients would be (40% vs 13.3% for cough; 41.67% vs 13.3% for sore throat;
required in each group to detect a difference in the inci- and 50% vs 25% for hoarseness) after 24 hours. Many
dence with a power of 90% and ␣ ⫽ 0.05 by using the Epi factors including airway management, female sex, younger
Info Web site (www.cdc.gov/epiinfo/). Statistical analysis patients, gynecological procedure, and succinylcholine ad-
was performed with JMP software (version 4, SAS Institute, ministration predict postoperative sore throat.3 All patients
Cary, NC). Statistical significance for differences was tested in this trial were young females and were candidates for
by Student t test and ␹2 test when appropriate. A P value cesarean delivery. Therefore, age and sex were eliminated
⬍0.05 was considered statistically significant. as possible confounding factors.
Researchers recognizing the potential role of inflammation
RESULTS in these postoperative airway sequelae have described the use
As shown in Table 2, there were no significant differences of inhaled and topical steroids.5–7 Stride8 concluded that 1%
in age, height, weight, body mass index, duration of hydrocortisone water-soluble cream was ineffective for
surgery, intubation, and grade of laryngeal exposure be- reducing the incidence of postoperative sore throat.
tween the 2 groups. The total incidence of sore throat, Sumathi et al.9 showed that the widespread application of
cough, and hoarseness was significantly lower in group F betamethasone gel on the tracheal tube decreased the
compared with group C (P ⬍ 0.05) not only in the first incidence and severity of postoperative sore throat, cough,
postoperative hour but also 24 hours after surgery (Table and hoarseness. The differences in the findings between
3). During the first postoperative hour, none of the patients these 2 studies may be attributable to the fact that Stride

896 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

 Table 3. Postoperative Complications by Grade After 1 and 24 h
After 1 h After 24 h
Fluticasone Control Fluticasone Control
propionate group group P propionate group group P
Sore throat
1 2 (3.33) 19 (31.67) 6 (10) 15 (25)
2 0 (0) 2 (3.33) 2 (3.33) 7 (11.67)
3 0 (0) 1 (1.67) 0 (0) 2 (3.33)
Total 2 (3.33) 22 (36.67) ⬍0.0001 8 (13.33) 24 (40) 0.001
Cough
1 2 (3.33) 8 (13.33) 8 (13.33) 14 (23.33)
2 0 (0) 3 (5) 0 (0) 9 (15)
3 0 (0) 0 (0) 0 (0) 2 (3.33)
Total 2 (3.33) 11 (18.33) 0.008 8 (13.33) 25 (41.67) 0.005
Hoarseness
1 2 (3.33) 16 (26.67) 13 (21.67) 23 (38.33)
2 0 (0) 4 (6.67) 2 (3.33) 4 (6.67)
3 0 (0) 1 (1.67) 0 (0) 3 (5)
Total 2 (3.33) 21(35) ⬍0.0001 15 (25) 30 (50) 0.005

lubricated the tube only to the 5-cm mark, whereas Sumathi
et al. lubricated the tube to the 15-cm mark. Thus, because
of more widespread application of steroid gel to the tube,
more gel came in contact with the posterior pharyngeal
wall, vocal cords, and trachea and was not just confined to
the tip and cuff of the tracheal tube.
Our study showed that inhaled steroid could be simi-
larly effective in decreasing postoperative sore throat,
cough, and hoarseness, and is similar to the study by
Sumathi et al.9 In the 2 studies, the patient populations and
the type of surgery were different. Their patients were
either sex, aged between 18 and 50 years, and undergoing
elective surgery.
Inhaled fluticasone delivers the drug in smaller doses
and in a shorter time to the patient’s airway compared with
widespread lubrication of the tube with betamethasone,
which may increase the dose of drug that comes in contact
with the mucosa of the oropharynx, larynx, and trachea,
resulting in higher systemic absorption and a possible
aggravation of local subtle infection, especially in pregnant
patients.
Coughing, wheezing, and shortness of breath are symp-
toms of asthma. Asthma treatment includes inhaled bron-
chodilators, which reverse airflow obstruction, and inhaled
corticosteroids to prevent asthma exacerbations by damp-
ing the inflammatory processes that underlie asthma at-
tacks. Inhaled fluticasone propionate is a relatively new
inhaled corticosteroid for the treatment of asthma.
In many studies, fluticasone propionate was used for
improved outcomes in children or adults who were at risk
of asthma. Treatment with fluticasone propionate or a
combination of this drug with a long-acting ␤-2 agonist
such as salmeterol was significantly effective in asthmatic
patients.10 –12 A study of the safety of intranasal corticoste-
roids such as fluticasone propionate did not identify any
systemic adverse events, which suggests that this drug can
be safely administered.13
Some studies evaluated the effect of treatment with
fluticasone propionate nasal spray in rhinitis and inhaled
fluticasone during pregnancy. The authors did not report
any adverse effects on maternal and fetal health.14 –16 It is
therefore unlikely that a single dose of fluticasone will
affect the fetus. In this study, we used fluticasone propi-
onate in healthy women and did not observe any side
effects.
A limitation of our study was the infrequency of data
collection, which we could also have performed at 6 and 12
hours. Also, we did not use an inert inhaler in the control
group, leading to potential bias.
In conclusion, inhaled fluticasone propionate (500 ␮g)
decreases the incidence and severity of postoperative sore
throat, cough, and hoarseness in patients undergoing cesar-
ean delivery under general anesthesia.
REFERENCES
1. McHardy FE, Chung F. Postoperative sore throat: cause, pre-
vention and treatment. Anaesthesia 1999;54:444 –53
2. Joshi GP, Inagaki Y, White PF, Taylor-Kennedy L, Wat LI,
Gevirtz C, McCraney JM, McCulloch DA. Use of laryngeal
mask airway as an alternative to the tracheal tube during
ambulatory anesthesia. Anesth Analg 1997;85:573–7
3. Higgins PP, Chung F, Mezei G. Postoperative sore throat after
ambulatory surgery. Br J Anaesth 2002;88:582– 4
4. Maruyama K, Sakai H, Miyazawa H, Tida N, Iinuma Y,
Mochizuki N, Hara K, Otagiri T. Sore throat and hoarseness
after total intravenous anaesthesia. Br J Anaesth 2004;92:541–3
5. Ayoub MC, Ghobashy A, McGrimely L, Koch ME, Gadir S,
Silverman DG. Wide spread application of topical steroids to
decrease sore throat, hoarseness and cough after tracheal
intubation. Anesth Analg 1998;87:714 – 6
6. Selvaraj T, Dhanpal R. Evaluation of the application of topical
steroids on the endotracheal tube in decreasing postoperative
sore throat. J Anaesthesiol Clin Pharmacol 2002;18:167–70
7. El-Hakim M. Beclomethasone prevents postoperative sore
throat. Acta Anaesthesiol Scand 1993;37:250 –2
8. Stride PC. Postoperative sore throat: topical hydrocortisone.
Anaesthesia 1990;45:968 –71
9. Sumathi PA, Shenoy T, Ambareesha M, Krishna HM. Con-
trolled comparison between betamethasone gel and lidocaine
jelly applied over tracheal tube to reduce postoperative sore
throat, cough, and hoarseness of voice. Br J Anaesth 2008;
100:215– 8
10. Bacharier LB, Guilbert TW, Zeiger RS, Strunk RC, Morgan WJ,
Lemanske RF Jr, Moss M, Szefler SJ, Krawiec M, Boehmer S,
Mauger D, Taussig LM, Martinez FD. Patient characteristics
associated with improved outcomes with use of an inhaled
corticosteroid in preschool children at risk for asthma. J
Allergy Clin Immunol 2009;123:1077– 82

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 897

Postoperative Sore Throat and Fluticasone
11. de Blic J, Ogorodova L, Klink R, Sidorenko I, Valiulis A, Hofman
J, Bennedbaek O, Anderton S, Attali V, Desfougeres JL, Poterre
M. Salmeterol/fluticasone propionate vs. double dose flutica-
sone propionate on lung function and asthma control in
children. Pediatr Allergy Immunol. 2009;20:763–71
12. Markham A, Jarvis B. Inhaled salmeterol/fluticasone propi-
onate combination: a review of its use in persistent asthma.
Drugs 2000;60:1207–33
13. Demoly P. Safety of intranasal corticosteroids in acute rhino-
sinusitis. Am J Otolaryngol 2008;29:403–13
898 www.anesthesia-analgesia.org
14. Choi S, Han JY, Kim MY, Velázques-Armenta EY, Nava-Ocampo
AA. Pregnancy outcomes in women using inhaled fluticasone
during pregnancy: a case series. Allergol Immunopathol 2007;
35:239 – 42
15. Rahimi R, Nikfar S, Abdollahi M. Meta-analysis finds use of
inhaled corticosteroids during pregnancy safe: a systematic
meta-analysis review. Hum Exp Toxicol 2006;25:447–52
16. Ellegård EK, Hellgren M, Karlsson NG. Fluticasone propionate
aqueous nasal spray in pregnancy rhinitis. Clin Otolaryngol
Allied Sci 2001;26:394 – 400
ANESTHESIA & ANALGESIA
 International Society for Anaesthetic Pharmacology
Anesthetic Pharmacology and Preclinical Pharmacology Section Editor: Marcel E. Durieux
Clinical Pharmacology Section Editor: Tony Gin

Repinotan, a Selective 5-HT1A-R-Agonist, Antagonizes

Morphine-Induced Ventilatory Depression in
Anesthetized Rats
U. Guenther, MD,* H. Wrigge, PhD,* N. Theuerkauf, MD,* M. F. Boettcher, MD,† G. Wensing, MD,†
J. Zinserling, PhD,* C. Putensen, PhD,* and A. Hoeft, PhD*

BACKGROUND: Spontaneous breathing during mechanical ventilation improves arterial oxygen-

ation and cardiovascular function, but is depressed by opioids during critical care. Opioid-induced
ventilatory depression was shown to be counteracted in anesthetized rats by serotonin(1A)-
receptor (5-HT1A-R)-agonist 8-OH-DPAT, which cannot be applied to humans. Repinotan hydro-
chloride is a selective 5-HT1A-R-agonist already investigated in humans, but the effects on
ventilation and nociception are unknown. In this study, we sought to establish (a) the effects of
repinotan on spontaneous breathing and nociception, and (b) the interaction with the standard
opiate morphine.
METHODS: The dose-dependent effects of repinotan, given alone or in combination with
morphine, on spontaneous minute ventilation (MV) and nociceptive tail-flick reflex latencies
(TFLs) were measured simultaneously in spontaneously breathing anesthetized rats. An addi-
tional series with NaCl 0.9% and the 5-HT1A-R-antagonist WAY 100 135 served as controls.
RESULTS: (a) Repinotan dose-dependently activated spontaneous breathing (MV, mean [95%
confidence interval]; 53% [29%–77%]) of pretreatment level) and suppressed nociception (TLF,
91% maximum possible effect [68%–114%]) with higher doses of repinotan (2–200 ␮g/kg). On
the contrary, nociception was enhanced with a small dose of repinotan (0.2 ␮g/kg; TFL, ⫺47%
maximum possible effect [⫺95% to 2%]). Effects were prevented by 5-HT1A-antagonist WAY 100
135. (B) Morphine-induced depression of ventilation (MV, ⫺72% [⫺100% to ⫺44%]) was
reversed by repinotan (20 ␮g/kg), which returned spontaneous ventilation to pretreatment levels
(MV, 18% [⫺40% to 77%]). The morphine-induced complete depression of nociception was
sustained throughout repinotan and NaCl 0.9% administration. Despite a mild decrease in mean
arterial blood pressure, there were no serious cardiovascular side effects from repinotan.
CONCLUSIONS: The 5-HT1A-R-agonist repinotan activates spontaneous breathing in anesthetized
rats even in morphine-induced ventilatory depression. The potency of 5-HT1A-R-agonists to stimulate
spontaneous breathing and their antinociceptive effects should be researched further. (Anesth Analg
2010;111:901–7)

O pioids are potent analgesics, but their clinical admin-

istration is limited by the intrinsic risk of fatal apnea.
Hence, pain therapy involving opioids must be bal-
anced against respiratory depression.1 The serotonin(1A)-
receptor (5-HT1A-R)-agonist buspirone has been shown to
stimulate spontaneous breathing in cats2 and to overcome
neurogenic breathing disturbances in humans.3,4 Another
5-HT1A-R-agonist, 8-hydroxy-2-(di-n-propylamino)tetralin
(8-OH-DPAT), a substance not approved for use in humans,
From the *University Hospital of Bonn, Clinic of Anaesthesiology and
Intensive Care Medicine, Bonn; and †Department of Pharmacological Re-
search, Bayer Schering Pharma AG, Wuppertal, Germany.
Accepted for publication May 19, 2010.
Supported by Bayer Schering Pharma AG, Germany, and departmental
funding.
Address correspondence and reprint requests to Ulf Guenther, MD, Univer-
sity Hospital of Bonn, Clinic of Anaesthesiology and Intensive Care Medi-
cine, Sigmund-Freud-Strasse 25, 53105 Bonn, Germany. Address e-mail to
u.guenther@uni-bonn.de.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181eac011
has been demonstrated to overcome opioid-induced ventila-
tory depression in anesthetized rats.5
Nociception, another target of 5-HT1A-R-agonists, was
reported either to be depressed6 – 8 or enhanced by 5-HT1A-
R-agonists.9,10 Later, 5-HT1A-R-agonist F13640 was found
to exert a dual effect, hyperalgesic and analgesic, depend-
ing on plasma and brain concentrations.11 Recently,
enhancement of nociceptive reflexes by small doses of
8-OH-DPAT and suppression by higher doses were con-
firmed in 2 different experimental models.12
Repinotan (R-(⫺)-2-{4-[(chroman-2-ylmethyl)-amino]-
butyl}-1,1-dioxobenzo[d]-isothiazolone hydrochloride) is a
highly effective, selective, full 5-HT1A-R-agonist.13,14 Unlike
other 5-HT1A-R-agonists, repinotan is approved for IV use in
humans and has already undergone a series of clinical inves-
tigations into the effects of neuroprotection after traumatic
brain injury and stroke.15–18 However, its effects on nocicep-
tion and ventilation are not yet established. The aim of this
study was to verify the effects of repinotan on spontaneous
breathing and nociception simultaneously, and to determine
the interaction with the standard opiate morphine on sponta-
neous breathing and nociception in anesthetized rats. Two

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 901

Repinotan Antagonizes Ventilatory Depression

hypotheses were tested: (1) repinotan at a dose to stimulate

A Repinotan
spontaneous breathing does not enhance a nociceptive reflex,
and (2) repinotan antagonizes morphine-induced depression
Repinotan 0.02µg/kg WAY 100135 1µg/kg NaCl 0.9%
of spontaneous breathing.

Repinotan 0.2µg/kg Repinotan 0.2µg/kg NaCl 0.9%

METHODS
Animals
This study was performed with approval from the local Repinotan 2µg/kg Repinotan 2µg/kg NaCl 0.9%

Institutional Animal Review Board for animal research and in

accordance with the “Guide for the Care and Use of Labora- Repinotan 20µg/kg Repinotan 20µg/kg NaCl 0.9%
tory Animals.” Animals were housed in standard laboratory
conditions with a 12-hour light/dark schedule and free access Repinotan 200µg/kg Repinotan 200µg/kg NaCl 0.9%
to food and water. Thirty-nine male Sprague-Dawley rats
weighing 260 g (244 –277 g) (mean, 95% confidence interval n=8 n=4 n=8
[CI]) were deeply anesthetized with sodium-pentobarbitone B Repinotan / Morphine
(60 mg/kg) intraperitoneally and placed supine on a heating
pad to maintain rectal temperature constantly at 37°C ⫾
0.5°C. The right inguinal vessels were cannulated via a Morphine Morphine Morphine
12[6-18]mg/kg 11[7-15]mg/kg 13[7-18]mg/kg
small surgical incision for continuous monitoring of
arterial blood pressure and systemic administration of
Repinotan 0.2µg/kg Repinotan 0.02µg/kg NaCl 0.9%
study drugs. Anesthesia was maintained with sevoflu-
rane, leveled at an inspiratory concentration of 1.5 to 2.5
Repinotan 2µg/kg Repinotan 6µg/kg NaCl 0.9%
vol% to ensure immobility, stable spontaneous breath-
ing, and detectable tail-flick reflex (TFR).
Repinotan 20µg/kg Repinotan 60µg/kg NaCl 0.9%

Measurements
Animals breathed spontaneously via a tracheotomy tube
(inner diameter, 1.2 mm). The expired air was led through
the flowhead (order number, MLT1L; ADInstruments
GmbH, Spechbach, Germany) of a spirometer (ML141)
connected to an A/D-interface (PowerLab 4/25®; all de-
vices from ADInstruments GmbH) to record respiratory
rate (RR) and tidal volume (Vt) by integration of ventila-
tory airflow over time. Minute ventilation (MV) was calcu-
lated as MV [mL/min] ⫽ RR [1/min] ⫻ Vt [mL].
The TFR was evoked by a 100-W light beam source
mounted 15 mm over the base of the tail to reach
maximum temperature within a second. The latency of
the reflex response (TFR latency, TFL) was recorded with
a strain gauge attached to the tail distal to the heating
spot. A shortened TFL indicates enhanced nociceptive
responsiveness; an elongated TFL indicates depressed
nociception. Heating was stopped when the tail flicked or
after a maximum heating time of 15 seconds (TFLoffset) to
prevent damage to the tail. TFLs were calculated as change
in percent of the maximum possible effect [% MPE] accord-
ing to the formula8: % MPE ⫽ 100 ⫻ [TFLtreatment ⫺
TFLpretreatment] ⫻ [TFLoffset ⫺ TFLpretreatment]⫺1). A 100%
MPE means complete suppression of nociception. Three
sweeps were recorded and averaged. A blood pressure
transducer, temperature probe, and strain-gauge trans-
ducer were also connected to the same A/D-interface such
as the spirometer (PowerLab 4/25®; ADInstruments
GmbH).
Drug Administration Protocols
Two different sets of experiments were performed: (a) the
first set was aimed at determining the effects of repinotan
on spontaneous breathing, and (b) the second set assessed
interactions of repinotan with the opiate morphine (see Fig.
1 for schematic overview).
Repinotan 200µg/kg NaCl 0.9%
n=8 n=5 n=6
Figure 1. A, Serotonin(1A)-receptor (5-HT1A-R)-agonist repinotan was
injected every 15 minutes at increasing doses. In a second series,
the selective 5-HT1A-R-antagonist WAY 100 135 was given before
repinotan. A third series involving only NaCl 0.9% served as controls.
B, Morphine was given at increments of 5 mg/kg until a target
depression of respiratory rate of ⬎50% was established. The mean
required morphine dosing is given as mean (95% confidence inter-
val). Repinotan was given in 2 distinct series to cover all doses
required to delineate the top of the bell-shaped dose-response curve
and to maintain experiments at comparable length.
Repinotan
Repinotan was injected IV every 15 minutes with doses
ranging from 0.02 through 200 ␮g/kg (Fig. 1A). The doses
were chosen because it was concluded from preliminary
dose-finding experiments that the 20 ␮g/kg dose was the
most efficient to counteract opioid-induced ventilatory
depression. The wide range of dosage was necessary to
verify whether repinotan also possesses dose-dependent
pro- and antinociceptive effects similar to the standard
5-HT1A-R-agonist, 8-OH-DPAT. The number of experi-
ments involving repinotan (n ⫽ 8) was chosen based on our
previous experience with 8-OH-DPAT, in which the small-
est effective dose increased spontaneous MV by 46% with a
standard deviation of 35%. With an ␣ set at 0.05, the power
was calculated as 0.93 with n ⫽ 8 experiments in the
double-sided power analysis.
Before the first drug administration, a series of 3 TFL
sweeps was averaged and taken as the pretreatment level.
Subsequent TFLs were taken 10 minutes after each drug
administration. For control experiments, NaCl 0.9% was
injected every 15 minutes instead of study drugs (n ⫽ 8). In
another series of 4 experiments, the selective 5-HT1A-R-

902 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

antagonist WAY 100 135 (1 mg/kg) was injected before
administration of repinotan (n ⫽ 4).
Repinotan/Morphine Coadministration
Morphine was injected at increments of 5 mg/kg until
respiratory frequency was depressed to at least 50% of the
pretreatment level. Thereafter, repinotan was added cumu-
latively at the same doses as in the first set (Fig. 1B). Control
experiments were again performed by injection of 200 ␮L of
NaCl 0.9% (n ⫽ 6). After completion of this series, the
ventilatory dose-response curve had a bell shape. To delin-
eate the top of the bell shape more precisely, 5 additional
experiments were performed with repinotan concentra-
tions that were between the initially intended measuring
points (Fig. 1). To verify integrity of the TFR at the end of
experiments, naloxone (1 mg/kg) was given (not shown).
Statistical Analysis
All data were tested for normal distribution (Kolmogorov-
Smirnov test). Pretreatment levels of matched groups were
compared with the Student t test. All ventilatory variables
were calculated as change in percent of pretreatment level
(% change). Results of experiments without morphine were
compared with the values before the first administration of
study drugs (pretreatment level). Results of experiments
involving morphine/repinotan coadministration were com-
pared with the variables obtained after morphine administra-
tion. The results of the ventilatory experiments were analyzed
by 1-way repeated-measures analysis of variance. Each repi-
notan concentration was compared with the pretreatment
level (i.e., 0% change) by Dunnett multiple comparison post
test.19 TFLs were calculated as change in % MPE as stated
above and also analyzed by comparison of each drug concen-
tration to pretreatment levels by Dunnett multiple compari-
son test. Data were processed with the Chart 4.0 and Scope 4.0
software package (ADInstruments GmbH); statistical analyses
were performed using Prism4® software package for Macin-
tosh (GraphPad Software Inc., San Diego, CA). Power analy-
ses were done with the Simple Interactive Statistical Analysis
(SISA) online software package (http://www.quantitativeskills.
com/sisa/calculations/power.htm).
Drugs
Repinotan was provided by the manufacturer, Bayer
Healthcare AG (Wuppertal, Germany). Morphine-sulfate
was purchased from Merck KG (Darmstadt, Germany),
with permission from the German Institute for Pharmacy
and Medical Products. WAY 100 135 was obtained from
Tocris (Bristol, UK). Naloxone-HCl was purchased from
Ratiopharm (Ulm, Germany). All compounds were diluted
in isotonic saline at the respective concentrations; injection
volumes were 200 ␮L.
RESULTS
Effects of Repinotan
Spontaneous Ventilation
The mean pretreatment RR was 58 breaths/min (95% CI,
46 –70 breaths/min) in the repinotan group versus 61
breaths/min (54 – 69 breaths/min) in the control group, and
the mean arterial blood pressure (MAP) was 111 mm Hg
(100 –122 mm Hg) in the repinotan group versus 110 mm
October 2010 • Volume 111 • Number 4
A
0.02 0.2 2 20 200
Repinotan (µg/kg)
Repinotan
NaCl 0.9%
B
0.02 0.2 2 20 200
Repinotan (µg/kg)
Repinotan
NaCl 0.9%
Figure 2. Effects of the serotonin(1A)-receptor (5-HT1A-R)-agonist
repinotan (n ⫽ 8) on spontaneous minute ventilation (MV) and
tail-flick reflex (TFR) latencies (TFLs). Control experiments were
performed with NaCl 0.9% (n ⫽ 8). Values of MV are given as percent
change from pretreatment level (% change), results of TFLs are
shown as percent of maximum possible effect (% MPE; mean [95%
confidence interval]). *P ⬍ 0.05, ***P ⬍ 0.001, compared with
pretreatment level, 1-way repeated-measures analysis of variance.
A, Repinotan dose dependently increased MV to a maximum of 53%
(10%–29%) above the pretreatment level with the 200 ␮g/kg dose.
B, Repinotan effects on nociception were dose dependent. Initial
shortening of TFL (⫺47 [⫺95 to 2], indicating enhanced nociceptive
responsiveness) with small doses (0.2 ␮g/kg) was followed by
elongation of TFL to 91% MPE (68%–114% MPE) with the highest
dose (200 ␮g/kg), meaning that TFR was profoundly suppressed.
Hg (100 –120 mm Hg) in the control group. The mean TFL
was 7 seconds (5–9 seconds) in both groups. All data were
normally distributed, and pretreatment values did not
differ statistically. Repinotan dose dependently increased
spontaneous MV (Fig. 2A), reaching the maximum effect
(MV; 53% [29%–77%]) with the 200 ␮g/kg dose.
Tail-Flick Reflex Latency
The repinotan effects on nociception were dose dependent:
a small dose of repinotan (0.2 ␮g/kg) shortened TFL,
whereas a high dose of repinotan (200 ␮g/kg) elongated
TFL (Fig. 2B), meaning that nociception was enhanced with
small doses and suppressed with higher doses.
Controls
The 5-HT1A-antagonist WAY 100 135 (1 mg/kg, n ⫽ 4) did
not significantly alter MV itself, but prevented repinotan
from activating spontaneous MV (data not shown). Injec-
tions of NaCl 0.9% had neither detectable effects on venti-
lation nor nociception.
www.anesthesia-analgesia.org 903
Repinotan Antagonizes Ventilatory Depression
A Pre-treatment
10
mL/s
0
10 s heat on
B Morphine 10 mg/kg
10
mL/s
0 levels, giving the dose-response curve a bell shape.
10 s
heat on
C Morphine 10 mg/kg, Repinotan 20 µg/kg
10
mL/s
0 rity of the TFR, because it returned with a TFL of 8% MPE
10 s
heat on
Figure 3. Left panel, Expiratory airflow (mL/s). Animal was breathing
spontaneously through a tracheostomy tube. Right panel, Tracings of
the tail-flick reflex (TFR). Arrows indicate onset and offset of heat to
the tail, and the circle marks the artifact evoked by the tail flick. A,
Pretreatment level. Left panel, Spontaneous respiratory frequency,
51 breaths/min. Right panel, Heat to the tail evoked a tail flick with
a latency (TFL) of 3.9 seconds. B, Morphine (10 mg/kg) completely
depressed spontaneous ventilation; the last 3 breaths before apnea
are shown (left panel). The TFR was abolished (right panel). C,
Repinotan (20 ␮g/kg) reestablished spontaneous ventilation (left
panel), whereas TFR remained suppressed (right panel).
MV (%change [95% CI])
n = 13 5 8 8 5 8 5 8
MO 0.02 0.2 2 20 200
Repinotan (µg/kg)
Morphine/ Repinotan
Morphine/ NaCl 0.9%
Figure 4. Effects of morphine (MO) and subsequent serotonin(1A)-
receptor (5-HT1A-R)-agonist repinotan on spontaneous minute venti-
lation (MV). NaCl 0.9% served as controls (n ⫽ 6). Values are given
as mean (95% confidence interval) and percent change from pre-
treatment level (% change). *P ⬍ 0.05, **P ⬍ 0.01, compared with
morphine levels, 1-way repeated-measures analysis of variance.
Morphine (11 mg/kg [8 –16 mg/kg]) depressed MV to ⫺72%
(⫺100% to ⫺44%) of pretreatment level. Repinotan dose depen-
dently activated MV to a maximum of 18% (⫺40% to 77%). Repino-
tan dosage ⬎20 ␮g/kg re-decreased MV, giving the dose-response
curve a bell shape (see Discussion).
Repinotan/Morphine Coadministration
Figure 3 shows a representative experiment on the antago-
nization of morphine-induced ventilatory depression, dur-
ing which nociceptive TFR remains completely depressed by
904 www.anesthesia-analgesia.org
the sustained action of morphine. Morphine depressed spon-
taneous MV to ⫺72% (⫺100% to ⫺44%) and always abolished
the TFR with the first dosing increment (Fig. 4). Morphine
TFR consumption did not differ between groups (Fig. 1).
2s
Spontaneous Breathing
heat off Repinotan (after morphine) dose dependently activated spon-
taneous breathing (Fig. 4), resulting in a maximum MV of 18%
(⫺40% to 77%) above the pretreatment level with the 20
␮g/kg dose (P ⬍ 0.01, compared with morphine level).
Further increases to doses ⬎20 ␮g/kg returned MV to lower
2s
Nociception
TFR was abolished with the first bolus of morphine (TFL,
100% MPE), and remained completely suppressed through-
out administrations of repinotan (n ⫽ 13) and control drugs
(n ⫽ 6; P ⬍ 0.001, compared with pretreatment levels).
Naloxone-HCl at the end of experiments verified the integ-
2s (⫺18% to 33% MPE) (not shown).
Cardiovascular Side Effects
Repinotan depressed MAP with higher doses (Table 1, n ⫽
8), but this did not have deleterious effects on the experi-
ments. Likewise, although morphine (12 mg/kg [8 –16
mg/kg], n ⫽ 13) markedly depressed MAP, repinotan did
not further aggravate arterial hypotension, and neither did
NaCl 0.9% (Table 1). There were no serious cardiovascular
complications.
DISCUSSION
This study was performed to clarify whether the 5-HT1A-
R-agonist repinotan also antagonizes opioid-induced ven-
tilatory depression similar to other 5-HT1A-R-agonists. It
was verified that (a) higher doses (2–200 ␮g/kg) of repino-
tan stimulated spontaneous breathing, small doses (0.2
␮g/kg) enhanced nociception, and the highest dose (200
␮g/kg) depressed nociception; and (b) morphine-induced
depression of spontaneous breathing was antagonized by
higher doses of repinotan, whereas depression of nocicep-
tion persisted. Despite a mild depression of MAP, repino-
tan did not produce serious cardiovascular complications.
These findings confirm previous work in which the
5-HT1A-R-agonist 8-OH-DPAT was shown to antagonize
an opioid-induced ventilatory depression without impair-
ing antinociception.12 8-OH-DPAT, however, is not ap-
proved for human use. Buspirone, the only commercially
available 5-HT1A-R-agonist for use in humans, has been
shown to stabilize apneustic breathing disturbances.3 Bu-
spirone, being only a partial 5-HT1A-R-agonist, failed to
counteract a morphine-induced ventilatory depression in
healthy volunteers,20 nor did it cause antinociceptive ef-
fects in healthy volunteers.21 In a direct comparison with
8-OH-DPAT, only a weak ventilatory stimulation in coad-
ministration with fentanyl was found for buspirone in an in
situ perfused brainstem–spinal cord preparation, whereas
8-OH-DPAT proved to be an effective ventilatory stimulant.22
Both enhancement and depression of nociception by
5-HT1A-R-agonists, given alone or in combination with
ANESTHESIA & ANALGESIA
 Table 1. Effects of Morphine and Repinotan on Mean Arterial Blood Pressure
Repinotan NaCl 0.9%
Mean (95% CI) P value Mean (95% CI) P value
Repinotan (% change of pretreatment)
Repinotan 0.02 ␮g/kg 10 (4–17) ⬍0.05 0 (⫺13 to 13) NS
Repinotan 0.2 ␮g/kg 4 (⫺6 to 14) NS ⫺3 (⫺21 to 16) NS
Repinotan 2 ␮g/kg 1 (⫺7 to 8) NS ⫺5 (⫺25 to 15) NS
Repinotan 20 ␮g/kg ⫺17 (⫺29 to ⫺4) ⬍0.001 ⫺10 (⫺25 to 10) NS
Repinotan 200 ␮g/kg ⫺19 (⫺29 to ⫺9) ⬍0.001 ⫺7 (⫺21 to 10) NS
Morphine, repinotan (% change of pretreatment)
Morphine ⫺30 (⫺46 to ⫺14) ⬍0.001 ⫺21 (⫺54 to ⫺13) ⬍0.05
Repinotan 0.02 ␮g/kg ⫺16 (⫺31 to ⫺2) NS ⫺21 (⫺54 to ⫺11) NS
Repinotan 0.2 ␮g/kg ⫺24 (⫺40 to ⫺8) NS ⫺22 (⫺50 to ⫺6) NS
Repinotan 2 ␮g/kg ⫺40 (⫺55 to ⫺25) NS ⫺21 (⫺53 to ⫺11) NS
Repinotan 6 ␮g/kg ⫺28 (⫺50 to ⫺5) NS
Repinotan 20 ␮g/kg ⫺30 (⫺46 to ⫺14) NS ⫺19 (⫺37 to 0) NS
Repinotan 60 ␮g/kg ⫺21 (⫺60 to ⫺17) NS
Repinotan 200 ␮g/kg ⫺25 (⫺64 to 15) NS ⫺19 (⫺49 to ⫺11) NS
CI ⫽ confidence interval; NS ⫽ not significant; MAP ⫽ mean arterial blood pressure.
MAP is given as mean (95% CI). Repinotan depressed MAP with higher doses (20, 200 ␮g/kg, n ⫽ 8, repeated-measures analysis of variance, P ⬍ 0.001).
Morphine (12 mg/kg 关8 –16 mg/kg兴) markedly depressed MAP (n ⫽ 13, P ⬍ 0.001, compared with pretreatment level). Subsequent repinotan did not further
aggravate hypotension. There were no serious cardiovascular complications.

opioids, have been variously reported.8,9,23–26 More re-
cently, the highly selective 5-HT1A-R-agonist F13640 was
reported to induce both hyperalgesia and/or analgesia
depending on the blood and brain concentration time
course.11 Most notably, F13640 was also shown to alleviate
opioid-induced hyperallodynia and neuropathic pain in
rats.27,28 The dose-dependent pro- and antinociceptive ef-
fects of repinotan found in this study contribute to recon-
ciling the past contradictory findings, which were at least in
part attributable to different experimental models, drug
administration routes, and dosing ranges.
The dose-response curve of 5-HT1A-R stimulation of spon-
taneous breathing after morphine-induced ventilatory depres-
sion is inversely U shaped or “bell shaped,” meaning that
stimulatory effects subsided with high concentrations.29 Also,
repinotan produced a combination of low-dose stimulation of
the TFR followed by high-dose inhibition. This dose-response
characteristic is generally referred to as “hormesis.”30 More
than 30 receptor systems, including opioid and adrenergic
receptors, were identified to have hormetic dose responses,
and the serotonin (5-HT) receptor system is among them.31
The neuroprotective effects of 5-HT1A-R-agonists have been
shown to have bell-shaped dose responses.32 It is proposed
that the basic biological principle behind this is that a mild
stress may promote function or action, and extreme stress
may promote depressive or toxic action.30
Although hormesis can be observed in a wide range of
receptor systems and agents, there is no one-for-all molecu-
lar mechanism. The 5-HT1A-R are variously located and
involved at different levels in the modulation of opioider-
gic effects on nociceptive pathways.9,33 Activation of cen-
tral 5-HT1A-R, for instance, has been shown to enhance
opioidergic inhibition of spinal reflexes,33 whereas systemic
(intraperitoneal, IV) administration of 5-HT1A-R-agonists
produced both pro- and antinociceptive effects.7,11 Directly
applied onto the spinal cord, activation of 5-HT1A-R inhib-
ited nociceptive neural responses only with the highest
studied dose of 8-OH-DPAT.8,34 We speculate that IV
repinotan overpowered possible pronociceptive effects
(mediated by spinal 5-HT1A-R) by actions via central
5-HT1A-R, once the administered dose was high enough to
establish sufficient brain tissue concentrations. The under-
lying mechanisms of hormetic dose responses clearly de-
serve further research.30,35
The cardiovascular depression after repinotan adminis-
tration was much less severe than that by 8-OH-DPAT. In
previous work, we reported severe, occasionally fatal,
cardiocirculatory depression with the highest dose of
8-OH-DPAT (100 ␮g/kg) in anesthetized rats.12 Others saw
that 8-OH-DPAT prevented arterial hypotension induced
by the short-acting opioid remifentanil in conscious, non-
anesthetized rats.36 Unlike repinotan,13 8-OH-DPAT also
stimulates 5-HT7-R,37 which are critical for activation of
cardiac vagal input.38 For instance, blockade of central
5-HT7-R attenuates the bradycardia and pressor response
to both chemoreflex activation (induced by intracisternal
injection of potassium cyanide) and baroreflex activation
(induced by IV phenylephrine).39 Activation of 5-HT7-R in
turn might add to the depression of MAP seen in this study
after morphine administration (Table 1), which was likely
induced by peripheral vasodilation.40 Furthermore, it has
been shown in anesthetized animals that the 5-HT1A-
R-agonist F13640 markedly reduced the intraoperative re-
quirement of the volatile anesthetic.36 The concentration of
the anesthetic was maintained constant in this study ac-
cording to our protocol, which certainly contributed to
arterial hypotension caused by increasing Pco2 as the
consequence of hypoventilation.
Some limitations of this study warrant comment. First,
repinotan, a 5-HT1A-R-agonist, was developed as an antide-
pressant, and was also found to exert neuroprotective effects
on in vivo rats.15 Despite promising clinical data in humans,16
multicenter studies failed to show favorable effects on neuro-
logical outcomes in patients with stroke and traumatic brain
injury.17,18 Specific serotonergic complications of repinotan,
such as headache, nausea and vomiting, flush, tachycardia,
and agitation, in humans were reported.17,41 These symptoms
may even be aggravated in coadministration with mor-
phine.20 Specific serotonergic side effects were not seen in this
study because of the experimental setup, but they could,

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 905

Repinotan Antagonizes Ventilatory Depression
however, limit the clinical applicability of repinotan at least in
conscious patients.
Second, it should be highlighted that the morphine
concentrations in this investigation were much higher than
in studies aiming solely at nociception.42,43 This happened
because morphine dosage was targeted to produce venti-
latory depression, requiring higher dosing. The TFR was
always abolished with the first bolus of morphine and
always before ventilatory depression occurred. The pos-
sible attenuation of morphine-induced antinociception by
small doses of repinotan was presumably overpowered by
the strong morphine effect. We did not investigate to
determine whether small pronociceptive doses of repinotan
would interfere with more moderate doses of morphine.
This should be considered for further research.
Third, the TFR is an acute, polysynaptic nociceptive
spinal reflex.44 Although pronociceptive effects were seen
only with very small doses of repinotan, but not within the
dosing range to stimulate breathing, it is conceivable that
small pronociceptive doses of repinotan could alleviate mor-
phine antinociception. It was shown by others that 5-HT1A-R
influence nociceptive processing differently, according to the
type of noxious stimulus.45 Thus, nociceptive modalities other
than the one investigated here may be activated by small
doses of 5-HT1A-R-agonists, which may not be treated with
opioids. This will be clarified by further investigations.
In conclusion, this work confirmed that the 5-HT1A-R-
agonist repinotan activates spontaneous breathing and
suppresses nociception with higher doses, and that it
antagonizes morphine-induced ventilatory depression in
anesthetized rats. Selective 5-HT1A-R-agonists thus are
promising candidates for research into the stabilization of
spontaneous breathing and pain therapy.
AUTHOR CONTRIBUTIONS
UG, MFB, GW, HW, CP, and AH helped with study design;
UG, NT, JZ, and GW helped with study conduction; UG, NT,
and JZ helped with data collection; UG, NT, and JZ helped
with data analysis; and UG, HW, MFB, NT, CP, and AH helped
with manuscript preparation. All authors read and approved
the final manuscript. UG and MFB reviewed the original study
data and data analysis. UG maintains the study records.
DISCLOSURE
Bayer Schering Pharma AG, Germany, provided the study
drug and funded part of this study. MFB and GW are
employees of Bayer Schering Pharma AG.
REFERENCES
1. Couzin J. Medicine: a sigh of relief for painkillers. Science
2003;301:150
2. Garner SJ, Eldridge FL, Wagner PG, Dowell RT. Buspirone, an
anxiolytic drug that stimulates respiration. Am Rev Respir Dis
1989;139:946 –50
3. Wilken B, Lalley P, Bischoff AM, Christen HJ, Behnke J,
Hanefeld F, Richter DW. Treatment of apneustic respiratory
disturbance with a serotonin-receptor agonist. J Pediatr
1997;130:89 –94
4. El-Khatib MF, Kiwan RA, Jamaleddine GW. Buspirone treat-
ment for apneustic breathing in brain stem infarct. Respir Care
2003;48:956 – 8
5. Sahibzada N, Ferreira M, Wasserman AM, Taveira-DaSilva AM,
Gillis RA. Reversal of morphine-induced apnea in the anesthe-
tized rat by drugs that activate 5-hydroxytryptamine(1A) recep-
tors. J Pharmacol Exp Ther 2000;292:704 –13
906 www.anesthesia-analgesia.org
6. Bardin L, Tarayre JP, Malfetes N, Koek W, Colpaert FC.
Profound, non-opioid analgesia produced by the high-efficacy
5-HT(1A) agonist F 13640 in the formalin model of tonic
nociceptive pain. Pharmacology 2003;67:182–94
7. Bardin L, Tarayre JP, Koek W, Colpaert FC. In the formalin
model of tonic nociceptive pain, 8-OH-DPAT produces
5-HT1A receptor-mediated, behaviorally specific analgesia.
Eur J Pharmacol 2001;421:109 –14
8. Nadeson R, Goodchild CS. Antinociceptive role of 5-HT1A
receptors in rat spinal cord. Br J Anaesth 2002;88:679 – 84
9. Clarke RW, Ogilvie J, Houghton AK. Enhancement and depression
of spinal reflexes by 8-hydroxy-2-(di-n-propylamino) tetralin in the
decerebrated and spinalized rabbit: involvement of 5-HT1A-
and non-5-HT1A-receptors. Br J Pharmacol 1997;122:631– 8
10. Zhang YQ, Gao X, Ji GC, Huang YL, Wu GC, Zhao ZQ.
Expression of 5-HT1A receptor mRNA in rat lumbar spinal
dorsal horn neurons after peripheral inflammation. Pain
2002;98:287–95
11. Bardin L, Assie MB, Pelissou M, Royer-Urios I, Newman-
Tancredi A, Ribet JP, Sautel F, Koek W, Colpaert FC. Dual,
hyperalgesic, and analgesic effects of the high-efficacy
5-hydroxytryptamine 1A (5-HT1A) agonist F 13640 [(3-chloro-
4-fluoro-phenyl)-[4-fluoro-4-{[(5-methyl-pyridin-2-ylmethyl)-
amino]-methyl}piperidin-1-yl]methanone, fumaric acid salt]:
relationship with 5-HT1A receptor occupancy and kinetic param-
eters. J Pharmacol Exp Ther 2005;312:1034 – 42
12. Guenther U, Manzke T, Wrigge H, Dutschmann M, Zinserling
J, Putensen C, Hoeft A. The counteraction of opioid-induced
ventilatory depression by the serotonin 1A-agonist 8-OH-
DPAT does not antagonize antinociception in rats in situ and
in vivo. Anesth Analg 2009;108:1169 –76
13. De Vry J, Schohe-Loop R, Heine HG, Greuel JM, Mauler F,
Schmidt B, Sommermeyer H, Glaser T. Characterization of the
aminomethylchroman derivative BAY ⫻ 3702 as a highly
potent 5-hydroxytryptamine1A receptor agonist. J Pharmacol
Exp Ther 1998;284:1082–94
14. Schwarz T, Beckermann B, Buehner K, Mauler F, Schuhmacher
J, Seidel D, Steinke W, Weinz C, Zimmerd D. Pharmacokinetics
of repinotan in healthy and brain injured animals. Biopharm
Drug Dispos 2005;26:259 – 68
15. Harkany T, Mulder J, Horvath KM, Keijser J, van der
Meeberg EK, Nyakas C, Luiten PG. Oral post-lesion admin-
istration of 5-HT(1A) receptor agonist repinotan hydrochlo-
ride (BAY ⫻ 3702) attenuates NMDA-induced delayed
neuronal death in rat magnocellular nucleus basalis. Neu-
roscience 2001;108:629 – 42
16. Ohman J, Braakman R, Legout V. Repinotan (BAY ⫻ 3702): a
5HT1A agonist in traumatically brain injured patients. J Neuro-
trauma 2001;18:1313–21
17. Teal P, Silver FL, Simard D. The BRAINS study: safety,
tolerability, and dose-finding of repinotan in acute stroke. Can
J Neurol Sci 2005;32:61–7
18. Teal P, Davis S, Hacke W, Kaste M, Lyden PD, Fierus M. A
randomized, double-blind, placebo-controlled trial to eval-
uate the efficacy, safety, tolerability, and pharmacokinetic/
pharmacodynamic effects of a targeted exposure of intrave-
nous repinotan in patients with acute ischemic stroke: modi-
fied Randomized Exposure Controlled Trial (mRECT). Stroke
2009;40:3518 –25
19. Motulsky HJ. Analyzing Data with GraphPad Prism. San
Diego: GraphPad Software Inc., 1999
20. Oertel BG, Schneider A, Rohrbacher M, Schmidt H, Tegeder I,
Geisslinger G, Lotsch J. The partial 5-hydroxytryptamine1A
receptor agonist buspirone does not antagonize morphine-
induced respiratory depression in humans. Clin Pharmacol
Ther 2007;81:59 – 68
21. Pavlakovic G, Tigges J, Crozier TA. Effect of buspirone on
thermal sensory and pain thresholds in human volunteers.
BMC Clin Pharmacol 2009;9:12
22. Guenther U, Bischoff A, Kettler D, Richter DW. 5-HT1A-
agnoists protect against opioidergic depression of respiration.
In: Urban BW, Barann M, eds. Molecular und basic mecha-
nisms of anesthesia D-49525 Lengerich: Papst Science Publish-
ers, 2001:400 –3
ANESTHESIA & ANALGESIA
23. Colpaert FC, Wu WP, Hao JX, Royer I, Sautel F, Wiesenfeld-
Hallin Z, Xu XJ. High-efficacy 5-HT1A receptor activation
causes a curative-like action on allodynia in rats with spinal
cord injury. Eur J Pharmacol 2004;497:29 –33
24. Galeotti N, Ghelardini C, Bartolini A. 5-HT1A agonists induce
central cholinergic antinociception. Pharmacol Biochem Behav
1997;57:835– 41
25. Bardin L, Colpaert FC. Role of spinal 5-HT(1A) receptors in
morphine analgesia and tolerance in rats. Eur J Pain
2004;8:253– 61
26. Graeff FG. Serotonin, the periaqueductal gray and panic.
Neurosci Biobehav Rev 2004;28:239 –59
27. Deseure K, Breand S, Colpaert FC. Curative-like analgesia in a
neuropathic pain model: parametric analysis of the dose and
the duration of treatment with a high-efficacy 5-HT(1A) recep-
tor agonist. Eur J Pharmacol 2007;568:134 – 41
28. Colpaert FC, Deseure K, Stinus L, Adriaensen H. High-efficacy
5-hydroxytryptamine 1A receptor activation counteracts opi-
oid hyperallodynia and affective conditioning. J Pharmacol
Exp Ther 2006;316:892–9
29. Calabrese EJ, Baldwin LA. U-shaped dose-responses in biol-
ogy, toxicology, and public health. Annu Rev Public Health
2001;22:15–33
30. Calabrese EJ. Hormesis: basic, generalizable, central to toxicol-
ogy and a method to improve the risk-assessment process. Int
J Occup Environ Health 2004;10:466 –7
31. Calabrese EJ, Baldwin LA. Hormesis: the dose-response revo-
lution. Annu Rev Pharmacol Toxicol 2003;43:175–97
32. Mauler F, Horvath E. Neuroprotective efficacy of repinotan
HCl, a 5-HT1A receptor agonist, in animal models of stroke
and traumatic brain injury. J Cereb Blood Flow Metab
2005;25:451–9
33. Clarke RW, Ward RE. The role of 5-HT(1A)-receptors in
fentanyl-induced bulbospinal inhibition of a spinal withdrawal
reflex in the rabbit. Pain 2000;85:239 – 45
34. Gjerstad J, Tjolsen A, Hole K. The effect of 5-HT1A receptor
stimulation on nociceptive dorsal horn neurones in rats. Eur
J Pharmacol 1996;318:315–21
35. Manzke T, Dutschmann M, Schlaf G, Morschel M, Koch UR,
Ponimaskin E, Bidon O, Lalley PM, Richter DW. Serotonin targets
inhibitory synapses to induce modulation of network functions.
Philos Trans R Soc Lond B Biol Sci 2009;364:2589 – 602
October 2010 • Volume 111 • Number 4
36. Dutschmann M, Waki H, Manzke T, Simms AE, Pickering AE,
Richter DW, Paton JF. The potency of different serotonergic
agonists in counteracting opioid evoked cardiorespiratory dis-
turbances. Philos Trans R Soc Lond B Biol Sci 2009;364:2611–23
37. Eriksson TM, Golkar A, Ekstrom JC, Svenningsson P, Ogren
SO. 5-HT7 receptor stimulation by 8-OH-DPAT counteracts the
impairing effect of 5-HT(1A) receptor stimulation on contex-
tual learning in mice. Eur J Pharmacol 2008;596:107–10
38. Kellett DO, Ramage AG, Jordan D. Central 5-HT7 receptors are
critical for reflex activation of cardiac vagal drive in anaesthe-
tized rats. J Physiol 2005;563:319 –31
39. Damaso EL, Bonagamba LG, Kellett DO, Jordan D, Ramage
AG, Machado BH. Involvement of central 5-HT7 receptors in
modulation of cardiovascular reflexes in awake rats. Brain Res
2007;1144:82–90
40. White DA, Reitan JA, Kien ND, Thorup SJ. Decrease in
vascular resistance in the isolated canine hindlimb after graded
doses of alfentanil, fentanyl, and sufentanil. Anesth Analg
1990;71:29 –34
41. Lutsep HL. Repinotan, a 5-HT1A agonist, in the treatment of
acute ischemic stroke. Curr Drug Targets CNS Neurol Disord
2005;4:119 –20
42. Hutto CW Jr, Crowder WF. Dosage choices of rats for mor-
phine, for heroin, and between morphine and heroin. Pharma-
col Biochem Behav 1997;58:133– 40
43. Eisenach JC, Gebhart GF. Intrathecal amitriptyline: antino-
ciceptive interactions with intravenous morphine and intra-
thecal clonidine, neostigmine, and carbamylcholine in rats.
Anesthesiology 1995;83:1036 – 45
44. Sandkuhler J, Gebhart GF. Characterization of inhibition of a
spinal nociceptive reflex by stimulation medially and laterally
in the midbrain and medulla in the pentobarbital-anesthetized
rat. Brain Res 1984;305:67–76
45. Bonnefont J, Chapuy E, Clottes E, Alloui A, Eschalier A. Spinal
5-HT1A receptors differentially influence nociceptive process-
ing according to the nature of the noxious stimulus in rats:
effect of WAY-100635 on the antinociceptive activities of
paracetamol, venlafaxine and 5-HT. Pain 2005;114:482–90
www.anesthesia-analgesia.org 907
Society for Technology in Anesthesia
Section Editor: Dwayne Westenskow

Goal-Directed Fluid Management Based on the Pulse

Oximeter–Derived Pleth Variability Index Reduces
Lactate Levels and Improves Fluid Management
Patrice Forget, MD,* Fernande Lois, MD,* and Marc de Kock, MD, PhD*

BACKGROUND: Dynamic variables predict fluid responsiveness and may improve fluid manage-
ment during surgery. We investigated whether displaying the variability in the pulse oximeter
plethysmogram (pleth variability index; PVI) would guide intraoperative fluid management and
improve circulation as assessed by lactate levels.
METHODS: Eighty-two patients scheduled for major abdominal surgery were randomized into 2
groups to compare intraoperative PVI-directed fluid management (PVI group) versus standard
care (control group). After the induction of general anesthesia, the PVI group received a 500-mL
crystalloid bolus and a crystalloid infusion of 2 mL 䡠 kg⫺1 䡠 h⫺1. Colloids of 250 mL were
administered if the PVI was ⬎13%. Vasoactive drug support was given to maintain the mean
arterial blood pressure above 65 mm Hg. In the control group, an infusion of 500 mL of
crystalloids was followed by fluid management on the basis of fluid challenges and their effects
on mean arterial blood and central venous pressure. Perioperative lactate levels, hemodynamic
data, and postoperative complications were recorded prospectively.
RESULTS: Intraoperative crystalloids and total volume infused were significantly lower in the
goal-directed PVI group. Lactate levels were significantly lower in the PVI group during surgery and
48 hours after surgery (P ⬍ 0.05).
CONCLUSIONS: PVI-based goal-directed fluid management reduced the volume of intraoperative
fluid infused and reduced intraoperative and postoperative lactate levels. (Anesth Analg 2010;
111:910 –4)

H ypovolemia occurs frequently in the operating

room. Its diagnosis remains difficult, but assessment
of the adequacy of intravascular volume is of prime
importance to maintain cardiac output and thus avoid tissue
hypoxia. In 1 meta-analysis the authors observed that periop-
erative hemodynamic optimization reduced mortality.1
For many years, cardiac filling pressures were used to
guide intravascular volume therapy. This, however, is not a
reliable predictor of fluid responsiveness.2 Dynamic vari-
ables (indices evaluating the response to a cyclic preload
variation) provide a better prediction of fluid responsive-
ness.3 Among these, the arterial pulse pressure variation
induced by mechanical ventilation has been demonstrated
as one of the best tools to guide volume therapy.3 Lopes et
al. showed an improvement in postoperative outcome after
high-risk surgery when the pulse pressure variation was
used to guide intraoperative fluid therapy.4 Natalini et al.
and Cannesson et al. demonstrated that respiratory varia-
tions in the amplitude of the pulse oximeter plethysmo-
graphic waveform and in the pulse pressure both predict
fluid responsiveness.5– 8 Zimmerman et al. showed that
pleth variability index (PVI) predicts fluid responsiveness
From the *Department of Anesthesiology, Université catholique de Louvain,
St.-Luc Hospital, Brussels, Belgium.
Accepted for publication May 27, 2010.
Address correspondence and reprint requests to Patrice Forget, Department
of Anesthesiology, St.-Luc Hospital, av. Hippocrate 10 –1821, 1200 Brussels,
Belgium. Address e-mail to forgetpatrice@yahoo.fr.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181eb624f
as accurately as does stroke volume variation.9 Neverthe-
less, it remains unknown whether the optimization of the
plethysmogram variability that occurs intraoperatively im-
proves fluid management and circulation. To investigate
this, we used a pulse oximeter to continuously monitor the
PVI.7 We measured the impact of PVI-based goal-directed
fluid management on perioperative lactate levels.
METHODS
After approval of the Ethics Committee of St.-Luc Hos-
pital (Brussels, Belgium) (www.clinicaltrials.gov, no.
NCT00816153), and after obtaining written informed
consent, a pilot study including 20 patients (10 per group)
was conducted for the power analysis. Results showed an
improvement of 20% of the primary outcome (whole blood
lactate levels) with the use of the PVI. A sample size of 37
patients per group was calculated for a 0.05 difference
(2-sided) with a power of 80%.
Between May and September 2008, we obtained written
informed consent from 86 patients who met the inclusion
criteria: older than 18 years and the absence of cardiac
arrhythmias, ultrasonographic cardiac ejection fraction
⬍30%, lung pathology prohibiting mechanical ventilation
with tidal volumes larger than 6 mL 䡠 kg⫺1, and kidney
dialysis. They were scheduled for esophagectomy, gastric
resection/suture, hepatectomy, pancreatectomy, or intesti-
nal and colorectal surgeries. Patients were randomized to
either the PVI group or the control group.
Heart rate, arterial blood pressure, oxygen saturation,
inhaled gas concentrations, and temperature were measured

910 www.anesthesia-analgesia.org October 2010 • Volume 111 • Number 4

continuously by a Datex S/5 monitor (Datex Ohmeda®, GE
Healthcare). A Masimo Set version V7.1.1.5 pulse oximeter
(Masimo Co., Irvine, California) was placed on the patient’s
finger for the continuous monitoring of PVI. A 20-G radial
arterial catheter and a central venous access catheter were
inserted at the end of the induction phase. A thoracic epidural
catheter was placed before the induction of the general
anesthesia. Anesthesia was induced with propofol 2 to 4 mg 䡠
kg⫺1 and atracurium or rocuronium 0.4 to 0.6 mg 䡠 kg⫺1 and
maintained with sevoflurane or desflurane. The lungs were
ventilated with 6 to 8 mL 䡠 kg⫺1 of tidal volume, I:E ⫽ 1:2. The
Figure 1. Trial profile. PVI group: pleth variability index– guided fluid
frequency was set to maintain normocapnia (Paco2 target ⫽ management.
40 ⫾ 3 mm Hg).
In the PVI group, 500 mL of crystalloids (NaCl 0.9% or
P-Lyte®, Baxter) was infused during induction, followed by Table 1. Preoperative Characteristics, Incidence
a 2 mL 䡠 kg⫺1 䡠 h⫺1 continuous infusion. If PVI was higher of Chronic Diseases, Type and Duration of Surgery
than 13% for ⬎5 minutes, we gave a 250-mL bolus of and Anesthesia, Use of Epidural Analgesia in the
colloid (hydroxyethyl starch 6%, Voluven®, Fresenius Pleth Variability Index (PVI) Group (PVI-Guided
Kabi). The dose was repeated every 5 minutes if PVI was Fluid Management) and Control Group
still higher than 13%. Norepinephrine was given as needed PVI group Control group
(N ⴝ 41) (N ⴝ 41)
to maintain a mean arterial blood pressure ⬎65 mm Hg.
Age (years) 59 ⫾ 14 61 ⫾ 12
In the control group, 500 mL of crystalloids was infused Weight (kg) 71 ⫾ 15 68 ⫾ 16
during induction, followed by a continuous infusion of Height (cm) 169 ⫾ 9 170 ⫾ 9
crystalloids (4 to 8 mL 䡠 kg⫺1 䡠 h⫺1). A bolus of colloids was Sex (female/male) 16/25 (39/61) 16/25 (39/61)
given if acute blood loss of ⬎50 mL occurred, if the mean ASA score
2 22 (54) 22 (54)
arterial blood pressure decreased below 65 mm Hg, or if the 3 19 (46) 19 (46)
central venous pressure decreased below 6 mm Hg. A Chronic diseases
repeat bolus was given after waiting 5 minutes if any one of Cirrhosis 3 (7) 0 (0)
the criteria was met. If the mean arterial blood pressure Chronic obstructive pulmonary 2 (5) 2 (5)
decreased below 65 mm Hg and remained unresponsive to disease
Hypertension 18 (44) 13 (32)
fluids, norepinephrine was given to maintain the mean Peripheral vascular disease 7 (17) 7 (17)
arterial blood pressure above 65 mm Hg. Coronary artery disease 5 (12) 2 (5)
Arterial blood samples were taken at the time of skin Other cardiomyopathy 2 (5) 4 (10)
incision, each hour during surgery and 6, 12, 18, 24, 36, and Diabetes mellitus 4 (10) 2 (5)
Preoperative biological values
48 hours after the end of surgery. The lactate concentration Hemoglobin (g 䡠 dL⫺1) 12.5 ⫾ 2 12.7 ⫾ 2
was measured using an ABL 620 analyzer (Radiometer, Serum creatinine (mg 䡠 dL⫺1) 0.96 ⫾ 0.2 0.97 ⫾ 0.3
Copenhagen, Denmark). Serum creatinine concentrations Type of surgery
were measured 24 and 48 hours after surgery. The anesthe- Upper gastrointestinal 7 (17) 5 (12)
Hepato-biliary 11 (27) 15 (37)
siologist identified and recorded instances of intraoperative Lower gastrointestinal 24 (59) 22 (54)
hypotension (systolic blood pressure 20% below the value Laparoscopic approach 5 (12) 5 (12)
measured the day before surgery, while the patient was Duration of surgery (minutes) 295 ⫾ 125 301 ⫾ 154
resting quietly for at least 15 minutes) and oliguria (urine Duration of anesthesia 346 ⫾ 125 356 ⫾ 158
output ⬍0.5 mL 䡠 kg⫺1 for ⬎2 hours). (minutes)
Epidural analgesia 33 (81) 29 (71)
During the first 30 days after surgery, a blinded postop-
erative care team member identified, collected, and re- One patient per group had two types of surgery. Data are presented as
mean ⫾ SD or number (%). P ⬎ 0.05 for all the data.
corded instances of postoperative infection, pulmonary
embolism, acute myocardial infarction, acute lung
injury/acute respiratory distress syndrome, pulmonary patients completed the protocol and were analyzed. No
edema, arrhythmia, stroke, cardiac arrest, coagulopathy patient met abandon criteria.
(platelets ⬍100,000 ␮L⫺1, international normalized ratio Student’s t test was used to compare normally distrib-
⬎2), hepatic dysfunction, nausea or vomiting necessitating uted continuous variables and ␹2 for categorical variables.
treatment, upper digestive hemorrhage, leakage of anasto- Homogeneity of variances was verified by the Levene’s
mosis, and mortality. test. A P value ⬍0.05 was considered statistically signifi-
cant. Data are expressed as mean (⫾sd), mean [95% confi-
dence interval], or number (percentage). STATISTICA (data
Statistical Analysis analysis software system) version 7 (Statsoft, Inc., 2004) was
Data were analyzed by comparing the patients in the PVI used for all analyses.
group with those in the control group using a modified
intention-to-treat analysis (4 patients were excluded after RESULTS
the randomization for intraoperative arrhythmia or cancel- Table 1 lists the patients’ history and surgery. There were
lation of the surgery, 2 per group; Fig. 1). The remaining 82 no preoperative differences between the goal-directed fluid

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 911

Pleth Variability Index and Fluid Management

Table 2. Fluids Administered, Blood Loss, Hemodynamic Status, Physiologic Status, and Renal Function
During and After Surgery in the Pleth Variability Index (PVI) Group (PVI-Guided Fluid Management) and in
the Control Group
PVI group Control group
(N ⴝ 41) (N ⴝ 41) P value
Intraoperative fluids (mL)
Crystalloids 1363 关1185–1540兴 1815 关1568–2064兴 0.004
Colloids 890 关709–1072兴 1003 关779–1227兴 0.43
Blood products 141 关53–230兴 99 关20–179兴 0.48
Total of intraoperative fluids 2394 关2097–2692兴 2918 关2478–3358兴 0.049
Blood losses 349 关230–468兴 440 关242–637兴 0.43
Postoperative fluids (24 hours)
Crystalloids 3107 关2760–3454兴 3516 关3009–4024兴 0.17
Colloids 268 关126–409兴 358 关175–540兴 0.43
Blood products 8 关⫺8–25兴 44 关⫺45–133兴 0.41
Lactate levels (mMol 䡠 L⫺1)
Maximum intraoperative 1.2 关1–1.4兴 1.6 关1.2–2兴 0.04
At 24 hours 1.4 关1.3–1.5兴 1.8 关1.5–2.1兴 0.02
At 48 hours 1.2 关1–1.3兴 1.4 关1.2–1.5兴 0.03
Lactate levels ⬎1.7 mMol 䡠 L⫺1
Intraoperatively 7 (17) 4 (10) 0.33
At 24 hours 2 (5) 28 (68) <0.0001
At 48 hours 0 8 (20) 0.003
Lactate levels ⬎5 mMol 䡠 L⫺1
Intraoperatively 0 1 (2) 0.31
At 24 hours 0 1 (2) 0.31
At 48 hours 0 1 (2) 0.31
Intraoperative hypotension 22 (54) 28 (68) 0.17
Continuous infusion of norepinephrine
Intraoperative 9 (22) 9 (22) 1.0
At 24 hours 3 (7) 1 (2) 0.31
Renal function diuresis
Intraoperative oliguria 13 (32) 17 (42) 0.34
Postoperative oliguria (24 hours) 3 (8) 3 (8) 0.97
Serum creatinine (mg 䡠 dL⫺1)
At 24 hours 1.01 关0.9–1.1兴 1.12 关0.9–1.3兴 0.32
At 48 hours 0.91 关0.8–1兴 1.09 关0.9–1.3兴 0.11
Initiation of dialysis 1 (2) 0 (0) 0.32
Lactate levels: normal value 0.9 –1.7 mMol 䡠 L⫺1. Oliguria was defined as a urinary output ⬍0.5 mL 䡠 kg⫺1 for more than 2 hours. Data are presented as
mean 关95% confidence interval兴 or number (%).
P ⬍ 0.05 was considered as statistically significant (boldface numerical entries).

that lactate levels were lower in the PVI group during and
after surgery. There were no statistically significant differ-
ences in the incidence of hypotension, cardiovascular res-
cue, or renal dysfunction. Two patients in the PVI group
died from septic shock 20 days and 33 days after surgery
because of a failed anastomosis (Table 3).

Figure 2. Lactate levels during and after surgery in the pleth
variability index (PVI)– guided group (PVI-guided fluid management)
and in the control group. Intraoperative: maximum intraoperative
value. Data are presented as mean ⫾ SEM. *P ⬍ 0.05.
management group and the control group. Table 2 shows
that during surgery, patients in the PVI-directed fluid
management group were given less total fluid and less
crystalloid intraoperatively than was the control group.
There were no differences postoperatively. Figure 2 shows
DISCUSSION
We found that PVI-guided fluid management resulted in
less crystalloid administered perioperatively and reduced
lactate levels during and after major abdominal surgery.
Lactate levels provide an indirect but sensitive measure of
organ perfusion. Lactate is clearly correlated with the
adequacy of intravascular volume, tissue hypoxia, and
energy failure due to bloodflow redistribution.10 Lactate
levels can be improved by the optimization of the fluid
status and cardiac preload.2,4
Our results confirm the conclusion of Lopes et al. The
use of the noninvasive PVI, or the invasively obtained
pulse pressure variation, improves perioperative fluid
management.4 In Lopes et al.’s study, the average
amount of fluids was larger in the group guided by pulse
pressure variation, in contrast with our results. The
difference in results may be explained by the presence of
hypovolemia in some patients and hypervolemia in
others. These results therefore argue the superiority of

912 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

 Table 3. Postoperative Complications and Intensive Care Unit/Hospital Stay in the Pleth Variability Index
(PVI) Group (PVI-Guided Fluid Management) and in the Control Group
PVI group Control group
(N ⴝ 41) (N ⴝ 41) P Value
Postoperative complications
Infection of surgery site 8 (20) 8 (20) 1.0
Other infections (pulmonary, line-related, other abdominal) 6 (15) 7 (17) 0.77
Cardiovascular complications (acute myocardial infarction, 4 (10) 8 (20) 0.26
acute lung injury/acute respiratory distress
syndrome, pulmonary edema, arrhythmia)
Coagulopathy 5 (12) 6 (15) 0.75
Nausea and/or vomiting 0 (0) 4 (16) 0.08
Upper digestive hemorrhage 4 (10) 3 (7) 0.78
Leakage of anastomosis 5 (12) 5 (12) 1.0
Morbidity (event per patient) 1.2 ⫾ 1.8 1.5 ⫾ 2.2 0.46
Mortality 2 (5) 0 (0) 0.16
Length of stay
Postoperative mechanical ventilation 1 (2) 3 (7) 0.31
Intensive care unit (days) 2.2 ⫾ 5.7 1.8 ⫾ 7.2 0.71
Hospital (days) 15.1 ⫾ 14.3 16.0 ⫾ 17.8 0.78
Data are presented as mean ⫾ SD or number (%).

goal-directed fluid management over simplistic restrictive
or liberal approaches for fluid management, avoiding hy-
povolemia and hypervolemia.11,12
Unlike Lopes et al., we did not find an improvement
in terms of the number of complications. The much
higher incidence of hypovolemia reported by Lopes et al.
may account for this difference. The clinical significance of
lower lactate levels in our relatively small study may be
questioned. Additionally, fluid management in the control
group was different by design, favoring greater fluid
crystalloid administration (2 mL 䡠 kg⫺1 䡠 h⫺1 in the PVI
group vs. 4 to 8 mL 䡠 kg⫺1 䡠 h⫺1 in the control group), and
it was possibly influenced by the fact that the control group
had a greater blood loss (440 [242 to 637] mL vs. 349 [230 to
468] mL) (although this was not statistically significant).
When mean arterial blood pressure decreased to ⬍65
mm Hg, the PVI group received norepinephrine, whereas
the control group received norepinephirne and a bolus of
crystalloid. In our study a “learning contamination bias”
may have blunted the differences between the groups. This
bias occurs when a team member gains experience with
pulse pressure variation and begins, intuitively, to use
respiratory variations of the arterial pressure curve to treat
patients in the control group. However, small variations are
difficult to see without using a device that makes the
calculations from the curve.
The PVI was calculated by the new Masimo Set pulse
oximeter (Masimo Co., Irvine, California) from the respira-
tory variations in the perfusion index (PI). The PI is the
percentage amplitude difference between the pulsatile in-
frared signal and the nonpulsatile infrared signal. The PVI
is calculated by measuring changes in the PI during the
respiratory cycle: PVI ⫽ [(PImax ⫺ PImin)/PImax] ⫻ 100.
Cannesson et al. have demonstrated that the PVI predicts
fluid responsiveness in the operating room. They showed
that the cutoff value to distinguish responders from non-
responders to intravascular volume expansion (in terms of
an increase of cardiac index) was a PVI ⬎14%.7 We
confirmed their results in a preliminary study (data not
shown) and consequently chose 14% as the threshold for
fluid loading.
Whereas the PVI may be useful in most patients, our
exclusion criteria limit the application of our results in
some patients. To maintain homogeneity between the 2
study groups, we did not include patients with severe
cardiac insufficiency (ejection fraction ⬍30%) or chronic
dialysis. Moreover, the dynamic variables must not be
calculated in the presence of arrhythmia. One patient per
group was excluded because of an intraoperative arrhyth-
mia. Additionally, these results cannot be extrapolated to
other devices that calculate the respiratory variation of the
plethysmographic curve. The algorithm used to process
the signal may explain the poor accuracy observed by
others.13 Moreover, we did not measure the possible
impact of the use of epidural analgesia and thoracotomy
in some patients.
In conclusion, the use of PVI-guided fluid management
was associated with lower lactate levels during major
abdominal surgery. Patients in the PVI-guided group were
given less crystalloid. Reduced lactate levels in PVI-guided
patients suggests that PVI-guided fluid management may
lead to fluid administration that is tailored to each indi-
vidual patient’s needs.
ACKNOWLEDGMENTS
Masimo Corporation graciously provided devices during the
study protocol.
REFERENCES
1. Poeze M, Greve JWM, Ramsay G. Meta-analysis of hemody-
namic optimisation: relationship to methodological quality.
Crit Care 2005;9:R771–9
2. Cavallaro F, Sandroni C, Antonelli M. Functional hemody-
namic monitoring and dynamic indices of fluid responsive-
ness. Minerva Anestesiol 2008;74:123–35
3. Michard F, Teboul JL. Predicting fluid reponsiveness in ICU
patients. A critical analysis of the evidence. Chest 2002;
121(6):2000 – 8

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 913

Pleth Variability Index and Fluid Management
4. Lopes MR, Oliveira MA, Pereira VO, Lemos IP, Auler JO Jr,
Michard F. Goal-directed fluid management based on pulse
pressure variation monitoring during high-risk surgery: a pilot
randomized controlled trial. Crit Care 2007;11(5):R100
5. Desebbe O, Cannesson M. Using ventilation-induced plethys-
mographic variations to optimize patient fluid status. Curr
Opin Anaesthesiol 2008;21:772– 8
6. Natalini M, Rosano A, Taranto M, Faggian B, Vittorielli E,
Bernardini A. Arterial versus plethysmographic dynamic
indices to test responsiveness for testing fluid administra-
tion in hypotensive patients: a clinical trial. Anesth Analg
2006;103(6):1478 – 84
7. Cannesson M, Desebbe O, Rosamel P, Delannoy B, Robin J,
Bastien O, Lehot JJ. Pleth variability index to monitor the
respiratory variations in the pulse oximeter plethysmographic
waveform amplitude and predict fluid responsiveness in the
operating theatre. Br J Anaesth 2008;101(2):200 – 6
8. Cannesson M, Attof Y, Rosamel P, Desebbe O, Joseph P,
Metton O, Bastien O, Lehot JJ. Respiratory variations in
pulse oximetry plethysmographic waveform amplitude to
predict fluid responsiveness in the operating room. Anes-
thesiology 2007;106(6):1105–11
914 www.anesthesia-analgesia.org
9. Zimmermann M, Feibicke T, Keyl C, Prasser C, Moritz S, Graf
BM, Wiesenack C. Accuracy of stroke volume variation com-
pared with pleth variability index to predict fluid responsive-
ness in mechanically ventilated patients undergoing major
surgery. Eur J Anaesthesiol 2009; [Epub ahead of print]
10. Valenza F, Aletti G, Fossali T, Chevallard G, Sacconi F, Irace M,
Gattinoni L. Lactate as a marker of energy failure in critically ill
patients: hypothesis. Crit Care 2005;9(6):588 –93
11. Bundgaard-Nielsen M, Holte K, Secher NH, Kehlet H.
Monitoring of peri-operative fluid administration by indi-
vidualized goal-directed therapy. Acta Anaesthesiol Scand
2007;51(3):331– 40
12. Bundgaard-Nielsen M, Ruhnau B, Secher NH, Kehlet H. Flow-
related techniques for preoperative goal-directed fluid optimi-
sation. Br J Anaesth 2007;98(1):38 – 44
13. Landsverk SA, Hoiseth LO, Kvandal P, Hisdal J, Skare O,
Kirkeboen KA. Poor agreement between respiratory variations
in pulse oximetry photoplethysmographic waveform ampli-
tude and pulse pressure in intensive care unit. Anesthesiology
2008;109(5):849 –55
ANESTHESIA & ANALGESIA
A Long-Term Clinical Evaluation of AutoFlow During
Assist-Controlled Ventilation: A Randomized
Controlled Trial
Sigismond Lasocki, MD, PhD, Françoise Labat, MD, Gaetan Plantefeve, MD, Mathieu Desmard, MD,
and Hervé Mentec, MD

BACKGROUND: Many new mechanical ventilation modes are proposed without any clinical
evaluation. “Dual-controlled” modes, such as AutoFlow™, are supposed to improve patient–
ventilator interfacing and could lead to fewer alarms. We performed a long-term clinical
evaluation of the efficacy and safety of AutoFlow during assist-controlled ventilation, focusing on
ventilator alarms.
METHODS: Forty-two adult patients, receiving mechanical ventilation for more than 2 days with
a Dräger Evita 4 ventilator were randomized to conventional (n ⫽ 21) or AutoFlow (n ⫽ 21)
assist-controlled ventilation. Sedation was given using a nurse-driven protocol. Ventilator-
generated alarms were exhaustively recorded from the ventilator logbook with a computer. Daily
blood gases and ventilation outcome were recorded.
RESULTS: A total of 403 days of mechanical ventilation were studied and 45,022 alarms were
recorded over a period of 8074 hours. The course of respiratory rate, minute ventilation, FIO2,
positive end-expiratory pressure, PaO2/FIO2, PaCO2, and pH and doses and duration of sedation
did not differ between the 2 groups. Outcome (duration of mechanical ventilation, ventilator-
associated pneumonia, course of Sequential Organ Failure Assessment score, or death) was not
different between the 2 groups. The number of alarms per hour was lower with AutoFlow
assist-controlled ventilation: 3.3 [1.5 to 17] versus 9.1 [5 to 19], P ⬍ 0.0001 (median [quartile
range]). In multivariate analysis, a low alarm rate was associated with activation of AutoFlow and
a higher midazolam dose.
CONCLUSIONS: This first long-term clinical evaluation of the AutoFlow mode demonstrated its
safety with regard to gas exchange and patient outcome. AutoFlow also allowed a very marked
reduction in the number of ventilator alarms. (Anesth Analg 2010;111:915–21)

D ual-controlled” ventilation modes1,2 are reported

to combine the advantages of both volume and
pressure-controlled ventilation modes. The only
studies comparing dual modes to conventional assist-
controlled ventilation (ACV) focused on short-term effects
(several hours). They report a reduction of inspiratory
pressure.3,4 Dräger’s AutoFlow™ (AF) is one of these
dual-controlled ventilation modes that also allows sponta-
neous breathing throughout the respiratory cycle. Drager
claims that AF is expected to improve patient–ventilator
interfacing and could decrease the number of times patients
fight the ventilator. AF could consequently reduce the
number of ventilator alarms. Alarms are partly responsible
for the high noise level in intensive care units (ICU),5–7 and
a reduction of ventilator alarms would therefore be benefi-
cial to both patients and ICU staff. ICU staff also fail to
correctly identify many of these alarms.8 Reducing the
From the Réanimation Polyvalente, Centre Hospitalier Victor Dupouy,
Argenteuil, France.
Accepted for publication June 19, 2010.
Sigismond Lasocki, MD, PhD, is currently affiliated with Réanimation
Chirurgicale, APHP, CHU Bichat Claude Bernard, Paris, France.
The Hospital Victor Dupouy, Argenteuil, France, supported this study.
Drägger SA provided only technical assistance for the recording of a
ventilator logbook, but had no access to the data and was not involved in
the preparation of this manuscript.
Address correspondence to Sigismond Lasocki, MD, PhD, Réanimation
Chirurgicale, CHU Bichat, 46 rue Henri Huchard, 75018 Paris, France.
Address e-mail to sigismond@lasocki.com.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181f00015
number of alarms may therefore improve alarm efficiency.9
We hypothesize that activation of the AF would reduce the
number of ventilation-generated alarms, without impairing
the patient’s ventilation. Because no clinical evaluation of
this ventilation mode is available, this study was also
designed to clinically evaluate AF activation.
The aim of this randomized controlled study was to
clinically evaluate AF activation during ACV, with regard
to ventilator-generated alarms (primary aim) and to gas
exchange and patient outcome (secondary aims).
METHODS
The study protocol was approved by the Comité Consul-
tatif de Protection des Personnes dans la Recherche
Biomédicale (independent ethics committee) of Saint-
Germain-en-Laye, France (ClinicalTrial.gov identifier:
NCT0092774). Written informed consent was obtained from
the patient or next of kin.
Patients
Adult patients admitted to the ICU of Victor Dupouy
Hospital, Argenteuil, France, were eligible when they re-
quired ACV with an Evita4 ventilator (Dräger Medical,
Antony, France) for an expected duration of ⬎2 days.
Patients were not included in the case of coma, ventilation
for ⬎12 hours before inclusion, pregnancy, or inclusion in
another study.
AutoFlow
AF is a dual-control mechanical ventilation mode associ-
ated with ACV.1,10 All breaths are pressure-controlled, with

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 915

Evaluation of AutoFlow Controlled Ventilation
Figure 1. Sedation algorithm used by nurses to conduct patient sedation (VAS ⫽ visual analgesia scale; Ramsay ⫽ sedation score according
to the Ramsay scale11).
a delivered level of pressure support that varies from
breath to breath to deliver the set tidal volume (Vt). AF
uses a feedback loop that regulates inspiratory flow. Dy-
namic compliance is measured breath by breath, and the
required ␦ pressure for the next breath is calculated by
dividing the desired Vt by dynamic compliance. Changes
of inspiratory pressure from breath to breath are limited to
3 mbar. When inspiratory pressure reaches the upper
pressure limit minus 5 mbar, inspiratory time is increased
within the limits defined by the set respiratory rate.
Mechanical Ventilation
Patients were randomly assigned to the control group
(AF⫺) or the AF group (AF⫹) by opening a sealed enve-
lope. Attending physicians chose all respiratory settings,
without intervention by the study investigators. The upper
limit of inspiratory pressure alarm was initially set at 50 cm
H2O. Other alarm limits were set at the manufacturer’s
default values, which varied according to the patient’s
body weight. Attending physicians were allowed to change
any alarm limit and ventilator mode when clinically indi-
cated, except that AF was always used with ACV in the
AF⫹ group, and AF was never used with ACV in the AF⫺
group. Discontinuation of mechanical ventilation was per-
formed according to our ICU standard protocol, by the
gradual reduction of pressure support. Blood gases were
obtained at least once daily during the first days. Morning
values of ventilator settings, highest Fio2, highest positive
end-expiratory pressure, highest Paco2, lowest Pao2/Fio2
ratio, and lowest pH were recorded daily.
Sedation
Patients were sedated by continuous infusions of midazo-
lam and fentanyl according to a nurse-driven protocol (Fig.
1). The sedation goal was a Ramsay score11 of 2 or 3.
916 www.anesthesia-analgesia.org
Ventilator-Generated Alarms
Ventilators were periodically connected to a personal
computer to download the number of alarms, changes in
ventilator settings, alarm limit settings, and the number
of times alarms were manually silenced (silence knob
activation).
Outcome
Organ failure was assessed daily using the Sequential
Organ Failure Assessment (SOFA) score.12 Duration of
mechanical ventilation, ICU and hospital survival, the
incidence of pneumothorax, and the incidence of ventilator-
associated pneumonia were recorded.
Statistical Analysis
Results are expressed as median[Q1–Q3] or mean ⫾ sd.
Comparisons were performed with a Mann–Whitney test
or Student’s t test, a ␹2 test with Yates’ correction, or an
analysis of variance (ANOVA) for repeated measures as
appropriate. Analysis was performed by either intention to
treat on the whole study period or per protocol (for the
period during which patients were on ACV, with or
without AF), depending on the variable considered. Thus
we report the alarm rate for the different ventilation mode
(ACV or non-ACV) in both groups. Obviously, in the AF⫹
group, AF was not activated during non-ACV ventilation
(because it was not available).
To estimate the sample size of the study, we assumed
that the total alarm rate in an ICU would be 36.5
alarms/hr,13 and that ventilator alarms would account for
38% of all alarms.14 The alarm rate during ACV was
assumed to be 14 alarms/hr. A 50% reduction when AF
was added to ACV was considered to be clinically relevant.
Twenty-one patients in each group were needed to achieve
a 90% power, with an ␣ risk of 5%. To assess factors
ANESTHESIA & ANALGESIA
 Table 1. Patient Characteristics Table 2. Ventilator Settings and Gas
AFⴙ (n ⴝ 21) AFⴚ (n ⴝ 21) P Exchange Variables
Age (years) 71 关25–90兴 65 关32–88兴 0.70 AFⴙ (n ⴝ 21) AFⴚ (n ⴝ 21) P
Body weight (kg) 73 关53–101兴 64 关38–150兴 0.75 RR (cycles/min) 18 关14–22兴 20 关15–22兴 0.87
Height (cm) 168 关150–191兴 167 关150–185兴 0.49 VT (mL) 500 关400–650兴 500 关350–600兴 0.40
Gender M/F 14/7 14/7 0.99 PEEP (cm H2O) 5 关0–15兴 5 关0–13兴 0.97
COPD (%) 2 (10) 6 (29) 0.24 FIO2 1 关0.35–1兴 0.8 关0.40–1兴 0.76
Type of patient 0.99 pH 7.31 关6.98–7.60兴 7.28 关7.11–7.57兴 0.76
Medical 15 15 PaCO2 (mm Hg) 39 关31–61兴 42 关29–110兴 0.51
Urgent surgery 6 6 PaO2 (mm Hg) 141 关62–492兴 152 关49–286兴 0.87
Indication of 0.31 PaO2/FIO2 (mm Hg) 192 关62–532兴 242 关49–356兴 0.98
mechanical
Values are expressed as median 关min–max兴.
ventilation
AF⫹ ⫽ assist-controlled ventilation with AutoFlow activation; AF⫺ ⫽ assist-
ARF 15 12 controlled ventilation without AutoFlow activation; RR ⫽ respiratory rate; VT ⫽
Ventilation 4 3 tidal volume; PEEP ⫽ positive end-expiratory pressure; FIO2 ⫽ inspiratory
without oxygen fraction.
ARFa
ARF on COPD 1 3
Postoperative 1 3
Ventilator Alarms and Interventions
ventilation
Admission 60 关25–101兴 51 关19–105兴 0.33
A total of 403 days (8074 hours) of mechanical ventilation
SAPS II were studied. ACV was used for 3997 hours. AF was used
Admission SOFA 9 关2–17兴 9 关4–19兴 0.57 for 2133 hours. A total of 45,022 alarms were recorded;
Values are expressed as median 关min–max兴 or numbers (%). nearly 1 alarm every 10 minutes. During ACV, 7060 alarms
AF⫹ ⫽ assist-controlled ventilation with AutoFlow activation; AF⫺ ⫽ were recorded in the AF⫹ group, and 16,817 alarms were
assist-controlled ventilation without AutoFlow activation; COPD ⫽ chronic recorded in the AF⫺ group. Figure 3 shows that the
obstructive pulmonary disease; ARF ⫽ acute respiratory failure; SAPS II ⫽
simplified acute physiologic score (34); SOFA ⫽ Sequential Organ Failure ventilator alarm rate was lower when AF was used in
Assessment (12). conjunction with ACV (3.3 [1.5 to 17] alarms/hr with AF vs.
a
Shock and PaO2/FIO2 ⬎250 mm Hg. 9.1 [5.2 to 19] without AF [P ⬍ 0.0001]). In the AF⫺ group,
the alarm rate was lower for ventilation modes other than
ACV (mainly pressure support) than they were for ACV
influencing the alarm rate, we divided patients into 2 (without AF), but no difference was observed between
groups— higher and lower than the median alarm these ventilation modes and ACV for the AF⫹ group (Fig.
rate—and we performed a logistic regression, including all 3). The number of alarm setting modifications per hour was
the parameters that had a P value of ⬍0.1 in the univariate not different between groups (0.07 [0.02 to 0.23] vs. 0.09 [0.02
analysis (i.e., indexed midazolam, indexed fentanyl and AF to 0.23] modifications per hour for AF⫹ and AF⫺ patients,
group, with indexed midazolam and indexed fentanyl respectively; P ⫽ 0.85). Setting the high-pressure limit above
being the total dose of midazolam or fentanyl indexed to 50 cmH2O was less frequent in the AF⫹ group (0 [0 to 2] vs.
the patient’s body weight and the duration of drug infu- 2 [0 to 8] times per patient; P ⫽ 0.0007).
sion). Statistical analysis was performed with SPSS 15.0. The type of alarm differed between the 2 groups. ACV
P ⬍ 0.05 was considered statistically significant. plus AF generated fewer pressure alarms than did ACV
alone (P ⬍ 0.0001) (Fig. 4). The silence knob activation rate
was lower in the AF⫹ group during ACV (0.26 [0.1 to 1.1]
RESULTS vs. 0.72 [0.26 to 2] activation per hour; P ⫽ 0.0013).
Patients The median alarm rate during ACV was 6.37 alarms/hr.
Forty-two patients were included. No statistically signifi-
Patients with alarm rates lower than 6.37 alarms/hr were
cant differences in demographic data were observed be-
more frequently randomized to the AF⫹ group (P ⫽
tween the 2 groups (Table 1). Indications for mechanical
0.0002) and had a higher dosage of sedative drugs (fenta-
ventilation were not statistically different between groups
nyl, P ⫽ 0.02; midazolam, P ⫽ 0.07) (Table 4). In multivar-
(P ⫽ 0.31).
iate analysis, an alarm rate lower than 6.37 alarms/hr was
associated with activation of AF (OR [95% CI], 90 [5 to
Ventilation and Gas Exchange 1570], P ⫽ 0.002) and a higher midazolam dosage (OR 1801
Baseline ventilator settings and blood gases were not [3 to 1.1 106] per mg/d/kg, P ⫽ 0.02).
different between the 2 groups (Table 2). An ANOVA for
repeated measures from day 1 to day 5 showed no signifi- Outcome
cant difference between the 2 groups for either ventilator No patient suffered from pneumothorax in the AF⫹ group
settings or blood gases (Fig. 2). in comparison with 2 patients in the AF⫺ group (P ⫽ 0.48).
Four cases of ventilator-associated pneumonia were ob-
Sedation served in the AF⫹ group in comparison with 8 in the AF⫺
Sedation was not different between the 2 groups (Table 3). group (P ⫽ 0.16). The median duration of ventilation was 6
Total doses of midazolam (114 [0 to 163] vs. 150 [0 to 316] [2 to 25] versus 9 [2 to 36] days in AF⫹ and AF⫺ groups,
mg, P ⫽ 0.5) and fentanyl (6600 [0 to 55,480] vs. 8000 [0 to respectively (P ⫽ 0.33), and the median number of days free
21,500] ␮g, P ⫽ 0.8) were not different between the 2 groups of mechanical ventilation at day 28 were 15 [0 to 26] and 13
(AF⫹ and AF⫺ groups, respectively). [0 to 26], respectively (P ⫽ 0.55). An ANOVA for repeated

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 917

Evaluation of AutoFlow Controlled Ventilation

Figure 2. Time course of ventilatory settings, gas exchange, and Sequential Organ Failure Assessment (SOFA) score12 during the first 5 days
(D1 to D5) of mechanical ventilation. Solid squares represent the Dräger’s AutoFlow (AF)⫹ group, and open circles represent the control (AF⫺)
group. Mean ⫾ SEM. An analysis of variance (ANOVA) for repeated measures was not significant. PEEP ⫽ positive end-expiratory pressure.

measures from day 1 to day 5 showed no significant DISCUSSION

difference between the 2 groups for SOFA score (Fig. 2). Six This study is the first long-term clinical evaluation of AF
patients in the AF⫹ group and 9 patients in the AF⫺ group during ACV. Clinical outcome and blood gas variables
died in the ICU (P ⫽ 0.39), whereas 8 and 10 patients died were not different with and without AF. There were fewer
in the hospital, respectively (P ⫽ 0.62). ventilator alarms when AF was used.

918 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

 Table 3. Sedation Variables
AFⴙ (n ⴝ 21) AFⴚ (n ⴝ 21) P
Sedation duration (days) 3 关1–10兴 5 关0–12兴 0.44
Indexed midazolam 0.38 关0–1.13兴 0.44 关0–0.78兴 0.73
(mg/d/kg)a
Indexed fentanyl (␮g/d/kg)b 28 关0–93兴 25 关10–52兴 0.46
Sedation modifications 1.5 关0–3.3兴 1.4 关0–3.8兴 0.66
(per day)
Ramsay score of 2 or 3 2.9 关0.5–6.3兴 2.2 关0–5.5兴 0.29
(per day)c
Values are expressed as median 关min–max兴.
AF⫹ ⫽ assist-controlled ventilation with AutoFlow activation; AF⫺ ⫽ assist-
controlled ventilation without AutoFlow activation.
a
Indexed midazolam: total dose of midazolam indexed to the patient’s body
weight and the duration of drug infusion.
b
Indexed fentanyl: total dose of fentanyl indexed to the patient’s body weight Figure 4. Distribution of alarm types in the 2 groups during assist-
and the duration of drug infusion. controlled ventilation (ACV). AutoFlow ACV group (AF⫹) is repre-
c
Ramsay score (11) of 2 or 3, the mean number of Ramsay scoring not equal sented with dark gray bars, and conventional ACV group (AF⫺) is
to 2 or 3 per day. represented with light gray bars. The left side of the chart shows the
total number of all volume alarms generated during ACV (i.e., low or
high tidal volume (VT) and low or high minute ventilation alarms) and
the total “partially delivered VT” alarm (this alarm is generated when
the pressure limit is reached, leading to a stopping of inspiration),
and the right side shows the total number of high and low airway
pressure alarms. ACV without AF generated many more pressure
alarms than did ACV with AutoFlow. Mean ⫾ SEM.

Table 4. Univariate Analysis of Factors Influencing

Alarm Rate
Low alarm High alarm
ratea (n ⴝ 21) ratea (n ⴝ 21) P
Age (years) 66 关26–90兴 66 关25–88兴 0.89
Gender (M/F) 15/6 13/8 0.11
COPD (%) 2 (10) 5 (24) 0.41
SAPS II (points) 59 关23–101兴 54 关29–105兴 0.27
Tracheal aspirate 1.5 关0.7–2.6兴 1.9 关1.0–2.8兴 0.27
volume
Figure 3. Hourly alarm rates. Rectangular boxes represent the (points)
median hourly alarm rate during assist-controlled ventilation (ACV) Nebulizations 0 关0–36兴 0 关0–22兴 0.25
and diabolo boxes represent the median hourly alarm rate during the (total number)
other mechanical ventilation modalities, mainly during pressure Indexed 0.5 关0.2–1.1兴 0.3 关0–0.8兴 0.07
support (non-ACV). Upper and lower edges represent the 75th and midazolam
25th percentiles. Dark gray shapes represent rates observed for (mg/d/kg)b
patients in the AutoFlow group (AF⫹), and light gray shapes repre- Indexed fentanyl 30 关7–93兴 21 关0–53兴 0.02
sent rates observed in the conventional ACV group (AF⫺). The (␮g/d/kg)c
dashed line represents the overall median hourly alarm rate, which Sedation 4 关2–10兴 4 关1–12兴 0.72
was almost 1 alarm every 10 minutes. The lowest alarm rate was duration
observed during ACV using AutoFlow, and the highest alarm rate was (days)
observed during ACV without AutoFlow. ACV duration 85 关15–284兴 54 关11–316兴 0.16
(hours)
AF is based on an attractive principle: to guarantee a set Silence knob 0.26 关0.1–1.1兴 0.80 关0.26–2兴 0.001
Vt and minute ventilation while maintaining the advan- activation (per
tages of pressure-controlled ventilation.1,2 Despite this hour)
Study group 17/4 4/17 0.0002
potential advantage, clinical evaluations have not been (AF⫹/ AF⫺)
performed. Even clinical efficiency in comparison with
Values are expressed as median 关min–max兴 or number.
conventional ACV has not been formally demonstrated.
COPD ⫽ chronic obstructive pulmonary disease; SAPS II ⫽ Simplified Acute
This is the first report of around-the-clock observation Physiology Score II (34); ACV ⫽ assist-controlled ventilation; AF⫹ ⫽ assist-
and long-term bedside evaluation of this ventilation controlled ventilation with AutoFlow activation; AF⫺ ⫽ assist-controlled
ventilation without AutoFlow activation.
modality in the context of standard care. We found no a
The patients were divided into 2 groups according to whether their ventilator
differences with or without AF for gas exchange (P/F alarm rate was lower or higher than was the median alarm rate of the overall
ratio or Paco2) during ACV. No other studies are avail- population (6.37 alarms/hr).
b
able for comparison. Previous studies have only reported Indexed midazolam, total dose of midazolam indexed to the patient’s body
weight and the duration of drug infusion.
the physiological advantages associated with pressure- c
Indexed fentanyl, total dose of fentanyl indexed to the patient’s body weight
regulated ACV in comparison with volume-controlled and the duration of drug infusion.

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 919

Evaluation of AutoFlow Controlled Ventilation
ventilation, such as lower inspiratory pressures3,4,15 and
lower Paco2.3,15
One concern regarding AF is that the level of support
(i.e., the level of pressure delivered) could theoretically
decrease as patient demand increases. Indeed, the work of
breathing increases when pressure support decreases.16
This has been recently confirmed in a lung simulator.17
However, the absence of altered gas exchange in the AF⫹
group and the trend towards a shorter duration of ventila-
tion do not support this hypothesis, in vivo. In fact, for a set
Vt, pressure-controlled ventilation reduces the work of
breathing in comparison with volume-controlled ventila-
tion.18 In the present study, 2133 hours of AF ACV were
recorded, and no patient experienced any signs of respira-
tory distress. However, our study, with a sample size of 42
patients, is not powered to definitively assess these clinical
aspects. Indeed, it was powered to compare the ventilation-
related alarm rate with and without AF during ACV.
An original tool was used to record all ventilator-
generated alarms. A higher alarm rate was observed than
was previously reported by Chambrin et al. (0.6 alarm/hr
in14). Gabor et al. found a higher rate of sound increase
(37 ⫾ 20 to 72 ⫾ 13 times per hour of sleep),13 but they did
not identify the source of each sound. The high alarm rate
observed in the present study is certainly due to the
method used, but could also be due to the target sedation
level (Ramsay score of 2 to 3). A patient with less sedation
may fight the ventilator. However, this low target is widely
recommended and validated.19 –24 The lowest alarm rate
was observed during ACV with AF, and the highest rate
was observed during conventional ACV. It is not surprising
that a pressure-controlled mode, such as AF, is associated
with a reduction in pressure alarms. This is in accordance
with studies demonstrating a reduction of peak inspiratory
pressure during pressure-controlled ventilation.3,4,15 Mul-
tivariate analysis found only 2 factors associated with a
lower alarm rate: midazolam doses and AF activation.
It may be beneficial to reduce the number of alarms in an
ICU, because alarms are partly responsible for the high
noise level in an ICU.5–7 The first consequence of alarm
noise could be sleep disruption.25,26 In addition, an excess
of false positive alarms may decrease alarm efficiency: Only
26.8% of ventilator-generated alarms lead to an action.14
Only one half of critical alarms are correctly identified by
ICU staff.8 Reducing the total number of alarms should
reduce noise in the ICU and improve alarm efficiency.9 We
also showed that AF activation was associated with a
reduction in a surrogate marker of care interruption, acti-
vation of the silence knob, which could help reduce cross-
infections, because the ICU environment can be a reservoir
for pathogens.27,28 Further studies will be needed to con-
firm the alarm reduction observed during AF ACV.
The major limitations of our study are the absence of
fixed alarm limits and the unblinded design. The attending
physician was allowed to modify alarm limits as usual,
because no clear recommendations have been published.29
Physicians are unlikely to have set high alarm limits in
view of the very high alarm rate observed in this study.
Furthermore, no difference was observed for changes of
alarm limits, apart from the upper-pressure alarm limit set
above 50 cmH2O (more frequent in the AF⫺ group), which
920 www.anesthesia-analgesia.org
should have reduced the alarms rate in that group. The trial
could not be blinded because the ventilator screen displays
ventilator settings. However, the end points (gas exchange,
alarms, and outcome) were assessed objectively. Lastly, no
modification of clinical outcome such as decreased dura-
tion of mechanical ventilation or mortality was observed. It
is not surprising because no new mechanical ventilation
mode has improved clinical outcome.10 In particular, a
large-scale study comparing pressure-controlled ventila-
tion with volume-controlled ventilation did not find any
direct benefit.30 In the present study, sedation was con-
trolled by a nurse-driven protocol, but it can be assumed
that if physicians had prescribed sedation, they would have
increased sedation during conventional ACV to decrease
the alarm rate, which could have accentuated the trends
observed in this study.
CONCLUSIONS
ACV with AF appears to be safe in terms of gas exchange
and clinical outcome in this first long-term around-the-
clock clinical evaluation. AF was associated with a marked
decrease in ventilator-generated alarms. The beneficial ef-
fect of such a reduction of alarm rate on the comfort of
patients and ICU staff (quality of sleep and stress) deserves
further evaluation.
REFERENCES
1. Branson RD, Davis K Jr. Dual control modes: combining
volume and pressure breaths. Respir Care Clin N Am
2001;7:397– 408
2. Campbell RS, Davis BR. Pressure-controlled versus volume-
controlled ventilation: does it matter? Respir Care 2002;
47:416 –24
3. Alvarez A, Subirana M, Benito S. Decelerating flow ventilation
effects in acute respiratory failure. J Crit Care 1998;13:21–5
4. Guldager H, Nielsen SL, Carl P, Soerensen MB. A comparison
of volume control and pressure-regulated volume control
ventilation in acute respiratory failure. Crit Care (London)
1997;1:75–7
5. Freedman NS, Gazendam J, Levan L, Pack AI, Schwab RJ.
Abnormal sleep/wake cycles and the effect of environmental
noise on sleep disruption in the intensive care unit. Am J
Respir Crit Care Med 2001;163:451–7
6. Meyer TJ, Eveloff SE, Bauer MS, Schwartz WA, Hill NS,
Millman RP. Adverse environmental conditions in the respira-
tory and medical ICU settings. Chest 1994;105:1211– 6
7. Walder B, Francioli D, Meyer JJ, Lancon M, Romand JA. Effects
of guidelines implementation in a surgical intensive care unit
to control nighttime light and noise levels. Crit Care Med
2000;28:2242–7
8. Cropp AJ, Woods LA, Raney D, Bredle DL. Name that tone.
The proliferation of alarms in the intensive care unit. Chest
1994;105:1217–20
9. Siebig S, Sieben W, Kollmann F, Imhoff M, Bruennler T,
Rockmann F, Gather U, Wrede CE. Users’ opinions on inten-
sive care unit alarms—a survey of German intensive care units.
Anaesth Intens Care 2009;37:112– 6
10. Branson RD, Johannigman JA. What is the evidence base for
the newer ventilation modes? Respir Care 2004;49:742– 60
11. Ramsay MA, Savege TM, Simpson BR, Goodwin R. Controlled
sedation with alphaxalone-alphadolone. Br MedJ 1974;2:656 –9
12. Vincent JL, de Mendonca A, Cantraine F, Moreno R, Takala J,
Suter PM, Sprung CL, Colardyn F, Blecher S. Use of the SOFA
score to assess the incidence of organ dysfunction/failure in
intensive care units: results of a multicenter, prospective study.
Working group on “sepsis-related problems” of the European
Society of Intensive Care Medicine. Crit Care Med 1998;26:
1793– 800
ANESTHESIA & ANALGESIA
13. Gabor JY, Cooper AB, Crombach SA, Lee B, Kadikar N,
Bettger HE, Hanly PJ. Contribution of the intensive care unit
environment to sleep disruption in mechanically ventilated
patients and healthy subjects. Am J Respir Crit Care Med
2003;167:708 –15
14. Chambrin MC, Ravaux P, Calvelo-Aros D, Jaborska A, Chopin
C, Boniface B. Multicentric study of monitoring alarms in the
adult intensive care unit (ICU): a descriptive analysis. Intens
Care Med 1999;25:1360 – 6
15. Edibam C, Rutten AJ, Collins DV, Bersten AD. Effect of
inspiratory flow pattern and inspiratory to expiratory ratio on
nonlinear elastic behavior in patients with acute lung injury.
Am J Respir Crit Care Med 2003;167:702–7
16. Kreit JW, Capper MW, Eschenbacher WL. Patient work of
breathing during pressure support and volume-cycled me-
chanical ventilation. Am J Respir Crit Care Med 1994;149:
1085–91
17. Mireles-Cabodevila E, Chatburn RL. Work of breathing in
adaptive pressure control continuous mandatory ventilation.
Respir Care 2009;54:1467–72
18. Cinnella G, Conti G, Lofaso F, Lorino H, Harf A, Lemaire F,
Brochard L. Effects of assisted ventilation on the work of
breathing: volume-controlled versus pressure-controlled ven-
tilation. Am J Respir Crit Care Med 1996;153:1025–33
19. Brook AD, Ahrens TS, Schaiff R, Prentice D, Sherman G,
Shannon W, Kollef MH. Effect of a nursing-implemented
sedation protocol on the duration of mechanical ventilation.
Crit Care Med 1999;27:2609 –15
20. Heffner JE. A wake-up call in the intensive care unit. New Engl
J Med 2000;342:1520 –2
21. Jacobi J, Fraser GL, Coursin DB, Riker RR, Fontaine D, Witt-
brodt ET, Chalfin DB, Masica MF, Bjerke HS, Coplin WM,
Crippen DW, Fuchs BD, Kelleher RM, Marik PE, Nasraway SA
Jr., Murray MJ, Peruzzi WT, Lumb PD. Clinical practice
guidelines for the sustained use of sedatives and analgesics in
the critically ill adult. Crit Care Med 2002;30:119 – 41
October 2010 • Volume 111 • Number 4
22. Kollef MH, Levy NT, Ahrens TS, Schaiff R, Prentice D,
Sherman G. The use of continuous i.v. sedation is associated
with prolongation of mechanical ventilation. Chest 1998;114:
541– 8
23. Kress JP, Pohlman AS, Hall JB. Sedation and analgesia in the
intensive care unit. Am J Respir Crit Care Med 2002;166:
1024 – 8
24. Kress JP, Pohlman AS, O’Connor MF, Hall JB. Daily interrup-
tion of sedative infusions in critically ill patients undergoing
mechanical ventilation. New Engl J Med 2000;342:1471–7
25. Freedman NS, Kotzer N, Schwab RJ. Patient perception of
sleep quality and etiology of sleep disruption in the intensive
care unit. Am J Respir Crit Care Med 1999;159:1155– 62
26. Parthasarathy S, Tobin MJ. Sleep in the intensive care unit.
Intens Care Med 2004;30:197–206
27. Bures S, Fishbain JT, Uyehara CF, Parker JM, Berg BW.
Computer keyboards and faucet handles as reservoirs of
nosocomial pathogens in the intensive care unit. Am J Infect
Control 2000;28:465–71
28. Eggimann P, Pittet D. Infection control in the ICU. Chest
2001;120:2059 –93
29. Les recommandations des experts de la Société de Réanimation
de Langue Française: monitorage de la ventilation mécanique.
Réanim Urgences 2000;9:407–12
30. Esteban A, Alia I, Gordo F, de Pablo R, Suarez J, Gonzalez G,
Blanco J. Prospective randomized trial comparing pressure-
controlled ventilation and volume-controlled ventilation in
ARDS. For the Spanish Lung Failure Collaborative Group.
Chest 2000;117:1690 – 6
www.anesthesia-analgesia.org 921
Oxygen Delivery During Transtracheal Oxygenation:
A Comparison of Two Manual Devices
François Lenfant, MD, PhD,* Didier Péan, MD,† Laurent Brisard, MD,† Marc Freysz, MD, PhD,*
and Corinne Lejus, MD, PhD†

BACKGROUND: The Manujet™ and the ENK Oxygen Flow Modulator™ (ENK) deliver oxygen
during transtracheal oxygenation. We sought to describe the ventilation characteristics of these
2 devices.
METHODS: The study was conducted in an artificial lung model consisting of a 15-cm ringed
tube, simulating the trachea, connected via a flow analyzer and an artificial lung. A 15-gauge
transtracheal wire reinforced catheter was used for transtracheal oxygenation. The ENK and
Manujet were studied for 3 minutes at respiratory rates of 0, 4, and 12 breaths/min, with and
without the artificial lung, in a totally and a partially occluded airway. Statistical analysis was
performed using analysis of variance followed by a Fisher exact test; P ⬍ 0.05 was considered
significant.
RESULTS: Gas flow and tidal volume were 3 times greater with the Manujet than the ENK
(approximately 37 vs 14 L 䡠 min⫺1 and 700 vs 250 mL, respectively) and were not dependent on
the respiratory rate. In the absence of ventilation, the ENK delivered a 0.6 ⫾ 0.1 L 䡠 min⫺1
constant gas flow. In the totally occluded airway, lung pressures increased to 136 cm H2O after
3 insufflations with the Manujet, whereas the ENK, which has a pressure release vent, generated
acceptable pressures at a low respiratory rate (4 breaths/min) (peak pressure at 27.7 ⫾ 0.7 and
end-expiratory pressure at 18.8 ⫾ 3.8 cm H2O). When used at a respiratory rate of 12
breaths/min, the ENK generated higher pressures (peak pressure at 95.9 ⫾ 21.2 and
end-expiratory pressure at 51.4 ⫾ 21.4 cm H2O). In the partially occluded airway, lung pressures
were significantly greater with the Manujet compared with the ENK, and pressures increased with
the respiratory rate with both devices. Finally, the gas flow and tidal volume generated by the
Manujet varied proportionally with the driving pressure.
DISCUSSION: This study confirms the absolute necessity of allowing gas exhalation between 2
insufflations and maintaining low respiratory rates during transtracheal oxygenation. In the case
of total airway obstruction, the ENK may be less deleterious because it has a pressure release
vent. Using a Manujet at lower driving pressures may decrease the risk of barotrauma and allow
the safe use of higher respiratory rates. (Anesth Analg 2010;111:922–4)

T ranstracheal oxygenation may be lifesaving in “can-

not intubate/cannot ventilate” patients. The French
Society of Anesthesiologists has recently published
updated guidelines for difficult airway management and
characterize these 2 devices in terms of oxygen flow, tidal
volumes, and airway pressures.

recommends the use of dedicated devices for transtracheal METHODS

oxygen delivery.1 Two devices are available, the ENK
Oxygen Flow Modulator™ (ENK) and the Manujet™. Both
have been studied in a pig model.2 Both maintain oxygen-
ation, but the ENK seemed to achieve better ventilation
because of a continuous flow that provides CO2 washout
between insufflations. Very little is known concerning the
lung pressures generated with these 2 devices.3 Because it
is technically impossible to assess these variables in clinical
situations, we conducted this study in an artificial lung to
From the *Department of Anesthesiology and Intensive Care, CHU de Dijon,
General Hospital, Dijon; and †Department of Anesthesiology, CHU de
Nantes, Hotel Dieu, Nantes, France.
Accepted for publication June 8, 2010.
Supported by departmental sources except for the PF-300™ flow analyzer,
which was provided free of charge by SEBAC (Gennevilliers, France).
Disclosure: The authors report no conflicts of interest.
Address correspondence and reprint requests to Dr. François Lenfant,
Department of Anesthesiology, CHU de Dijon, General Hospital, 3 rue du
Faubourg Raines, 21033 Dijon Cedex, France. Address e-mail to
francois.lenfant@chu-dijon.fr.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181ee81b0
The Manujet III™ jet ventilator (VBM威; Vitrolles, France)
was connected to a 4-bar oxygen source and the driving
pressure was set at 3 bar. Manual activation of the trigger
delivered oxygen to a transtracheal catheter. The ENK
Oxygen Flow Modulator™ (COOK威, Charenton, France)
was connected to an oxygen wall flow regulator set at 15
L/min. Oxygen was delivered when the 5 holes in the
ENK’s perforated tube were manually occluded.
The 2 devices were evaluated by connecting them to a
15-gauge (2-mm internal diameter, 75-mm length) wire
reinforced transtracheal catheter (COOK). The catheter was
inserted into the simulated trachea, a 15-cm ringed tube
(Intersurgical, Fontenay sous Bois, France) occluded at one
end by a plastic cork. The other end of the tube was
connected to an IMT Medical Flow Analyzer PF-300™ (IMT
Medical威, Buchs, Switzerland) to measure flow rate, tidal
volume, peak pressure, mean pressure, and positive end-
expiratory pressure. An adult SmartLung™ (IMT Medical)
was connected to the analyzer output. The SmartLung
settings were as follows: total lung volume ⫽ 1000 mL;
airway resistance ⫽ 5 mbar/L/s; and lung compliance ⫽ 30
mbar/mL (Fig. 1).

922 www.anesthesia-analgesia.org October 2010 • Volume 111 • Number 4


Figure 1. Photograph showing the ENK Oxygen Flow Modulator™
device connected to the simulated trachea, the PF-300™ flow
analyzer, and the artificial lung. The trachea is occluded at one end
by a plastic cork.
Two investigators manually delivered 1-second insuffla-
tions for 3 minutes at the rate of 4 breaths/min and 12
breaths/min with the plastic cork in place, simulating total
upper airway obstruction, and after making a 2-mm hole in
the plastic cork, simulating partial upper airway obstruc-
tion. A metronome was used to keep the rate and the
duration of the insufflations constant. Data were recorded
for 3 minutes with an insufflation rate of 4 breaths/min
with the Manujet driving pressure set at 0.5, 1.0, 1.5, 2.0, 2.5,
and 3 bar. The SmartLung was removed and the gas flow
delivery rates were recorded without insufflations (0
breaths/min).
All of the variables were sampled and recorded every 10
milliseconds using Flowlab™ software and transferred to a
personal computer as Excel™ files. Statistical analysis was
performed using StatPlus威 software (AnalystSoft, StatPlus
Mac, Version 2008; http://www.analystsoft.com/fr). Con-
tinuous data were expressed as the mean ⫾ SD. Data were
analyzed using analysis of variance followed by a Fisher
exact test. P ⬍ 0.05 was considered significant.
RESULTS
The mean duration of the manually delivered insufflations
was 1.01 ⫾ 0.06 second at 4 breaths/min and 0.95 ⫾ 0.05
second at 12 breaths/min for the ENK, and 0.97 ⫾ 0.08
second at 4 breaths/min and 0.95 ⫾ 0.08 second at 12
breaths/min for the Manujet. During insufflations, peak
flow was 36.0 ⫾ 6.3 L 䡠 min⫺1 with the Manujet and 13.7 ⫾
2.42 L 䡠 min⫺1 with the ENK. When holes were not occluded,
the ENK delivered oxygen at the rate of 0.7 ⫾ 0.1 L 䡠 min⫺1.
Table 1 shows the airway pressure delivered by the ENK
during total airway occlusion. With the Manujet, airway
pressure was 34 cm H2O after the first insufflation and 136
cm H2O after the second, therefore the experiment was
stopped. During total airway occlusion, airway pressures
were significantly lower with the ENK.
October 2010 • Volume 111 • Number 4
Table 1. Airway Pressure Delivered by the ENK
Oxygen Flow Modulator™ During Total
Airway Obstruction
Mean pressure Mean pressure
Respiratory during during
rate (breaths/ inspiration Peak pressure expiration
min) (cm H2O) (cm H2O) (cm H2O)
0 13.2 ⫾ 0.0
4 21.9 ⫾ 4.4 27.7 ⫾ 0.7 18.8 ⫾ 3.8
12 62.7 ⫾ 24.3* 95.9 ⫾ 21.2* 51.4 ⫾ 21.4*
* P ⬍ 0.05, 4 vs 12 breaths/min.
Table 2 shows tidal volumes, minute volumes, and
mean and peak airway pressures during partial upper
airway obstruction. Values were significantly higher
with the Manujet. Table 3 shows how the Manujet gas
flows and tidal volumes increased with an increase in
driving pressure.
DISCUSSION
The Manujet delivers oxygen at a higher peak flow rate
than the ENK. As a result, 1-second insufflations with the
Manujet result in larger tidal and minute volumes and
higher airway pressures than with the ENK. Our results
confirm the results of Flint et al.3 The tidal volumes we
measured for the Manujet (650 mL) are close to those
calculated from the minute volumes reported by Flint et al.3
(627 mL). For the ENK, we measured a 240-mL tidal
volume, whereas Flint et al. measured a 156-mL tidal
volume. This discrepancy may be explained by the differ-
ences in both the model and the methodologies used in the
2 studies.
Yildiz et al.4 found that the oxygen that flows from the
ENK between manual insufflations enhances oxygenation,
especially at low respiratory rates. Preussler et al.2 found
that the constant flow improves CO2 elimination. Because
the Manujet generates higher airway pressure, it may result
in better alveolar recruitment and better lung ventilation.
Both investigators found that decreasing the Manujet driv-
ing pressure allows it to deliver higher respiratory rates
without excessive airway pressure.2,4
Because the main risk of transtracheal oxygenation is
barotrauma, the Manujet should not be used in case of
total upper airway obstruction. The ENK seems to be less
problematic because it delivers gas at a lower pressure
and has a pressure release vent that allows gas to escape
between insufflations.5 However, even with the ENK, the
peak and mean airway pressures may reach dangerously
high levels at high respiratory rates (12 breaths/min)
during total airway obstruction. Increasing the respira-
tory rate results in short expiratory times and, as a
consequence, an increase in the volume of the gas
trapped in the lung. Our results confirm the absolute
necessity of ensuring that the insufflated gas is exhaled
during the expiratory period.6 Sufficient exhalation can
easily be checked by watching the chest fall before
delivering a second tidal volume.
When using the Manujet during partial upper airway
obstruction, great attention should be given to delivering
insufflations that last ⬍1 second and using low respiratory
www.anesthesia-analgesia.org 923
Comparison of Two Transtracheal Oxygenation Manual Devices

Table 2. Gas Flow Rate During Inspiration, Tidal Volume, Minute Volume, and Mean and Peak Airway
Pressures Measured with the Manujet and the ENK Oxygen Flow Modulator™ (ENK) During Partial
Airway Obstruction
Respiratory rate Flow Tidal volume Minute volume Mean pressure
(breaths/min) (L 䡠 minⴚ1) (mL) (L 䡠 minⴚ1) (cm H2O) Peak pressure (cm H2O)
ENK 4 12.4 ⫾ 2.0 263.0 ⫾ 26.6 0.91 ⫾ 0.01 11.3 ⫾ 1.8 14.1 ⫾ 0.3
Manujet 4 39.9 ⫾ 4.9 676.6 ⫾ 62.3 2.67 ⫾ 2.04 16.7 ⫾ 6.8 27.7 ⫾ 5.0
ENK 12 12.0 ⫾ 1.6 223.9 ⫾ 27.3 2.69 ⫾ 0.07 12.1 ⫾ 2.0 18.2 ⫾ 1.2
Manujet 12 38.8 ⫾ 5.8 753.4 ⫾ 88.9 9.06 ⫾ 0.02 16.7 ⫾ 6.8 58.3 ⫾ 25.0
All differences between Manujet and ENK were significant at P ⬍ 0.05.

2. Preussler NP, Schreiber T, Huter L, Gottschall R, Schubert H,

Table 3. Gas Flow Rate During Inspiration and Rek H, Karzai W, Schwarzkopf K. Percutaneous transtracheal
Tidal Volume Versus Driving Pressure for the ventilation: effects of a new oxygen flow modulator on oxygen-
Manujet at a Respiratory Rate of 4 breaths/min ation and ventilation in pigs compared with a hand triggered
Driving pressure Gas flow Tidal volume emergency jet injector. Resuscitation 2003;56:329 –33
(bar) (L 䡠 minⴚ1) (mL) 3. Flint NJ, Russell WC, Thompson JP. Comparison of different
0.5 12.6 ⫾ 1.6 246.0 ⫾ 15.9 methods of ventilation via cannula cricothyroidotomy in a
1.0 17.4 ⫾ 2.5 304.0 ⫾ 34.9 trachea-lung model. Br J Anaesth 2009;103:891–5
1.5 23.7 ⫾ 2.9 356.0 ⫾ 25.2 4. Yildiz Y, Preussler NP, Schreiber T, Hueter L, Gaser E, Schubert
2.0 28.2 ⫾ 4.2 485.3 ⫾ 18.0 H, Gottschalll R, Schwarzkopf K. Percutaneous transtracheal
2.5 33.9 ⫾ 3.5 626.3 ⫾ 5.5 emergency ventilation during respiratory arrest: comparison of
3.0 37.1 ⫾ 5.8 653.3 ⫾ 50.3 the oxygen flow modulator with a hand-triggered emergency jet
injector in an animal model. Am J Emerg Med 2006;24:455–9
5. Hamaekers A, Borg P, Enk D. The importance of flow and
pressure release in emergency jet ventilation devices. Paediatr
rates, to avoid excessive airway pressure. The most impor-
Anaesth 2009;19:452–7
tant goal is to deliver oxygen to the lung, as recommended 6. Bourgain JL, Desruennes E, Fischler M, Ravussin P. Transtra-
by Cook and Alexander.7 cheal high frequency jet ventilation for endoscopic airway
surgery: a multicentre study. Br J Anaesth 2001;87:870 –7
REFERENCES 7. Cook TM, Alexander R. Major complications during anaesthesia
1. Combes X, Pean D, Lenfant F, Francon D, Marciniak B, Legras for elective laryngeal surgery in the UK: a national survey of the
A. Difficult airway management devices: establishment and use of high-pressure source ventilation. Br J Anaesth
maintenance— question 4. Societe Francaise d’Anesthesie et de 2008;101:266 –72
Reanimation. Ann Fr Anesth Reanim 2008;27:33– 40

924 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

Two Serial Check Valves Can Prevent
Cross-Contamination Through Intravenous Tubing
During Total Intravenous Anesthesia
Oliver C. Radke, MD, PhD,*† Katrin Werth, MD,‡ Margarete Borg-von-Zepelin, MD, PhD,§
Petra Saur, MD, PhD,储 and Christian C. Apfel, MD, PhD†

BACKGROUND: Nonsterile handling of propofol for anesthesia has been linked with severe
sepsis and death. Placing a single check valve in the IV tubing does not prevent retrograde
ascension of pathogens into propofol-filled syringes, so we designed an IV tubing set with
multiple check valves. To estimate the efficacy of this design, we measured the concentration of
pathogens detected upstream in the IV tubing in relation to the pathogen concentration in a
model of a contaminated patient.
METHODS: A glass container with a rubber sealed port was filled with a suspension of either
bacteria or phagocytes and kept at 37°C (“contaminated patient” model). A bag of normal saline was
connected to an IV cannula, punctured through the rubber sealed port of the patient model. Two
additional sidestream infusion lines were connected to syringes in 2 standard infusion pumps. One
of the syringes contained propofol and the other contained normal saline as a substitute for an opioid
preparation. After 5 hours of infusion, we obtained samples from different parts of the infusion lines
and syringes. The samples were streaked out on blood agar plates and incubated at 37°C for 24
hours. We repeated this experiment with 6 different pathogens.
RESULTS: We incubated 825 agar plates. Whereas the concentration of bacteria and phago-
cytes in the “patient” had significantly increased during the 5-hour experiments (positive control),
no bacterial growth could be detected in any of the incubated plates.
CONCLUSION: The data from this experimental setting suggest that the design with multiple
check valves in paired configuration prevents retrograde contamination. Of note, this does
not permit the reuse of propofol syringes because reusing is against the manufacturer’s
recommendations. (Anesth Analg 2010;111:925–8)

P ropofol infusions have been identified as a medium

supporting the growth of microorganisms.1 Contami-
nation can occur during the preparation of the drug2
and during its administration.3–9 Clinicians often put a single
1-way valve into the IV line to prevent backflow but, unfor-
tunately, these valves may not prevent ascension of bacteria
from the patient into the tubing and the syringes.10
To overcome the shortcomings of single check valves,
we propose placing 2 check valves in series.11 We hypoth-
esized that the serial placement of check valves working in
pairs would improve their function as a barrier to blood-
borne pathogens. Because fluids are not compressible, the 2
From the *Department of Anesthesiology and Critical Care Medicine,
University Hospital Carl Gustav Carus Dresden at the TU Dresden,
Dresden, Germany; †Perioperative Clinical Research Core, Department of
Anesthesia and Perioperative Care, University of California at San Francisco,
UCSF Medical Center at Mount Zion, San Francisco, California; ‡Depart-
ment of Pediatrics, Klinikum Aschaffenburg, Aschaffenburg, Germany;
§Institute of Infectiology & Pathobiology, Hufeland Klinikum GmbH, Stan-
dort Mühlhausen, Mühlhausen, Germany; and 储Department of Anesthesia,
Intensive Care Medicine and Pain Medicine, Sana Kliniken Lübeck GmbH,
Lübeck, Germany.
Accepted for publication May 28, 2010.
Funded by departmental resources. Medex Medical, now Smith Medical
International, Watford, UK, donated the IV tubings. B.Braun, Melsungen,
Germany, donated the infusion pumps for the experiments. AstraZeneca
GmbH, Wedel, Germany, donated the propofol. None of these companies
provided financial support.
Address correspondence and reprint requests to Oliver C. Radke, MD,
PhD, DEAA, Klinik und Poliklinik für Anästhesiologie und Intensiv-
therapie, Fetscherstr. 74, 01307 Dresden, Germany. Address e-mail to
oliver.radke@uniklinikum-dresden.de.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181eb7194
valves in each pathway would work in conjunction, ensur-
ing that the column of fluid between them did not move
backward even if one of the valves failed.
To estimate the efficacy of this design, we conducted a
series of experiments with a model of a contaminated
patient. Our main objective was to measure the concentra-
tion of pathogens detected upstream in the IV tubing in
relation to the pathogen concentration in the patient model.
METHODS
Tests for Bacterial Contamination
As a model of a contaminated patient, a 3000-mL glass
container with a rubber-sealed port was filled with 2500 mL
of a liquid medium (Fig. 1). The medium consisted of 25 g
peptone, 5 g NaCl, 10 g dry meat extract, and 10 g glucose
in distilled water. For each experimental run, we infected
the medium with a bacteria suspension to achieve a con-
centration of at least 1 䡠 106 mL⫺1 colony-forming units
(CFUs). The bacteria were strains isolated from patients of
our university hospital (Staphylococcus aureus, Staphylococ-
cus epidermidis, Escherichia coli, Proteus mirabilis, and Pseudo-
monas aeruginosa).
We assembled the IV tubing set (“TIVA-Set”; Smith
Medical Deutschland GmbH, Grasbrunn, Germany) shown
in Figure 2 with 4 strategically placed check valves.11 A
500-mL bag of 0.9% NaCl was connected via the TIVA-Set
to an 18-gauge IV cannula. This cannula was punctured
through the rubber-sealed port of the patient model (Fig. 2).
The 2 infusion lines of the TIVA-Set were connected to
50-mL syringes (B.Braun, Melsungen, Germany) in Perfu-
sor威 fm and perfusor compact infusion pumps (B.Braun).

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 925

Can Serial Check Valves Prevent Cross-Contamination?
Figure 1. Photograph of the “contaminated patient” model. The
rubber disc that seals the connector at the bottom has been
punctured with a standard 16-gauge IV cannula. The thermometer at
the top is connected to the heated stirrer to ensure a constant
temperature of 37°C.
One of the syringes contained propofol 1% (Disoprivan威;
AstraZeneca, Wedel, Germany) and the other contained
normal saline as a substitute for an opioid preparation. The
infusion pumps were approximately 30 cm above the level
of the port in the patient model.
The flow of the normal saline (Table 1) was regulated
by a drip pump (Infusomat fmS; B.Braun). All pumps
were operated by a microprocessor-controlled fluid man-
ager (fm controller; B.Braun) that was programmed to
run a 5-hour infusion program (Table 1). This program
was designed to simulate an actual anesthetic. For each
pathogen, the experimental run was repeated at least 8
times (Table 2).
926 www.anesthesia-analgesia.org
Table 1. Infusion Program
Carrier Propofol Saline
Duration drip rate infusion infusion
(min) (mL/h) rate (mL/h) rate (mL/h)
Before induction 30 100 0 0
Maintenance I 120 100 10 10
Fluids run dry 30 0 10 10
Maintenance II 120 100 10 10
Standardized infusion program, controlled by a microprocessor. The program
consisted of 4 parts mimicking a total IV anesthesia in clinical practice with
low drug infusion rates. Total duration of each experimental run was 5 hours.
Before and after the infusion program, we took a sample
of 100 ␮L from the bacteria suspension in the patient model
and prepared a dilution series to estimate the bacteria
concentrations. Additional undiluted samples of 500 ␮L
were obtained under sterile conditions from the following
locations (Fig. 2): IV tubing close to the cannula; both
microtubings leading from the drug pumps to the main IV
line; and leftover drugs in both 50-mL syringes.
Each sample was spread out on 3 blood agar plates (100 ␮L
per plate), and the plates were incubated at 37°C for 24 hours.
At the end of the incubation, we manually counted the
number of CFUs on the plates. Bacteria concentrations in the
patient model were compared using the Student t test for
repeated measurements (SPSS 16.0.2; SPSS Inc., Chicago, IL).
Test for Nonbacterial Contamination
Because nonbacterial pathogens such as viruses or prions are
much smaller than bacteria, we used a bacteriophage (Bacte-
riophage T3) as a sample of such a small infectious unit. The
Bacteriophage T3 infects E coli bacteria and can be detected by
their ability to lyse the bacteria. The experimental runs with
the phage were conducted almost identically to the runs with
bacteria, but because a phage cannot replicate on its own, we
added E coli B14 to the patient model. The blood agar plates
for detection of pathogens in the samples were also flooded
with E coli B14 to provide the hosts for the phages’ replication.
After incubation, the concentration of phages in a sample was
calculated from the number of lytic holes in the bacterial
spread on the blood agar plate.
Figure 2. Placement of check valves and
sample sites. Schematic of the IV tubing.
Four check valves are placed in paired
configurations to prevent retrograde flow.
The 5 sample sites are marked.
ANESTHESIA & ANALGESIA
 Table 2. Bacteria Concentrations in the Patient Model
Before
106 · mL⫺1 CFUs
Staphylococcus aureus 12.3 ⫾ 6.3
Staphylococcus epidermidis 9.4 ⫾ 1.2
Escherichia coli 50.7 ⫾ 38.8
Proteus mirabilis 1.0 ⫾ 0.0
Pseudomonas aeruginosa 20.7 ⫾ 4.0
106 · mL⫺1 PFUs
Bacteriophage T3 1.4 ⫾ 1.7
CI ⫽ confidence interval; CFU ⫽ colony-forming unit; PFU ⫽ plaque-forming unit.
Concentrations are given as mean ⫾ standard deviation in 106 CFUs or 106 PFUs per mL. Growth rate is the CFUafter divided by CFUbefore. Total number of plates: 825.
RESULTS
The test runs were repeated 8 to 12 times with each of the
5 bacterial isolates and the phage. We performed a total of
55 runs. For each bacterial strain, we found a statistically
significant increase in concentration in the patient model
(Table 2). Proteus mirabilis had the highest increase in
concentration (67-fold). S aureus had the second highest
increase (10-fold), whereas the others increased their start-
ing concentration 3- to 6-fold. The phages’ replication factor
was even higher, ranging from 10 to 333 depending on the
starting concentration.
In 55 experimental runs, we obtained 275 samples from
different parts of the IV tubing and the drug syringes and
incubated 825 plates. Even with the significant increase of
the bacteria and phage concentrations in the patient model,
no plates grew colonies or had lytic holes.
DISCUSSION
Soon after propofol was introduced into clinical practice,
there were several cases of severe sepsis resulting from the
use of propofol that was contaminated with bacteria (e.g.,
see Veber et al.12). Contamination by the anesthesia pro-
vider can be found in 8% to 11% of syringes whenever
propofol is drawn up for clinical use,13 even when the
manufacturer’s recommendations are closely adhered to.3
Because bacterial growth in propofol is slow at room
temperature and begins after a latency period of 5 to 6
hours,14,15 the manufacturer recommends discarding any
propofol within 6 hours of opening the vial. Serious infec-
tious outbreaks of sepsis were always attributed to propo-
fol that had been opened several hours ahead of use
(usually the day before); therefore, bacterial contamination
from opening the vials could grow to achieve bacterial
counts that were high enough to cause severe sepsis.
The risk of bacterial contamination by the anesthesia
provider can be addressed by hand washing, aseptic tech-
niques, and using propofol within 6 hours after opening.
Transfer of pathogens from one patient to another by
means of propofol syringes has not yet been documented,
but in vitro studies have shown that retrograde contami-
nation occurs easily.16
With our experimental setup, we wanted to measure the
incidence and the magnitude, if any, of retrograde bacterial
contamination through a special IV tubing set for total IV
anesthesia (TIVA).
October 2010 • Volume 111 • Number 4
Growth rate
After P value (95% CI) Runs Plates
106 · mL⫺1 CFUs
110.6 ⫾ 4.0 ⬍0.001 10.4 (8.1–12.6) 12 180
34.0 ⫾ 25.2 0.028 3.6 (1.5–5.8) 8 120
249.4 ⫾ 135.4 ⬍0.001 5.9 (3.8–8.0) 9 135
67.3 ⫾ 14.2 ⬍0.001 67.3 (56.5–78.2) 8 120
66.6 ⫾ 12.0 ⬍0.001 3.4 (2.4–4.5) 9 135
106 · mL⫺1 PFUs
37.8 ⫾ 43.7 0.015 97.2 (21.4–173.0) 9 135
Validity of the Patient Model
The pathogens tested in this study are representative of
infections found in the patients of our hospital. S aureus was
used because of its ubiquitous presence, accounting for 12%
of all nosocomial infections.17,18 S aureus strains with resis-
tance to a wide spectrum of common antibiotics (not only
methicillin-resistant S aureus) are an increasing problem in
modern health care.19,20 S epidermidis is an important cause
of infection in patients with a compromised immune sys-
tem, or who have indwelling catheters. This bacterium has
a special ability to produce a matrix (“biofilm”) that allows
them to adhere to artificial surfaces, such as those of
medical prostheses.21,22 E coli is frequently found in noso-
comial infections, and strains that are highly resistant to
antibiotics are increasingly isolated.20 Proteus mirabilis has
an exceptionally high intrinsic motility and could poten-
tially swarm actively into IV lines. Pseudomonas aeruginosa
has a high intrinsic resistance to antibiotics and disinfec-
tants,20 and some strains can also generate a biofilm.23
The Bacteriophage T3 is a well-known phage without any
pathogenic potency for humans. It has approximately the
same size as a virus and therefore the same potential for
contamination. Its lytic effect on E coli bacteria makes it
easy to detect.
The bacterial load in our contaminated patient model was
at least 1 䡠 106 CFUs mL⫺1. Studies in patients with endocar-
ditis measured a maximum of 173 CFUs per mL of blood,24
and children with manifested infections (meningitis, peritoni-
tis) had bacterial loads of up to 104 CFUs per mL of blood.25
Even blood samples from infected indwelling catheters mea-
sured only up to 300 CFUs per mL.26 Therefore, our contami-
nated patient model was ⬎300 times more infectious than a
worst case patient in real-life conditions.
Validity of the IV Setup
With our experimental setup, we tried to model real-life
conditions during TIVA. We expected the risk for contami-
nation to increase with duration of infusion. Also, high
infusion rates would flush the IV tubing and make it less
likely for pathogens to ascend. To facilitate the migration of
bacteria from our contaminated patient model into the
tubing, we ran an infusion pattern with low infusion rates
over a period of 5 hours.
It is obvious that an IV line without any check valves
would not protect against contamination. In a survey of IV
tubings without check valves in the operating room, the
www.anesthesia-analgesia.org 927
Can Serial Check Valves Prevent Cross-Contamination?
authors found traces of blood from the patient in 3.3% of the
cases.4 In an experiment with a contaminated patient model
similar to our setup but where the IV tubing was filled with
liquid culture medium instead of saline and propofol, the
researchers found a bacterial contamination of the line close to
the patient in 87% of the cases,27 even though the line was
“protected” against backflow with a 1-way valve. The re-
searchers even increased the “venous” pressure of their model
to 150 mm Hg to simulate occlusion caused by noninvasive
arterial blood pressure measurements.16 This maneuver did
not cause increased ascension of the pathogens into the
infusion tubing farther away from the patient. In our model,
we did not increase the venous pressure to such high read-
ings. Because the authors of the aforementioned article did not
find an increase in contamination rates even with only a single
check valve, we do not expect to see an increase in contami-
nation rates with our paired valves configuration.
The lipid preparation containing propofol supports bac-
terial growth, but only after a latency period of 5 to 6
hours.14,15 Additionally, the propofol preparation we used
contained EDTA, an additive that suppresses bacterial
growth.28 The bacteriostatic effect of the EDTA could have
inhibited the growth of small amounts of bacteria that in
fact entered the IV tubing and thus prevented their detec-
tion. However, running a propofol/saline mixture in the IV
tubing is closer to real-life conditions than running liquid
medium. Any bacteria that could not grow on our plates
with optimal conditions for the growth of human patho-
gens probably would not have been capable of growing in
a human organism either.
CONCLUSION
It is highly unlikely that during 5 hours of propofol anesthesia
the syringe filled with propofol and EDTA will become
contaminated by pathogens from the patient’s blood if 2 check
valves are placed in series in the IV line. We did not find a
single trace of contamination in any of our 825 samples.
AUTHOR CONTRIBUTIONS
OCR: Inventor of the TIVA-Set, initiator of the study, and the
main author of the manuscript. KW: Performed all the micro-
biological testing and did additional literature research.
MB-v-Z: Designed and supervised the microbiological testing.
PS: Coauthor of the manuscript and the advisor of study
design. CCA: Senior author of the manuscript, statistical
analysis and interpretation of data. All authors contributed to
the drafting of the article and gave final approval of the version
to be published.
DISCLOSURE
The University of Göttingen, Germany, holds a patent on the
IV tubing set and is required by German law to share a fraction
of all royalties with the inventor, Dr. Oliver C. Radke. To avoid
a conflict of interest, Dr. Radke was not involved in the
microbiological experiments.
REFERENCES
1. Thomas DV. Propofol supports bacterial growth. Br J Anaesth
1991;66:274
2. Lorenz IH, Kolbitsch C, Lass-Florl C, Gritznig I, Vollert B, Lingnau
W, Moser PL, Benzer A. Routine handling of propofol prevents
contamination as effectively as does strict adherence to the manu-
facturer’s recommendations. Can J Anaesth 2002;49:347–52
928 www.anesthesia-analgesia.org
3. Lessard MR, Trepanier CA, Gourdeau M, Denault PH. A
microbiological study of the contamination of the syringes
used in anaesthesia practice. Can J Anaesth 1988;35:567–9
4. Trepanier CA, Lessard MR, Brochu JG, Denault PH. Risk of
cross-infection related to the multiple use of disposable sy-
ringes. Can J Anaesth 1990;37:156 –9
5. Zacher AN, Zornow MH, Evans G. Drug contamination from
opening glass ampules. Anesthesiology 1991;75:893–5
6. Bach A. Syringe— or lead change for TCI? [in German].
Anaesthesist 1998;47:434 – 6
7. el Mikatti N, Dillon P, Healy TE. Hygienic practices of consul-
tant anaesthetists: a survey in the north-west region of the UK.
Anaesthesia 1999;54:13– 8
8. Halkes MJ, Snow D. Re-use of equipment between patients
receiving total intravenous anaesthesia: a postal survey of
current practice. Anaesthesia 2003;58:582–7
9. Vonberg RP, Gastmeier P. Infection control measures in anaes-
thesia [in German]. Anasthesiol Intensivmed Notfallmed
Schmerzther 2005;40:453– 8
10. Gretzinger DT, Cafazzo JA, Ratner J, Conly JM, Easty AC.
Validating the integrity of one-way check valves for the
delivery of contrast solution to multiple patients. J Clin Engl
1996;21:375– 82
11. Radke O. Das TIVA-Set. Anasthesiol Intensivmed 2003;44:787– 8
12. Veber B, Gachot B, Bedos JP, Wolff M. Severe sepsis after
intravenous injection of contaminated propofol. Anesthesiol-
ogy 1994;80:712–3
13. Magee L, Godsiff L, Matthews I, Farrington M, Park GR.
Anaesthetic drugs and bacterial contamination. Eur J Anaes-
thesiol Suppl 1995;12:41–3
14. Langevin PB, Gravenstein N, Doyle TJ, Roberts SA, Skinner S,
Langevin SO, Gulig PA. Growth of Staphylococcus aureus in
Diprivan and intralipid: implications on the pathogenesis of
infections. Anesthesiology 1999;91:1394 – 400
15. Sosis MB, Braverman B. Growth of Staphylococcus aureus in
four intravenous anesthetics. Anesth Analg 1993;77:766 – 8
16. Sim J, Choi Y, Yoon M, Lee D, Leem J. The peripheral venous
pressure changes during non-invasive blood pressure mea-
surement. Can J Anaesth 1999;46:711–2
17. Bouza E. Intravascular catheter-related infections: a growing
problem, the search for better solutions. Clin Microbiol Infect
2002;8:255
18. Emori TG, Gaynes RP. An overview of nosocomial infections,
including the role of the microbiology laboratory. Clin Micro-
biol Rev 1993;6:428 – 42
19. Gastmeier P, Geffers C. Nosocomial infections in Germany:
what are the numbers, based on the estimates for 2006? [in
German] Dtsch Med Wochenschr 2008;133:1111–5
20. Toniolo A, Endimiani A, Luzzaro F. Microbiology of postop-
erative infections. Surg Infect (Larchmt) 2006;7:S13– 6
21. Ziebuhr W, Hennig S, Eckart M, Kranzler H, Batzilla C,
Kozitskaya S. Nosocomial infections by Staphylococcus epider-
midis: how a commensal bacterium turns into a pathogen. Int
J Antimicrob Agents 2006;28:S14 –20
22. Pascual A. Pathogenesis of catheter-related infections: lessons
for new designs. Clin Microbiol Infect 2002;8:256 – 64
23. O’Toole GA, Kolter R. Flagellar and twitching motility are
necessary for Pseudomonas aeruginosa biofilm development.
Mol Microbiol 1998;30:295–304
24. Werner AS, Cobbs CG, Kaye D, Hook EW. Studies on the
bacteremia of bacterial endocarditis. JAMA 1967;202:199 –203
25. Santosham M, Moxon ER. Detection and quantitation of bac-
teremia in childhood. J Pediatr 1977;91:719 –21
26. Flynn PM, Shenep JL, Barrett FF. Differential quantitation with
a commercial blood culture tube for diagnosis of catheter-
related infection. J Clin Microbiol 1988;26:1045– 6
27. Eichler W, Schumacher J, Ohgke H, Klotz KF. Reuse of a set for
total intravenous anaesthesia: safe against bacterial contami-
nation? Eur J Anaesthesiol 2004;21:501–3
28. Hart B. ‘Diprivan’: a change of formulation. Eur J Anaesthesiol
2000;17:71–3
ANESTHESIA & ANALGESIA
 TECHNICAL COMMUNICATION

Robot-Assisted Airway Support: A Simulated Case

Patrick J. Tighe, MD, S. J. Badiyan, MD, I. Luria, BS, MS, S. Lampotang, PhD, and S. Parekattil, MD

Recent advances in telemedicine and robotically assisted telesurgery may offer advanced
surgical care for the geographically remote patient. Similar advances in tele-anesthesia will be
necessary to optimize perioperative care for these patients. Although many preliminary
investigations into tele-anesthesia are underway, none involves remote performance of
anesthesia-related procedures. Here we describe simulated robotically assisted fiberoptic
intubations using an airway simulation mannequin. Both oral and nasal approaches to
fiberoptic intubation were successful, but presented unique opportunities and challenges
inherent to the robot’s design. Robotically assisted airway management is feasible using
multipurpose surgical robotic systems. (Anesth Analg 2010;111:929 –31)

T elemedicine has improved access to consultant-level

medical care by minimizing geographic obstacles.
Similar advances in telesurgery could offer expert
surgical care for the geographically remote patient. Such
surgical advances have been historically preceded by simi-
lar advances in anesthesia. True to this paradigm, early
investigations into tele-anesthesia are already underway.1
However, none of these efforts has tackled one of the
central tenets of anesthesiology: airway management.
In this report we present what we believe to be the first
description of a simulated robotically assisted fiberoptic
intubation. Instead of a procedure-specific device, we used
the multipurpose DaVinci Surgical System type S (DVS)
(Intuitive Surgical, Sunnyvale, California). This system
incorporates 4 separate robotic arms, 1 of which is mated to
a high-definition stereoscopic camera. The workstation
allows the person performing the procedure to view the
robot’s camera output, control the limbs, and receive simul-
taneous video input from third-party sources.2,3 The DVS is
already in widespread clinical use for a variety of urologic,
gynecologic, and cardiothoracic surgical procedures.4 In
this study involving an airway mannequin, we successfully
used the DVS for both oral and nasal fiberoptic intubation.
METHODS
An adult airway mannequin was placed at the head of a
standard operating room bed. A stereoscopic video camera
(DaVinci Surgical System, Intuitive Surgical) was mated to
the first robotic arm. This arm was situated above the
mannequin in a sagittal plane, angled to view the manne-
quin from a caudal to rostral fashion. The second and third
arms were equipped with large and small graspers (Figs. 1
and 2). The fourth arm was equipped with a standard
From the University of Florida College of Medicine, Gainesville, Florida.
Accepted for publication June 8, 2010.
Funding was provided by the University of Florida College of Medicine,
Department of Anesthesiology, as well as by National Institutes of Health
grant UL1 RR029890 Clinical and Translational Science Award, NIH
(NCRR).
Supplemental digital content is available for this article. Direct URL citations
appear in the printed text and are provided in the HTML and PDF versions
of this article on the journal’s Web site (www.anesthesia-analgesia.org).
Address correspondence to Patrick J. Tighe, MD, University of Florida
College of Medicine, Department of Anesthesiology, PO Box 100254, Gaines-
ville, FL 32610-0254. Address e-mail to ptighe@anest.ufl.edu.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181ef73ec
fiberoptic bronchoscope (Karl Storz Endoscopy, El Seg-
undo, California). The trocars used for mounting standard
robotic instruments were removed from this arm, allowing
the bronchoscope handle to slide into the mounting
bracket. The bronchoscope handle was oriented to keep the
tip actuator lever facing away from the arm, with the
suction port positioned to the side (Fig. 3). An external
brace was required to keep the bronchoscope firmly seated
within the arm during manipulation of the actuator.
We manually loaded the endotracheal tube onto the
bronchoscope. A camera was attached to the bronchoscope
and connected to the DVS Tilepro multivideo input system
(Intuitive Surgical, Sunnyvale, California), allowing simul-
taneous viewing of both the robot camera and broncho-
scope camera in a single three-dimensional view (Fig. 4).
Before attempting intubation, the urologist spent ap-
proximately 2 hours performing robotic dexterity exercises
with the robotic surgical instruments and operator console.
After training was complete, an anesthesiologist manually
placed the bronchoscope tip within the oropharnyx, and a
urologist (S. Parekattil) used robot arms 2 and 3 to adjust
the bronchoscope actuator and steer the bronchosc-
ope tip into the hypopharynx and through the vocal
cords (Video 1; see Supplemental Digital Content 1,
http://links.lww.com/AA/A171; see the Appendix for
video legend). Next the bronchoscope tip was placed
external to the nare, an anesthesiologist manually ad-
vanced and rotated the bronchoscope, and an urologist
used the robot to steer the tip into the hypopharynx and
through the vocal cords.
RESULTS
Two intubation attempts were completed for this dem-
onstration. During oral intubation, it took 75 seconds to
advance the bronchoscope tip from the oropharynx to
carinal visualization. For nasal intubation, it took 67
seconds to advance the tip from nasal entry to carinal
visualization.
DISCUSSION
Fiberoptic intubation is feasible with robotic equipment.
We did not encounter significant differences between the
nasal and oral intubation routes. Even if optimized for
anesthetic practice, robotic-assisted anesthetic procedures
are not likely to become a part of routine anesthetic

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 929

 TECHNICAL COMMUNICATION
Figure 1. Schematic depicting the arrangement of the DaVinci
Surgical System type S (DVS), bronchoscope, and airway
mannequin.
Figure 2. Schematic depicting the arrangement of the DaVinci
Surgical System type S (DVS), bronchoscope, and airway
mannequin.
Figure 3. The handle of the bronchoscope was positioned into a
robotic arm. Two other robotic arms were fitted with graspers and
used to control the flexion and extension of the bronchoscope tip.
930 www.anesthesia-analgesia.org
Figure 4. Image from the DaVinci workstation during the roboti-
cally assisted fiberoptic intubation. The robot operator can simul-
taneously visualize video output from both the bronchoscope and
the robot’s main camera without turning attention away from the
workstation.
practice. Their optimal role may ultimately be for environ-
ments hazardous to routine anesthetic practice, such as the
battlefield or space-based environments.5
Our initial attempts at robotic-assisted intubation fo-
cused on direct laryngoscopy. However, we had consider-
able difficulty using the DVS robotic graspers to lift and
manipulate both Macintosh and Miller laryngoscope light
handles. Direct endotracheal tube manipulation with the
DVS was quite challenging. On the other hand, the focus of
the DVS on small-scale manipulation, coupled with its
multiaxis flexibility and unifying video input, suggested
that fiberoptic approaches to airway management would
capitalize on the capabilities of the DVS.
The robot operator controls the DVS through a dedi-
cated workstation in a corner of the operating room. The
workstation could easily be placed in another room, build-
ing, or continent.6,7 Traditionally, such distances have been
limited by latencies between user input, robot action, and
robot-user feedback.8 Progress in adapting data packet
transmission along existing telecommunication systems has
minimized such limitations,9 allowing long-distance ro-
botic telesurgery with acceptable latencies. 10
Ideally, this exercise would have required zero nonro-
botic interventions. However, because of the cost of the
bronchoscope, we elected to accept a loss in simulation
fidelity to minimize potential damage to the costly equip-
ment. Further work will be necessary to explore how well
the DVS can rotate and advance a bronchoscope tip.
Our experience indicates that because of the support and
preparation required for robotically assisted intubation, cur-
rent systems are not ready for such deployment into complex
operating environments. The bedside presence of the anesthe-
siologist was necessary even in this focused simulation to
assist with important maneuvers necessary during robotically
assisted fiberoptic intubation. Furthermore, this exercise did
ANESTHESIA & ANALGESIA
Robot-Assisted Airway Support
not include important factors such as patient preparation,
positioning, topical and systemic medication administration,
and patient monitoring.
This study demonstrated that a multipurpose surgical
robot could be adapted for use in airway management.
Although limited in its approach to direct laryngoscopy,
the DVS was able to assist with both oral and nasal
fiberoptic intubation. Future studies will be necessary to
optimize robotic interfaces with other airway management
techniques.
APPENDIX: VIDEO CAPTIONS
Video 1. This video demonstrates how the robotic manipu-
lation arms adjusted the flexion and extension of the
bronchoscope tip, and how the bronchoscope itself was
secured to a third arm. Both oral and nasal intubations are
demonstrated. The bronchoscope advancement was per-
formed manually to avoid damage to the bronchoscope and
required very minimal human input. Advancement was
performed at the direction of the robot operator.
REFERENCES
1. Hemmerling TM. Automated anesthesia. Curr Opinion Anaes-
thesiol 2009;103:811– 6
2. Bhayani S, Snow D. Novel dynamic information integration
during da Vinci robotic partial nephrectomy and radical ne-
phrectomy. J Robotic Surg 2008;2:67–9
October 2010 • Volume 111 • Number 4
3. Rogers CG, Laungani R, Bhandari A, Krane LS, Eun D, Patel
MN, Boris R, Shrivastava A, Menon M. Maximizing console
surgeon independence during robot-assisted renal surgery by
using the fourth arm and TilePro. J Endourol 2009;23:115–22
4. Palep JH. Robotic assisted minimally invasive surgery. J Minim
Access Surg 2009;5:1–7
5. Harnett BM, Doarn CR, Rosen J, Hannaford B, Broderick TJ.
Evaluation of unmanned airborne vehicles and mobile robotic
telesurgery in an extreme environment. Telemed e-Health
2008;14:539 – 44
6. Sterbis JR, Hanly EJ, Herman BC, Marohn MR, Broderick TJ,
Shih SP, Harnett B, Doarn C, Schenkman NS. Transcontinental
telesurgical nephrectomy using the da Vinci robot in a porcine
model. Urology 2008;71:971–3
7. Marescaux J, Leroy J, Gagner M, Rubino F, Mutter D, Vix M,
Butner SE, Smith MK. Transatlantic robot-assisted telesurgery.
Nature 2001;413:379 – 80
8. Anvari M, Broderick T, Stein H, Chapman T, Ghodoussi M,
Birch DW, Mckinley C, Trudeau P, Dutta S, Goldsmith CH.
The impact of latency on surgical precision and task comple-
tion during robotic-assisted remote telepresence surgery.
Comp Aided Surg 2005;10:93–9
9. Rayman R, Primak S, Patel R, Moallem M, Morady R, Tavakoli
M, Subotic V, Galbraith N, van Wynsberghe A, Croome K.
Effects of latency on telesurgery: an experimental study. Medi-
cal image computing and computer-assisted intervention.
MICCAI 2005;8:57– 64
10. Rayman R, Croome K, Galbraith N, McClure R, Morady R,
Peterson S, Smith S, Subotic V, Wynsberghe AV, Primak S.
Long-distance robotic telesurgery: a feasibility study for care in
remote environments. Intl Med Robotics Comp Assist Surg
2006;2:216 –24
www.anesthesia-analgesia.org 931
 Anesthesia Patient Safety Foundation
Section Editor: Sorin J. Brull
SPECIAL ARTICLE

Personal Protective Equipment for Care of Pandemic

Influenza Patients: A Training Workshop for the
Powered Air Purifying Respirator
Bonnie M. Tompkins, MD,* and John P. Kerchberger, MD†

Virulent respiratory infectious diseases may present a life-threatening risk for health care
professionals during aerosol-generating procedures, including endotracheal intubation. The
2009 Pandemic Influenza A (H1N1) brings this concern to the immediate forefront. The
Centers for Disease Control and Prevention have stated that, when performing or participating
in aerosol-generating procedures on patients with virulent contagious respiratory diseases,
health care professionals must wear a minimum of the N95 respirator, and they may wish to
consider using the powered air purifying respirator (PAPR). For influenza and other diseases
transmitted by both respiratory and contact modes, protective respirators must be combined
with contact precautions.
The PAPR provides 2.5 to 100 times greater protection than the N95, when used within the
context of an Occupational Safety and Health Administration– compliant respiratory protec-
tion program. The relative protective capability of a respirator is quantified using the assigned
protection factor. The level of protection designated by the APF can only be achieved with
appropriate training and correct use of the respirator.
Face seal leakage limits the protective capability of the N95 respirator, and fit testing does not
assure the ability to maintain a tight face seal. The protective capability of the PAPR will be defeated
by improper handling of contaminated equipment, incorrect assembly and maintenance, and improper
don (put on) and doff (take off) procedures. Stress, discomfort, and physical encumbrance may impair
performance. Acclimatization through training will mitigate these effects.
Training in the use of PAPRs in advance of their need is strongly advised. “Just in time”
training is unlikely to provide adequate preparation for groups of practitioners requiring
specialized personal protective equipment during a pandemic. Employee health departments
in hospitals may not presently have a PAPR training program in place. Anesthesia and critical
care providers would be well advised to take the lead in working with their hospitals’
employee health departments to establish a PAPR training program where none exists.
User instructions state that the PAPR should not be used during surgery because it
generates positive outward airflow, and may increase the risk of wound infection. Clarifica-
tion of this prohibition and acceptable solutions are currently lacking and need to be
addressed. The surgical hood system is not an acceptable alternative.
We provide on line a PAPR training workshop. Supporting information is presented here.
Anesthesia and critical care providers may use this workshop to supplement, but not
substitute for, the manufacturers’ detailed use and maintenance instructions. (Anesth Analg
2010;111:933–45)

E xposure to virulent respiratory pathogens during in-

vasive airway procedures may present a life-
threatening risk for health care providers. The Institute
of Medicine (IOM) of the National Academies Committee on
From the *Department of Anesthesiology, University of Wisconsin School of
Medicine and Public Health, Madison, Wisconsin; and †Department of
Anesthesiology, Rush University Medical Center, Chicago, Illinois.
Accepted for publication May 6, 2010.
Disclaimer: Neither author has a financial relationship with any device
manufacturer.
Disclosure: The authors report no conflicts of interest.
Address correspondence and reprint requests to Bonnie M. Tompkins, MD,
Department of Anesthesiology, University of Wisconsin Hospitals, 600 High-
land Ave., Madison, WI 53792-0001. Address e-mail to bmtompki@wisc.edu.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181e780f8
PPE (personal protective equipment) for Healthcare Workers
stated that, “There is an urgent need to address the lack of
preparedness regarding effective PPE for use in an influenza
pandemic.” (Table 1, Ref. 1). In failing to anticipate and
address PPE issues, health care workers may threaten not
only personal, colleague, and family safety, but patient safety
as well.1 Safe and effective use of PPE can only be optimized
by practicing correct procedures and by being aware of the
operational factors that defeat its protective capacity.
VIRAL PANDEMICS AND OUTBREAKS—PAST
AND PRESENT
Influenza pandemics typically occur several times each
century. The 1957 and 1968 pandemics (Pandemic Severity
Index [PSI] 2, case fatality ratio [CFR] 0.1⫺⬍0.5%) did not

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 933

 SPECIAL ARTICLE

Table 1. Website References

Reference no. Website
1 Institute of Medicine (IOM) 2008. Preparing for an Influenza Pandemic: Personal Protective Equipment for Healthcare Workers.
Washington, DC: The National Academies Press. Available at: http://www.nap.edu/catalog.php?record_id⫽11980.
Accessed February 21, 2010
2 Interim Pre-pandemic Planning Guidance: Community Strategy for Pandemic Influenza Mitigation in the United States. Available
at: http://pandemicflu.gov/professional/community/commitigation.html#IV. Accessed February 21, 2010
3 Department of Heath and Human Services (U.S. DHHS) Pandemic Influenza Plan 2005. Available at: http://hhs.gov/
pandemicflu/plan/appendixb.html. Accessed February 22, 2010
4 Occupational Safety and Health Administration (OSHA). Pandemic Influenza Preparedness and Response Guide 2007.
Available at: http://www.osha.gov/Publications/3328-05-2007-English.html. Accessed February 22, 2010
5 Centers for Disease Control and Prevention (CDC). Estimates of 2009 H1N1 Influenza Cases, Hospitalizations, and Deaths in the
United States. Available at: http://www.cdc.gov/h1n1flu/estimates/results_2009_h1n1.htm. Accessed January 16, 2010
6 Cumulative Number of Confirmed Human Cases of Avian Influenza A/(H5N1) Reported to World Health Organization (WHO).
Available at: http://www.who.int/csr/disease/avian_influenza/country/cases_table_2010_02_17/en/index.html. Accessed
February 18, 2010
7 WHO/Pandemic Influenza Preparedness and Response. Available at: http://www.who.int/csr/disease/influenza/
PIPGuidance09.pdf. Accessed February 21, 2010
8 Center for Infectious Disease Research and Policy (CIDRAP). Avian Influenza (Bird Flu): Implications for Human Disease. Last
updated April 2, 2010. Available at: http://www.cidrap.umn.edu/cidrap/content/influenza/avianflu/biofacts/
avflu_human.html. Accessed February 21, 2010
9 WHO Summary of Probable SARS Cases with Onset of Illness from 1 November 2002 to 31 July 2003. Available at: http://
www.who.int/csr/sars/country/table2004_04_21/en/index.html. Accessed February 22, 2010
10 The SARS Commission Executive Summary: Spring of Fear. Available at: http://www.health.gov.on.ca/english/public/pub/
ministry_reports/campbell06/online_rep/V1.html. Accessed September 28, 2009
11 Adalja AA. Will the D225G Mutation Herald More Severe Illness in Patients with 2009 H1N1 Influenza? Clinician’s Biosecurity
Network Report. Available at: cbn_editor@upmc-biosecurity.org. Accessed December 11, 2009
12 1918 Polymorphism in Ukraine H1N1? Available at: http://www.recombinomics.com/News/11090902/Ukraine_1918.html.
Accessed December 11, 2009
13 Siegel JD, Rhinehart E, Jackson M, Chiarello L, and the Hospital Infection Control Practices Advisory Committee (HICPAC).
2007 Guideline for Isolation Precautions: Preventing Transmission of Infectious Agents in Healthcare Settings. Available at:
http://www.cdc.gov/ncidod/dhqp/pdf/guidelines/Isolation2007.pdf. Accessed February 20, 2010
14 Interim Domestic Guidance on the Use of Respirators to Prevent Transmission of SARS May 3, 2005. Available at: http://
www.cdc.gov/ncidod/sars/respirators.htm. Accessed February 19, 2010
15 IOM 2009. Respiratory Protection for Healthcare Workers in the Workplace Against Novel H1N1 Influenza A: A Letter Report.
Washington, DC: The National Academies Press. Available at: http://www.nap.edu/catalog/12748.html. Accessed February
22, 2010
16 IOM 2006. Reusability of Facemasks During an Influenza Pandemic. Board of Health Sciences Policy. Washington, DC: The
National Academies Press. Available at: http://books.nap.edu/openbook.php?record_id⫽11637&page⫽R1. Accessed
February 22, 2010
17 Interim Guidance on Planning for the Use of Surgical Masks and Respirators in Health Care Settings during an Influenza
Pandemic. Oct. 2006 Available at: http://www.flu.gov/professional/hospital/maskguidancehc.html#airborne. Accessed Feb
20, 2010
18 Proposed Guidance on Workplace Stockpiling of Respirators and Facemasks for Pandemic Influenza. Available at: http://
www.osha.gov/dsg/guidance/proposedGuidanceStockpilingRespirator.pdf. Accessed February 21, 2010
19 Infection Prevention and Control in Health Care for Confirmed or Suspected Cases of Pandemic (H1N1) 2009 and Influenza-
Like Illnesses 25 June 2009. Available at: http://www.who.int/csr/resources/publications/SwineInfluenza_infectioncontrol.pdf.
Accessed February 21, 2010
20 Interim Guidance on Infection Control Measures of 2009 H1N1 Influenza in Healthcare Settings, Including Protection of
Healthcare Personnel. October 14, 2009. Available at: http://www.cdc.gov/h1n1flu/guidelines_infection_control.htm.
Accessed February 21, 2010
21 OSHA Instruction Directive #: CPL-02-02-075 Effective Nov. 11, 2009. Enforcement Procedures for High to Very High
Occupational Exposure Risk to 2009 H1N1 Influenza. Available at: http://www.osha.gov/OshDoc/Directive_pdf/
CPL_02_02–075.pdf. Accessed February 18, 2010
22 OSHA, Appendix C, Pandemic Influenza Preparedness and Response Guide 2007 Implementation and Planning for Respiratory
Protection Programs in Healthcare Settings. Available at: http://www.osha.gov/Publications/OSHA_pandemic_health.pdf.
Accessed February 22, 2010
23 OSHA State Occupational Safety and Health Plans. Available at: http://www.osha.gov/dcsp/osp/index.html. Accessed
February 22, 2010
24 Assigned Protection Factors for the Revised Respiratory Protection Standard OSHA 3352–02 2009. Available at: http://
www.osha.gov/Publications/3352-APF-respirators.html. Accessed February 22, 2010
25 King JH. Assessing Powered Air Purifying and Supplied Air Respirator Performance: An Employer’s Guide to OSHA’s Final Rule
on Assigned Protection Factors for Respirators. An E.D. Bullard Company White Paper. September 21, 2006. Available at:
http://www.bullard.com/V3/resources/downloads/respiratory_dwnlds.php#Technical_Publications. Accessed February 22,
2010
26 3M™ Technical Data Bulletin #160; Reusable Respirator Facepieces and Powered Air Purifying Respirator Systems (PAPRs) in
the HealthCare Environment: Considerations for Use; and TDB #175 Assigned Protection Factors. 3M Occupational Health
& Environmental Safety Respiratory Protection Media Library. Available at: http://www.3M.com/occsafety. Accessed
February 21, 2010
(Continued)

934 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

Personal Protective Equipment for Pandemic Influenza

Table 1. (Continued)
Reference no. Website
27 Understanding Respiratory Protection Against SARS. Available at: http://www.cdc.gov/niosh/npptl/topics/respirators/
factsheets/respsars.html. Accessed February 26, 2010
28 Adalja AA. Update on Personal Protective Equipment. Available at: http://www.upmc-cbn.org/report_archive/2008/
12_December_2008/cbnreport_12192008.html. Accessed February 21, 2010
29 General Procedures for Putting On and Taking Off a Disposable Respirator. Available at: http://www.cdc.gov/flu/
freeresources/2009-10/pdf/n95respirator_instructions.pdf. Accessed February 19, 2010
30 3M Air-Mate™. Available at: http://www.3m.com/occsafety. Accessed February 21, 2010
31 3M Resource CD. Available on request from 3M Occupational Health and Environmental Safety Division Technical Service at
800-243-4630 and from jfbrachmann1@mmm.com. Available at: http://www.3M.com/occsafety. Accessed October 11,
2009
32 Powered Air Purifying System PA20™. Available at: http://www.bullard.com/Respiratory/papr/pa20page.shtml. Accessed
October 11, 2009
33 Bullard EVA™ Competitive Comparison. Available at: http://www.bullard.com/V3/resources/downloads/respiratory_
dwnlds.php#EVA. Accessed February 5, 2010
34 PAPR Training Workshop. Available at: http://www.med.wisc.edu/papr-workshop. Accessed October 12, 2009
35 OSHA Best Practices for Hospital-Based First Receivers of Victims from Mass Casualty Incidents Involving the Release of
Hazardous Substances. Available at: http://www.osha.gov/dts/osta/bestpractices/html/hospital_firstreceivers.html.
Accessed October 11, 2009
36 Full Barrier Personal Protective Equipment (PPE) with Powered Air Purifying Respirator (PAPR). Available at: http://
www.health.state.mn.us/divs/idepc/dtopics/infectioncontrol/ppe/ppepapr.html. Accessed September 28, 2009
37 3M Technical Data Bulletin #178: Maintenance and Care of 3M Powered Air Purifying Respirator (PAPR) Batteries. Published
March 2007, Revised November 2008. Available at: http://www.3M.com/occsafety. Accessed February 21, 2010
38 Maintenance of Battery Packs for Bullard Powered Air-Purifying Respirators (PAPRs). Available at: http://www.bullard.com.
Accessed February 21, 2010
39 Bullard Technical Advisory: Release of Filtered Particles. Available at: http://www.bullard.com. Accessed February 21, 2010
40 Stryker T4 Surgical Helmet System Filtration Testing Summary Report. Available at: www.sars.medtau.org/strykerreport.doc.
Accessed August 31, 2009
41 Using the Stryker T4 Personal Protection System for High-Risk Procedures During SARS Outbreaks. Available at:
sars.medtau.org/strykertraining.htm. Accessed April 16, 2010
42 Questions and Answers About CDC’s Interim Guidance on Infection Control Measures for 2009 H1N1 Influenza in Healthcare
Settings, Including Protection of Healthcare Personnel. December 1, 2009. Available at: http://www.cdc.gov/h1n1flu/
guidance/control_measures_qa.htm. Accessed February 21, 2010
43 Personal Protective Equipment in Pandemic/Avian Influenza/SARS: N95 or PAPR for Intubation? ASA Newsletter. Available at:
http://www.asahq.org/Newsletters/2008/01-08/tompkins01-08.html. Accessed September 28, 2009

approach the catastrophic nature of the 1918 Great Influ-
enza, “Spanish Flu” (PSI 5, CFR ⱖ2%), wherein one-third
(approximately 500 million) of the world’s population
became ill and 50 to 100 million died (Table 1, Refs. 2– 4).2,3
The Centers for Disease Control (CDC) midrange esti-
mates for the 2009 Pandemic Influenza A (H1N1) (hereafter
2009 H1N1) in the United States from April 2009 to January
2010 are 57 million total cases, 257,000 (0.45%) hospitaliza-
tions, and 11,700 deaths (overall CFR 0.02%, hospitalized
CFR 4.5%) (Table 1, Ref. 5). In seasonal influenza, the
greatest mortality is among the very young and the very
old; in contrast, the 1918 and 2009 H1N1 pandemics
predominantly affected children and younger adults (Table
1, Refs. 3 and 5).2
The World Health Organization (WHO) reported that 478
human confirmed cases and 286 deaths (CFR 60%) occurred
from 2003 through February 2010 from the H5N1 highly
pathogenic avian influenza (HPAI) (“bird flu”). Many cases
were in children and young adults (Table 1, Ref. 6). This virus
has not currently developed a high affinity for human respi-
ratory tract receptors; therefore, human-to-human transmis-
sion is nonsustained and cases have occurred only in small
clusters (WHO pandemic phase 3). Should H5N1 HPAI and
2009 H1N1 occur simultaneously in the same individual, a
highly pathogenic reassortant pandemic strain could emerge
(Table 1, Refs. 7 and 8).
The severe acute respiratory syndrome coronavirus
(SARS-CoV) epidemics of 2002 to 2003 incurred ⬎8000
cases and 700 deaths in 29 countries (CFR 9.6%) (Table 1,
Ref. 9). Twenty percent of the cases were in health care
workers.4 In Ontario, Canada, ⬎50% of the 438 SARS cases
and 3 of the 43 deaths were in health care workers (Table 1,
Ref. 10).5–7 The SARS experience is of particular import to
anesthesiology and critical care specialists.1
VIRULENCE FACTORS IN SEVERE VIRAL
RESPIRATORY ILLNESS
The history and pathology of influenza have been studied
and presented effectively by Taubenberger and Morens.8
Both pandemic and seasonal influenza viruses may repli-
cate throughout the respiratory tract. Disease is limited to
the upper respiratory tract and trachea in nonfatal cases,
but fatal cases show lung involvement.8,9 The 1918 and
2009 H1N1 pandemic and H5N1 HPAI viruses exhibit a
greater propensity than seasonal influenza to bind to viral
receptors in the lower respiratory tract.3,8 –11 The histopa-
thology from autopsy of influenza victims is that of a
primary viral hemorrhagic bronchitis and pneumonia with
diffuse alveolar damage and destruction.8,9,12 Secondary
bacterial pneumonia contributes prominently to the mor-
tality in seasonal and pandemic influenza.8,9,12

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 935

 SPECIAL ARTICLE
The genomes of the 1918 and present-day influenza
viruses have been reconstructed, studied, and compared,
presenting opportunity for identifying preventive and
therapeutic approaches. This work has clarified how the
virus binds to lower as well as upper respiratory tract
receptors and has identified gene components associated
with increased virulence of the 1918 pandemic, H5N1
avian, and 2009 H1N1 pandemic influenza viruses (Table 1,
Refs. 11 and 12).3,8,11,13 Identified genetic amino acid se-
quences associated with host specificity and high virulence
may provide a predictive monitoring tool.14
In addition to diffuse lung damage, altered immune
mechanisms and a viral-associated extreme proinflammatory
cytokine response, sometimes with hemophagocytosis, have
been observed in young, previously healthy influenza pa-
tients who died or were critically ill.8,15 These pathologic
features also have been described in animal models inocu-
lated with the reconstructed 1918 virus,11 in animals and
humans with H5N1 avian influenza,13,15,16 in SARS-CoV,17–19
and in 2009 H1N1 victims (Table 1, Ref. 12).9
TRANSMISSION MODES
The predominant transmission modes for influenza and
SARS are respiratory droplet and direct and indirect con-
tact (fomite). (Table 1, Refs. 13 and 14). Epidemiologic and
investigational evidence is strongly suggestive of near-
range airborne transmission (Table 1, Refs. 1, 9, 10, and
13–15).20 –22 Remote airborne transmission is supported by
laboratory research and theoretical modeling of aerosol
behavior and is suspected in specific outbreaks, but be-
lieved to be unusual (Table 1, Refs. 1 and 15).21–25
Contact Mode
The sources of contact transmission are often overlooked
(contaminated surfaces, clothing, equipment, personal pro-
tective equipment, exposed skin) (Table 1, Refs. 1 and 13).
The infectivity of virus on surfaces, skin, and hands decays,
and the time that the fomite remains infective may vary
(Table 1, Ref. 13).22
Droplet and Airborne Modes
Newer investigations on aerosol transmission of influenza
have challenged the traditional belief that airborne trans-
mission does not occur. Droplet and airborne modes are
now viewed as a continuum, with particle sizes ranging
from large to fine droplet or aerosol (Table 1, Refs. 1, 14,
and 15).20 –22 The term “droplet” is consistent with trans-
mission by larger disease-bearing particles that settle out of
the air onto surfaces within shorter distances from the
source or are inhaled into the upper airway and trachea.
Larger droplets may evaporate to small “droplet nuclei”
that behave as aerosol (Table 1, Refs. 1, 13, and 16).
“Near-range airborne” transmission is through smaller
droplet particles (inspirable) that remain suspended in air
for relatively short but variable distances (1 m with breath-
ing, 2 m with coughing, 6 m with sneezing), and when
inhaled, reach only the trachea and bronchi (Table 1, Refs.
1, and 13–15).20 –22 Near-range airborne transmission
through a spectrum of disease-bearing particle sizes is now
acknowledged by the CDC and others to be operational in
influenza and SARS (Table 1, Refs. 1, 8, 13, and 14).22
936 www.anesthesia-analgesia.org
The term “airborne transmission” traditionally refers to
the remote transmission and inhalation of yet smaller
droplets and aerosol-sized (respirable) particles that may
access the alveoli as well as the upper airway and may
remain suspended in the air for an indeterminate time and
distance (Table 1, Ref. 13).20 Airborne transmission of any
respiratory infectious disease is difficult to prove or to
definitively exclude.23 Isolated specific outbreaks of SARS
(aerosolized remote fecal source) and influenza (airplane
outbreaks) are believed to be examples of unusual, or
opportunistic, airborne transmission; aerosol research sup-
ports the possibility of remote airborne transmission of
influenza and SARS (Table 1, Refs. 1, 14, 15, and 17).20,22,23
Transmission and subsequent infection is influenced by
additional factors that include viability of the agent, expi-
ratory force, distance from and duration of exposure to the
source, and environmental conditions (humidity, tempera-
ture, and wind) (Table 1, Ref. 1).26 –28
AEROSOL-GENERATING PROCEDURES IN
ANESTHESIA AND CRITICAL CARE
Aerosol particles are generated during all invasive airway
procedures, noninvasive and positive pressure ventilatory
support modes, suction, sputum induction, high-flow oxy-
gen delivery, aerosolized or nebulized medication delivery,
interventions that stimulate coughing, and autopsies (Table
1, Refs. 1, 4, 13, 15, and 17).21,22 Infection is established with
a smaller quantity of aerosol than nasal instillation (Table 1,
Ref. 1).22,29 Health care professionals who perform and
assist with aerosol-generating procedures and therapies are
included in the Occupational Safety and Health Adminis-
tration (OSHA) “very high occupational exposure risk”
category (Table 1, Refs. 3, 4, and 18), and are at a higher risk
of infection than others.
TRANSMISSION-BASED PRECAUTIONS
Close Patient Care
Confusion persists about whether a facemask or respirator
is indicated for close care of influenza patients and those
with a flu-like illness. For 2009 H1N1 influenza, because of
the limited access to N95s in many countries, the WHO has
stated that the N95 (or European equivalent, EU FFP2) is
only indicated during aerosol-generating procedures, and
that a surgical mask in addition to face shield and eye
protection, with other contact precautions, is satisfactory
for all other patient care activity (Table 1, Ref. 19). In
contrast to the WHO, the CDC, OSHA, and the IOM of the
National Academies all state that the N95 respirator, or
higher ([N100], or powered air purifying respirator
[PAPR]), should be used during close contact with influ-
enza or SARS patients. Contact precautions (gown, gloves,
hat, close-fitting eye protection, shoe covers) are also rec-
ommended for influenza and other virulent diseases that
are transmitted by both the respiratory and contact modes.
The CDC describes close contact as within 6 to 10 ft. of the
patient or in the patient’s room (Table 1, Ref. 20).
Aerosol-Generating Procedures
The CDC, OSHA, and the IOM further suggest that health
care personnel consider using the higher level of protection
provided by the PAPR when performing or assisting with
ANESTHESIA & ANALGESIA
Personal Protective Equipment for Pandemic Influenza
aerosol-generating procedures (Table 1, Refs. 1, 3, 4, 13, 15,
and 20). Some state and hospital pandemic influenza,
SARS, and tuberculosis protocols assert definitively that
the PAPR is to be used for endotracheal intubation and
bronchoscopy. The CDC states that these procedures
should be done in an airborne infection isolation room
(Table 1, Ref. 20).
OSHA STANDARDS APPLICABLE TO HEALTH
CARE FACILITIES
The General Duty Clause of the Occupational Safety and
Health Act requires that employers abate or address recog-
nized workplace hazards. OSHA standards for employee
safety were developed initially for the industrial work-
place; subsequently, standards for the health care venue
were developed as well (Table 1, Refs. 4 and 21). The OSHA
Blood-borne Pathogens, Personal Protective Equipment,
and Respiratory Protection standards, as listed in the Code
of Federal Regulations, are of particular importance in
pandemic influenza preparedness and response (Table 1,
Ref. 4). Airborne pathogens are included in the hazardous
airborne contaminants covered by the standards. OSHA
has posted inspection, investigation, and enforcement pro-
cedures that address 2009 H1N1 Influenza in health care
workers in the “high” and “very high occupational expo-
sure risk” categories (Table 1, Ref. 21).
Under the Respiratory Protection standard (29 CFR
1910.134), all respirators must be deployed within the
context of a written, worksite-specific respiratory protec-
tion program. The requisite respiratory protection program
director oversees and documents respirator selection,
maintenance and cleaning procedures, employee medical
clearance to wear a respirator, fit testing of tight-fitting
respirators, instruction, and periodic program review
(Table 1, Ref. 22).
States may elect to develop their own standards for
employee safety, which the employers must follow, but
these plans must be approved by OSHA (Table 1, Ref. 23).
When applied, the OSHA standards maximize safety by
supporting the correct and consistent use of respirators.
QUANTIFICATION OF RESPIRATORY PROTECTION:
THE ASSIGNED PROTECTION FACTOR
The assigned protection factor (APF), applied to industrial
respirators initially, is a value given or assigned to each
respirator by OSHA and the National Institute of Occupa-
tional Safety and Health (NIOSH) denoting the factor by
which a respirator reduces the contaminating substance in the
ambient air (Table 1, Ref. 24). The APF is derived from
laboratory simulated workplace (SWPF) and actual work-
place protection factors (WPF) computed from photometric
determinations of the test contaminant outside (Co) the respi-
rator relative to inside (Ci) the respirator facemask.30 For each
respirator tested, the lowest fifth percentile of test Co/Ci
values are averaged and further divided by a safety factor of
25. This final value is assigned to the respirator as the APF.
APFs range from 10 to 10,000. This number represents the
minimum factor by which exposure to contaminants is re-
duced when wearing the respirator. Hence, a higher number
indicates greater protection (Table 1, Refs. 25 and 26).
October 2010 • Volume 111 • Number 4
The APF of the N95 respirator is 10, meaning that the
wearer will expect to inhale no more than one-tenth of the
hazardous airborne particles present (Table 1, Ref. 24).
OSHA has set a minimum APF of 1000 that a PAPR must
achieve to obtain NIOSH certification, but charges the
prospective buyer with obtaining confirmatory testing evi-
dence of the APF from each PAPR manufacturer (Table 1,
Refs. 24 and 25).30
TERMINOLOGY AND CATEGORIZATION OF
FACEMASKS AND RESPIRATORS
Medical Masks
Medical masks of several different types are worn, as
intended, for protecting the patient and the environment
from the respiratory secretions of health care workers
(Table 3A). Masks also provide the wearer with droplet and
contact protection for the skin area covered by the mask.
The degree of protection corresponds to the masks’ fluid
resistance; surgical masks are fluid resistant, procedure
masks are less so, and isolation masks generally are not.
The filtering fabric of some masks is tested to 0.1-␮m
particle size but this is not required for Food and Drug
Administration certification. Medical masks do not permit
a tight face seal, do not provide a barrier to aerosol
particles, and therefore do not meet the criteria for a
respirator. Gauze masks are an effective droplet barrier
only with 6 or more layers (Table 1, Ref. 16).
The infection rate was reduced in Hong Kong during the
SARS outbreak with consistent use of medical masks plus
contact barrier garb that consisted of gowns, gloves, and
eye protection.6,26 Public use of masks, with hand hygiene,
also reduced the infection rate (Table 1, Refs. 1 and 16). A
comparison of the infection rate between nurses wearing
surgical masks or N95s found the influenza rate to be no
different, but still considerable in both groups (approxi-
mately 23%).31
Discussion and study continue on evidence-based choice
of respiratory personal protective equipment (PPE) for health
care workers exposed to influenza.32,33 In the absence of a
respirator or when supplies are limited, the surgical mask,
with contact transmission precautions, should be used for
routine bedside care of SARS and influenza patients (Table 1,
Ref. 17). Masks must be removed carefully and discarded,
followed by hand washing (Table 1, Refs. 13 and 20).
Respirators
In the context of PPE, a “respirator” reduces exposure to
inhaled environmental contaminants (Table 1, Ref. 16).
Respirator classifications consider the air supply, power,
filter type and efficiency, facepiece description, and pres-
ence of an exhalation valve. Additional descriptive fea-
tures, such as air flow direction, clarify the respirator’s
protective mechanism (Tables 2 and 3B) (Table 1, Refs. 1, 4,
15, and 16).34 The particulate respirator filter removes
droplet and aerosol-sized airborne particles and an absor-
bent respirator filter removes chemical vapors and gases.
The nonpowered category of air purifying respirators
includes the elastomeric and the filtering facepiece respira-
tors. The “elastomeric” respirator has a full or half facepiece
that is nondisposable, tight fitting, and made of silicone or
rubber with either a replaceable particulate or vapor/gas
www.anesthesia-analgesia.org 937
 SPECIAL ARTICLE
Table 2. Categorizing Respirators
Type of respirators
Air purifying
Nonpowered: Depend on the wearer drawing air in through filters
or cartridges
Powered air purifying respirators: A blower draws air through the
filter and delivers it to the wearer
Air supplying
Self-contained breathing apparatus
Type of filters
Particulate filters
P (oil proof; can survive oil exposure for ⬎1 work shift)
R (oil resistant; can be used for oil exposure in 1 shift)
N (not oil resistant; used for oil-free environments)
Gas-vapor respirator
Combination particulate and gas vapor
Filtering efficiency
Certified for a range of efficiency classes (e.g., 95%, 99%, 100%)
Type of facepiece
Filtering facepieces
Replaceable filter components: half-mask and full-mask
elastomeric respirators
Loose-fitting facepieces
Use or nonuse of an exhalation valve
Adapted from: Institute of Medicine. Preparing for an Influenza Pandemic:
Personal Protective Equipment for Health Care Workers. Washington, DC: The
National Academies Press, 2008. Reprinted with permission from the Na-
tional Academies Press.
absorbing filter, or a combination filter. Examples are the
“gas mask,” “MOPP” respirator (Mission Oriented Protec-
tive Posture) (dual filter), the painter’s respirator (vapor),
and the particulate elastomeric respirator for use in health
care if N95 stores are exhausted.
The filtering facepiece respirators (N95 and higher) are
classified by the resistance to degradation by oil, and by the
percent filtration efficiency of 0.3-␮m saline test particles,
the most penetrating size and substance (Table 1, Ref. 4).
Both the N95 and the elastomeric are negative pressure
respirators; that is, they are dependent on the maintenance
of a tight face seal to prevent contaminated air from
bypassing the filter (Table 1, Refs. 4, 14, and 27).
The notable advantages of the N95 are simplicity and
accessibility.35 The predominant disadvantages are the
well-known increased resistance to breathing36 and the
propensity for a gap to occur in the face seal (face seal
leakage) (Table 4). Prior fit testing does not assure success
in attaining and maintaining a tight face seal. Only those
wearing the N95 in their daily work were found to achieve
a face seal with some consistency, irrespective of when
prior fit testing had occurred (Table 1, Ref. 28).37 The N95
don (put on) and doff (take off) procedure posted by the
CDC includes a positive and negative pressure face seal
check, which should be performed when using any tight-
fitting respirator (Table 1, Refs. 4, 14, and 29).
Hospitals are urged to stockpile supplies. OSHA has
predicted that every nurse, using 4 N95s per shift, would
use 480 over a 12-week pandemic wave, at a cost of $240
per worker (Table 1, Ref. 18). All disposables will present a
substantial expense and will be depleted rapidly.38 The
CDC recommends prioritization of N95 use based on risk of
exposure, to be initiated when N95 supplies are limited
(Table 1, Ref. 20).
938 www.anesthesia-analgesia.org
Table 3A. Medical Masks
Types of Isolation; Surgical; Laser;
medical masks a Procedure Dental
Water resistance b Variable Yes
Transmission mode
protection c
For wearer:
Contact Yes Yes
Droplet Very little Yes
Aerosol (airborne) No No
For wearer’s contacts:
Contact Yes Yes
Droplet Very little Yes
Aerosol (airborne) No No
Chemical protection d None None
a
Medical masks include various types of face masks used in the healthcare
venue; they provide no protection from airborne biologic aerosol and are not
classified as respirators. Masks are intended to protect patients and other
close contacts from the wearer’s secretions.
b
Only water resistant masks and respirators provide the wearer (patient or
healthcare provider) with droplet protection from others. Water resistance is
not a requirement for certification of procedure and isolation masks and may
vary among different brands. Surgical masks are water resistant and protect
the wearer and their contacts from droplets and splashes.
c
Contact and droplet protection for the wearer is only over the area covered
by the mask or respirator.
d
Medical masks and respirators are particulate filters only and do not provide
protection from chemical agents.
Reuse of the N95 is strongly discouraged but permissible if
supplies are limited, provided it is not soiled, creased, dam-
aged, moist, or wet (Table 1, Refs. 14 and 16). Reaerosolization
of disease particles from the N95 does not occur (Table 1, Ref.
16).34 However, the used respirator is a fomite and presents a
contamination risk. Covering the N95 with a face shield will
protect it from droplets and splashes. A fluid-resistant N95 is
available and should be used in surgery (Table 1, Refs. 14 and
16). The used N95 should be stored in a paper, not plastic, bag
to avoid condensation (Table 1, Ref. 26). When donning a used
mask, extreme care should be taken not to contaminate
oneself, and hand washing should follow. The N95 filtering
facepiece respirator provides a 10-fold factor of protection
relative to ambient air (APF 10) (Table 1, Refs. 24 and 26).
THE PAPR FOR BIOLOGICAL HAZARDS
The PAPR provides a higher level of protection than the
N95 respirator because it supplies maximally filtered air,
eliminates face seal leak, and provides contact protection
for the head (Table 3B). It is composed of a belt-worn case
that houses a battery, fan, and filter. The fan, referred to as
the “blower,” draws ambient air through a high-efficiency
particulate air (HEPA) filter (99.97% efficient) and blows it at
⬎170 L/min (6 cu. ft./min) through a flexible tube and into a
Tyvek or Tychem hood with a plastic face shield (Table 1, Refs.
4 and 30–33). The high flow of air exiting the hood prevents the
wearer from entraining contaminated ambient air.
The PAPR hoods are available in 2 styles: the double-
shrouded hood and the loose-fitting face covering. The former
completely covers the head and shoulders and does not
permit use of a conventional stethoscope. The inner shroud
tucks beneath the neck of a gown, and the filtered air exits
from under the inner shroud and gown. The PAPR used with
a double-shrouded hood provides more than a 1000-fold
protection relative to ambient air (APF 1000) and 100 times the
ANESTHESIA & ANALGESIA
Personal Protective Equipment for Pandemic Influenza

Table 3B. Features of Masks and Respirators

a
Self-contained breathing apparatus.
b
Remote supplied air respirator.
c
Available with particulate filter, gas/vapor absorbent canister, or combination.
d
Powered air purifying respirator.
e
Water resistant N95 respirators are available and should be used during surgery.
f
Mission Oriented Protective Posture: Uniformed Services elastomeric respirator and chem warfare agent resistant clothing.
g
Chemical absorbent filters are agent specific; painter’s respirator is not protective against chemical and nerve agents.
h
The loose fitting face covering helmet covers the top of the head, the face and the chin, leaving the ears and neck exposed.
i
The double shrouded hood covers the entire head and shoulders.
j
Assigned Protection Factor (factor by which the test contaminant is reduced by the respirator).
k
Contact protection is only for the skin area covered by the mask or respirator.
l
Only with particulate filter.
m
Only with absorbent filter.
n
Storage sites and containers for biologic PAPRs should be labeled prominently, “Not for chemical protection”.
o
High Efficiency Particulate Air filter.
p
APF 50 for full face elastomeric.

protection of the N95. The double-shrouded hood offers the
best protection for aerosol-generating procedures.
The loose-fitting face covering protects the face, chin,
and top of the head but leaves the neck and ears exposed,
allowing use of a stethoscope. Illustrations are available
online (Table 1, Refs. 30 –34). The high-flowing air exits
through perforations beneath the chin. This hood is
appropriate for continuous bedside care in the absence of
aerosol-generating conditions. The protection factor of
this hood is 25-fold greater than no protection (APF 25)
and 2.5 times the protection of the N95 respirator (APF
10) (Table 1, Refs. 30 –33).
Because the PAPR is a loose-fitting respirator, fit testing
is not required. The full hood, but not the loose-fitting face
cover, may be used by those with a beard. Facial hair can
permit unfiltered air to be entrained under the elastic band
border of the loose-fitting face cover.
The PAPR does not increase the work of breathing and
is more comfortable for extended wear than an N95 (Table
1, Refs. 1 and 4) (Table 5). The PAPR and the nondisposable

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 939

 SPECIAL ARTICLE
elastomeric air purifying respirator are the only realistic
alternatives for respirators in the hospital during a pan-
demic if N95 stocks are exhausted.
THE PAPR FOR CHEMICAL OR
HAZARDOUS MATERIAL
The particulate HEPA filter in PAPRs used for protection
against biological hazard, e.g., 3M™ Air-Mate™, Bullard
PA20™, or Bullard EVA™, do not protect against hazard-
ous chemicals, gases, or vapors. PAPRs for biological
protection should be prominently labeled “not for chemical
Table 4. N95 Respirator Advantages
and Disadvantages
Advantages
● Filters 95% of aerosol test particles
● 10-fold protection (APF 10)
● Easily accessible
● Disposable
● No setup required
● No interference to using a stethoscope
● Not powered
● Noiseless
Disadvantages
● Face seal leak common
● Increases resistance to breathing (exhalation valve may reduce
discomfort but must cover with surgical mask during surgery)
● Requires fit testing (costly, labor intensive, does not ensure a
tight face seal)
● Facial features may preclude a satisfactory fit with any model
● Headaches, dizziness, shortness of breath, possible CO2
retention
● Supplies rapidly depleted when demand is high
● Reuse risks self-contamination (use face shield if reuse is likely)
(store in paper bag)
● Facial hair inhibits a face seal, defeats protection
● Ineffective when moist, wet, creased, or damaged
Table 5. Powered Air Purifying Respirator (PAPR) Advantages and Disadvantages
PAPR advantages
● HEPA filtered air
● Positive inside to outside air flow (170 L/min, 6
cfm 关cu. ft./min兴)
● Highest level of protection for aerosol-generating
procedures
● Contact protection of head and neck: double-
shrouded hood better than the loose-fitting face
covering hood; both more than N95
● Protection factor, relative to no protection:
1000 ⫻ for PAPR with double-shrouded hood
25 ⫻ for PAPR with loose-fitting face covering hood
10 ⫻ for N95 respirator
● No fit testing required
● No entrainment of contaminated air
● Comfortable for extended wear (bedside care,
reprocessing worker, invasive airway procedures)
● Hoods are disposable (reusable, by single user
only, after reprocessing)
● Facial hair is acceptable, with double-shrouded
hood only, not with loose-fitting face covering hood
HEPA ⫽ high-efficiency particulate air.
940 www.anesthesia-analgesia.org
or vapor protection.” The hazardous-material PAPRs, e.g.,
3M Breathe-Easy™, Bullard PA30™, and Bullard EVA™,
have vapor-absorbing and dual (combination) cartridges
and chemical-resistant hoods; these must be worn with
additional full-body chemical protective suits (Table 1,
Refs. 31, 33, and 35) (Appendix 1).
To be certified to wear chemical protective equipment, the
OSHA standard 29 CFR 1910.120 on Hazardous Waste Op-
erations and Emergency Response (HAZWOPER) requires a
combination of 8 hours of OSHA-compliant awareness-level
(informational) and operations-level (practical) training (Table
1, Ref. 35). Hospital security services and emergency depart-
ments strictly adhere to, and enforce, this mandate.
PAPR ISSUES AND LIMITATIONS
Attention to many details of PAPR use and maintenance is
essential for the assurance of effective protection (Table 6).
The Donning and Doffing Sequence Is Important
Exposed contaminated skin and PPE constitute indirect
sources of infection (fomite) for health care workers and
potentially their contacts. Using fluorescein dye as a surrogate
marker of contamination, a comparison of the N95 and PAPR
Table 6. Reasons for Failure of Powered Air
Purifying Respirator (PAPR) Protection
● Neglect of battery maintenance procedures
● Failure to check equipment and flow rate before use
● Decreased air flow during use (likely incomplete battery charge)
● Absence or incorrect placement of filter and gasket
● Incorrect don/doff sequence
● Neglect of hood care/inspection and unit cleaning/reprocessing
● Removal of PAPR during a procedure (failure to practice
procedures while wearing the equipment)
Caution: A biological PAPR does not provide chemical or vapor protection.
PAPR disadvantages
● Initial cost high (but may be cost effective compared with N95 stockpiling/
use)
● Larger requirement for storage space
● Self-contamination from used equipment possible
● Reprocessing of hoods required between work breaks
● Longer training time than N95 (required by Occupational Safety and Health
Administration, essential for safety: awareness of equipment; limitations;
assembly; preuse check; maintenance/reprocessing; don/doff sequence;
acclimatization and procedure practice)
● Battery recharging schedule is crucial
● Biannual discharge of unused units needed so battery maintains its capacity
to accept a charge
● Some models do not have real-time air flow indicator
● Care needed in seating of HEPA filter gasket after changing battery to avoid
misplacement
● Equipment may impede performance
● Hearing reduced because of fan noise; speech muffled because of hood
● PAPR use contraindicated during surgery (issue has not been addressed,
remains unresolved)
● The surgical hood system is not certified as a respiratory protection device
ANESTHESIA & ANALGESIA
Personal Protective Equipment for Pandemic Influenza
PPE ensembles showed that skin contamination was greater
with the N95 ensemble. Errors in don/doff steps were more
frequent with the PAPR ensemble, thereby increasing the
chance for self-contamination or exposure.39 However, in
following the CDC proscribed sequence for removal of the
N95 and contact garb, viral contamination of hands and
clothing occurred. In that study, a fluorescein dye surrogate
marker was not a reliable indicator of viral contamination.
Double gloving and the customary sequence for removal of
gowns and mask after surgery were recommended.40 The Min-
nesota Health Department has posted an illustrated don/doff
sequence for the PAPR and contact garb (Table 1, Ref. 36).
Initial Cost Is High and Storage Space Is Challenged
Budget constraints and scarce storage space limit supplies of
PAPRs and hoods in most hospitals. Hospitals may purchase
PPE using federal funds for disaster preparedness.
Battery Maintenance Is Crucial
The 3M Air-Mate has an interchangeable rechargeable
nickel-cadmium battery. A fully charged and maintained
battery will run for longer than 8 hours when properly
charged. Close attention to battery-charging procedures
will avoid failure of the PAPR during use (Table 1, Refs. 37
and 38).
The Bullard PA20 does not have an interchangeable
battery so it must be removed from service for recharging.
A rapid charger is available (Table 1, Refs. 32 and 33). The
battery runs for 9 to 10 hours.
A Flow Indicator Is Desirable
The 3M Air-Mate has no real-time flow or battery charge
indicator to signal when airflow or charge is reduced (Table
1, Refs. 31 and 33). The air flow must be checked with all
PAPRs with the test float each time the unit is turned on.
The 3M Breathe-Easy PAPR for chemical protection does
have an external air-flow gauge. The Bullard PA20 PAPRs
have a low battery charge and flow sound indicator.
Reprocessing Is Essential
The PAPR hoods are disposable and intended for single
use. Reuse of a hood is acceptable, but only by the same
individual. The hood must be cleaned every time it is
removed, because reuse of a soiled hood presents a contact
transmission risk. Reprocessing procedures must follow
manufacturers’ instructions, should be documented in
writing, and should be approved by the hospital infection
control practitioner. Reprocessing personnel and users
must understand the equipment and be thoroughly versed
in procedural details.
The HEPA Filter and Gasket Placement
Require Care
The 3M Air-Mate filter and gasket should be checked for
integrity and correct placement. This should be done in the
reprocessing room while wearing gloves and before the
unit is returned to service. The PAPR reprocessing steps
require removal of the filter to change the battery and to
confirm that the filter gasket is present and aligned cor-
rectly in the groove under the filter. The PAPR unit must
not be used without an intact and correctly placed filter
gasket. The used filter, as does a used N95, presents a risk
October 2010 • Volume 111 • Number 4
of contact but not respiratory transmission, because
disease-bearing particles do not reaerosolize from the filter
material (Table 1, Refs. 13, 16, and 39).
The filter seldom needs to be changed unless it becomes
wet or moist, which destroys its filtering capability, or
unless the flow decreases despite a fully charged battery.
Large industrial particles, such as wood or asbestos, may
clog a filter quickly, but accumulated biological particles
rarely cause the flow to decrease (Table 1, Ref. 31). A
decreased flow is likely to be the result of an incompletely
charged battery.
PAPR Training Is Essential for Safe Use
The requisite PAPR training is more involved and less avail-
able than N95 fit testing and training. As with the N95, the
PAPR training must be in cooperation with a hospital OSHA-
compliant respiratory protection program director, usually
the director of employee health. This individual may be
overextended with required N95 fit testing and may not have
the time or expertise to undertake PAPR training.
Although instructions may be sought on a “just in time”
basis, OSHA strongly encourages users to achieve compe-
tency with a respirator in advance. The PAPR competency
can be included in the Department of Anesthesiology initial
and periodic required demonstration of competencies with
various other equipment.
PPE Impedes Performance
PPE impedes performance of manual clinical procedures,
impairs hearing and communication, and may trigger
claustrophobia in some users. Although this is a recognized
issue with military and industrial protective gear, the same
could apply with the PAPR used in health care (Table 1,
Ref. 1).41 Practicing airway procedures on a mannequin
while dressed in the PAPR is strongly advised so that
patient care is not compromised in the clinical setting.
Training is effective in improving compliance with PPE use
and in reducing distressing symptoms while wearing pro-
tective equipment (Table 1, Ref. 1).35,41
PAPR Use Is Contraindicated During Surgery
The 3M Air-Mate user instructions state that the PAPR
should not be used during surgery (Table 1, Ref. 31). The
presumed rationale is that positive (outward) air flow
could increase the risk of wound infection. The user in-
structions do not specify whether the prohibition applies to
both the double-shrouded full hood and the loose-fitting
face cover hood styles, or only to the latter style. The air exit
holes of the loose-fitting face cover are located under the
chin; thus, if the wearer were standing at the operating
table, air would exit directly onto the surgical field. In
contrast, the air exits from the double-shrouded hood
under the accompanying surgical gown and close to the
floor.
OSHA has acknowledged the stated prohibition of PAPR
use during surgery but has not addressed the potential impact
on the safety of operating room personnel, most notably
anesthesia providers (Table 1, Ref. 4); neither has the CDC
expressed an opinion on PAPR use in surgery. Therefore, in
consort with their hospitals’ infection control practitioner and
www.anesthesia-analgesia.org 941
 SPECIAL ARTICLE
respiratory protection program and employee health direc-
tors, anesthesiology departments should address PAPR use in
the operating suite and develop procedures that maximize
safety for both patients and health care workers. At one
author’s hospital (Rush), a Department of Anesthesiology
policy for using the PAPR with the double-shrouded hood in
the surgical suite was developed, approved and documented
in cooperation with the hospital’s infectious disease and
infection control sections.
Outcome Data
A comparison of infection rates in health care workers who
used either a PAPR or an N95 for aerosol-generating
procedures would be pertinent, as would data on PAPR use
issues and problems. However, no such data are available
at this time.
The Surgical Hood System Is Not an Acceptable
Alternative to the PAPR
During SARS outbreaks, some health care workers used the
surgical hood system, and added goggles and an N95 upon
recommendation from Stryker (Table 1, Refs. 40 and 41).39
Although the surgical hood system used in orthopedics and
spine surgery would seem to be a satisfactory substitute for
the PAPR, it is neither classified nor certified as a respirator,
nor has NIOSH evaluated it for that purpose. The Stryker
surgical hood system draws ambient air through the material
of the hood and gown, not through a HEPA filter. An
evaluation of 2 surgical hood systems found both to have a
lower filtration efficiency than either the N95 or the PAPR.42
There is further concern that the incoming surgical hood
system airflow blows directly over the wearer’s eyes.
CRITICAL POINTS IN PAPR MAINTENANCE AND USE
The respiratory and contact protection provided by the
PAPR will fail if critical factors are neglected (Table 6).
EQUAL PROTECTION FOR HEALTH CARE
WORKERS AND PATIENTS
The health and safety of health care workers and patients
alike are supported by a “hierarchy of controls” that reduce
exposure, use engineering solutions, enact administrative
policies, and lastly, facilitate the consistent and effective use
of personal protective equipment and transmission-based
precautions (Table 1, Refs. 1, 20, 42, and 43).4 A culture of
safety is created when all parties participate with mutual
respect and support in a program that uses effective
measures known to minimize nosocomial and workplace-
acquired infections (Table 1, Ref. 1).
PROTOCOL FOR TRACHEAL INTUBATION
OF PATIENTS WITH TRANSMISSIBLE
RESPIRATORY DISEASE
A practical guide provides a template for anesthesia depart-
ment preparedness (Appendix 2). A written protocol for
endotracheal intubation of patients with a virulent transmis-
sible respiratory disease should be in place.1,43,44 The protocol
should emphasize the use of appropriate PPE within the
context of an OSHA-compliant respiratory protection pro-
gram. Rehearsing the procedure, avoiding emergent intuba-
tion through early intervention, and minimizing a mask leak
942 www.anesthesia-analgesia.org
or avoiding manual ventilation will reduce exposure risk. A
technique that prevents coughing and avoids nebulized medi-
cation delivery will minimize aerosolization.
Noninvasive ventilation, an aerosol-generating proce-
dure, is generally avoided in patients with a virulent
transmissible respiratory disease; however, noninvasive
ventilation, in conjunction with infection control air han-
dling systems, was used successfully during the SARS
epidemic without an increase in health care worker infec-
tions.17 Early noninvasive ventilation may reduce the risk
of health care worker exposure by avoiding tracheal intu-
bation and ventilatory support.17
OPERATING ROOM POLICY AND PROCEDURES
FOR INFLUENZA
Previously published recommendations for the manage-
ment of SARS and tuberculosis patients in the surgical suite
may be adapted to avian/pandemic/H1N1 influenza
plans. These include recommendations on scheduling of
operations, air handling controls, isolation procedures,
protection of the anesthesia machine, disinfecting and
cleaning procedures, and communication with the director
of plant engineering (Table 1, Refs. 14 and 34).17,45 During
the course of a pandemic, operating room schedules could
be profoundly disrupted by staff and equipment shortages
and policies that defer elective operations. Anesthesia
machines might be enlisted as surge capacity ventilators
and anesthesia professionals as critical care providers.
Consideration of these possibilities is encouraged.
A TRAINING WORKSHOP
A PAPR workshop in PowerPoint format is available for
training purposes (Table 1, Refs. 34 and 43). This article and
PowerPoint presentation support the workshop’s practical
component that incorporates PPE don/doff sequence and
procedure practice. The workshop was developed to
supplement, not replace, the manufacturers’ training mate-
rials, and must be a component of an OSHA-compliant
respiratory protection program.
SUMMARY
Pandemic influenza is highly contagious and potentially
many times more lethal than seasonal influenza. Those
who participate in aerosol-generating procedures, in-
cluding endotracheal intubation, are in the OSHA “very
high occupational exposure risk” category. OSHA stan-
dards require that employers provide safe working con-
ditions in a hazardous environment. We must work in
concert with our hospitals and our fellow health care
professionals to create a culture of safety.
The N95 respirator is limited by face seal leakage.
Wearing an N95 in daily work improved the success of
maintaining a face seal. Using an N95 without having been
fit tested voids the assigned protective factor (APF 10).
When used within the context of an OSHA-compliant
respiratory protection program, the PAPR offers a higher level
of respiratory and contact protection than the N95. The PAPR
has an APF of ⬎1000 when used with the double-shrouded
head- and shoulders-covering hood, and an APF of 25 when
used with the loose-fitting face covering hood. That is, the
ANESTHESIA & ANALGESIA
Personal Protective Equipment for Pandemic Influenza

wearer may expect a 1000-fold (or more) or a 25-fold reduc- protection. The surgical hood system is not a respira-
tion in contaminant within the respirator, depending on the tor and is not a satisfactory alternative.
hood choice. Anesthesia providers must be able to perform
procedures comfortably while wearing a PAPR. The double-
shrouded hood is most appropriate for aerosol-generating Recommendations
procedures and the loose-fitting face covering hood for usual Prepare for Safety
bedside care. • Get vaccinated and practice sound respiratory,
The PAPR carries substantial use and maintenance hand, personal, and workplace hygiene.
issues that must be addressed by health care workers and • Communicate with:
institutions alike. The package user instructions and OSHA • Hospital director of OSHA-compliant respiratory
state that the PAPR is not to be used during surgery protection program regarding respirator choice
because it could increase the risk of wound infection. This and training
dilemma is as yet unresolved and should be addressed • Infection control practitioner regarding proce-
with the hospital infection control practitioner. dures for personal protective equipment (PPE)
don/doff sequence, and discuss/receive clear-
ance for use of PAPR with full hood in the
operating room
APPENDIX 1. Powered Air Purifying Respirator • Plant engineer regarding air flow/filtration controls
(PAPR) Brand Name and Type of Protective Filter • Identify an interested anesthesia department coor-
Name Protection Filter type dinator to lead pandemic (and other) emergency
“Medical” PAPRsa preparedness/response.
1. 3M™ Air-Mate™ Biologic Particulate (HEPA)
• Select a respirator for aerosol-generating procedures.
2. Bullard PA20TM Biologic Particulate
3. Bullard EVA™ Biologic Particulate
• If N95, be fit tested; remember your brand and
“Chemical” PAPRs size; wear an N95 daily to improve (but not
1. 3M Breathe-Easy™ Chemical (gas/vapor) Absorbent assure) your ability to maintain a tight face seal;
Bbiologic and Combination address reuse protocol with hospital infection
chemical control practitioner; use a face shield to protect
2. Bullard PA30™ Chemical Absorbent the N95 if reuse is intended.
Biologic and Combination • If PAPR, become familiar with chosen hospital
chemical brand, or recommend a brand; know manufac-
3. Bullard EVA Biologic and Combination
turer’s use and maintenance procedures; practice
chemical
4. MSA OptimAir姞b TL Chemical Absorbent
don/doff sequence of PAPR and associated con-
Biologic and Combination tact PPE; practice procedures while dressed in
chemical PPE; secure same protection for assisting staff.
4. North Chemical Absorbent
CompactAir®b Biologic and Combination
Prepare for Invasive Airway Procedure
chemical
• Anticipate/avoid the need for emergent intuba-
HEPA ⫽ high-efficiency particulate air. tion: use noninvasive ventilatory support; intu-
See Table 1, Refs. 30, 32, and 33.
a bate preemptively.
Industrial use also.
b
Industrial PAPR. • Use air infection isolation room where available or
high-efficiency particulate air (HEPA) filtered room
exhaust.
APPENDIX 2 • Experienced individual should intubate. Excuse:
pregnant; nonessential personnel.
• Don PPE (PAPR plus contact, or N95 plus contact)
PRACTICAL GUIDE: AIRWAY MANAGEMENT per protocol, training, and hospital respiratory pro-
OF PATIENTS WITH CONTAGIOUS tection plan before entering the room.
RESPIRATORY DISEASE • Confirm that equipment and medications are imme-
Background diately available.
• Anesthesiology and Critical Care providers are in • Use closed suction when possible.
the Occupational Safety and Health Administration • Use disposable or dedicated monitoring equipment.
Remove unnecessary equipment.
(OSHA) “very high exposure risk” category.
• Use a HEPA filter between mask and resuscitation bag.
• Some influenza patients progress to fatal destructive • Plan a procedure that will avoid all of the following:
viral pneumonia and extreme inflammatory response, patient coughing, nebulized/topical/transtracheal
“cytokine storm.” lidocaine, forceful bag-mask ventilation.
• Airborne (droplet and aerosol) and contact transmis- • Rehearse the procedure just before intubation.
sion precautions are indicated.
• The limiting features of the N95 respirator are an Conduct Urgent/Emergent Intubation
increased work of breathing and leakage around the
face seal. • Patients in respiratory failure/arrest: Above, and
• The powered air purifying respirator (PAPR) has a intubate using the practitioner’s technique of great-
higher protection factor than the N95; however, inat- est success. A laryngeal mask airway may be used
tention to details of maintenance and use will void as a bridge or a conduit.

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 943

 SPECIAL ARTICLE
• Patients in impending respiratory failure: Above,
and perform focused history and physical examina-
tion; assess the airway; consider glycopyrrolate,
preoxygenate.
• For normal airway: Administer a muscle relax-
ant, ventilate gently if at all.
• For difficult airway: Consider deep sedation with
titrated midazolam, fentanyl, or ketamine. Use
lidocaine 1.5 mg/kg IV 1 minute before intuba-
tion. Administer muscle relaxant after intubation
is confirmed.
• For all cases: Remove PAPR per don/doff proto-
col outside the room, with a gloved assistant to
place reusable equipment, including laryngo-
scope, in a marked biohazard container for repro-
cessing. Reprocess PAPR equipment according to
manufacturer’s recommendation and hospital
protocol.
Adapt Above for Perioperative Management
• Defer elective surgery.
• Follow hospital infection control guidelines for pa-
tient transport.
• Use HEPA filters on both limbs of anesthesia circuit;
consider disposable circuit.
• Remove gloves and wash hands after each patient
contact.
• Limit contamination of the anesthesia equipment.
• Recover the patient in isolation.
• Change PPE before transporting the patient; used
PPE and exposed skin are fomites.
References:
(Table 1, Refs. 4, 13, 20, 22, 26, 29, 30 –34, and 36 –38)1,43,44,45
ACKNOWLEDGMENTS
The authors gratefully thank Chelsea Wanta and Traci
Nathans-Kelly for editing assistance.
REFERENCES
1. Kamming D, Gardam M, Chung FI. Anesthesia and SARS. Br J
Anaesth 2003;90:715– 8
2. Barry JM. The Great Influenza: The Epic Story of the Deadliest
Plague in History. New York: Penguin, 2004
3. Taubenberger JK, Morens DM. 1918 influenza: the mother of
all pandemics. Emerg Infect Dis 2006;12:15–22
4. Lau AC, Yip IK, Li MC, Wan M, Sit AW, Lee RA, Yung RW,
Yam LY. Response to SARS as a prototype for bioterrorism:
lessons in a regional hospital in Hong Kong. In: McIsaac JH, ed.
Hospital Preparation for Bioterror: A Medical and Biomedical
Systems Approach. Boston: Elsevier Academic Press, 2006:
281–94
5. Caputo KM, Byrick R, Chapman MG, Orser BJ, Orser BA.
Intubation of SARS patients: infection and perspectives of
healthcare workers. Can J Anaesth 2006;53:122–9
6. Nicolle L. SARS safety and science. Can J Anaesth
2003;50:983– 8
7. Shaw K. The 2003 SARS outbreak and its impact on infection
control practices. Public Health 2006;120:8 –14
8. Taubenberger JK, Morens DM. The pathology of influenza
virus infections. Annu Rev Pathol 2008;3:499 –522
9. Gill JR, Sheng ZM, Ely SF, Guinee DG, Beasley MB, Suh J,
Deshpande C, Mollura DJ, Morens DM, Bray M, Travis WD,
Taubenberger JK. Pulmonary pathologic findings of fatal 2009
pandemic influenza A/H1N1 viral infections. Arch Pathol Lab
Med 2010;134:235– 43
10. Yeh E, Luo RF, Dyner L, Hong DK, Banaei N, Baron EJ, Pinsky
BA. Preferential lower respiratory tract infection in swine-
origin 2009 A(H1N1) influenza. Clin Infect Dis 2010;50:391– 4
944 www.anesthesia-analgesia.org
11. Watanabe T, Watanabe S, Shinya K, Kim JH, Hatta M,
Kawaoka Y. Viral RNA polymerase complex promotes optimal
growth of 1918 virus in the lower respiratory tract of ferrets.
Proc Natl Acad Sci USA 2009;106:588 –92
12. Gupta RK, George R, Nguyen-Van-Tam JS. Bacterial pneumo-
nia and pandemic influenza planning. Emerg Infect Dis
2008;14:1187–92
13. Stevens J, Blixt O, Tumpey TM, Taubenberger JK, Paulson JC,
Wilson IA. Structure and receptor specificity of the hemagglu-
tinin from an H5N1 influenza virus. Science 2006;312:404 –10
14. Allen JE, Gardner SN, Vitalis EA, Slezak TR. Conserved amino
acid markers from past influenza pandemic strains. BMC
Microbiol 2009;9:77– 87
15. de Jong MD, Simmons CP, Thanh TT, Hien VM, Smith GJ,
Chau TN, Hoang DM, Chau NV, Khanh TH, Dong VC, Qui PT,
Cam BV, Ha do Q, Guan Y, Peiris JS, Chinh NT, Hien TT,
Farrar J. Fatal outcome of human influenza A (H5N1) is
associated with high viral load and hypercytokinemia. Nat
Med 2006;12:1203–7
16. Baskin CR, Bielefeldt-Ohmann H, Tumpey TM, Sabourin PJ,
Long JP, Garcia-Sastre A, Tolnay AE, Albrecht R, Pyles JA,
Olson PH, Aicher LD, Rosenzweig ER, Murali-Krishna K,
Clark EA, Kotur MS, Fornek JL, Proll S, Palermo RE, Sabourin
CL, Katze MG. Early and sustained innate immune response
defines pathology and death in nonhuman primates infected
by highly pathogenic influenza virus. Proc Natl Acad Sci USA
2009;106:3455– 60
17. Lau AC, Yam LY, So LK. Management of critically ill patients
with severe acute respiratory syndrome (SARS). Int J Med Sci
2004;1:1–10
18. Ng PC, Lam CW, Li AM, Wong CK, Cheng FW, Leung TF, Hon
EK, Chan IH, Li CK, Fung KS, Fok TF. Inflammatory cytokine
profile in children with severe acute respiratory syndrome.
Pediatrics 2004;113:e7–14
19. Pang BS, Wang Z, Zhang LM, Tong ZH, Xu LL, Huang XX,
Gao WJ, Zhu M, Wang C. Dynamic changes in blood cytokine
levels as clinical indicators in severe acute respiratory syn-
drome. Chin Med J (Engl) 2003;116:1283–7
20. Nicas M, Jones RM. Relative contributions of four exposure
pathways to influenza infection risk. Risk Anal 2009;
29:1292–303
21. Tellier R. Review of aerosol transmission of influenza A virus.
Emerg Infect Dis 2006;12:1657– 62
22. Tellier R. Aerosol transmission of influenza A virus: a review
of new studies. J R Soc Interface 2009;6(Suppl 6):S783–90
23. Roy CJ, Milton DK. Airborne transmission of communicable
infection: the elusive pathway. N Engl J Med 2004;350:1710 –2
24. Tong TR, Liang C, Nicastri E, Petrosillo N, Puro V, Fowler RA,
Scales DC, Ilan R, Yu ITS, Li Y. Evidence of airborne transmis-
sion of SARS [correspondence]. N Engl J Med 2004;351:609 –11
25. Yu IT, Li Y, Wong TW, Tam W, Chan AT, Lee JH, Leung DY,
Ho T. Evidence of airborne transmission of the severe acute
respiratory syndrome virus. N Engl J Med 2004;350:1731–9
26. Gamage B, Moore D, Copes R, Yassi A, Bryce E. Protecting
health care workers from SARS and other respiratory patho-
gens: a review of the infection control literature. Am J Infect
Control 2005;33:114 –21
27. Scales DC, Green K, Chan AK, Poutanen SM, Foster D, Nowak
K, Raboud JM, Saskin R, Lapinsky SE, Stewart TE. Illness in
intensive care staff after brief exposure to severe acute respi-
ratory syndrome. Emerg Infect Dis 2003;9:1205–10
28. Varia M, Wilson S, Sarwal S, McGeer A, Gournis E, Galanis E,
Henry B. Investigation of a nosocomial outbreak of severe
acute respiratory syndrome (SARS) in Toronto, Canada. CMAJ
2003;169:285–92
29. Alford RH, Kasel JA, Gerone PJ, Knight V. Human influenza
resulting from aerosol inhalation. Proc Soc Exp Biol Med
1966;122:800 – 4
30. Cohen HJ, Hecker LH, Mattheis DK, Johnson JS, Biermann AH,
Foote KL. Simulated workplace protection factor study of
powered air-purifying and supplied air respirators. AIHAJ
2001;62:595– 604
ANESTHESIA & ANALGESIA
Personal Protective Equipment for Pandemic Influenza
31. Loeb M, Dafoe N, Mahony J, John M, Sarabia A, Glavin V,
Webby R, Smieja M, Earn DJ, Chong S, Webb A, Walter SD.
Surgical mask vs N95 respirator for preventing influenza
among health care workers: a randomized trial. JAMA
2009;302:1865–71
32. Radonovich LJ Jr, Perl TM, Davey V, Cohen H. Preventing the
soldiers of health care from becoming victims on the pandemic
battlefield: respirators or surgical masks as the armor of choice.
Disaster Med Public Health Prep 2009;3(Suppl 2):S203–10
33. Srinivasan A, Perl TM. Respiratory protection against influ-
enza. JAMA 2009;302:1903– 4
34. Fennelly KP. The role of masks in preventing nosocomial
transmission of tuberculosis. Int J Tuberc Lung Dis
1998;2:S103–9
35. Yassi A, Lockhart K, Copes R, Kerr M, Corbiere M, Bryce E,
Danyluk Q, Keen D, Yu S, Kidd C, Fitzgerald M, Thiessen R,
Gamage B, Patrick D, Bigelow P, Saunders S. Determinants of
healthcare workers’ compliance with infection control proce-
dures. Healthc Q 2007;10:44 –52
36. Li Y, Tokura H, Guo YP, Wong AS, Wong T, Chung J, Newton
E. Effects of wearing N95 and surgical facemasks on heart rate,
thermal stress and subjective sensations. Int Arch Occup
Environ Health 2005;78:501–9
37. Lee MC, Takaya S, Long R, Joffe AM. Respirator-fit testing:
does it ensure the protection of healthcare workers against
respirable particles carrying pathogens? Infect Control Hosp
Epidemiol 2008;29:1149 –56
October 2010 • Volume 111 • Number 4
38. Phin NF, Rylands AJ, Allan J, Edwards C, Enstone JE, Nguyen-
Van-Tam JS. Personal protective equipment in an influenza
pandemic: a UK simulation exercise. J Hosp Infect 2009;
71:15–21
39. Zamora JE, Murdoch J, Simchison B, Day AG. Contamination:
a comparison of two personal protective systems. CMAJ
2006;175:249 –54
40. Casanova L, Alfano-Sobsey E, Rutala WA, Weber DJ, Sobsey
M. Virus transfer from personal protective equipment to
healthcare employees’ skin and clothing. Emerg Infect Dis
2008;14:1291–3
41. Barach P, Rivkind A, Israeli A, Berdugo M, Richter ED.
Emergency preparedness and response in Israel during the
Gulf War. Ann Emerg Med 1998;32:224 –33
42. Derrick JL, Gomersall CD. Surgical helmets and SARS infec-
tion. Emerg Infect Dis 2004;10:277–9
43. Cooper A, Joglekar A, Adhikari N. A practical approach to
airway management in patients with SARS. CMAJ 2003;
169:785–7
44. Peng PW, Wong DT, Bevan D, Gardam M. Infection control
and anesthesia: lessons learned from the Toronto SARS out-
break. Can J Anaesth 2003;50:989 –97
45. Stackhouse RA. Severe acute respiratory syndrome and tuber-
culosis. Anesthesiol Clin North America 2004;22:437–55
www.anesthesia-analgesia.org 945
CME

Early Postoperative Subcutaneous Tissue Oxygen

Predicts Surgical Site Infection
Raghavendra Govinda, MD,*† Yusuke Kasuya, MD,†‡ Endrit Bala, MD,§ Ramatia Mahboobi, MD,§
Jagan Devarajan, MD,§ Daniel I. Sessler, MD,§ and Ozan Akça, MD†储

BACKGROUND: Subcutaneous oxygen partial pressure is one of several determinants of surgical

site infections (SSIs). However, tissue partial pressure is difficult to measure and requires
invasive techniques. We tested the hypothesis that early postoperative tissue oxygen saturation
(StO2) measured with near-infrared spectroscopy predicts SSI.
METHODS: We evaluated StO2 in 116 patients undergoing elective colon resection. Saturation
was measured near the surgical incision, at the upper arm, and at the thenar muscle with an
InSpectra™ tissue spectrometer model 650 (Hutchinson Technology Inc., Hutchinson, MN) 75
minutes after the end of surgery and on the first postoperative day. An investigator blinded to
StO2 assessed patients daily for wound infection. Receiver operating characteristic curves were
used to analyze the performance of StO2 measurements as a predictor of SSI.
RESULTS: In 23 patients (⬇20%), SSI was diagnosed 9 ⫾ 5 days (mean ⫾ SD) after surgery.
Patients who did and did not develop an SSI had similar age (48 ⫾ 14 vs 48 ⫾ 15 years,
respectively; P ⫽ 0.97) and gender (female:male, 15:8 vs 46:47, respectively), but patients who
developed SSI weighed more (body mass index 32 ⫾ 7 vs 27 ⫾ 6 kg/m2; P ⬍ 0.01). StO2 at the
upper arm was lower in patients who developed SSI than in those who did not develop SSI (52 ⫾
22 vs 66 ⫾ 21; P ⫽ 0.033), and these measurements had a sensitivity of 71% and specificity
of 60% for predicting SSI, using StO2 of 66% as the cutoff point.
CONCLUSION: StO2 measured at the upper arm only 75 minutes after colorectal surgery
predicted development of postoperative SSI, although the infections were typically diagnosed
more than a week later. Although further testing is required, StO2 measurements may be able to
predict SSI and thus allow earlier preventive measures to be implemented. (Anesth Analg 2010;
111:946 –52)

S urgical site infections (SSIs) are perhaps the most

common serious complication of anesthesia and sur-
gery. They cause considerable morbidity and are
expensive to treat.1,2 The transition from contamination to
established infection occurs during a decisive period that
probably lasts only a few hours, even though infections are
typically detected a week or longer after surgery.3,4 If
antibiotics are administered during this decisive period,
they are more effective in reducing infection risk than when
given later.5 Tissue oxygen tension levels are one of the best
factors established for predicting SSI.6,7 Oxidative burst
function of neutrophils is one of the primary defenses
against SSI.8 Oxidative burst depends on the Po2 over the
entire physiological range of tissue values.9 Interventions to
improve tissue oxygenation during or immediately after
surgery are thus most likely to reduce the morbidity and
mortality associated with SSI.10,11
Tissue oxygenation has traditionally been measured
with Clark-type electrodes or similar systems. However,
these methods are invasive, expensive, and require exper-
tise to use.12,13 Near-infrared spectroscopy (NIRS) is an
alternative noninvasive technique.14 –17 We tested the hy-
pothesis that tissue oxygen saturation (Sto2) measured soon
after surgery with NIRS predicts ultimate development of
SSI. Confirming this hypothesis would allow early clinical
interventions, which might reduce the risk of infection.

From the *Department of Anesthesiology, Tufts Medical Center, Boston,

Massachusetts; †Department of Anesthesiology and Perioperative Medicine,
and 储Neuroscience ICU, University of Louisville, Louisville, Kentucky;
‡Tokyo Women’s Medical University, Tokyo, Japan; and §Department of
Outcomes Research, Cleveland Clinic, Cleveland, Ohio.
Accepted for publication May 6, 2010.
Supported in part by Hutchinson Technology Inc. (Hutchinson, MN) and the
Joseph Drown Foundation (Los Angeles, CA).
Data were presented in part at the American Society of Anesthesiologists
Annual Meeting in Orlando, FL (October 2008).
All authors are also affiliated with the Outcomes Research Consortium.
Disclosure: The authors report no conflicts of interest.
Address correspondence and reprint requests to Ozan Akça, MD, Depart-
ment of Anesthesiology and Perioperative Medicine, University of Louis-
ville Hospital, 530 S. Jackson St., Louisville, KY 40202. Address e-mail to
ozan.akca@louisville.edu.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181e80a94
METHODS
With approval by the Human Studies Committees at the
University of Louisville and the Cleveland Clinic, and
written informed consent by the participants, we included
116 colorectal surgery patients. Forty-three of these patients
had laparoscopic-assisted colorectal procedures; the re-
mainder had laparotomies (Table 1).
In a preliminary study, we found that patients who
developed SSI had lower Sto2 measurements at the thenar
eminence ⬃15% ⫾20% (mean ⫾ SD) than those who
remained uninfected postoperatively; the analogous abso-
lute difference was ⬇20% ⫾30% at the upper arm.18 Using
these estimates, a sample size of 80 measurements at the
thenar eminence and 132 at the upper arm was calculated
to achieve 90% power with an ␣ of 0.05. We therefore

946 www.anesthesia-analgesia.org October 2010 • Volume 111 • Number 4

 Table 1. Demographic and Morphometric Characteristics, and Potential Confounding Factors
Patients with Patients without
SSI (n ⴝ 23) SSI (n ⴝ 93) P value
Demographic and morphometric characteristics
Age (y) 48 ⫾ 14 48 ⫾ 15 0.971
Sex (female/male) 15/8 46/47 0.175
Weight (kg) 92 ⫾ 21 78 ⫾ 20 0.006
BMI (kg/m2) 32 ⫾ 6.8 27 ⫾ 5.5 ⬍0.001
ASA physical status I/II/III, n 2/15/6 10/57/26 0.426
Smokers, n (%) 4 (17%) 18 (19%) 0.830
Cardiac disease, n (%) 7 (30%) 25 (27%) 0.732
Diabetes mellitus, n (%) 3 (13%) 4 (4%) 0.115
Previous chemotherapy, n (%) 2 (9%) 4 (4%) 0.394
Preoperative Hct (%) 40 ⫾ 4 40 ⫾ 5 0.612
SENIC I/II/III (%) 17/78/5 9/80/11 0.373
NNISS I/II/III (%) 43.5/43.5/13 39/48/13 0.824
Intraoperative confounders
Crystalloids (mL) 3273 ⫾ 1414 3274 ⫾ 1500 0.996
Crystalloids (mL)a 3150 (2550–3875) 3000 (2400–4000)
Colloids (mL) 620 ⫾ 532 439 ⫾ 452 0.101
Colloids (mL)a 500 (0–1000) 500 (0–813)
Blood loss (mL) 549 ⫾ 511 373 ⫾ 378 0.078
Blood loss (mL)a 350 (200–800) 250 (150–463)
PRBC transfusion (U) 0.1 ⫾ 0.4 0.2 ⫾ 0.5 0.541
Duration of surgery (h) 3.4 ⫾ 1.6 3.5 ⫾ 1.5 0.784
Intraoperative core temperature (°C) 36.1 ⫾ 0.5 36.1 ⫾ 0.6 0.911
Intraoperative mean FIO2 62 ⫾ 18 61 ⫾ 19 0.816
Mean arterial blood pressure (mm Hg) 85 ⫾ 8 86 ⫾ 9 0.470
Laparoscopic-assisted procedures, n (%) 6 (26%) 37 (40%) 0.335
Epidural analgesia, n (%) 0 (0%) 5 (5%) 0.581
SSI ⫽ surgical site infection; BMI ⫽ body mass index; SENIC ⫽ Study on the Effect of Nosocomial Infection Control; NNISS ⫽ National Nosocomial Infection
Surveillance System; Hct ⫽ hematocrit; PRBC ⫽ packed red blood cell; FIO2 ⫽ fraction of inspired oxygen.
Data are presented as mean ⫾ standard deviation, count (percentage), or a Median (25th–75th quartile).

enrolled 116 patients to complete the study with sufficient
power for both outcomes.
The study enrollment period was from July 2007 to May
2008. Adults between 18 and 80 years of age with an ASA
physical status I to III were included in the study. Patients
with intestinal obstruction, those in whom the surgeon did
not anticipate a primary wound closure, and those with a
diagnosed or suspected intraabdominal abscess were ex-
cluded from the study. Patients were also excluded if they
had severe chronic obstructive pulmonary disease, recent
myocardial infarction, unstable angina, or required oxygen
preoperatively.
Protocol
All patients received antibiotics before surgery according to
the protocols of the 2 participating institutions, each of
which required administration of appropriate prophylactic
antibiotic within an hour before surgical incision. Patients
were given midazolam for premedication, propofol or
etomidate for induction of anesthesia, and succinylcholine
or rocuronium for initiation of muscle relaxation. Muscle
relaxation was maintained with rocuronium or vecuronium.
Anesthesia was maintained with sevoflurane, desflu-
rane, or isoflurane in 30% to 80% oxygen, supplemented
with fentanyl or morphine. Intraoperative IV fluid manage-
ment was at the discretion of the attending anesthesiologist,
and consisted of 8 to 10 mL/kg. Core temperature was
maintained near 36°C by using forced-air warming blan-
kets and fluid warmers. Hair was clipped from the surgical
site immediately preoperatively, and the skin was prepared
with a chlorhexidine-based antiseptic kit. Postoperative
analgesia was provided with IV intermittent boluses or via
patient-controlled analgesia with morphine or hydromor-
phone. If an epidural catheter was placed preoperatively,
epidural analgesia with fentanyl (⬃25 to 50 ␮m/h) was
used intraoperatively and approximately for the first 2
hours of the recovery if it was the preference of the
attending anesthesiologist. A combination of local anes-
thetic and opioid via the epidural was only administered
after the study was completed.
Spontaneously breathing patients were given oxygen
via nasal cannula or Venturi mask. The inspired oxygen
concentration was adjusted to maintain pulse oximeter
saturation (Spo2) ⱖ95%. Sto2 was measured 15 minutes
after weaning from oxygen and thereafter patients received
supplemental oxygen as required to maintain Spo2 ⱖ92%.
Measurements
The duration of surgery, blood loss, intraoperative crystal-
loid and colloid administration, hemodynamic variables,
blood transfusion requirements, and urine output were
recorded.
Two systems were used to evaluate the SSI risk: the
Study on the Efficacy of Nosocomial Infection Control
(SENIC) and National Nosocomial Infection Surveillance
System (NNISS). The SENIC scoring system assigns 1 point
for each of the following factors: ⱖ3 underlying diagnoses,
surgery that lasts ⱖ2 hours, an abdominal site of surgery,
and the presence of a contaminated or infected wound.19
The NNISS predicts risk on the basis of the contamination

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 947

Tissue Oxygen and Surgical Site Infection
Figure 1. InSpectra™ StO2 tissue oxygenation monitor.
of surgery, the rating of physical status on a scale devel-
oped by the American Society of Anesthesiologists, and the
duration of surgery.20
Sto2 and tissue hemoglobin index (THI) were measured
by an InSpectra™ tissue spectrometer model 650 (Hutchinson
Technology Inc., Hutchinson, MN) 75 minutes after surgery
and on the first postoperative day (Fig. 1). A time point of
75 minutes postoperatively was chosen because patients are
usually weaned from supplemental oxygen at the end of the
first postoperative hour, and waiting an additional 15 minutes
generally allows full washout of supplemental oxygen.
The InSpectra tissue spectrometer measures Sto2 using
wide-gap second-derivative NIRS.21,22 The InSpectra spec-
trometer makes use of the characteristic absorption prop-
erties of hemoglobin in the near-infrared wavelength range
between 680 and 800 nm. The absorption spectrum of light
remitted from the tissue sample varies mainly with oxyhe-
moglobin and deoxyhemoglobin concentration; other
chromophores have minimal effect. Sto2 is a measure of
hemoglobin oxygen saturation of the blood contained in the
volume of tissue illuminated by the near-infrared light. The
maximum depth of the tissue sampled is the distance
between the probe’s send and receive fibers. Mean mea-
surement depth is half the probe’s spacing. We used a
15-mm probe that measures Sto2 of 5- to 8-mm tissue depth.
For the forearm and wound sites, this corresponds to
subcutaneous tissue above the skeletal muscle. For the
thenar muscle site, this depth corresponds to muscle.
Sto2 near the wound was measured 2.5 cm lateral to the
incision at the upper, middle, and lower third of the incision.
The upper lateral arm site was chosen for measurement
because this site reflects the Sto2 of operative wounds in the
chest and abdomen even though it is approximately 10 mm
Hg higher than in the wound area.6,12 The thenar muscle site
was chosen because it is one of the best established sites for
tissue oximetry measurement.23,24 At each of these sites, the
probe was placed for 15 to 30 seconds until a stable oxygen
saturation value was obtained.
THI was measured simultaneously throughout the study
from the same probe as the Sto2 was monitored. THI is not yet
a well-established value. As with several others, this NIRS-
based device can provide a hemoglobin value obtained from
948 www.anesthesia-analgesia.org
the probe’s detection site, but this hemoglobin value does not
necessarily represent the body’s total hemoglobin concentra-
tion. THI monitors the total amount of hemoglobin in the
tissue site where the NIRS probe is applied. Therefore, it may
be considered more of a perfusion variable.
During each set of measurements, mean arterial blood
pressure, heart rate, and Spo2 were recorded simultaneously.
We asked the patients to rate their pain using a 10-cm visual
analog scale.
An independent investigator not aware of the Sto2
measurements evaluated patients’ wounds daily through-
out their hospitalization. SSI was diagnosed according to
the surgical wound infection definitions from the Centers
for Disease Control and Prevention (CDC).19,25
Data Analysis
Primary outcomes were the Sto2 measured around the
surgical incision, at the upper arm, and at the thenar
muscle. Surgical site Sto2 measurements were summarized
as a mean value and presented as a post hoc analysis.
Specifically, subcutaneous Sto2 values at the upper one-
third, middle one-third, and lower one-third of the surgical
incision were averaged into a single “incision” value for
each patient.
Normally distributed data are presented as means ⫾ SD.
Skewed data are presented as medians and interquartile
ranges. After descriptive analysis of all parameters, univar-
iate analysis was performed using the ␹2 test for categorical
variables and unpaired, 2-tailed t and Kruskal-Wallis tests
for continuous variables. P values ⬍0.05 were considered
statistically significant. Additionally, a multivariate analy-
sis was performed to assess the independent contribution
of each potential variable (SPSS Inc., Chicago, IL).
Receiver operating characteristic (ROC) curves were
developed for Sto2 in the upper arm and Spo2 on the first
postoperative day to predict surgical wound infections. We
compared our predictions based only on early postopera-
tive upper arm Sto2, using an Sto2 of 66% as the cutoff
point, with SENIC predictions. The CDC in the SENIC
developed a predictive model for the risks of SSI that has
become the standard.19
RESULTS
Of the 116 patients enrolled, 23 developed SSI (20%). If the
diverticulitis cases were excluded, the SSI rate would
decrease to 14%. All of the 23 patients who developed SSI
had superficial incisional site infections. Three of these
patients also developed deep incisional site infections, and
4 developed peritoneal infections as defined by CDC crite-
ria. Infections were diagnosed an average of 9 ⫾ 5 days
after surgery.
Age, ASA physical status, and SENIC and NNISS risk
scores were similar in the patients who developed SSI and
those who remained uninfected (Table 1). Patients with SSI
had a greater body mass index. There were no statistically
significant or clinically significant differences in the dura-
tion of surgery, IV fluid administration, intraoperative
temperature, or blood transfusion requirements. Surgical
technique and use of epidural analgesia were also similar in
the patients who developed SSI and those who did not
(Table 1).
ANESTHESIA & ANALGESIA
 Table 2. Major Study Outcomes: Tissue Oxygen
Saturation Given as Percentage
Patients Patients
with SSI without SSI P
Measurements (n ⴝ 23) (n ⴝ 93) value
75 min after surgery (early
postoperative period)
StO2 at surgical incision (%) 44 ⫾ 16 51 ⫾ 20 0.184
THI at surgical incision 3.8 ⫾ 1.1 4.3 ⫾ 2.0 0.229
StO2 at upper arm (%) 52 ⫾ 22 66 ⫾ 21 0.033
THI at upper arm 3.8 ⫾ 1.3 5.3 ⫾ 3.2 0.041
StO2 at thenar eminence (%) 69 ⫾ 17 74 ⫾ 16 0.210
THI at thenar eminence 8.2 ⫾ 3.6 8.9 ⫾ 3.1 0.403
Figure 3. Surgical wound infection rates (%) stratified by early
SpO2 (%) 97 ⫾ 3 98 ⫾ 2 0.193
postoperative tissue oxygen saturation (StO2) (%) values measured
MAP (mm Hg) 94 ⫾ 15 93 ⫾ 14 0.831
at the upper arm. The bars represent the difference between the
Pain (cm VAS) 7.6 ⫾ 1.5 6.3 ⫾ 2.6 0.048
observed infection rates and those expected based on the Study on
Pain (cm VAS)a 8.0 (7.0–8.3) 6.0 (4.0–8.0) 0.043
the Effect of Nosocomial Infection Control (SENIC) multivariate risk
No. of patients on room air 13 (57%) 48 (52%) 0.403
index.
No. of patients receiving 8 (35%) 35 (38%) 0.746
nasal O2 at 2–3 L/min
No. of patients with FIO2 2 (8%) 10 (10%) 0.467 measurements were done under supplemental oxygen be-
⬎0.60
cause of the oxygenation needs of the patients as specified
First postoperative day
StO2 at surgical incision (%) 44 ⫾ 25 50 ⫾ 24 0.300 above.
THI at surgical incision 4.5 ⫾ 1.3 5.0 ⫾ 2.1 0.410 In the early postoperative period (Table 2), Sto2 at the
StO2 at upper arm (%) 61 ⫾ 13 60 ⫾ 22 0.825 surgical incision did not differ significantly between pa-
THI at upper arm 4.1 ⫾ 0.9 4.9 ⫾ 2.6 0.193 tients who eventually developed SSI and those who re-
StO2 at thenar eminence (%) 77 ⫾ 12 79 ⫾ 13 0.538
THI at thenar eminence 11.0 ⫾ 2.7 10.9 ⫾ 2.8 0.824
mained uninfected. The THI near the surgical incision was
SpO2 (%) 95 ⫾ 3 97 ⫾ 2 0.001 25% lower in patients who developed SSI, but this differ-
MAP (mm Hg) 89 ⫾ 11 84 ⫾ 13 0.068 ence did not reach statistical significance.
Pain (cm VAS) 5.8 ⫾ 2.2 4.9 ⫾ 2.5 0.152 In contrast, Sto2 at the upper lateral arm was signifi-
a
Pain (cm VAS) 5.0 (4.6–7.8) 5.0 (3.0–6.5)
cantly lower in the patients who developed SSI (52 ⫾ 22
Hct (%) 35 ⫾ 5 32 ⫾ 5 0.436
No. of patients on room air 16 (70%) 69 (74%) mm Hg) than in those who did not (66 ⫾ 21 mm Hg; P ⫽
No. of patients receiving 6 (26%) 23 (25%) 0.033). ROC curve for Sto2 at the upper lateral arm had a
nasal O2 at 2–3 L/min sensitivity of 71% and specificity of 60% using an Sto2 of
No. of patients with FIO2 1 (4%) 1 (1%) 66% as the cutoff point for predicting SSI (Fig. 2). The
⬎0.60
positive predictive value of this cutoff value was 29%, with
SSI ⫽ surgical site infection; MAP ⫽ mean arterial blood pressure; Hct ⫽ a negative predictive value of 90%. The THI measured in
hematocrit; VAS ⫽ 10-cm-long visual analog scale for pain; StO2 ⫽ tissue
oxygen saturation; THI ⫽ tissue hemoglobin index; SpO2 ⫽ pulse oximeter the early postoperative period at the upper arm was also
oxygen saturation; FIO2 ⫽ fraction of inspired oxygen. statistically lower in the patients who developed SSI (3.8 ⫾
Data are presented as mean ⫾ SD, count (percentage), or a Median 1.3 vs 5.5 ⫾ 3.2 g/dL). ROC curves for THI at the upper
(25th–75th quartile).
lateral arm in the early postoperative period had a sensi-
tivity of 88% and specificity of 55% for predicting SSI at a
A significant proportion of the patients (35%–38%) THI of 4.3. Figure 3 shows the difference between observed
could not be weaned from supplemental oxygen at the and expected (based on SENIC scores) infection risk as a
75th-minute measurement period (Table 2). Additionally, function of early postoperative Sto2 at the upper arm.
from this group of patients, there were 14 patients (3 with Because intraoperative oxygen concentrations were not
SSI and 11 with no SSI) who required oxygen rates of more controlled per protocol, we calculated whether there was
than 2 to 3 L/min. During the first postoperative day any correlation between inspired intraoperative oxygen
measurements, 70% to 75% of the patients were weaned concentrations and postoperative Sto2 at the upper lateral
from supplemental oxygen. Therefore, some of the study arm. There was no correlation (r2 ⫽ 0.007).

Figure 2. A, Upper arm tissue oxygen saturation

(StO2) measured 75 minutes after surgery in pa-
tients who did and did not develop surgical site
infections (SSIs). B, The associated receiver oper-
ating characteristic curve.

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 949

Tissue Oxygen and Surgical Site Infection
Table 3. Independent Contributors to Surgical
Site Infections (Multivariate Analysis)
Odds 95% Confidence
ratio interval
Body mass index (kg/m2) 1.081 0.984–1.188
Blood loss 1.001 0.998–1.002
Upper arm StO2 0.975 0.945–1.006
(postoperative 75 min)
Pain VAS (postoperative 1.292 0.944–1.767
75 min)
SpO2 (postoperative day 1) 0.765 0.603–0.971
StO2 ⫽ tissue oxygen saturation; VAS ⫽ visual analog scale; SpO2 ⫽ pulse
oximeter oxygen saturation.
Thenar muscle oxygenation and THI values were both
higher than those of the subcutaneous tissue values in both
groups of patients. Neither the Sto2 nor the THI measured
at the thenar eminence 75 minutes after surgery differed
significantly in patients who did or did not develop SSI
(Table 2). Early postoperative visual analog scale pain
scores were slightly, but statistically significantly, higher in
patients who developed SSI (Table 2).
Patients who developed SSI had lower Spo2 values on
the first postoperative day (95% ⫾ 3% vs 97% ⫾ 2%; P ⫽
0.001). The ROC curve for the first postoperative day Spo2
had a sensitivity of 75% and specificity of 73% using an
Spo2 of 95% as the cutoff point for predicting SSI.
Multivariate logistic regression analyses indicated that
the first postoperative day Spo2 value was the only statis-
tically significant independent factor contributing to SSI.
Although body mass index, postoperative pain, intraopera-
tive blood loss, and upper arm Sto2 data provided a good
clinical and statistical difference in univariate analysis, they
did not reach independent significance levels with multi-
variate statistics (Table 3).
DISCUSSION
The link between tissue oxygenation and surgical wound
infection is well established. This concept, developed by
Thomas Hunt in the 1960s and 1970s26,27 led to a landmark
article by Hopf et al.6 more than a decade ago. Hopf et al.
used a subcutaneous needle-guided tonometric silicon
catheter system into which they inserted a polarographic
Clark-type electrode to monitor tissue oxygen partial pres-
sure. They obtained measurements at 3 different times:
within 6 hours of surgery, on the first postoperative day,
and on the second postoperative day. They observed that
tissue oxygenation was a strong predictor of SSI.
A subcutaneous tissue oxygen monitoring system (LICOX;
Medical Systems Corp., Greenvale, NY) has been used by
various investigators, including us, for decades and is consid-
ered the “gold standard” for tissue oxygen monitoring. How-
ever, the LICOX system is invasive, expensive, and requires
approximately 45 minutes of calibration and equilibration
with tissue as well as considerable operator experience to be
accurate. Therefore, a simple and noninvasive tissue oxygen
monitoring system might be easier and more feasible for
perioperative use. The InSpectra Sto2 system is noninvasive
and relatively easy to use. Our main study question was
whether NIRS Sto2, measured immediately postoperatively,
would be able to predict SSI. Our principal finding was that
950 www.anesthesia-analgesia.org
patients who eventually developed SSI according to the CDC
criteria had lower Sto2 at the upper arm only 75 minutes after
surgery, which was approximately 9 days before the diagno-
sis of SSI was made clinically.
P value
Upper arm Sto2 provided a “fair” area under the ROC
0.106
0.157 curve and was a better predictor of infection than the
0.108 established SENIC or NNISS risk scores. This point is
important because the high-risk patients we identified
0.109 would generally not have been detected using the SENIC
risk score or other routinely available clinical systems.
0.044
As we have shown previously, subcutaneous tissue in
the upper lateral arm is relatively well oxygenated and
perfused under general anesthesia and in the awake
state.1,28,29 An average of 48 ⫾ 19 perforator arterioles from
15 vascular territories supply the integument of the upper
extremity. Septocutaneous arteries predominate in the
shoulder, elbow, distal forearm, and hand regions. Muscu-
locutaneous perforators are more numerous in the upper
arm and proximal forearm. The average perforator size in
the upper extremity is approximately 0.7 ⫾ 0.2 mm in
diameter and supplies an average area of 35 cm2.30,31
Therefore, we can extrapolate that our upper arm Sto2
measurement likely included a site perfused with at least 1
perforator artery.
Leukocyte-mediated oxidative burst and collagen for-
mation require oxygen partial pressures of at least 40 mm
Hg.32 Upper arm subcutaneous tissue oxygen tension is
typically ⬎50 mm Hg under a fraction of inspired oxygen
⬍0.4 even during sympathetic vasoconstriction. Our tissue
oxygen tension values were lower than those reported
previously (27 to 35 mm Hg). This difference might result
from differences in oxygen monitoring techniques. Sto2
measures hemoglobin’s oxygen saturation in a focal area.
Sto2 systems calculate the hemoglobin’s oxygen saturation
in the volume of tissue illuminated by near-infrared light.
Because in the majority of the peripheral tissues tested
(surgical wound area and upper arm subcutaneous tissue)
the THI values were low, we can extrapolate that there was
less perfusion in the site of interest. Therefore, Sto2 pro-
vides a different oxygenation value than tissue oxygen
tension, which monitors the partial pressure of free oxygen
in the tissue. Free oxygen was reported to be the main
source of oxygen available for the tissues.33 This interesting
concept of tissue oxygenation will continue to be an in-
tensely debated topic until applicability of monitoring,
reproducibility of oxygenation data, and until it is proven
with clinical outcomes.
We are not the first to use Sto2 to predict SSI. Ives et
al.16,34 reported that Sto2 of 53% at the surgical incision site,
measured 12 hours postoperatively, provided 71% sensitiv-
ity and 76% specificity as a test to predict SSI. Our study
extends previous work by showing that infection can be
predicted shortly after surgery, during the decisive period.
But interestingly, Ives et al. could not find any difference in
the upper arm Sto2 measurements between the patients
who did and did not develop SSI. A possible explanation is
that Ives et al. measured tissue oxygenation 10 cm below
the shoulder tip over the bulk of the biceps muscle. In
contrast, we used the lateral upper arm, which was ap-
proximately 15 cm below the shoulder tip, the area between
the biceps and triceps brachii’s lateral head. This is the site
ANESTHESIA & ANALGESIA
used in most previous studies.6,28,35 The reason behind our
failure to find a statistically significant difference at the
surgical site can be explained (1) by the lack of power and
small sample size, and (2) because of the stretch of the
peri-incisional tissues during surgery, there might have
been tissue edema to blunt the extent of oxygenation
difference. If the difference between the groups continued
consistently, 170 to 180 patients would have provided a
statistically significant difference.
Another interesting finding of the current trial is that
Spo2 values of the first postoperative day apparently
predicted SSI. Before taken into consideration, some major
limitations of these data need to be addressed: (1) Spo2 was
not planned as one of the a priori outcomes of this study;
(2) the data were gathered from only a few values from a
snapshot period; and (3) although the difference was
statistically significant, clinical meaning may be of limited
value.
Early detection of patients at special risk of wound
infection is important because there are well-established
interventions that improve tissue oxygenation and may
thus reduce infection rate and possibly improve surgical
outcomes. For example, sympathetic nervous system acti-
vation triggers vasoconstriction and reduces tissue oxygen-
ation.36 A major mediator of sympathetic activity, and one
that can often be treated, is surgical pain. In fact, it is well
established that adequate analgesia improves tissue oxy-
genation,37,38 which is supported by the fact that patients in
our study who eventually developed SSI had higher pain
scores in the early postoperative period.
Thermoregulatory vasoconstriction is another factor
that decreases tissue oxygen tension and perfusion.39,40
As might thus be expected, maintaining perioperative
normothermia41 and local warming42 decreases SSI rate.
Supplemental fluid administration increases tissue oxy-
genation,43,44 but it does not necessarily improve the SSI
rate.45 However, supplemental fluids may have short-
term tissue oxygenation benefits that should be consid-
ered despite the ongoing perioperative fluid debates.46
Finally, supplemental oxygen (i.e., 80% inspired oxygen)
approximately doubles tissue oxygen partial pressure1
without causing atelectasis.37,47
There are thus at least 4 established interventions that
improve tissue oxygenation and could be implemented
relatively easily in patients found to have low Sto2 during
the decisive period. We caution, however, that although the
link between tissue oxygenation and SSI is clear, whether
titrating interventions perioperatively to improve Sto2 ac-
tually reduces infection risk is yet to be proven.
Despite the promising data provided in this trial sup-
porting the link between oxygenation and SSI, there are
some important limitations that need to be addressed. The
values of the upper arm Sto2 were relatively weak (low
ROC area), which were also apparent with the nonsignifi-
cant odds ratio obtained from the multivariate analysis. A
larger trial aiming to reconfirm the current results in a
larger and wider surgical population will be needed. More
importantly, no study in human subjects has validated the
accuracy of Sto2 by comparing it with the “gold standard”
tissue oxygen tension monitoring. Additionally, it is impor-
tant to recognize that the reflectance spectroscopy-based
October 2010 • Volume 111 • Number 4
oximeters calculate the mean value of oxyhemoglobin across
all the vessels of the microcirculation of the skin. Therefore,
the derived oximeter saturation is a mean blood oxygen
saturation across arterioles, capillaries, and venules.48
Another important limitation is in the claims made by
diagnosing potential infections in a decisive period. This
decisive period was established for antibiotic prophy-
laxis as the intervention. Therefore, the decisive period
responding to increased oxygenation needs to be tested
before considering the validity of the proposed interven-
tions. Using an animal model, Knighton et al.7 showed
that oxygen might have a decisive period of at least 6
hours.
In summary, Sto2 measured at the upper arm only 75
minutes postoperatively predicts development of surgical
wound infection after colorectal surgery, even though
infections were typically diagnosed more than a week later.
Because high-risk patients can potentially be identified
noninvasively, it may also be possible to intervene to
improve tissue oxygenation.
ACKNOWLEDGMENTS
Samual Chen, MD, Luke Reynolds, MSc, Adrian Alvarez, MD,
and Gena Harrison, BA (Department of Outcomes Research,
Cleveland Clinic) are acknowledged for their assistance with
data acquisition. We appreciate the efforts of Nancy Alsip,
PhD, and Gilbert Haugh, MS (OCRSS, University of Louis-
ville), in medical editing and statistical assistance; and Joseph
Ortner and Hutchinson Technology Inc. (Hutchinson, MN) for
technical support and for providing tissue oximeters and their
probes.
REFERENCES
1. Greif R, Akça O, Horn EP, Kurz A, Sessler DI, Outcomes
Research™ Group. Supplemental perioperative oxygen to re-
duce the incidence of surgical wound infection. N Engl J Med
2000;342:161–7
2. Olsen MA, Chu-Ongsakul S, Brandt KE, Dietz JR, Mayfield J,
Fraser VJ. Hospital-associated costs due to surgical site infec-
tion after breast surgery. Arch Surg 2008;143:53– 60
3. Miles AA, Miles EM, Burke J. The value and duration of
defence reactions of the skin to the primary lodgement of
bacteria. Br J Exp Pathol 1957;38:79 –96
4. Polk HC Jr. The prophylaxis of infection following operative
procedures. J Ky Med Assoc 1974;72:139 – 43
5. Classen DC, Evans RS, Pestotnik SL, Horn SD, Menlove RL,
Burke JP. The timing of prophylactic administration of antibi-
otics and the risk of surgical-wound infection. N Engl J Med
1992;326:281– 6
6. Hopf HW, Hunt TK, West JM, Blomquist P, Goodson WH III,
Jensen JA, Jonsson K, Paty PB, Rabkin JM, Upton RA, von
Smitten K, Whitney JD. Wound tissue oxygen tension predicts
the risk of wound infection in surgical patients. Arch Surg
1997;132:997–1004
7. Knighton DR, Halliday B, Hunt TK. Oxygen as an antibiotic:
the effect of inspired oxygen on infection. Arch Surg
1984;119:199 –204
8. Allen DB, Maguire JJ, Mahdavian M, Wicke C, Marcocci L,
Scheuenstuhl H, Chang M, Le AX, Hopf HW, Hunt TK.
Wound hypoxia and acidosis limit neutrophil bacterial killing
mechanisms. Arch Surg 1997;132:991– 6
9. Babior BM. Oxygen-dependent microbial killing by phagocytes
(first of two parts). N Engl J Med 1978;298:659 – 68
10. Leaper D, Burman-Roy S, Palanca A, Cullen K, Worster D,
Gautam-Aitken E, Whittle M. Prevention and treatment of
surgical site infection: summary of NICE guidance. BMJ
2008;337:a1924
www.anesthesia-analgesia.org 951
Tissue Oxygen and Surgical Site Infection
11. Lee JT. A new surgical site infection (SSI) prevention guideline.
Surg Infect (Larchmt) 2000;1:127–31
12. Gottrup F, Firmin R, Chang N, Goodson WH III, Hunt TK.
Continuous direct tissue oxygen tension measurement by a
new method using an implantable silastic tonometer and
oxygen polarography. Am J Surg 1983;146:399 – 403
13. Hopf HW, Hunt TK. Comparison of Clark electrode and
optode for measurement of tissue oxygen tension. Adv Exp
Med Biol 1994;345:841–7
14. Soller BR, Idwasi PO, Balaguer J, Levin S, Simsir SA, Vander
Salm TJ, Collette H, Heard SO. Noninvasive, near infrared
spectroscopic-measured muscle pH and PO2 indicate tissue
perfusion for cardiac surgical patients undergoing cardiopul-
monary bypass. Crit Care Med 2003;31:2324 –31
15. Soller BR, Ryan KL, Rickards CA, Cooke WH, Yang Y, Soyemi
OO, Crookes BA, Heard SO, Convertino VA. Oxygen satura-
tion determined from deep muscle, not thenar tissue, is an
early indicator of central hypovolemia in humans. Crit Care
Med 2008;36:176 – 82
16. Ives CL, Harrison DK, Stansby GS. Tissue oxygen saturation,
measured by near-infrared spectroscopy, and its relationship
to surgical-site infections. Br J Surg 2007;94:87–91
17. Ives CL, Harrison DK, Stansby G. Prediction of surgical site
infections using spectrophotometry: preliminary results. Adv
Exp Med Biol 2006;578:149 –54
18. Govinda R, Kasuya Y, Mahboobi R, Devarajan J, Akca O.
Oxygen saturation at thenar eminence predicts surgical wound
infections. American Society of Anesthesiologists Annual
Meeting, Orlando, FL, 2008
19. Haley RW, Culver DH, Morgan WM, White JW, Emori TG,
Hooton TM. Identifying patients at high risk of surgical wound
infection: a simple multivariate index of patient susceptibility
and wound contamination. Am J Epidemiol 1985;121:206 –15
20. Culver DH, Horan TC, Gaynes RP, Martone WJ, Jarvis WR,
Emori TG, Banerjee SN, Edwards JR, Tolson JS, Henderson TS,
Hughes JM; National Nosocomial Infections Surveillance Sys-
tem. Surgical wound infection rates by wound class, operative
procedure, and patient risk index. Am J Med 1991;91:152S–7S
21. Myers DE, Anderson LD, Seifert RP, Ortner JP, Cooper CE,
Beilman GJ, Mowlem JD. Noninvasive method for measuring
local hemoglobin oxygen saturation in tissue using wide gap
second derivative near-infrared spectroscopy. J Biomed Opt
2005;10:034017
22. Myers DE, Cooper CE, Beilman GJ, Mowlem JD, Anderson LD,
Seifert RP, Ortner JP. A wide gap second derivative NIR
spectroscopic method for measuring tissue hemoglobin oxy-
gen saturation. Adv Exp Med Biol 2006;578:217–22
23. Creteur J. Muscle StO2 in critically ill patients. Curr Opin Crit
Care 2008;14:361– 6
24. Cohn SM, Nathens AB, Moore FA, Rhee P, Puyana JC, Moore
EE, Beilman GJ. Tissue oxygen saturation predicts the devel-
opment of organ dysfunction during traumatic shock resusci-
tation. J Trauma 2007;62:44 –54
25. Horan TC, Gaynes RP, Martone WJ, Jarvis WR, Emori TG.
CDC definitions of nosocomial surgical site infections, 1992: a
modification of CDC definitions of surgical wound infections.
Infect Control Hosp Epidemiol 1992;13:606 – 8
26. Hunt TK, Dunphy JE. Effects of increasing oxygen supply to
healing wounds. Br J Surg 1969;56:705
27. Hunt TK, Linsey M, Sonne M, Jawetz E. Oxygen tension and
wound infection. Surg Forum 1972;23:47–9
28. Akca O, Sessler DI, Delong D, Keijner R, Ganzel B, Doufas AG.
Tissue oxygenation response to mild hypercapnia during car-
diopulmonary bypass with constant pump output. Br J An-
aesth 2006;96:708 –14
952 www.anesthesia-analgesia.org
29. Akça O, Doufas A, Morioka N, Iscoe S, Fisher J, Sessler D.
Hypercapnia improves tissue oxygenation. Anesthesiology
2002;97:801– 6
30. Morris S, Tang M, Geddes C. Vascular anatomical basis of
perforator skin flaps. Cir Plast Iberlatinamer 2006;32:1–5
31. Offman SL, Geddes CR, Tang M, Morris SF. The vascular basis
of perforator flaps based on the source arteries of the lateral
lumbar region. Plast Reconstr Surg 2005;115:1651–9
32. Hopf HW, Holm J. Hyperoxia and infection. Best Pract Res
Clin Anaesthesiol 2008;22:553– 69
33. Bitterman H. Bench-to-bedside review: oxygen as a drug. Crit
Care 2009;13:1– 8
34. Ives CL, Harrison DK, Stansby GS. Prediction of surgical site
infections after major surgery using visible and near-infrared
spectroscopy. Adv Exp Med Biol 2007;599:37– 44
35. Hopf HW, Viele M, Watson JJ, Feiner J, Weiskopf R, Hunt TK,
Noorani M, Yeap H, Ho R, Toy P. Subcutaneous perfusion and
oxygen during acute severe isovolemic hemodilution in
healthy volunteers. Arch Surg 2000;135:1443–9
36. Jensen JA, Jonsson K, Goodson WH III, Hunt TK, Roizen MF.
Epinephrine lowers subcutaneous wound oxygen tension.
Curr Surg 1985;42:472– 4
37. Akça O, Melischek M, Scheck T, Hellwagner K, Arkiliç C, Kurz
A, Kapral S, Heinz T, Lackner FX, Sessler DI. Postoperative
pain and subcutaneous oxygen tension. Lancet 1999;354:41–2
38. Buggy DJ, Kerin MJ. Paravertebral analgesia with levobupiva-
caine increases postoperative flap tissue oxygen tension after
immediate latissimus dorsi breast reconstruction compared
with intravenous opioid analgesia. Anesthesiology 2004;100:
375– 80
39. Rabkin JM, Hunt TK. Local heat increases blood flow and
oxygen tension in wounds. Arch Surg 1987;122:221–5
40. Sheffield CW, Sessler DI, Hopf HW, Schroeder M, Moayeri A,
Hunt TK, West JM. Centrally and locally mediated thermo-
regulatory responses alter subcutaneous oxygen tension.
Wound Repair Regen 1996;4:339 – 45
41. Kurz A, Sessler DI, Lenhardt R. Perioperative normothermia to
reduce the incidence of surgical-wound infection and shorten
hospitalization. Study of Wound Infection and Temperature
Group. N Engl J Med 1996;334:1209 –15
42. Melling AC, Ali B, Scott EM, Leaper DJ. Effects of preoperative
warming on the incidence of wound infection after clean
surgery: a randomised controlled trial. Lancet 2001;358:876 – 80
43. Arkilic CF, Taguchi A, Sharma N, Ratnaraj J, Sessler DI, Read
TE, Fleshman JW, Kurz A. Supplemental perioperative fluid
administration increases tissue oxygen pressure. Surgery
2003;133:49 –55
44. Jonsson K, Jensen JA, Goodson WH III, West JM, Hunt TK.
Assessment of perfusion in postoperative patients using tissue
oxygen measurements. Br J Surg 1987;74:263–7
45. Kabon B, Akca O, Taguchi A, Nagele A, Jebadurai R, Arkilic
CF, Sharma N, Ahluwalia A, Galandiuk S, Fleshman J, Sessler
DI, Kurz A. Supplemental intravenous crystalloid administra-
tion does not reduce the risk of surgical wound infection.
Anesth Analg 2005;101:1546 –53
46. Liu B, Finfer S. Intravenous fluids in adults undergoing
surgery. BMJ 2009;338:b2418
47. Edmark L, Kostova-Aherdan K, Enlund M, Hedenstierna G.
Optimal oxygen concentration during induction of general
anesthesia. Anesthesiology 2003;98:28 –33
48. Thorn CE, Matcher SJ, Meglinski IV, Shore AC. Is mean blood
saturation a useful marker of tissue oxygenation? Am J Physiol
Heart Circ Physiol 2009;296:H1289 –95
ANESTHESIA & ANALGESIA
 American Society of Critical Care Anesthesiologists
Section Editor: Michael J. Murray

High-Fidelity Simulation Demonstrates the Influence

of Anesthesiologists’ Age and Years from Residency
on Emergency Cricothyroidotomy Skills
Lyndon W. Siu, MBBS, FANZCA,* Sylvain Boet, MD,* Bruno C. R. Borges, MD,*
Heinz R. Bruppacher, MD, FMH,* Vicki LeBlanc, PhD,† Viren N. Naik, MD, MEd, FRCPC,*
Nicole Riem, MD,* Deven B. Chandra, MD, MEd, FRCPC,* and Hwan S. Joo, MD, FRCPC*

BACKGROUND: Age-related deterioration in both cognitive function and the capacity to control fine
motor movements has been demonstrated in numerous studies. However, this decline has not been
described with respect to complex clinical anesthesia skills. Cricothyroidotomy is an example of a
complex, lifesaving procedure that requires competency in the domains of both cognitive processing
and fine motor control. Proficiency in this skill is vital to minimize time to reestablish oxygenation
during a “cannot intubate, cannot ventilate” scenario. In this prospective, controlled, single-blinded study,
we tested the hypothesis that age affects the learning and performance of emergency percutaneous
cricothyroidotomy in a high-fidelity simulated cannot intubate/cannot ventilate scenario.
METHODS: Thirty-six staff anesthesiologists (19 aged younger than 45 years and 17 older than
45 years) managed a high-fidelity cannot intubate/cannot ventilate scenario in a high-fidelity
simulator before and after a 1-hour standardized training session. The group division cutoff age
of 45 years was based on the median age of our sample subject population before enrollment.
The scenarios required the insertion of an emergency percutaneous cricothyroidotomy. We
compared cricothyroidotomy skills in the older group with those in the younger group using
procedural time, 5-point task-specific checklist score, and global rating scale score. Correlation
based on age, years from residency, weekly clinical hours worked, previous continuing medical
education in airway management, and previous simulation experience was also performed.
RESULTS: In both prestandardization and poststandardization, age and years from residency
correlated with procedural time, checklist scores, and global rating scores. Baseline, prestandard-
ization variables were all better for the younger group, with a mean age of 37 years, compared with
the older group, with a mean age of 58 years. Procedural time was 100 (72–128) seconds versus
152 (120 –261) seconds. Checklist scores were 7.0 (6.1– 8.0) versus 6.0 (4.8 – 8.0). Global rating
scale scores were 22.0 (17.8 –29.8) versus 17.5 (10.4 –20.6). After the 1-hour standardized training
session, the younger group continued to perform better than the older group with procedural time of
75 (66 –91) seconds versus 87 (78 –123) seconds, checklist scores of 10.0 (9.1–10.0) versus 9.0
(8.0–10.0), and global rating scale scores of 35.0 (32.1–35.0) versus 32.0 (29.0–33.8). Regression
analysis was performed on the poststandardization data. Both age and years from residency indepen-
dently affected procedural time, checklist scores, and global rating scale scores (all P ⬍ 0.05).
CONCLUSIONS: Baseline proficiency with simulated emergency cricothyroidotomy is associated
with age and years from residency. Despite standardized training, operator age and years from
residency were associated with decreased proficiency. Further research should explore the
potential of using age and years from residency as factors for implementing periodic continuing
medical education. (Anesth Analg 2010;111:955–60)

A ge-related deterioration in cognitive functioning and

fine motor skills has been demonstrated in numerous
studies and reviews.1,2 The theoretical clinical signifi-
cance of aging has been extensively addressed in the litera-
ture.3–5 However, the specific impact of age on a particular
anesthetic procedure has never been objectively assessed.
From the *Department of Anesthesia, St. Michael’s Hospital, University of
Toronto; and †Wilson Centre, University Health Network, Department of
Medicine, University of Toronto, Toronto, Ontario, Canada.
Accepted for publication May 7, 2010.
Funded by departmental funds.
Disclosure: The authors report no conflicts of interest.
Address correspondence and reprint requests to Hwan S. Joo, MD, FRCPC,
St. Michael’s Hospital, 30 Bond St., Toronto, ON, M5B1W8, Canada. Address
e-mail to hwanjoomd@yahoo.com.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181ee7f4f
Cricothyroidotomy is a complex lifesaving procedure
that requires competency in the domains of both cognitive
processing and fine motor control. Proficiency in this skill is
vital because minimizing time to achieve oxygenation is
essential in a cannot intubate/cannot ventilate situation.
Age-related impact on this skill, and potentially other
important skills, should be recognized to facilitate the
development of appropriate educational strategies.
Using a low-fidelity static model, Wong et al.6 observed
that anesthesiologists aged younger than 45 years per-
formed cricothyroidotomies faster than those older than
45 years. In another study, John et al.7 demonstrated that
under stressful conditions in simulated cannot intubate/
cannot ventilate scenarios, procedural times were longer
compared with times achieved on static mannequin
models. However, neither of the 2 studies controlled for

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 955

Age and Percutaneous Cricothyroidotomy Skills
confounders such as previous procedural and/or simula-
tion experience.
The primary purpose of this prospective, controlled,
single-blinded study was to investigate whether age affects
learning and performance of simulated cannot intubate/cannot
ventilate emergency percutaneous cricothyroidotomies.
We hypothesize that age may affect the ability to learn
and perform emergency cricothyroidotomy in a high-
fidelity simulated setting.
METHODS
After institutional ethics approval, 36 attending anesthesi-
ologists in a tertiary care teaching hospital (19 aged
younger than 45 years and 17 older than 45 years) managed
a high-fidelity cannot intubate/cannot ventilate scenario
immediately before and after a standardization process that
included cricothyroidotomy training. The age 45 years was
chosen to divide the anesthesiologists into equal groups,
based on the median age of 39 possible participants before
study enrollment. To clarify for this article, we have named
the “older than 45 years” group the “older” group and the
“younger than 45 years” group the “younger” group. All 39
attending anesthesiologists in the Department of Anesthe-
sia of our institution were approached for recruitment; 3
declined to participate. Informed consent and confidential-
ity agreements were obtained from all participants.
Standardization Process
All participants received a 1-hour introduction of the
simulation center consisting of an orientation session to the
human patient simulator and participation in an introduc-
tory high-fidelity airway management scenario, which was
not part of the study.
The prestandardization scenario was a cannot intubate/
cannot ventilate scenario. In this videotaped session, an
actor playing a second-year resident calls for help after 2
unsuccessful intubation attempts and difficulty with face-
mask ventilation. The simulated patient’s oxygen satura-
tion was 89% when the subject arrived and decreased by
10% every minute. All alternative methods of intubation
were intended to be unsuccessful. This was accomplished
by changing the mannequin’s airway anatomy. The cervical
spine was immobilized, the tongue was made macroglos-
sic, and the vocal cords were adducted. When requested by
the participant, an ear-nose-throat surgeon and second
anesthesiologist were called but would not be available.
The scenario was designed to necessitate an emergency
percutaneous cricothyroidotomy using a 4.0-mm Melker
emergency cricothyroidotomy catheter set (C-TCCS-400;
Cook Inc., Bloomington, IN). The scenario only ended with
successful cricothyroidotomy, defined by positive capnog-
raphy on the monitor.
Immediately after the first cannot intubate/cannot ven-
tilate scenario, a teaching session including practical in-
structions on percutaneous cricothyroidotomy insertion
and video-assisted debriefing was provided. The same
individual conducted all debriefing sessions (LWS).
Immediately after the standardization session, all par-
ticipants managed the poststandardization scenario, an
identical cannot intubate/cannot ventilate scenario to that
presented during the prestandardization session. Subjects
956 www.anesthesia-analgesia.org
did not have prior knowledge of the content of any
scenario.
All scenarios were completed in a simulated operating
room environment containing a high-fidelity mannequin
(Sim Man; Laerdal, Kent, UK) equipped with an anatomi-
cally accurate larynx, properly designed for performance of
cricothyroidotomies, and standard monitors (electrocardio-
gram, noninvasive arterial blood pressure, oxygen satura-
tion as measured by pulse oximetry, and end-tidal CO2).
Appropriate equipment and airway devices in the room
included multiple-sized laryngoscopes, endotracheal tubes,
laryngeal mask airways, gum elastic bougie, and anesthesia
drug cart. Airway adjuncts kept outside the room included
a fiberoptic bronchoscope, a videolaryngoscope (Glide-
Scope®; Verathon, Inc., Bothell, WA), intubating laryngeal
mask airways, and a cricothyroidotomy kit. Two simula-
tion assistants played the scripted roles of a nurse and a
junior resident.
Sample Size Calculation
We hypothesized that there would be a difference in the
procedural time in favor of the younger group. One stan-
dard deviation for procedural times between the 2 groups
was defined to be a clinically important difference. We
calculated that 17 participants per group would be required
to achieve a difference of 1 SD between groups in the
poststandardization procedural times based on a 2-tailed ␣
of 0.05 and a power of 0.8.
Data Collection
Demographic data including age, the number of years after
graduation from anesthesia residency, the number of hours
of clinical practice per week, previous simulation and/or
airway simulation experience, and previous cricothyroid-
otomy experience on both patients and mannequins were
collected. All cricothyroidotomy performances (2 per sub-
ject) were video-recorded and later evaluated by 2 blinded
evaluators. The evaluators assigned were blinded to the
study outcome, each other’s scores, and whether the
video was from the pre- or posttest cannot intubate/
cannot ventilate session. Blinding to age was attempted
but may not have been possible as the approximate age
may have been guessed because participants may have
been recognized.
Outcome Measurement
The primary outcome was the comparison of cricothyroid-
otomy performances between the younger and the older
groups. Performance was assessed with 3 variables: proce-
dural time and 2 previously validated tools for procedural
skill evaluation (a 3-point task-specific checklist [Appendix
1] and global rating scale [GRS] [Appendix 2]).8 Procedural
time was measured during video review and was defined
as the time between the first instances when the subject
grasps any equipment from the cricothyroidotomy kit to
the time of successful cricothyroidotomy. The cricothyroid-
otomy checklist was based on observation of common but
important mistakes made by novice operators based on the
study by Friedman et al.8 A score of 0, 1, or 2 was given
when a stage was not performed, poorly performed, or
performed well, respectively. The GRS uses more general
ANESTHESIA & ANALGESIA
 Table 1. Demographics
Characteristics Age <45 y (n ⴝ 19) Age >45 y (n ⴝ 17) P value
Age 36.7 (⫾3.4) 58.0 (⫾8.3) ⬍0.001
Gender (male/female) 13 (68.4%)/6 (32.6%) 15 (88.2%)/2 (11.8%) NS
Years from residency graduation 4 (0.5–7.0) 28 (14.8–33.3) ⬍0.001
Hours of clinical work per week 43.4 (⫾3.8) 38.5 (⫾8.6) 0.031
Previous simulation experience 16 (84.2%) 5 (29.4%) 0.002
Previous simulation experience in airway 4 (21.1%) 2 (11.8%) NS
management
Attended airway lecture or continued medical 12 (63.1%) 11 (64.7%) NS
education within 10 y
Previous percutaneous cricothyroidotomy experience 15 (78.9%) 7 (41.2%) 0.039
Patient 3 (15.8%) 3 (17.7%) NS
Mannequin or pig 12 (63.2%) 4 (23.5%) 0.023
NS ⫽ not significant.
Values are mean (⫾standard deviation), median (interquartile range), or number (percentage).

Table 2. Cricothyroidotomy Performance Before and After Standardization

Prestandardization Poststandardization
Age <45 y Age >45 y P value Age <45 y Age >45 y P value
Procedural time (s) 100 (72–128) 152 (120–261) 0.003 75 (66–91)* 87 (78–123)* 0.018
Checklist scorea 7.0 (6.1–8.0) 6.0 (4.8–8.0) 0.024 10.0 (9.1–10.0)* 9.0 (8.0–10.0)* 0.005
Global rating scale scoreb 22.0 (17.8–29.8) 17.5 (10.4–20.6) 0.004 35.0 (32.1–35.0)* 32.0 (29.0–33.8)* 0.012
Data are median (interquartile range).
a
Checklist score ranges from 0 to 10.
b
Global rating scale ranges from 0 to 35.
* Significant difference with corresponding age group prestandardization data (P ⬍ 0.01).

descriptors and focuses on the overall performance of the dependent variables at P ⱕ 0.1 would be taken in
subject, not the specifics of the manual task. consideration in the final analysis of age as a predictor
variable.

Statistical Analysis
Analysis was performed using SPSS 11.0 software (SPSS,
Inc., Chicago, IL). To determine the reliability of assess-
ments provided by the 2 evaluators, intraclass correlation
coefficients (ICCs) were calculated for checklist and GRS
scores.9 We compared cricothyroidotomy performances
between younger and older groups both before and after
standardization. Procedural times, checklist scores, and
GRS scores were compared using the Mann-Whitney test.
Correlations between demographic variables and proce-
dural time, checklist scores, and GRS scores were per-
formed using Spearman’s correlation. All P values were 2
sided except for correlation, which was 1 sided. Signifi-
cance for correlation was set at a value of P ⬍ 0.05.
Significance for comparison between younger and older,
and between pre- and poststandardization variables was
set at a value of P ⬍ 0.025 to account for Bonferroni
correction.
We used multiple regression analysis to account for
differences between the 2 groups regarding variables that
could influence our primary outcomes. Initially, we entered
our primary outcomes measurements (procedural times,
GRS score, and checklist score) as dependent variables and
the other factors such as age, the number of hours of clinical
practice per week, previous simulation and/or airway
simulation experience, and previous cricothyroidotomy
experience on both patients and mannequins as predictor
variables. Any of these factors that correlated with the
RESULTS
Thirty-six attending anesthesiologists were recruited over 9
months. General demographics of the younger (aged 37 ⫾
3 years) and older (aged 58 ⫾ 8 years) groups are shown in
Table 1.
Interrater reliability was strong for both checklist and
GRS scores (checklist: ICC ⫽ 0.911; GRS: ICC ⫽ 0.837) (both
P ⬍ 0.05).
The performance of both age groups significantly im-
proved after teaching for all 3 variables (Table 2). Prestan-
dardization procedural time was longer and checklist and
GRS scores were lower in the older group (Table 2).
Poststandardization procedural time was longer for the
older group compared with the younger group. Both
checklist and GRS scores were lower in the older group
compared with the younger group (Table 2).
Age and years from residency correlated with proce-
dural time, checklist scores, and GRS scores, both before
and after standardization (Table 3). Previous simulation
experience also correlated with GRS scores, and with
procedural time but only before standardization. Weekly
clinical hours correlated with GRS scores, but only after
standardization (Table 3).
Multiple regression analysis was performed on post-
standardization data. Both age and years from residency
independently affected procedural time, checklist scores,
and GRS scores (all P ⬍ 0.05).

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 957

Age and Percutaneous Cricothyroidotomy Skills

Table 3. Predictor Variables: Correlation and P Values

Airway continued Previous simulation Weekly clinical Years from
medical education experience hours residency Age
Statistical analysis of
prestandardization data
Procedural time ⫺0.257 ⫺0.437 ⫺0.103 0.370 0.362
P ⫽ 0.065 P ⫽ 0.004* P ⫽ 0.275 P ⫽ 0.013* P ⫽ 0.015*
Checklist scores ⫺0.031 0.186 0.130 ⫺0.411 ⫺0.503
P ⫽ 0.43 P ⫽ 0.139 P ⫽ 0.225 P ⫽ 0.006* P ⫽ 0.001*
Global rating scale scores 0.257 0.437 0.007 ⫺0.454 ⫺0.497
P ⫽ 0.065 P ⫽ 0.004* P ⫽ 0.483 P ⫽ 0.003* P ⫽ 0.001*
Statistical analysis of
poststandardization data
Procedural time ⫺0.007 ⫺0.041 0.070 0.341 0.310
P ⫽ 0.484 P ⫽ 0.407 P ⫽ 0.344 P ⫽ 0.021* P ⫽ 0.033*
Checklist scores 0.096 0.177 0.187 ⫺0.333 ⫺0.333
P ⫽ 0.289 P ⫽ 0.151 P ⫽ 0.138 P ⫽ 0.024* P ⫽ 0.024*
Global rating scale scores 0.076 0.302 0.344 ⫺0.564 ⫺0.521
P ⫽ 0.330 P ⫽ 0.037* P ⫽ 0.02* P ⬍ 0.001* P ⫽ 0.001*
Values represent Spearman correlation coefficient and P values.
* Statistically significant (P ⬍ 0.05).

DISCUSSION include the ability to perform complex tasks rapidly, to

Before standardization, the younger group significantly
outperformed the older group as assessed by all 3 variables.
This difference in baseline (prestandardization) perfor-
mances between the 2 groups may reflect differences in
previous simulation exposure and/or cricothyroidotomy
experience on mannequins (Table 1). Of note, however,
those who had previously performed cricothyroidotomies
on mannequins had only done so once or twice and none in
the 6 months before the study. In addition, a previous study
suggests that cricothyroidotomy performance in low-
fidelity models declines to near baseline after 3 months.10
Both groups improved significantly in time and checklist
and GRS scores after standardization. This trend suggests
that training is effective in improving both the cognitive
and psychomotor aspects of emergency cricothyroidotomy
skills.
To investigate whether the difference in performance
was attributable to age, we focused our comparative anal-
ysis on the poststandardization data. By priming partici-
pants with a 1-hour standardization session consisting of 2
airway management simulation scenarios, debriefing, and
a practical instructional session on cricothyroidotomy im-
mediately before the study scenario, we aimed to minimize
bias caused by the variable simulation and cricothyroid-
otomy experience of participants. The results of this study
demonstrated that after standardized exposure to high-
fidelity simulation and after standardized teaching, an-
esthesiologists who are older and further away from
residency training require more time to perform emer-
gency percutaneous cricothyroidotomies and have lower
checklist and GRS scores.
We failed to find studies in the literature that addressed
the impact of age on the performance of a specific anes-
thetic procedure, except for the study by Wong et al.,6
which also used cricothyroidotomies but that study was
not primarily powered for this specific objective. However,
the literature is rich with articles that address patient safety
concerns and aging physicians, including surgeons and
anesthesiologists.3,4 Psychomotor and perceptual processes
that are required in aviation that deteriorate with age
process incoming information and make complex deci-
sions, and to perform effectively in a stressful environ-
ment.11 All of them are required for the practice of
anesthesiology as well.
Proficient performance of emergency cricothyroidotomy
requires both the cognitive ability to recall essential steps
for the procedure and psychomotor skills to execute
planned actions efficiently. Both aspects may be adversely
affected by increasing age as a result of the general slowing
of central cognitive processes.12–14 One probable neuro-
physiological mechanism is the loss of neural connectivity
or decreased levels of neurotransmitters in the aging
brain.13,14 The difference in poststandardization procedural
times, however, supports an age-related decline in psy-
chomotor skills required for this procedure. Another pos-
sibility is that years from residency or training is just as
important. Time from formal residency or training may
affect technical skills, even after standardized training.
Nonetheless, the clinical significance of the difference in
procedural times, measured within a simulation setting, is
unknown and may be difficult to determine.
It should be clear that our study was not focused on
finding a cutoff age beyond which more training on crico-
thyrotomies is necessary, rather, simply on showing that
age may well be a factor that affects learning and perfor-
mance of emergency cricothyroidotomies. The cutoff age of
45 years was used mainly as a median-age dividing point.
Further research should explore the potential of using
age and years from residency as factors for implementing
periodic continuing medical education. As shown in our
study, the older anesthesiologists benefited more from
standardized cricothyroidotomy training than the younger
anesthesiologists. However, this may also be attributable to
a ceiling effect because the younger anesthesiologists may
have been closer to their maximal potential in their pres-
tandardization scenario.
This study has several limitations. Even with our stan-
dardized training, we might not have equalized previous
knowledge base in performing cricothyroidotomies. Previ-
ous knowledge and recent airway training in many

958 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

younger participants may have contributed to better times
and scores. All participants were attending anesthesiolo-
gists recruited from a single tertiary academic center. A
multicenter study would render the results more general-
izable. Because our study was conducted using 4.0-mm
Melker cricothyroidotomy kits (C-TCCS-400; Cook Inc.),
the results might not reflect performance using other
commercial kits, especially those that do not use
Seldinger’s technique. In addition, as for all simulation
research, there are conflicting views on whether results can
be extrapolated to real-life clinical settings.
In conclusion, there may well be an age/number of
years from residency–related decline in proficiency for
emergency cricothyroidotomy skills in a high-fidelity simu-
lated setting. Because emergency cricothyroidotomies are
performed rarely, if at all during routine clinical practice,
lack of experience further contributes to failure in real
life.15–17 Further research should explore the potential of
using age and years from residency as factors for imple-
menting periodic specific continuing medical education.
AUTHOR CONTRIBUTIONS
LWS helped with study design, conduct of study, and manuscript
preparation; SB helped with conduct of study, data collection and
analysis, and manuscript preparation; BCRB helped with conduct
of study and manuscript preparation; HRB helped with data
collection; VL helped with data analysis; VNN helped with study
design; NR and DBC helped with manuscript preparation; and
HSJ helped with study design, conduct of study, and manuscript
preparation, and is responsible for archival.

Appendix 1. Task-Specific Checklist for Cricothyroidotomy

Score
Task 0 1 2
Aspiration to identify trachea Does not aspirate Performed inadequately Aspirates with air-filled
or fluid-filled syringe
Ventilation during Does not ventilate during the Ventilates for part of the Ventilates for the entire
cricothyroidotomy cricothyroidotomy duration of the duration of the
cricothyroidotomy cricothyroidotomy
Correct caudal angling during Cephalad angle 90° to trachea 45° caudad
guidewire insertion (not
during needle insertion)
Adequate skin and membrane Does not use the scalpel for Performed inadequately Cuts skin and
incision incision cricothyroid
membrane with the
scalpel
Correct use of dilator and Attempts to insert cricothyroidotomy Dilates separately and Railroad entire
cricothyroidotomy cannula without dilator in place then railroad entire assembly (dilator and
assembly cricothyroidotomy
cannula)

Appendix 2. Global Rating Scale for Cricothyroidotomy

Preparation for procedure
Respect for tissue
Time and motion
Instrument handling
Flow of procedure
Knowledge of procedure
Overall performance
1 2
Did not organize equipment well.
Had to stop procedure
frequently to prepare
equipment
Frequently used unnecessary
force on tissue or caused
damage
Many unnecessary moves
Repeatedly made tentative or
awkward moves with
instruments
Frequently stopped procedure
and seemed unsure of next
move
Deficient knowledge
Very poor
Score
3 4
Equipment generally organized.
Occasionally had to stop and
prepare items
Careful handling of tissue but
occasionally caused
inadvertent damage
Efficient time/motion but some
unnecessary moves
Competent use of instruments
but occasionally appeared
stiff or awkward
Demonstrated some forward
planning with reasonable
progression of procedure
Knew all important steps of
procedure
Competent
5
All equipment neatly organized,
prepared, and ready for use
Consistently handled tissues
appropriately with minimal
damage
Clear economy of movement
and maximum efficiency
Fluid moves with instruments
and no awkwardness
Obviously planned course of
procedure with effortless
flow from 1 move to the next
Demonstrated familiarity with
all aspects of procedure
Clearly superior

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 959

Age and Percutaneous Cricothyroidotomy Skills
REFERENCES
1. Krampe RT. Aging, expertise and fine motor movement. Neu-
rosci Biobehav Rev 2002;26:769 –76
2. Birren J, Schaie K. Handbook of the Psychology of Aging. 5th
ed. San Diego: Academic Press, 2001:313– 48
3. Katz JD. Issues of concern for the aging anesthesiologist. Anesth
Analg 2001;92:1487–92
4. Greenfield LJ, Proctor MC. When should a surgeon retire? Adv
Surg 1999;32:385–93
5. Boom-Saad Z, Langenecker SA, Bieliauskas LA, Graver CJ,
O’Neill JR, Caveney AF, Greenfield LJ, Minter RM. Surgeons
outperform normative controls on neuropsychologic tests, but
age-related decay of skills persists. Am J Surg 2008;195:205–9
6. Wong DT, Prabhu AJ, Coloma M, Imasogie N, Chung FF. What
is the minimum training required for successful cricothyroid-
otomy. Anesthesiology 2003;98:349 –53
7. John B, Suri I, Hillermann C, Mendonca C. Comparison of
cricothyroidotomy on manikin vs. simulator: a randomised
cross-over study. Anaesthesia 2007;62:1029
8. Friedman Z, You-Ten KE, Bould MD, Naik V. Teaching lifesav-
ing procedures: the impact of model fidelity on acquisition and
transfer of cricothyrotomy skills to performance on cadavers.
Anesth Analg 2008;107:1663–9
9. Landis JR, Koch GG. An application of hierarchical kappa-type
statistics in the assessment of majority agreement among mul-
tiple observers. Biometrics 1977;33:363–74
960 www.anesthesia-analgesia.org
10. Amrhein PC, Stelmach GE, Goggin NL. Age differences in the
maintenance and restructuring of movement preparation. Psy-
chol Aging 1991;6:451– 66
11. Eyraud MY, Borowsky MS. Age and pilot performance. Aviat
Space Environ Med 1985;56:553– 8
12. Birren J, Schaie K. Handbook of the Psychology of Aging. 5th
ed. San Diego: Academic Press, 2001:288 –312
13. Salthouse TA. The processing-speed theory of adult age differ-
ences in cognition. Psychol Rev 1996;103:403–28
14. Birren J, Schaie K. Handbook of the Psychology of Aging. 5th
ed. San Diego: Academic Press, 2001:135– 60
15. Schaumann N, Lorenz V, Schellongowski P, Staudinger T,
Locker GJ, Burgmann H, Pikula B, Hofbauer R, Schuster E,
Frass M. Evaluation of Seldinger technique emergency crico-
thyroidotomy versus standard surgical cricothyroidotomy in
200 cadavers. Anesthesiology 2005;102:7–11
16. Ravlo O, Bach V, Lybecker H, Moller JT, Werner M, Nielsen
HK. A comparison between two emergency cricothyroidotomy
instruments. Acta Anaesthesiol Scand 1987;31:317–9
17. Chang RS, Hamilton RJ, Carter WA. Declining rate of crico-
thyrotomy in trauma patients with an emergency medicine
residency: implications for skills training. Acad Emerg Med
1998;5:247–51
ANESTHESIA & ANALGESIA
Adaptive Support Ventilation with Protocolized
De-Escalation and Escalation Does Not Accelerate
Tracheal Extubation of Patients After Nonfast-Track
Cardiothoracic Surgery
Dave A. Dongelmans, MD, MSc,* Denise P. Veelo, MD, PhD,*†‡ Jan M. Binnekade, PhD,*
Bas A.J.M. de Mol, MD, PhD,§ Anna Kudoga, MS,* Frederique Paulus, MSc,*
and Marcus J. Schultz, MD, PhD*‡㛳

BACKGROUND: It is uncertain whether adaptive support ventilation (ASV) accelerates weaning

of nonfast-track cardiothoracic surgery patients. A lower operator set %-minute ventilation with
ASV may allow for an earlier definite switch from controlled to assisted ventilation, potentially
hastening tracheal extubation. We hypothesized that ASV using protocolized de-escalation and
escalation of operator set %-minute ventilation (ASV-DE) reduces time until tracheal extubation
compared with ASV using a fixed operator set %-minute ventilation (standard ASV) in uncompli-
cated patients after nonfast-track coronary artery bypass graft.
METHODS: We performed a randomized controlled trial comparing ASV-DE with standard ASV.
With ASV-DE, as soon as body temperature was ⬎35.0°C with pH ⬎7.25, operator set %-minute
ventilation was decreased stepwise to a minimum of 70%.
RESULTS: Sixty-three patients were randomized to ASV-DE, and 63 patients to standard ASV.
The duration of mechanical ventilation was not different between groups (10.8 [6.5–16.1] vs
10.7 [6.6 –13.9] hours, ASV-DE versus standard ASV; P ⫽ 0.32). Time until the first assisted
breathing period was shorter (3.1 [2.0 – 6.7] vs 3.9 [2.1–7.5] hours) and the number of assisted
ventilation episodes was higher (78 [34 –176] vs 57 [32–116] episodes), but differences did not
reach statistical significance. The duration of assisted ventilation episodes that ended with
tracheal extubation was different between groups (2.5 [0.9 – 4.6] vs 1.4 [0.3–3.5] hours, ASV-DE
versus standard ASV; P ⬍ 0.05).
CONCLUSION: Compared with standard ASV, weaning of patients after nonfast-track coronary
artery bypass graft using ASV with protocolized de-escalation and escalation does not shorten
time to tracheal extubation. (Anesth Analg 2010;111:961–7)

A daptive support ventilation (ASV) is an advanced

closed-loop mode of mechanical ventilation (MV)
that maintains an operator preset minute ventila-
tion. ASV adjusts respiratory rates and pressure levels
according to measured lung mechanics at each breath.1 In
addition, ASV automatically switches between controlled
and assisted ventilation according to the patient’s status.2
Previous randomized controlled trials have tested the effi-
cacy of ASV in patients after cardiothoracic surgery.3–5 In a
study of fast-track cardiothoracic surgery patients, ASV
compared with synchronized intermittent mandatory ven-
tilation or traditional pressure support ventilation short-
ened the time to tracheal extubation.5 This was confirmed
From the Departments of *Intensive Care Medicine, †Anesthesiology, and
§Cardiothoracic Surgery; and ‡Laboratory of Experimental Intensive
Care and Anesthesiology (L.E.I.C.A.), Academic Medical Center, Univer-
sity of Amsterdam; and 㛳HERMES Critical Care Group, Amsterdam, The
Netherlands.
Accepted for publication June 18, 2010.
Supported by the Department of Intensive Care Medicine, Academic Medi-
cal Center.
Presented, in part, at the ATS conference, San Diego, CA, May 18, 2009.
Disclosure: The authors report no conflicts of interest.
Address correspondence and reprint requests to Dave A. Dongelmans, MD,
MSc, Department of Intensive Care Medicine, G3-212, Academic Medical
Center, Meibergdreef 9, 1105 AZ Amsterdam, The Netherlands. Address
e-mail to D.A.Dongelmans@amc.uva.nl.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181efb316
in another study of fast-track cardiothoracic surgery pa-
tients in which ASV was compared with pressure-regulated
volume controlled ventilation.4 However, in a recent study
of nonfast-track cardiothoracic surgery patients, ASV com-
pared with traditional pressure support ventilation did not
shorten the time to tracheal extubation.3
A lower operator set %-minute ventilation with ASV
may allow for earlier and more frequent switches from
controlled to assisted ventilation. Indeed, patients whose
lungs are ventilated with lower minute volumes could be
forced to breathe spontaneously sooner because arterial
Pco2 thresholds for breathing are reached faster.6 Second,
in the above-mentioned study of nonfast-track cardiotho-
racic surgery patients,3 patients were able to trigger the
ventilator early in the weaning process, at least suggesting
that a lower operator set %-minute could push patients
toward longer periods of assisted ventilation and thereby
earlier tracheal extubation.
In a randomized controlled trial of patients after
planned and uncomplicated nonfast-track coronary artery
bypass graft (CABG), we compared ASV using protocol-
ized de-escalation and escalation of operator set %-minute
ventilation (ASV-DE) with ASV using a fixed operator set
%-minute ventilation (standard ASV). We hypothesized
that ASV-DE reduces time to tracheal extubation compared
with standard ASV.

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 961

Adaptive Support Ventilation Protocolized with De-Escalation After Cardiothoracic Surgery
METHODS
Patients and Setting
Consecutive patients after elective and uncomplicated
CABG admitted to the 28-bed intensive care unit (ICU) of
the Academic Medical Center, Amsterdam, The Nether-
lands, were eligible for inclusion. The study protocol was
approved by the local IRB, and preoperative written and
signed informed consent was obtained from eligible pa-
tients programmed for surgery.
Study Design
According to an open-label randomized controlled design,
patients were assigned to receive MV with either ASV-DE
or standard ASV.
Inclusion Criteria
Patients were included and randomized using sealed num-
bered envelopes on ICU arrival after surgery. We created a
homogeneous group of patients after elective and uncom-
plicated CABG, i.e., without a history of chronic obstructive
pulmonary disease or hemodynamic instability. Patients
with a history of chronic obstructive pulmonary disease or
a history of pulmonary surgery, and patients with an
intraaortic balloon pump or inotropics and/or vasopres-
sors at a more than usual rate (in milligrams per hour:
dopamine 20, norepinephrine 0.5, dobutamine 25, or epi-
nephrine [any rate]) on ICU arrival were excluded.
Cardiothoracic Surgery/Anesthesia Procedures
All patients in both groups were anesthetized according to
our standard institutional protocol, starting with 1 or 2 mg
lorazepam as premedication, followed by etomidate, sufen-
tanil, and rocuronium for induction of anesthesia and
facilitation of intubation. During the surgical procedure,
sufentanil was used as analgesic, and sevoflurane plus
propofol were used to maintain anesthesia. Muscle relax-
ants were not given during the surgical procedure. Mor-
phine and midazolam could be administered at the end of
the procedure.
Cardiopulmonary bypass was performed under moder-
ate hypothermia (28°C–35°C), using a membrane oxygen-
ator and nonpulsatile blood flow. At the end of anesthesia,
all patients were transferred to the ICU with tracheal
intubation. Anesthesiologists and surgeons at the operating
room were blinded for inclusion or randomization of
patients.
ICU Management
Unit policy comprised that a patient was cared for by a
dedicated ICU nurse, responsible for 1 or 2 patients.
Attending ICU nurses were constantly at the bedside, and
changes in treatment according to the postoperative ICU
protocol, based on observations by ICU nurses, were ex-
ecuted immediately.
The postoperative ICU protocol was similar for both
study groups and involved fluid resuscitation with normal
saline and starch solutions, blood transfusion to maintain
hemoglobin concentration (ⱖ5.0 mmol/L), norepinephrine
in continuous infusion to achieve mean arterial blood
pressure ⱖ70 mm Hg, and dobutamine and/or enoximone
962 www.anesthesia-analgesia.org
to achieve a cardiac index ⱖ2.5 L/min/m2 or a mixed
venous oxygenation ⬎60%.
Sedation and analgesics were given according to local
protocol for postoperative cardiothoracic surgery patients.
Propofol was given for sedation via continuous infusion.
Infusion of propofol was stopped instantly when core
temperature reached 35°C. The confusion assessment
method for the ICU was used to screen for delirium, and if
present, haloperidol was started. Acetaminophen (4 g/d)
was started in all patients. The requirement for additional
analgesia was assessed by attending ICU nurses. Morphine
was given in boluses of 1 to 2 mg IV until patients were free
of pain. The boluses were repeated as needed. Postopera-
tive shivering, if present, was treated with meperidine (25
mg IV). Muscle relaxants were not given in the ICU.
MV Protocols
All patients’ lungs were ventilated by a Hamilton Galileo
ventilator (software version GMP03.41f, GCP03.40a,
GTP01.00; Hamilton Medical AG, Rhäzüns, Switzerland).
Passive humidification of the ventilatory circuit was ap-
plied by means of an HME filter (Medisize Hygrovent S;
Medisize, Hillegom, The Netherlands).
In both groups, the initial levels of fraction of inspired
oxygen (Fio2) (50%) positive end-expiratory pressure
(PEEP) (5 cm H2O), peak airway pressure (35 cm H2O), and
%-minute ventilation (a theoretical value based on ideal
body weight, 100%) were set by the attending ICU physi-
cian. Flow trigger sensitivity was set at 2 L/s; active
patients could trigger the ventilator (i.e., actual minute
ventilation could exceed set %-minute ventilation). An
arterial blood gas analysis was performed 30 minutes after
connection to the ICU ventilator, and 30 minutes after each
modification of ventilator settings (except for Fio2), it was
advised to perform an additional arterial blood gas analy-
sis. Fio2 could be adjusted to maintain arterial oxygen
saturation of ⱖ95%.
In both groups, patients were tracheally extubated after
achieving general tracheal extubation criteria (i.e., respon-
sive and cooperative, urine output ⬎0.5 mL/kg/h, chest
tube drainage ⬍100 mL last hour, no uncontrolled arrhyth-
mia, and having a core temperature ⬎36.0°C and a respi-
ratory frequency of ⬎10 breaths per minute without
machine-controlled breaths for at least 30 minutes). T-piece
weaning was not used; patients were tracheally extubated
when they reached the above-described extubation criteria.
ASV-DE Versus Standard ASV
With ASV-DE, as soon as body temperature reached 35.0°C
with pH ⬎7.25, irrespective of arterial Pco2, %-minute
ventilation was decreased stepwise by 10% (de-escalation)
until 70% of the theoretical value based on ideal body
weight, only if pH declined ⬍7.25%-minute ventilation was
increased again (escalation) (Fig. 1). Indeed, by neglecting
the Pco2 safety limits as defined for standard ASV, we
created a span for de-escalation with ASV-DE.
With standard ASV, %-minute ventilation was only
changed if Pco2 was ⬍3.5 or ⬎5.5 kPa. Indeed, settings
with ASV were based on previously used safety limits to
guarantee sufficient minute ventilation at all times.3
ANESTHESIA & ANALGESIA
Figure 1. Flow sheet as used by intensive care unit (ICU) nurses and
ICU physicians [translated from Dutch]. ASV ⫽ adaptive support
ventilation; %-MV ⫽ percentage minute ventilation; F-spont ⫽ fre-
quency of spontaneous breath; ABG ⫽ arterial blood gas.
Data Collection
Collected data included the patient characteristics of gender,
age, weight, and height, and the operation characteristics of
number of bypasses, cardiopulmonary bypass time (pump
time), and aortic clamp time. Intraoperative and ICU sedative
and analgesic prescriptions; time from admission to ICU until
reaching a central body temperature of 36.0°C; ventilation
characteristics, tidal volume size, PEEP, inspiratory pressure
(defined as the maximum airway pressure minus the level of
PEEP) and respiratory rate, %-minute ventilation, and arterial
blood gas data in time were collected. Respiratory data were
collected by a data logger connected to the ventilator (Ham-
ilton data logger, version 3.27.1; Hamilton Medical AG) and
from our patient Data Management System (IMDsoft, Sassen-
heim, The Netherlands).
Outcome variables (see Definitions) were calculated for
each patient. Primary end point was total duration of
tracheal intubation; secondary end points were summed
single assisted ventilation episodes, time to first single
assisted ventilation episode, time to the assisted ventilation
episode that was followed by tracheal extubation, and
length of stay in the ICU.
Definitions
Total duration of tracheal intubation was defined as the
period from ICU admission until tracheal extubation, and a
single assisted ventilation episode was defined as an epi-
sode of ⱖ20 minutes during which the patient was breath-
ing at least 5 breaths per minute.
Opiate doses were all recalculated as morphine equipo-
tent doses with the following formula: 10 mg morphine ⫽
0.1 mg fentanyl ⫽ 0.01 mg sufentanil.7 Doses of benzodi-
azepines were similarly converted to equipotent doses of
diazepam using the following formula: 5 mg midazolam ⫽
10 mg diazepam ⫽ 50 mg oxazepam.8
Sample Size
The study was powered on total duration of tracheal
intubation. Sample size assumptions were based on results
of our previous study, i.e., a mean duration of ventilation of
October 2010 • Volume 111 • Number 4
Figure 2. Flow diagram showing the flow of the patients through each
stage of the clinical trial. CABG ⫽ coronary artery bypass graft;
COPD ⫽ chronic obstructive pulmonary disease; ASV-DE ⫽ adaptive
support ventilation de-escalation escalation.
16.4 hours in the ASV group.3 A reduction of approxi-
mately 10% was expected for the total duration of tracheal
intubation. A sample size of 61 patients in each group was
deemed to have 80% power to detect a difference in the
duration of MV of 10%, assuming a common standard
deviation of 3.3 hours, using a 2-sided t test with a 0.05
2-sided significance level.
Statistical Analysis
Descriptive statistics were used to summarize patient char-
acteristics. Categorical variables were compared between
groups by ␹2 tests. If normally distributed, continuous
values were expressed as means ⫾ SD; otherwise, medians
and interquartile ranges were used. All analyses were
performed in SPSS version 16.0 (SPSS, Inc., Chicago, IL).
RESULTS
Patients
We included 126 consecutive patients after elective and un-
complicated CABG: 63 patients were randomized to ASV-DE
and 63 to standard ASV (Fig. 2). Of patients enrolled in the
study, 2 patients were lost for analysis of the secondary end
points because of data logger failures: 1 patient randomized to
ASV-DE and 1 randomized to standard ASV.
Baseline, Perioperative, and ICU Characteristics
Groups were well balanced (Table 1). Arterial blood gas
analyses on ICU admission were not different (data not
shown). Core temperature on ICU admission was not
different (35.5°C ⫾ 1.1°C vs 35.7°C ⫾ 0.6°C, ASV-DE versus
standard ASV; P ⫽ 0.32). The number of patients with a
temperature ⬎36°C on ICU admission was also not differ-
ent (27 [44%] vs 32 [51%], ASV-DE versus standard ASV;
P ⫽ 0.24). There were no differences in time to rewarming
to 36°C (2.1 ⫾ 3.0 vs 1.7 ⫾ 2.1 hours, ASV-DE versus
standard ASV; P ⫽ 0.64). ICU survival was 100% for the 2
randomization groups.
www.anesthesia-analgesia.org 963
Adaptive Support Ventilation Protocolized with De-Escalation After Cardiothoracic Surgery

Table 1. Patient Characteristics, and Intraoperative and Intensive Care Characteristics

ASV-DE Standard ASV
Patient characteristics
Patients, n 63 63
Male gender, n (%) 56 (89) 55 (87)
Age (y), mean ⫾ SD 68 ⫾ 10 65 ⫾ 9
Actual body weight (kg), mean ⫾ SD 82 ⫾ 14 83 ⫾ 10
Ideal body weight (kg), mean ⫾ SD 69 ⫾ 8 70 ⫾ 8
Set body weight (kg), mean ⫾ SD 69 ⫾ 8 70 ⫾ 8
Height (cm), mean ⫾ SD 172 ⫾ 15 175 ⫾ 8
Intraoperative characteristics
No. of bypasses, mean ⫾ SD 3⫾1 3⫾1
ECC time (h), mean ⫾ SD 1.7 ⫾ 0.7 1.7 ⫾ 0.6
AOX time (h), mean ⫾ SD 1.1 ⫾ 0.5 1.1 ⫾ 0.5
Lowest core temperature (°C), mean ⫾ SD 34 ⫾ 1.2 34 ⫾ 1.3
Sufenta dose (␮g), n; median (IQR)a 63; 200 (135–250) 63; 200 (150–250)
Midazolam dose (mg), n; median (IQR)a 50; 15 (5–25) 49; 15 (5–20)
Morphine dose (mg), n; median (IQR)a 28; 20 (20–20) 26; 20 (20–20)
Clonidine dose (␮g), n; median (IQR)a 6; 150 (150–225) 3; 150 (150–300)
Intensive care unit characteristics
APACHE II score 17 ⫾ 5 16 ⫾ 5
Morphine dose (mg/kg) n; median (IQR)a 0.05 (0.03–0.11) 0.05 (0.02–0.07)
Propofol dose (mg) ICU, n; median (IQR)a 63; 1022 (621–1963) 63; 1125 (500–2122)
Sedation duration (h), median (IQR)a 3.6 (2.2–6.8) 4.6 (2.6–7.0)
Length of stay ICU (h), median (IQR) 27 (21–49) 27 (22–40)
ASV ⫽ adaptive support ventilation; ASV-DE ⫽ ASV de-escalation escalation; ECC time ⫽ duration of extracorporeal circulation; AOX time ⫽ duration of aortic
cross-clamping; IQR ⫽ interquartile range; APACHE ⫽ Acute Physiology and Chronic Health Evaluation; ICU ⫽ intensive care unit.
a
Only dose of patients who actually received medication is displayed.
There were no significant differences between the randomization groups.

Protocol Adherence
In the ASV-DE group, mean %-minute ventilation at tra-
cheal extubation was 92% ⫾ 13% (14% were tracheally
extubated at %-minute ventilation level of 70%, 77% at a
level between 70% and 100%, and 9% at a level ⬎100%). In
the standard ASV group, mean %-minute ventilation at the
time of tracheal extubation was 103% ⫾ 10% (2% at a level
⬍100%, 78% at a level of 100%, and 20% at a level ⬎100%)
(P ⬍ 0.05 versus ASV-DE).
Sedation and analgesic use was not different between
randomization groups (Table 1). There were no patients
who fulfilled the criteria for delirium, and haloperidol was
never started.

Duration of MV and Assisted
Ventilation Episodes
Duration of tracheal intubation was not different between
groups (10.8 [6.5–16.1] vs 10.7 [6.6 –13.9] hours, ASV-DE
versus standard ASV; P ⫽ 0.32) (Fig. 3; Table 2). Neither
time from admission to the first assisted breathing period
(3.1 [2.0 – 6.7] vs 3.9 [2.1–7.5] hours; P ⫽ 0.49) nor the
number of assisted ventilation episodes (78 [34 –176] vs 57
[32–116] episodes; P ⫽ 0.20) was different. However, dura-
tion of assisted ventilation episodes that ended with tra-
cheal extubation was longer with ASV-DE (2.5 [0.9 – 4.6] vs
1.4 [0.3–3.5] hours; P ⫽ 0.05).
In a per-protocol analysis in which only patients who
reached ⱕ80%-minute ventilation before tracheal extuba-
tion were compared with patients in the standard ASV
group, there were also no significant differences in time
until extubation: 10.5 (8.0 –16.0) vs 10.0 (6.0 –13.0) hours
(n ⫽ 16 vs n ⫽ 63). The choice for 80%-minute ventilation was
arbitrary; however, we believed that 90%-minute ventilation
Figure 3. Tracheally intubated patients (%) expressed as Kaplan-
Meier curve in the adaptive support ventilation (ASV) and in the ASV
de-escalation escalation (ASV-DE) groups.
was not clinically significant and with 70%-minute ventilation
the number of patients would result in too few patients.
Ventilator and Ventilation Variables
Ventilator and ventilation variables are presented in Figure
4. There were no differences between groups regarding
tidal volume, respiratory rate, arterial pH, Pco2, and Po2.
The highest levels of arterial Pco2 were similar in the 2
randomization groups (5.9 [range, 5.2– 6.4] vs 5.8 [range,
5.0 – 6.4] kPa, ASV-DE versus standard ASV; P ⫽ 0.82).
DISCUSSION
In this study of postoperative weaning of patients after
planned and uncomplicated nonfast-track CABG, we found
that ASV with protocolized de-escalation and escalation

964 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

 Table 2. Respiratory Characteristics
ASV-DE Standard ASV P value
Duration of tracheal intubation (h), median (IQR) 10.8 (6.5–16.1) 10.7 (6.6–13.9) 0.32
Time (h) from admission to first spontaneous breathing period, median (IQR) 3.1 (2.0–6.7) 3.9 (2.1–7.5) 0.49
No. of spontaneous breathing periods, median (IQR) 78 (34–176) 57 (32–116) 0.20
Spontaneous breathing/total time (%), median (IQR) 36 (21–61) 31 (21–50) 0.39
Controlled breathing/total time (%), median (IQR) 64 (39–79) 69 (50–79) 0.39
Last spontaneous breathing period before extubation (h), median (IQR) 2.5 (0.9–4.6) 1.5 (0.3–3.5) 0.045
Tidal volume (mL), mean ⫾ SD 573 ⫾ 105 580 ⫾ 105 0.62
Tidal volume (mL/kg ideal body weight), mean ⫾ SD 8.1 ⫾ 1.5 8.4 ⫾ 1.5 0.27
Respiratory rate during spontaneous breathing, mean ⫾ SD 12.4 ⫾ 1.4 12.7 ⫾ 1.3 0.22
PEEP level (cm H2O), mean ⫾ SD 5.3 ⫾ 1.2 5.3 ⫾ 1.0 0.60
ASV ⫽ adaptive support ventilation; ASV-DE ⫽ ASV de-escalation escalation; IQR ⫽ interquartile range; PEEP ⫽ positive end-expiratory pressure.

Figure 4. Graphical display of ventilator

and ventilation variables over time.
Minute ventilation in %, positive end-
expiratory pressure, tidal volume (VT) in
mL/kg ideal body weight (IBW), L/min,
mean pH, and PCO2 and PO2 in kPa; black
circles represent adaptive support venti-
lation (ASV) group; open circles represent
ASV de-escalation escalation (ASV-DE)
group. RR ⫽ respiratory rate in breaths
per minute; I ⫽ after stabilization; II ⫽
after 4 hours; III ⫽ before extubation.

compared with standard ASV did not shorten duration of
tracheal intubation. The time to the first assisted breathing
period was also not different between groups. There was,
however, a difference in the duration of the assisted
breathing period ending with extubation.
MV could harm all patients, including those whose lungs
are ventilated for only hours.9 Therefore, it is imperative to
strive for shorter duration of MV and tracheal intubation at all
times, including the weaning phase after surgery. In addition,
controlled forms of MV could rapidly cause muscle atrophy of
the diaphragm.10 To counteract this phenomenon, it is impor-
tant to allow patients to use their diaphragm as soon as
possible while still being mechanically ventilated. ASV allows
for automatic switches between controlled ventilation and
assisted ventilation, depending on the patient’s activity. As
such, ASV with protocolized de-escalation and escalation may
improve outcome because there was a trend to shorter time
until the first assisted breathing period, and more assisted
ventilation episodes. Our study, however, was underpowered
to show statistical difference regarding these secondary end
points.
Although we hypothesized that protocolized de-escalation
would prevent longer intubation times, our study showed
otherwise. Of note, improved patient-ventilator synchrony
with ASV could also lengthen duration of tracheal intubation.
Indeed, this could be attributable to increased comfort, fewer
alarms, and a gradual transmission from controlled to spon-
taneous ventilation thus not leading to apnea. Our study was
not designed to test this hypothesis.
In contrast to our study, a significant reduction of time
until tracheal extubation with ASV as compared with
synchronized intermittent mandatory ventilation/pressure
support was found in patients after fast-track cardiotho-
racic surgery.5 In this trial by Sulzer et al., initially ASV was
set at 100%-minute ventilation (phase 1). When spontane-
ous breathing occurred, %-minute ventilation was reduced
by 50% (phase 2), and if necessary again by 50% (phase 3).
This weaning approach can be described as much more
aggressive with respect to de-escalation, compared with
our study. We chose to de-escalate stepwise until %-minute
ventilation was 70%, as suggested on the Web site of the
manufacturer.¶ Our stepwise approach and the minimum
level of %-minute ventilation may have been a flaw.
¶Hamilton Medical. Available at: http://www.hamilton-medical.com/. Ac-
cessed May 14, 2009.

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 965

Adaptive Support Ventilation Protocolized with De-Escalation After Cardiothoracic Surgery
In a more recent study in patients after fast-track cardio-
thoracic surgery, time to extubation was also significantly
shorter with ASV as compared with pressure-regulated vol-
ume controlled with automode ventilation.4 In this trial by
Gruber et al., weaning consisted of 3 phases: controlled
ventilation (phase 1), assisted ventilation (phase 2), followed
by a T-piece trial (phase 3) that ended with extubation. This
trial did not use de-escalation. An important similarity be-
tween the trials by Sulzer et al. and Gruber et al., and in
contrast to our trial, is that the former trials were both
performed in fast-track cardiothoracic surgery patients,
whereas we explicitly included nonfast-track patients.
There were no differences in arterial blood gas variables,
including arterial pH and Pco2, which could be explained
by the fact that patients in the ASV-DE group could adjust
their minute ventilation when operator set %-minute ven-
tilation was decreased (resulting in higher actual minute
ventilation than the operator set %-minute ventilation).
Indeed, time until the first assisted breathing period was
shorter, and more assisted ventilation episodes were found
in the ASV-DE group. It is also important to note that the
highest levels of arterial Pco2 were not different between
the 2 randomization groups, indicating that ASV-DE is at
least as safe as standard ASV.
There are multiple reasons why we were unable to show
a difference between standard ASV and ASV-DE, including
standard of care in our setting and type of patients studied.
In a study comparing SmartCare® (a knowledge-based
weaning tool including an automatic gradual reduction of
pressure support, automatic performance of spontaneous
breathing trials, and generation of an incentive message
when a breathing trial was successfully passed) with con-
ventional weaning, Rose et al.11 found no differences
regarding duration of MV. This finding was in sharp
contrast to the results from a study by Lellouche et al.12
showing SmartCare to significantly reduce duration of MV.
One important difference, however, between the control
groups of the 2 studies was that duration of MV was
shorter in the study by Rose et al. The shorter duration of
MV could have masked any beneficial effect of the interven-
tion in the first study, whereas it allowed for an important
effect of the intervention in the second study. Differences in
duration of MV could have resulted from differences in
patient case mix and differences in standard care surrounding
the studied patient populations in these 2 studies. We may
have encountered a similar problem: in our study, duration of
tracheal intubation was rather long compared with other trials
of ASV. This may very well relate to the fact that we included
nonfast-track patients instead of most other studies of wean-
ing of cardiothoracic surgery patients. The rather long dura-
tion of MV may have precluded any effect of ASV-DE over
standard ASV in our study.
Because this was an open-label, i.e., not blinded, ran-
domized clinical trial, we could not exclude the possibility
that patients randomized to the standard ASV group also
benefited from early de-escalation. This could have mini-
mized contrast between the study groups. To promote
protocol adherence, nurses and physicians were able to
assess only 1 of the 2 flowcharts (as presented in Fig. 1), for
ASV-DE or standard ASV, depending on randomization
group. We could not always prevent both flowcharts being
966 www.anesthesia-analgesia.org
available in the unit. Notably, however, we noticed a signifi-
cant difference between the study groups regarding %-minute
ventilation at the time of extubation and the period of assisted
ventilation leading to tracheal extubation.
Apart from the fact that this was a single center study,
which limits the generalizability of our conclusions, there
are other limitations of the study. Sedation and analgesia
requirements were not reported using specific scales. How-
ever, we did not gradually wean patients off of sedation,
but stopped infusion of sedation completely when the core
temperature was ⬎35°C. Of note, sedation and analgesic
requirements were the same in the 2 study groups. Another
limitation is that we excluded patients with chronic ob-
structive pulmonary disease, also limiting generalizability.
Compared with a previous study of ASV by our group,
overall weaning time in the present study was considerably
shorter. Indeed, median time to tracheal extubation was
16.4 (12.5–20.8) hours in the previous study.3 The question
must be raised whether tracheal extubation of cardiothoracic
surgery patients is dependent on the ventilatory strategy
alone, or (also) on factors independent of the ventilation
strategy. The above-mentioned studies by Gruber et al.4 and
Sulzer et al.5 certainly show that the ventilator strategy
influences weaning time in these patients. One ventilatory
strategy factor that could have influenced weaning time in our
studies was the use of different levels of PEEP. Specifically, in
our previous study of ASV, patients received 10 cm H2O
PEEP in the first 4 hours after arrival in the ICU, and
thereafter 5 cm H2O PEEP until tracheal extubation. In the
present study, patients received 5 cm H2O PEEP throughout
the complete weaning phase. This extra step in the weaning
process could have accelerated weaning in the present study.
In addition, this change in practice could have resulted in a
change in use of sedatives because we continued sedatives for
at least the first 4 hours, or as long as the higher level of PEEP
was used, in the first study. This usually took longer than the
time needed to reach a core temperature ⬎35°C, which was
the time to stop sedatives in the present study. Factors
independent of ventilation strategy could also have a role.
Both the surgical/anesthesiological team and the ICU team
gained experience over time, whereas the local guidelines
(apart from the advice on PEEP) of these teams as well as their
composition did not change. Better understanding of the
needs of patients after cardiothoracic surgery could have led
to the use of less sedatives both intra- and postoperatively
despite the fact that no formal protocol changes were
implemented.13 Also, more experience with ASV in this
particular patient group could have led to more confidence
in earlier tracheal extubation in our department. Finally,
better awareness of long weaning times in our institution
could have led to a more proactive behavior with regard to
tracheal extubation.14 Although we performed a random-
ized controlled trial, and as such all these factors should not
have affected the primary outcome differently in the 2
study arms, one could certainly suggest that other factors
than the ventilatory strategy have key roles in time until
tracheal extubation in these patients.
In conclusion, compared with standard ASV, weaning of
patients after nonfast-track CABG using ASV with proto-
colized de-escalation and escalation does not shorten time
to tracheal extubation.
ANESTHESIA & ANALGESIA
AUTHOR CONTRIBUTIONS
DAD helped design and conduct the study, analyze the data,
and write the manuscript. This author has seen the original
study data, reviewed the analysis of the data, approved the
final manuscript, and is the author responsible for archiving
the study files. DPV helped conduct the study, analyze the
data, and write the manuscript. This author has seen the
original study data, reviewed the analysis of the data, and
approved the final manuscript. JMB helped design the study,
analyze the data, and write the manuscript. This author has
seen the original study data, reviewed the analysis of the data,
and approved the final manuscript. BAJMdM helped design
the study. This author has seen the original study data and
approved the final manuscript. AK helped conduct the study
and analyze the data. This author has seen the original study
data, reviewed the analysis of the data, and approved the final
manuscript. FP helped conduct the study. This author has seen
the original study data and approved the final manuscript. MJS
helped design the study, analyze the data, and write the
manuscript. This author has seen the original study data,
reviewed the analysis of the data, and approved the final
manuscript.
REFERENCES
1. Brunner JX, Iotti GA. Adaptive support ventilation (ASV).
Minerva Anestesiol 2002;68:365– 8
2. Arnal JM, Wysocki M, Nafati C, Donati S, Graniet I, Corno G,
Durrand-Gasselin J. Automatic selection of breathing pattern
using adaptive support ventilation. Intensive Care Med
2008;34:75– 81
3. Dongelmans DA, Veelo DP, Paulus F, de Mol BA, Korevaar JC,
Kudoga A, Middelhoek P, Binnekade JM, Schultz MJ. Weaning
automation with adaptive support ventilation: a randomized
controlled trial in cardiothoracic surgery patients. Anesth
Analg 2009;108:565–71
4. Gruber PC, Gomersall CD, Leung P, Joynt GM, Ng SK, Ho KM,
Underwood MJ. Randomized controlled trial comparing
adaptive-support ventilation with pressure-regulated volume-
controlled ventilation with automode in weaning patients after
cardiac surgery. Anesthesiology 2008;109:81–7
October 2010 • Volume 111 • Number 4
5. Sulzer CF, Chiolero R, Chassot PG, Mueller XM, Revelly JP.
Adaptive support ventilation for fast tracheal extubation after
cardiac surgery: a randomized controlled study. Anesthesiol-
ogy 2001;95:1339 – 45
6. Santiago TV, Edelman NH. Opioids and breathing. J Appl
Physiol 1985;59:1675– 85
7. Reisine T, Pasternak GW. Opioid analgesics and antagonists. In
Hardman JG, Limbird LE, eds. Goodman & Gilman’s The
Pharmacological Basis of Therapeutics. 9th ed. New York:
McGraw-Hill, 1996:521–55
8. Wilson WC, Smedira NG, Fink C, McDowell JA, Luce JM.
Ordering and administration of sedatives and analgesics dur-
ing the withholding and withdrawal of life support from
critically ill patients. JAMA 1992;267:949 –53
9. Schultz MJ, Haitsma JJ, Slutsky AS, Gajic O. What tidal
volumes should be used in patients without acute lung injury?
Anesthesiology 2007;106:1226 –3
10. Levine S, Nguyen T, Taylor N, Friscia ME, Budak MT, Rothen-
berg P, Zhu J, Sachdeva R, Sonnad S, Kaiser LR, Rubinstein
NA, Powers SK, Shrager JB. Rapid disuse atrophy of dia-
phragm fibers in mechanically ventilated humans. N Engl
J Med 2008;358:1327–35
11. Rose L, Presneill JJ, Johnston L, Cade JF. A randomised,
controlled trial of conventional versus automated weaning
from mechanical ventilation using SmartCare/PS. Intensive
Care Med 2008;34:1788 –95
12. Lellouche F, Mancebo J, Jolliet P, Roeseler J, Shortgen F, Dojat
M, Cabello B, Bouadma L, Rodriguez P, Maggiore S, Reynaert
M, Mersmann S, Brochard L. A multicenter randomized trial of
computer-driven protocolized weaning from mechanical ven-
tilation. Am J Respir Crit Care Med 2006;174:894 –900
13. De Hert SG, Van der Linden PJ, Cromheecke S, Meeus R, ten
Broecke PW, De Blier IG, Stockman BA, Rodrigus IE. Anesthe-
siology 2004;101:9 –20
14. Hawkes CA, Dhileepan S, Foxcroft D. Early extubation for
adult cardiac surgical patients. Cochrane Database Syst Rev
2003;4:CD003587
www.anesthesia-analgesia.org 967
Lung Recruitment and Positive End-Expiratory
Pressure Have Different Effects on CO2 Elimination in
Healthy and Sick Lungs
Gerardo Tusman, MD,* Stephan H. Bohm, MD,† Fernando Suarez-Sipmann, PhD,‡
Adriana Scandurra, Eng,§ and Göran Hedenstierna, PhD储

BACKGROUND: We studied the effects that the lung recruitment maneuver (RM) and positive
end-expiratory pressure (PEEP) have on the elimination of CO2 per breath (VTCO2,br).
METHODS: In 7 healthy and 7 lung-lavaged pigs at constant ventilation, PEEP was increased
from 0 to 18 cm H2O and then decreased to 0 in steps of 6 cm H2O every 10 minutes. Cycling
RMs with plateau pressure/PEEP of 40/20 (healthy) and 50/25 (lavaged) cm H2O were applied
for 2 minutes between 18-PEEP steps. Volumetric capnography, respiratory mechanics, blood
gas, and hemodynamic data were recorded.
RESULTS: In healthy lungs before the RM, VTCO2,br was inversely proportional to PEEP decreasing
from 4.0 (3.6 – 4.4) mL (median and interquartile range) at 0-PEEP to 3.1 (2.8 –3.4) mL at
18-PEEP (P ⬍ 0.05). After the RM, VTCO2,br increased from 3.3 (3–3.6) mL at 18-PEEP to 4.0
(3.5– 4.5) mL at 0-PEEP (P ⬍ 0.05). In lavaged lungs before the RM, VTCO2,br increased initially
from 2.0 (1.7–2.3) mL at 0-PEEP to 2.6 (2.2–3) mL at 12-PEEP (P ⬍ 0.05) but then decreased
to 2.4 (2–2.8) mL when PEEP was increased further to 18 cm H2O (P ⬍ 0.05). After the RM, the
highest VTCO2,br of 2.9 (2.1–3.7) mL was observed at 12-PEEP and then decreased to 2.5
(1.9 –3.1) mL at 0-PEEP (P ⬍ 0.05). VTCO2,br was directly related to changes in lung perfusion, the
area of gas exchange, and alveolar ventilation but inversely related to changes in dead space.
CONCLUSIONS: CO2 elimination by the lungs was dependent on PEEP and recruitment and showed
major differences between healthy and lavaged lungs. (Anesth Analg 2010;111:968 –77)

T he effect of positive end-expiratory pressure (PEEP)

on CO2 kinetics has been described. PEEP, at con-
stant ventilation and body metabolism, is related to a
decrement in the elimination of CO2 by the lungs because
of several factors: (1) a decrement in CO2 transport to the
lungs by a decrease in venous return and thus cardiac
output (CO),1–3 (2) a transient decrease in expired tidal
volume (Vt) caused by the sequential accumulation of air
within the lungs right after the increase in PEEP,3,4 (3) a
gain in functional residual capacity (FRC) leading to a
filling of the lungs with inspiratory gases free of CO2
thereby inducing a transient dilution of alveolar CO2,3–5
and (4) an increase in airway and alveolar dead space
causing a decrease in alveolar ventilation (V̇a).6,7
The collapse of healthy anesthetized and acutely injured
lungs of patients is well described and the main ventilatory
treatment of such collapse conditions is built upon
PEEP.8 –11 PEEP and low Vt ventilation are parts of a
lung-protective ventilatory strategy that aims to minimize
the injury caused by tidal recruitment and overdisten-
sion.12,13 However, because PEEP has been related to the
retention of CO2 within the body,14 –17 this protective
ventilatory strategy could lead to hypercapnia, especially in
the context of low Vt ventilation.12
Author affiliations are provided at the end of the article.
Accepted for publication June 22, 2010.
Disclosure: The authors report no conflicts of interest.
Reprints will not be available from the author.
Address correspondence to Gerardo Tusman, MD, Department of Anesthe-
siology, Hospital Privado de Comunidad, Mar del Plata, Argentina. Address
e-mail to gtusman@hotmail.com.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181f0c2da
We have observed in patients that after a lung recruitment
maneuver (RM), a ventilatory intervention aimed at restoring
pulmonary aeration, CO2 elimination increased despite the
use of high PEEP levels and low Vt values.18 –20 These results
contradict the classical understanding of the effects that PEEP
seems to have on CO2 kinetics.3–5 To our knowledge, a
systematic analysis of the elimination of CO2 during changes
in PEEP combined with a lung RM has not been performed.
These ventilatory interventions are likely to show different
responses in healthy and sick lungs because of the patho-
physiology of acute lung injury resulting from massive lung
collapse, lung edema, and tissue inflammation.
Therefore, the aim of this study was to describe such
changes in the elimination of CO2 in animals with healthy and
surfactant-depleted lungs and to determine whether lung
recruitment and PEEP induced retention of CO2 within the
blood.
METHODS
After approval by the Animal Research Committee of Upp-
sala University in Sweden, 14 Swedish mixed country breed
pigs (body weight ⫽ 24.5 ⫾ 3 kg) were anesthetized with IV
ketamine 25 to 50 mg/kg/h, midazolam 90 to 180 ␮g/kg/h,
fentanyl 3 to 6 ␮g/kg/h, and pancuronium 0.25 to 0.50
mg/kg/h. The trachea was intubated with a 7-mm inner
diameter cuffed endotracheal tube and air leaks were identi-
fied by incomplete flow-volume loops or changes in the
capnograms. The lungs were ventilated with a SERVO-i
(Maquet Critical Care, Wayne, NJ) using a volume control
mode with a Vt of 6 mL/kg, respiratory rate of 30
breaths/min, I:E ratio of 1:2, a fraction of inspired oxygen of 1,
and initially without PEEP. Intravenous saline solution was
maintained at a fixed rate of 4 mL/kg during the study. Body

968 www.anesthesia-analgesia.org October 2010 • Volume 111 • Number 4

temperature was maintained by a warm blanket to keep rectal
temperature within a range of 37.5°C ⫾ 0.5°C.
CO2 Data
Volumetric capnography was recorded on-line using the
NICO capnograph (Respironics, Wallingford, CT). The air-
way flow and CO2 mainstream sensor were placed at the “Y”
piece of the ventilator circuit and delivered data into a
custom-made MatLab program (MathWorks, Natick, MA),
which constructed volumetric capnograms by a Levenberg-
Marquardt fitting method, and all capnography-derived pa-
rameters were calculated from this mathematical function.21
The Vtco2,br is the amount of CO2 eliminated in 1 breath
obtained by integrating expired airway flow and Pco2 signals.
Petco2 is the partial pressure of CO2 at the end of
expiration and Peco2 is the mixed partial pressure of CO2
in 1 breath. Airway dead space was calculated as the
inflection point of the capnogram, which defines the
airway-alveolar interface or the limit between Vdaw and
the alveolar Vt (Vtalv).21
Dead space to Vt ratio (Vd/Vt) is an index that deter-
mines the global inefficiency of ventilation. It was calcu-
lated using the Bohr-Enghoff formula22:
Vd/Vt ⴝ (Paco2 ⴚ Peco2)/Paco2
The mean value of 10 slopes of phase III (SIII) was
calculated as previously described.21 SIII is a qualitative and
noninvasive marker of the global ventilation-perfusion
(V̇/Q̇) ratio, where low values are related to normal V̇/Q̇
ratios whereas high values are indicators of an increased
dispersion of V̇/Q̇.23–26
Respiratory Mechanics Data
Respiratory mechanics data were recorded using the flow
and pressure sensors of the same NICO capnograph. V̇a is
the effective portion of ventilation and is defined as the
product of Vtalv and respiratory rate.
Respiratory system dynamic compliance (Crs) was cal-
culated as ⌬Vt/⌬Paw. Changes in FRC (⌬FRC) induced by
PEEP were calculated by the following formula3:
⌬FRC ⴝ 冘
i⫽n
i⫽1
Vt(0) ⴚ Vt(i)
where Vt(0) was the reference exhaled Vt and n was the
number of breaths with decreased Vt [Vt(i)] after a PEEP
increase or increased Vt [Vt(i)] after a PEEP decrease,
respectively.
Gas Exchange Data
Arterial blood gases were monitored on-line using the
multiparameter intraarterial sensor TrendCare (Diametrics
Medical Ltd., High Newcombe, UK) inserted into the right
carotid artery. Independent arterial and mixed venous
blood gas samples were extracted at each protocol step and
analyzed within 5 minutes using an ABL 300 (Radiometer,
Copenhagen, Denmark).
Pao2, Paco2, and Pvco2 are the Po2 and CO2 in the
arterial and venous blood, respectively. The Pa-etco2 is the
difference between arterial and end-tidal partial pressure of
CO2 representing the area for eliminating CO2 through the
lungs where a low difference corresponds to a large area for
October 2010 • Volume 111 • Number 4
exchange with a better diffusion of CO2 and vice versa.20
Pv-aco2 is the gradient between venous and arterial CO2.
Shunt was calculated using a standard formula.27
Hemodynamic Data
Electrocardiogram and pulse oxymetry were recorded, and a
catheter for mean arterial blood pressure measurement was
placed in the femoral artery. A 7.5F pulmonary artery catheter
CCOmbo (Edwards Lifesciences, Irvine, CA) placed into the
right jugular vein provided continuous CO and pulmonary
pressures. CO was then subdivided into (1) an ineffective
shunting part (COSHUNT), calculated as the product between
shunt and CO to get the absolute value in L/min, and (2) an
effective pulmonary perfusion part (COEPP) or the portion of
the CO that participates in CO2 exchange calculated by
subtracting the COSHUNT value from CO. This last parameter
was used to represent the effect of pulmonary blood flow on
Vtco2,br during the protocol.
Protocol
Animals were randomly assigned to 2 groups: healthy (n ⫽
7) or surfactant-depleted lungs (n ⫽ 7). Lung lavages with
35 mL/kg warm isotonic saline solution were performed in
the lavaged lung group.28 Lavages were repeated every 5
minutes until Pao2 stabilized between 100 to 150 mm Hg at
pure oxygen and PEEP of 8 cm H2O.
The protocol consisted of 3 sequential parts:
1. An increasing PEEP limb, where PEEP was increased
in steps of 6 cm H2O from 0 to 18 cm H2O using a
volume control mode of ventilation. These data rep-
resent the isolated effect that PEEP would have on
the elimination of CO2.
2. An RM, which consisted of a 2-minute cycling RM in
pressure control ventilation using 40/20 cm H2O in
healthy lungs29 or 50/25 cm H2O in sick lungs30 for
plateau pressure and PEEP, respectively.
3. A decreasing PEEP limb, as a mere mirror image of part
1 of the protocol where PEEP was decreased from 18 to
0 PEEP in steps of 6 cm H2O. These data represent the
cumulative effect that PEEP in combination with a prior
RM would have on CO2 elimination.
Each level of PEEP in parts 1 and 3 was maintained for 10
minutes because previous publications3–5 and our own
results from pilot studies showed that ⬎90% of all changes
in Vtco2,br caused by PEEP occurred within this timeframe.
Data Analysis
Before starting the protocol, in vitro and in vivo calibrations of
devices were performed following the manufacturer’s guides.
Hemodynamic and on-line blood gas data were stored in a
laptop by an acquisition system programmed in LabView
(National Instruments, Austin, TX) while CO2 and respiratory
data were recorded in another laptop using the dedicated
software Aplus (Respironics, Wallingford, CT). Both laptops
were synchronized in time. CO2, respiratory mechanics, and
hemodynamic data belonging to the last minute of each PEEP
step (30 breaths ⫽ 30 data points), including the blood gas
samples taken at this time, were analyzed.
A descriptive statistical analysis was performed using the
MatLab program. Lilliefors test determined a non-Gaussian
www.anesthesia-analgesia.org 969
CO2 Elimination by Lung Recruitment and PEEP

Table 1. Ventilation-Related Data During the Protocol in Healthy Animals

PEEP (cm H2O)
Increasing limb of PEEP
0 6 12 18
VT (mL) 196 (184–208) 197 (187–207) 192 (180–204) 188 (165–211)
V̇A (L) 3.5 (3.1–3.9) 3.3 (2.8–3.8) 3.0 (2.6–3.4) 2.7 (2.3–3.1)
Pplat (cm H2O) 12 (10–14) 18 (17–19) 26 (25–27) 33 (31–35)
⌬FRC (mL) 96 (55–137) 279 (213–342) 570 (497–643)
Crs (mL/cm H2O) 17 (14–20) 16 (14–18) 15 (14–16) 15 (14–16)
VD/VT 0.43 (0.41–0.45) 0.49 (0.45–54) 0.48 (0.44–0.52) 0.52 (0.48–56)
SIII (mm Hg/mL) 0.022 (0.012–0.032) 0.018 (0.008–0.028) 0.023 (0.001–0.044) 0.036 (0.010–0.062)
pH 7.44 (0.43–0.45) 7.43 (0.38–0.48) 7.43 (0.37–49) 7.40 (0.33–0.47)
PaO2 (mm Hg) 496 (424–568) 483 (421–544) 514 (483–545) 552 (536–568)
Pa–ETCO2 (mm Hg) 3 (2–4) 2 (1–3) 1 (0–2) 1 (0–2)
Pv៮ –aCO2 (mm Hg) 9 (5–13) 9 (6–12) 12 (9–15) 13 (5–21)
Values are presented in median and interquartile range. Comparison of the same level of positive end-expiratory pressure (PEEP) before (increasing limb) and after
(decreasing limb) a lung recruitment maneuver.
VT ⫽ expired tidal volume; V̇A ⫽ alveolar ventilation; Pplat ⫽ plateau pressure; ⌬FRC ⫽ change in functional residual capacity; Crs ⫽ dynamic respiratory compliance;
VD/VT ⫽ ratio between physiological dead space and tidal volume; SIII ⫽ slope of phase III; PaO2 ⫽ arterial partial pressure of oxygen; Pa-ETCO2 ⫽ difference between
arterial and end-tidal partial pressure of carbon dioxide; Pv៮ -aCO2 ⫽ difference between mixed venous and arterial partial pressure of carbon dioxide.
* P ⬍ 0.05.

Table 2. Ventilation-Related Data During the Protocol in Surfactant-Depleted Animals

PEEP (cm H2O)
Increasing limb of PEEP
0 6 12 18
VT (mL) 176 (152–200) 177 (161–183) 175 (150–190) 172 (158–186)
V̇A (L) 3.0 (2.4–3.6) 2.6 (2.1–3.1) 2.5 (2–3) 2.3 (1.8–2.8)
Pplat (cm H2O) 27 (25–29) 26 (23–29) 27 (26–28) 32 (30–34)
⌬FRC (mL) 92 (76–108) 246 (206–286) 457 (406–508)
Crs (mL/cm H2O) 7 (6–8) 11 (9–11) 16 (14–18) 15 (12–18)
VD/VT 0.73 (0.62–0.84) 0.64 (0.54–0.74) 0.58 (0.44–0.72) 0.58 (0.53–0.63)
SIII (mm Hg/mL) 0.090 (0.070–0.110) 0.072 (0.047–0.097) 0.039 (0.013–0.065) 0.067 (0.027–0.107)
pH 7.29 (0.26–0.32) 7.28 (0.24–0.32) 7.34 (0.30–0.38) 7.36 (0.31–0.41)
PaO2 (mm Hg) 40 (18–62) 93 (32–154) 240 (155–325) 494 (421–565)
Pa-ETCO2 (mm Hg) 33 (25–41) 13 (7–19) 6 (2–10) 4 (2–6)
Pv៮ -aCO2 (mm Hg) 7 (4–10) 8 (6–10) 10 (7–13) 12 (10–14)
Values are presented in median and interquartile range. Comparison of the same level of positive end-expiratory pressure (PEEP) before (increasing limb) and after
(decreasing limb) a lung recruitment maneuver.
VT ⫽ expired tidal volume; V̇A ⫽ alveolar ventilation; Pplat ⫽ plateau pressure; ⌬FRC ⫽ change in functional residual capacity; Crs ⫽ dynamic respiratory compliance;
VD/VT ⫽ ratio between physiological dead space and tidal volume; SIII ⫽ slope of phase III; PaO2 ⫽ arterial partial pressure of oxygen; Pa-ETCO2 ⫽ difference between
arterial and end-tidal partial pressure of carbon dioxide; Pv៮ -aCO2 ⫽ difference between mixed venous and arterial partial pressure of carbon dioxide.
* P ⬍ 0.05.

distribution of the data. Friedman nonparametric test was
used to compare the results of the same level of PEEP before
with those after RM in a 2-way direction. The same test was
used to compare differences between results of consecutive
levels of PEEP. Values are expressed as median (interquartile
range) and P values ⬍0.05 were considered significant.
RESULTS
All pigs completed the protocol successfully. Absolute values
of the main variables belonging to the last minute for each
PEEP step are presented in Tables 1 to 3. In general, PEEP
applied after lung recruitment improved lung function when
compared with PEEP alone in both healthy and lavaged
lungs. The recruitment effect was characterized by a gain in
⌬FRC, an increase in Crs, and decreases in Vd/Vt and shunt,
paralleled by improvements in gas exchange.
In healthy pigs before recruitment, Vtco2,br decreased
from 4.0 (3.6 – 4.4) mL (0-PEEP) to 3.1 (2.8 –3.4) mL (18-
PEEP, P ⬍ 0.05) as PEEP increased (Fig. 1). Figure 2 shows
that the relative changes in the elimination of CO2 with
increasing PEEP levels were mainly related to a decrease in
the efficacy of ventilation (⬍V̇a) and the decrease in COEPP.
The area for eliminating CO2 (Pa-etco2) showed a small
increase with increasing PEEP levels but with little effect on
CO2 elimination. Dead space (Vd/Vt) and SIII increased
proportionally to PEEP (Table 1).
After recruitment, Vtco2,br increased from 3.3 (3–3.6)
mL (18-PEEP) to 4.0 (3.5– 4.5) mL (0-PEEP, P ⬍ 0.05) as
PEEP was reduced. This increased CO2 elimination was
associated with an increase in V̇a and COEPP. Initially,
Pa-etco2 decreased when going from 18 to 12 cm H2O of
PEEP, but progressively increased again when going fur-
ther down to 0-PEEP (Table 1).
Vtco2,br presented a different behavior in lavaged ani-
mals (Figs. 1 and 2). Before recruitment, Vtco2,br initially
increased from 2.0 (1.7–2.3) mL (0-PEEP) to 2.6 (2.2–3) mL
(12-PEEP) (P ⬍ 0.05). This increment in CO2 elimination
went along with an increased COEPP and a decreased

970 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

 Table 1. (Continued)
PEEP (cm H2O)
Decreasing limb of PEEP
18 12 6 0
191 (169–213) 195 (175–215) 200 (178–222) 201 (179–223)
2.9 (2.5–3.3) 3.0 (2.5–3.5) 3.6 (3.2–4.0)* 3.9 (3.4–4.4)*
29 (27–31) 19 (18–20)* 13 (12–14)* 9.6 (7–12)
⫺628 (⫺597 to 659)* ⫺401 (⫺370 to 432)* ⫺406 (⫺371 to 441)* ⫺181 (⫺151 to 211)
18 (16–20) 30 (27–33)* 31 (28–34)* 21 (18–24)*
0.51 (0.48–0.54) 0.44 (0.39–49)* 0.41 (0.38–0.44)* 0.42 (0.39–45)
0.031 (0.006–0.056)* 0.033 (0.023–0.043)* 0.014 (0.009–0.019)* 0.023 (0.010–0.036)
7.39 (0.32–0.46) 7.42 (0.37–0.47) 7.43 (0.39–47) 7.42 (0.38–0.47)
591 (570–612)* 576 (553–599)* 556 (540–572)* 476 (424–528)
1 (0–2) 0.6 (0.4–1)* 1 (0.7–2.1)* 5 (3–7)*
13 (8–17) 15 (7–22)* 12 (8–16) 11 (8–14)*

Table 2. (Continued)
PEEP (cm H2O)
Decreasing limb of PEEP
18 12 6 0
171 (158–184) 172 (149–185) 174 (150–198) 177 (153–201)
2.4 (1.4–3.4) 2.8 (2.2–3.4)* 2.8 (2.3–3.3)* 2.9 (2.3–3.5)
29 (27–31)* 22 (19–25)* 20 (18–22)* 22 (19–25)*
⫺607 (⫺541 to 673)* ⫺377 (⫺354 to 400)* ⫺384 (⫺353 to 415)* ⫺185 (⫺165 to 205)
19 (15–23)* 23 (17–29)* 15 (12–18)* 9 (8–10)
0.56 (0.50–0.62) 0.56 (0.48–0.64) 0.59 (0.56–0.62)* 0.63 (0.57–0.69)*
0.035 (0.010–0.069)* 0.024 (0.004–0.048)* 0.031 (0.011–0.051)* 0.050 (0.028–0.072)*
7.40 (0.32–0.48)* 7.40 (0.33–0.47)* 7.34 (0.23–45)* 7.25 (0.20–0.30)
527 (501–553)* 399 (354–444)* 179 (90–268)* 80 (50–110)*
2 (1–3)* 6 (3–9) 9 (3–15)* 17 (13–21)*
14 (12–16) 10 (9–11) 9 (8–10) 7 (3–11)

Pa-etco2. Vtco2,br then decreased from 2.6 (2.2–3) mL
(12-PEEP) to 2.4 (2–2.8) mL (18-PEEP) (P ⬍ 0.05). This time,
the impairment of CO2 elimination was associated with a
reduced V̇a and COEPP despite Pa-etco2 showing the
lowest value of the increasing limb of PEEP.
After recruitment, the highest Vtco2,br was observed at
12-PEEP (2.9 [2.1–3.7] mL, P ⬍ 0.05) which decreased to 2.5
(1.9 –3.1) mL at 0-PEEP (P ⬍ 0.05). The progressive decre-
ments in the area for CO2 exchange and in COEPP were
associated with a lower elimination of CO2 at 0-PEEP. Vd/Vt
and SIII increased with reductions in PEEP (Table 2).
Figure 3A represents the elimination of CO2 over time for
a single step change of PEEP from 6 to 12 cm H2O before
recruitment. Compared with 6-PEEP, the change in median
Vtco2,br was ⫺6% in healthy lungs and ⫹8% in lavaged lungs
at 12-PEEP (both P ⬍ 0.05). In both healthy and surfactant-
depleted animals, Vtco2,br decreased in the first breaths after
the PEEP increase with ⬎90% of the effect occurring within 5
minutes at the higher PEEP.
Figure 3B shows the effect of a PEEP change from 6- to
12-PEEP before recruitment on the main variables for 18
consecutive respiratory cycles. The effects were qualita-
tively similar, but quantitatively different in healthy and
in lavaged lungs. Vtco2,br decreased almost to zero in
the first breath but recovered within 5 to 6 consecutive
breaths in both healthy and lavaged lungs. This change
in Vtco2,br was related to a decrease in expired Vt right
after the PEEP change resulting in an expansion of ⌬FRC
by approximately 183 mL in healthy and 154 mL in
lavaged lungs. As opposed to Vtco2,br, CO was little
affected by the change in PEEP.
Pvco2, Paco2, and Petco2 are presented in Figure 1 and
Tables 1 and 2. In healthy lungs, no clinically significant
changes in those variables were observed. In surfactant-
depleted lungs, however, the difference between Paco2 and
Petco2 narrowed significantly but in an inverse relation to
the applied PEEP before and after lung recruitment with

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 971

CO2 Elimination by Lung Recruitment and PEEP

Table 3. Hemodynamic Data in Healthy and Surfactant-Depleted Animals

PEEP (cm H2O)
Increasing limb of PEEP
0 6 12 18
Healthy lungs
HR (bpm) 95 (68–122) 86 (48–124) 94 (66–122) 90 (68–112)
MAP (mm Hg) 74 (47–101) 82 (61–103) 69 (49–89) 65 (44–86)
MPAP (mm Hg) 21 (15–27) 25 (17–33) 25 (19–31) 28 (21–35)
CVP (mm Hg) 7 (5–9) 8 (6–10) 9 (5–13) 11 (7–15)
CO (L/min) 2.60 (2–3.2) 2.60 (2.1–3.1) 2.10 (1.6–2.6) 2.10 (1.4–2.8)
COEPP (L/min) 2.03 (1.1–3) 2.42 (2–2.8) 1.93 (1.3–2.5) 1.93 (1.2–2.6)
COSHUNT (L/min) 0.31 (0.1–0.5) 0.29 (0.1–0.4) 0.17 (0.1–0.3) 0.09 (0.05–0.15)
Lavaged lungs
HR (bpm) 106 (75–137) 104 (71–135) 101 (63–139) 114 (85–143)
MAP (mm Hg) 84 (62–106) 95 (70–120) 85 (65–105) 80 (55–105)
MPAP (mm Hg) 45 (24–66) 38 (29–47) 34 (26–42) 35 (28–42)
CVP (mm Hg) 9 (6–12) 9 (6–12) 9 (5–13) 11 (8–14)
CO (L/min) 5.50 (4.3–6.7) 5.40 (4.2–6.6) 4.31 (3.5–5.1) 4.30 (3.4–5.1)
COEPP (L/min) 3.03 (2–4) 3.08 (2.3–3.8) 3.77 (3.3–4.3) 3.50 (2.9–4.1)
COSHUNT (L/min) 2.48 (2–3) 1.70 (1.2–2.2) 0.84 (0.5–1.1) 0.30 (0.1–0.5)
Values are presented in median and interquartile range. Comparison of the same level of positive end-expiratory pressure (PEEP) before (increasing limb) and after
(decreasing limb) a lung recruitment maneuver.
HR ⫽ heart rate; MAP ⫽ mean systemic arterial pressure; MPAP ⫽ mean pulmonary artery pressure; CVP ⫽ central venous pressure; CO ⫽ cardiac output;
COEPP ⫽ effective part of CO; COSHUNT ⫽ ineffective part of CO.
* P ⬍ 0.05.

corresponding changes in pH. Both Pvco2 and Paco2

decreased with lung recruitment and PEEP.

DISCUSSION
The main findings of this study can be summarized as
follows:
1. Lung recruitment and PEEP have different effects on
CO2 elimination in healthy and surfactant-depleted lungs.
2. At constant metabolism and ventilatory settings, any
change in the elimination of CO2 can be explained by
a combination of changes in (a) the effectiveness of
lung perfusion, (b) the area for CO2 exchange, and (c)
the amount of V̇a and, thus, in the global V̇/Q̇
relationship of the lungs.
3. In healthy and lavaged lungs, changes in PEEP
altered the elimination of CO2 because of immediate
effects on both expired Vt values and ⌬FRC for 5 to
6 consecutive breaths. Stable new values for Vtco2,br,
lung perfusion, area of CO2 exchange, and V̇a were
reached within 5 minutes.
4. In lavaged lungs, the efficacy of CO2 elimination was
directly related to the recruitment/derecruitment ef-
fect. Lung recruitment and PEEP did not retain CO2
in the blood during the study periods.

Effect of PEEP and Lung Recruitment on

Figure 1. The elimination of CO2 per breath and tension-based CO2
values during the protocol. PVCO2 (dots), PaCO2 (squares) and PETCO2
(triangles). White columns in the lower panel are the elimination of
CO2 per breath (VTCO2,br) values. *Comparisons between values of the
same level of positive end-expiratory pressure (PEEP) before and after a
lung recruitment maneuver (RM); P ⬍ 0.05.
CO2 Elimination
Breen and Mazumdar3 and Johnson and Breen5 were the
first to analyze the effect of PEEP on the non–steady-state
CO2 kinetics in healthy lungs. Our main goal was to further
their work and to describe the same effect in the context of
a lung RM. Thus, to our knowledge, the data presented in
the current study are the first to describe the effect of PEEP
and lung recruitments on the elimination of CO2 in
non–steady-state conditions of healthy and surfactant-
depleted lungs.

972 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

 Table 3. (Continued)
PEEP (cm H2O)
Decreasing limb of PEEP
18 12 6 0

92 (67–117) 91 (67–115) 89 (68–102) 92 (64–120)

67 (45–89) 66 (38–94) 86 (69–103) 83 (65–101)*
20 (14–26)* 23 (15–31) 20 (12–28)* 26 (18–34)*
12 (6–18) 10 (7–13) 9 (6–12) 6 (4–8)
1.80 (1.3–2.3)* 2.00 (1–3) 3.10 (2.1–4.1)* 3.70 (2.5–5.2)*
1.63 (1.1–2.1)* 1.93 (0.9–2.9) 2.95 (2–4)* 3.29 (1.3–5.3)*
0.07 (0.05–0.1) 0.11 (0.1–0.2)* 0.19 (0.1–0.3)* 0.47 (0.1–1)*

113 (73–153) 119 (91–149)* 117 (92–142)* 120 (85–155)*

77 (53–104) 80 (59–99)* 87 (65–109)* 85 (65–105)
36 (25–47) 36 (26–46) 40 (32–48) 46 (38–54)
12 (8–16) 11 (9–13) 9 (7–11) 9 (7–11)
3.29 (2–4.4)* 4.00 (2.8–5.2)* 5.00 (3.8–6.2)* 5.80 (4.5–7.2)
2.99 (2.3–3.6)* 3.52 (2.6–4.4) 3.95 (3.1–4.8)* 2.96 (2.3–3.7)
0.22 (0.2–0.3)* 0.48 (0.2–0.8)* 1.07 (0.5–1.5)* 2.56 (2–3.2)

Figure 2. Relative changes in the CO2 elimination per breath (VTCO2,br) during sequential changes of positive end-expiratory pressure
(PEEP). The determinants of VTCO2,br are represented by the effective portion of cardiac output (COEPP), the area for CO2 exchange
(Pa-ETCO2), and the effective portion of ventilation (V̇A). The relative changes in each of the variables as they are induced by the PEEP
change are expressed in percentage, from zero to a maximum cutoff of 50%. An improvement is defined as a change toward more normal
values. Such improvements in CO2 elimination are characterized by increases in V̇A and COEPP and decreases in Pa-ETCO2 and are
represented as white bars above the zero line. Impairment is defined as a change toward more abnormal values. Such impairments in
CO2 elimination are characterized by decreases in V̇A and COEPP and increases in Pa-ETCO2 and are represented by black bars below the
zero line. *P ⬍ 0.05.

The elimination of CO2 by the lungs, at constant venti-
lation and body metabolism, depends on its transport by
the blood into the lungs, its diffusion through the alveolar-
capillary membrane, and its elimination by the V̇a.3,31
Therefore, the final Vtco2,br value measured at the airway
opening in response to a PEEP challenge with or without a
lung RM is the result of a complex interaction of several
factors. Figure 2 and Tables 1 to 3 show that these interactions
differ between healthy and lavaged lungs and, sometimes, the
influence on CO2 elimination of 1 factor differed from other
factors.
Figure 3 shows the change in the elimination of CO2
during an increase in PEEP from 6 to 12 cm H2O without RM.
The layout of this figure is similar to the ones presented
in the articles by Breen and Mazumdar3 and Johnson and
Breen5 to facilitate the reader’s comparison of the data.
According to the results shown in this figure, changes in
Vtco2,br, lung perfusion, CO2 exchange, and V̇a

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 973

CO2 Elimination by Lung Recruitment and PEEP

Figure 3. Time course of CO2 elimination during 2 consecutive positive end-expiratory pressure (PEEP) steps. A, Data of each of the animals is
displayed for two 10-minute periods at 6 and 12 cm H2O of PEEP, respectively. VTCO2,br is the elimination of CO2 per breath, V̇A the alveolar ventilation,
Pa-ETCO2 the arterial to end-tidal difference in PCO2 (calculated by the on-line PaCO2 value from the arterial catheter minus PETCO2 from capnograms),
CO the continuous cardiac output, and ⌬FRC the change in functional residual capacity. B, Baseline is the mean value of the last 10 breaths on 6
cm H2O of PEEP (here summarized in breath 0) followed by 18 consecutive breaths at 12-PEEP.

after a PEEP challenge vary between healthy and lavaged
lungs.
In healthy lungs, 12-PEEP decreased Vtco2,br because of
decreased V̇a and CO despite a slight improvement in gas
exchange (Fig. 3A). More than 90% of the effects on Vtco2,br
induced by 12-PEEP occurred within 5 minutes. These results
are similar to the findings of Breen and Mazumdar3 observed
in healthy dogs and those of Johnson and Breen5 in healthy
anesthetized humans. They also found that a PEEP challenge
decreased Vtco2,br because of a decrement in V̇a and CO,
with most of the recovery of Vtco2,br within 10 minutes.
Differences in the recovery of Vtco2,br observed between
these protocols could be explained by differences in the
species studied, differences in the levels of PEEP applied (11
cm H2O in the study by Breen and Mazumdar and 10 cm H2O
in the study by Johnson and Breen), or by differences in
hemodynamic status and treatments.
In lavaged lungs, Vtco2,br increased until 12-PEEP caused
by an increment in the area for CO2 exchange despite
decreasing V̇a and CO (Fig. 3A). The effects of a partial
alveolar recruitment induced by the increasing PEEP levels
are supported by concomitant increases in Pao2 and compli-
ance at reduced dead spaces (Table 2). Although CO was
reduced by 21%, its COEPP increased by 22% because of the
recruitment of shunt areas (Table 3). The final result was a
better elimination of CO2 because of an improved V̇/Q̇ ratio
as indicated by a decrement in SIII. The different findings in
lavaged and healthy lungs point toward differences in the
physiological effects that PEEP and recruitment have on
normally aerated and on collapsed lungs.
Figure 3B provides a zoomed view of the first breaths after
a change in a PEEP step. The decrement in Vtco2,br after an
increase in PEEP was caused by the trapped gas within lungs
at the onset of the higher PEEP, which diluted alveolar CO2.
The opposite results were observed in cases of PEEP reduc-
tion. After a PEEP change, a fast recovery in Vtco2,br was
found within 5 to 6 consecutive breaths. This finding is similar
to the one described by Johnson and Breen5 in anesthetized

974 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

Figure 3. Continued.
patients in which the recovery of Vtco2,br after a change of
PEEP from 0 to 10 cm H2O took 8 breaths.
Does PEEP and Lung Recruitment Retain CO2 in
the Blood?
A protective ventilatory management with low Vt and
plateau pressures is currently mandatory when treating
acutely injured lungs.12,13 Arterial hypercapnia is thus a
clinically tolerated negative consequence of an intentional
decrease in V̇a. This hypercapnic state could become even
worse if PEEP, as many authors have postulated, caused
CO2 retention in the body.14 –17
October 2010 • Volume 111 • Number 4
In healthy lungs, Vtco2,br decreased in proportion to PEEP
before the recruitment but increased after it. At constant
ventilation and metabolism, this decrease in CO2 elimination
should lead to retention of CO2 within the body when PEEP
is applied without a prior recruitment. However, Pvco2,
Paco2, and thus P៮v-aco2 were not affected much during the
protocol (Table 1 and Figs. 1 and 2), which must be inter-
preted that CO2 was not retained within the blood, at least
during the study period. After recruitment, the elimination of
CO2 and global lung physiology improved at PEEP levels
down to 12 cm H2O (Table 1 and Figs. 1 and 2) with chances
for CO2 retention even lower than before the recruitment.
www.anesthesia-analgesia.org 975
CO2 Elimination by Lung Recruitment and PEEP
Our results differ from those of Breen and Mazumdar3 and
Johnson and Breen5 who found increased Pvco2 and Paco2 in
healthy lungs at 10 and 11 cm H2O of PEEP, respectively.
Differences in the hemodynamic status, protocol time, and
experimental models might explain these differences.
In surfactant-depleted lungs, however, results were
totally different. Pv-aco2 increased whereas Pa-etco2
decreased with positive-pressure ventilation as long as
lung collapse was low and overall lung function pre-
served (Table 2 and Fig. 1). The increased area of gas
exchange leading to an augmented diffusion of CO2
across the alveolar-capillary membrane after a lung
recruitment can explain this effect and was confirmed by
parallel increments in Pao2 and Crs, 2 well-known mark-
ers of lung recruitment.32,33 COEPP increased with in-
creasing airway pressures because of the recruitment of
previously collapsed capillaries in the atelectatic areas
(Table 3). Because Vtco2,br is directly proportional to
lung perfusion, this increment in COEPP facilitates the
transport of CO2 from the body stores toward the lungs
where it is eliminated. Note that the absolute values of
Pvco2 and Paco2 decreased with both lung recruitment
and PEEP, thereby confirming that CO2 was not retained
within the blood but rather eliminated more efficiently.
In lavaged lungs, the highest recruitment pressures
applied in pressure control ventilation resulted in a mean
Vt of 7.4 mL/kg (221 [202–251] mL). This small and
transient increase in minute ventilation could have influ-
enced the elimination of CO2 beyond the actual lung
recruitment effect during the descending PEEP steps.
However, we believe that the decreased values for Pvco2
and Paco2 on the descending limb of the PEEP titration
were mainly attributable to the recruitment of alveoli
because: (1) changes in the area of gas exchange, and not
in V̇a, had a dominating influence on Vtco2,br (Fig. 2),
(2) variables representing the recruitment effect and
which are independent of V̇a (Crs, SIII, ⌬FRC, Pao2, or
shunt) improved after lung recruitment, and (3) a tran-
sient and marginal increase in minute ventilation for 2
minutes only would not have had a lasting impact on
Vtco2,br after a lapse of 10 minutes, the point in time
when the blood samples were taken.
Clinical Implications
In contrast to continuous positive airway pressure ma-
neuvers, cycling RMs have the following advantages: (a)
they are hemodynamically better tolerated,34 (b) the
step-wise and sequential increments in PEEP allow the
gained volumes of air to spread progressively instead of
abruptly throughout the lung parenchyma,35,36 and (c)
they allow real-time monitoring of respiratory variables
on a breath-by-breath basis. Today, the way to conduct
and optimize RM is a topic of much debate, where the
advent of volume-based capnographic monitoring may
add important new arguments in favor of such an
approach because of the noninvasive and real-time na-
ture of this methodology.
Because CO2 kinetics are context sensitive and highly
dependent on the sequence of steps during a cycling RM, in
our protocol, the levels of PEEP were not assigned in
random order. Had we randomized the protocol steps, we
976 www.anesthesia-analgesia.org
would not have been able to show reproducibly the sequen-
tial nature and the time dependence of CO2 elimination.
For the same reason, we chose to study the CO2 kinetics
during a short lapse of time because we were interested in
the CO2 kinetics during the non–steady-state conditions
induced by RMs and a PEEP titration process. This is
crucial new information, which should contribute to a
better understanding of CO2 kinetics, and we hope that it
will finally have clinical implications for the monitoring of
patients during mechanical ventilation.
Limitations
Surfactant-depleted lungs as used in our experimental
study do not adequately represent the complex nature of
acute lung injury in real patients, and thus our results
should be interpreted with caution.
It is well documented that healthy and sick human lungs
have different opening pressures29,30and therefore we
chose our recruitment pressures in health and disease
assuming that this difference would be true for animals
also. To achieve a complete recruitment effect in our
surfactant-depleted animals, we decided to use an arbitrary
and fixed opening pressure of 50 cm H2O for all animals
with sick lungs based on our own previous experience33
because we did not have access to lung imaging by
computed tomographic scan to determine an optimal open-
ing pressure individually for each animal. Had we used the
same pressure as we did for healthy animals, the risk of
having incompletely recruited diseased lungs and thus
inconclusive study results would have been high.
CONCLUSIONS
The results of this study show that lung recruitment and
PEEP have different effects on the elimination of CO2 in
healthy and lavaged lungs. These differences can be ex-
plained by a complex interaction between the key factors of
lung perfusion, diffusion through the alveolar-capillary
membrane, and V̇a. Our results suggest that sufficiently
high levels of PEEP applied after lung recruitment may
help decrease hypercapnia in patients treated with lung-
protective ventilation and low Vt strategies.
AUTHOR AFFILIATIONS
From the *Department of Anesthesiology, Hospital Privado de
Comunidad, Mar del Plata, Argentina; †CSEM Centre Suisse
d’Electronique et de Microtechnique SA, Research Centre for
Nanomedicine, Landquart, Switzerland; ‡Department of Criti-
cal Care Medicine, Fundación Jiménez Díaz-UTE, Madrid,
Spain; §Bioengineering Laboratory, Electronic Department,
University of Mar del Plata, Mar del Plata, Argentina; and
储Department of Medical Sciences, Clinical Physiology, Univer-
sity Hospital, Uppsala Sweden.
REFERENCES
1. Qvist J, Pontoppidan H, Wilson RS, Lowenstein E, Laver MB.
Hemodynamic response to mechanical ventilation with PEEP:
the effect of hypervolemia. Anesthesiology 1975;42:45–55
2. Berglund JE, Haldén E, Jakonson S, Landelius J. Echocardio-
graphic analysis of cardiac function during high PEEP venti-
lation. Intensive Care Med 1994;20:174 – 80
3. Breen PH, Mazumdar B. How does positive end-expiratory
pressure decrease CO2 elimination from the lung? Respir
Physiol 1996;103:233– 42
ANESTHESIA & ANALGESIA
 4. Elliott WR, Harris AE, Philip JH. Positive end-expiratory
pressure: implications for tidal volume changes in anesthesia
machine ventilation. J Clin Monit 1989;5:100 – 4
5. Johnson JL, Breen PH. How does positive end-expiratory
pressure decrease pulmonary CO2 elimination in anesthetized
patients? Respir Physiol 1999;118:227–36
6. Hedenstierna G, Lundberg S. Airway compliance during arti-
ficial ventilation. Br J Anaesth 1975;47:1277– 81
7. Coffey RL, Albert RK, Robertson HT. Mechanisms of physio-
logical dead space response to PEEP after acute oleic acid lung
injury. J Appl Physiol 1983;55:1550 –7
8. Lundquist H, Hedenstierna G, Strandberg A, Tokics L, Brismar
B. CT-assessment of dependent lung densities in man during
general anaesthesia. Acta Radiol 1995;36:626 –32
9. Brismar B, Hedenstierna G, Lundquist H. Pulmonary densities
during anesthesia with muscular relaxation: a proposal of
atelectasis. Anesthesiology 1985;62:422– 8
10. Gattinoni L, Caironi P, Valenza F, Carlesso E. The role of
CT-scan studies for the diagnosis and therapy of acute respi-
ratory distress syndrome. Clin Chest Med 2006;27:559 –70
11. Ware LB, Matthay MA. The acute respiratory distress syn-
drome. N Engl J Med 2000;342:1334 – 49
12. The National Heart, Lung, and Blood Institute ARDS Clinical
Trials Network. Ventilation with lower tidal volumes as com-
pared with traditional tidal volumes for acute lung injury and
the acute respiratory distress syndrome. The Acute Respira-
tory Distress Syndrome Network. N Engl J Med 2000;
342:1301– 8
13. Amato MBP, Barbas CSV, Medeiros DM, Magaldi RB, Schet-
tino GP, Lorenzi-Filho G, Kairalla RA, Deheinzelin D, Muñoz
C, Oliveira R, Takagaki TY, Carvalho CRR. Effect of a protec-
tive ventilation strategy on mortality in the acute respiratory
distress syndrome. N Engl J Med 1998;338:347–54
14. Dueck R, Wagner PD, West JB. Effects of positive end-
expiratory pressure on gas exchange in dogs with normal and
edematous lungs. Anesthesiology 1977;47:359 – 66
15. Pesenti A, Marcolin R, Prato P, Borelli M, Riboni A, Gattinoni
L. Mean airway pressure vs positive end-expiratory pressure
during mechanical ventilation. Crit Care Med 1985;13:34 –7
16. Isserles SA, Breen PH. Can changes in end-tidal PCO2 measure
changes in cardiac output? Anesth Analg 1991;73:808 –14
17. Hedenstierna G, White FC, Mazzone R, Wagner PD. Redistri-
bution of pulmonary blood flow in the dog with PEEP venti-
lation. J Appl Physiol 1979;46:278 – 87
18. Tusman G, Böhm SH, Suarez-Sipmann F, Turchetto E. Alveolar
recruitment improves ventilatory efficiency of the lungs dur-
ing anesthesia. Can J Anaesth 2004;51:723–7
19. Tusman G, Böhm SH, Suarez-Sipmann F, Maisch S. Lung
recruitment improves the efficiency of ventilation and gas
exchange during one-lung ventilation anesthesia. Anesth
Analg 2004;98:1604 –9
20. Tusman G, Suarez-Sipmann F, Böhm SH, Pech T, Reissmann
H, Meschino G, Scandurra A, Hedenstierna G. Monitoring
dead space during recruitment and PEEP titration in an
experimental model. Intensive Care Med 2006;32:1863–71
21. Tusman G, Scandurra A, Böhm SH, Suarez-Sipmann F, Clara F.
Model fitting of volumetric capnograms improves calculations
of airway dead space and slope of phase III. J Clin Monit
Comput 2009;23:197–206
October 2010 • Volume 111 • Number 4
22. Englhoff H. Volumen inefficax. Bemerkungen zur Frage des
schädlichen Raumes. Uppsala Läkareforen Forhandl 1938;
44:191–218
23. Harris B, Bailey DL, Chicco P, Bailey EA, Roach PJ, King GG.
Objective analysis of whole lung and lobar ventilation/
perfusion relationships in pulmonary embolism. Clin Physiol
Funct Imaging 2008;28:14 –26
24. Blanch LL, Fernandez R, Saura P, Baigorri F, Artigas A.
Relationship between expired capnogram and respiratory sys-
tem resistance in critically ill patients during total ventilatory
support. Eur Respir J 1999;13:1048 –54
25. Hofbrand BI. The expiratory capnogram: a measure of
ventilation-perfusion inequalities. Thorax 1966;21:518 –24
26. Strömberg NO, Gustafsson PM. Ventilatory inhomogeneity
assessed by nitrogen washout and ventilation-perfusion mis-
match by capnography in stable and induced airway obstruc-
tion. Pediatr Pulmonol 2000;29:94 –102
27. Berggren SM. The oxygen deficit of arterial blood caused by
non-ventilated parts of the lungs. Acta Physiol Scand
1942;Suppl 4:4 –9
28. Lachmann B, Jonson B, Lindroth M, Robertson B. Modes of
artificial ventilation in severe respiratory distress syndrome:
lung function and morphology in rabbits after wash-out of
alveolar surfactant. Crit Care Med 1982;10:724 –32
29. Rothen HU, Sporre B, Englberg G, Wegenius G, Hedenstierna
G. Reexpansion of atelectasis during general anaesthesia: a
computed tomography study. Br J Anaesth 1993;71:788 –95
30. Borges JB, Okamoto VN, Matos GF, Caramez MPR, Arantes
PR, Barros F, Souza CE, Victorino JA, Kacmarek RM, Barbas
CSV, Carvalho CRR, Amato MBP. Reversibility of lung col-
lapse and hypoxemia in early acute respiratory distress syn-
drome. Am J Respir Crit Care Med 2006;174:268 –78
31. Cherniack NS, Longobardo GS. Oxygen and carbon dioxide
gas stores of the body. Physiol Rev 1970;50:196 –243
32. Lachmann B. Open up the lung and keep the lung open.
Intensive Care Med 1992;18:319 –21
33. Suarez-Sipmann F, Böhm SH, Tusman G, Pesch T, Thamm O,
Reissmann H, Reske A, Magnusson A, Hedenstierna G. Use of
dynamic compliance for open lung positive end-expiratory
pressure titration in an experimental study. Crit Care Med
2007;35:214 –21
34. Celebi S, Köner O, Menda F, Korkut K, Suzer K, Cakar N. The
pulmonary and hemodynamic effects of two different recruit-
ment maneuvers after cardiac surgery. Anesth Analg
2007;104:384 –90
35. Hickling KG. Best compliance during a decremental, but not
incremental, positive end-expiratory pressure trial is related to
open-lung PEEP: a mathematical model of acute respiratory
distress syndrome lungs. Am J Respir Crit Care Med
2001;163:69 –78
36. Albaiceta GM, Taboada F, Parra D, Luyando LH, Calvo J,
Menendez R, Otero J. Tomographic study of the inflection
points of the pressure-volume curve in acute lung injury. Am J
Respir Crit Care Med 2004;170:1066 –72
www.anesthesia-analgesia.org 977
Society for Obstetric Anesthesia and Perinatology
Section Editor: Cynthia A. Wong

Pitfalls in Chronobiology: A Suggested Analysis Using

Intrathecal Bupivacaine Analgesia as an Example
Steven L. Shafer, MD,* Bjoern Lemmer, MD, PhD,† Emmanuel Boselli, MD, PhD,‡ Fabienne Boiste, MD,‡
Lionel Bouvet, MD,‡ Bernard Allaouchiche, MD, PhD,§ and Dominique Chassard, MD, PhD§

BACKGROUND: The duration of analgesia from epidural administration of local anesthetics to

parturients has been shown to follow a rhythmic pattern according to the time of drug administration.
We studied whether there was a similar pattern after intrathecal administration of bupivacaine in
parturients. In the course of the analysis, we came to believe that some data points coincident with
provider shift changes were influenced by nonbiological, health care system factors, thus incorrectly
suggesting a periodic signal in duration of labor analgesia. We developed graphical and analytical
tools to help assess the influence of individual points on the chronobiological analysis.
METHODS: Women with singleton term pregnancies in vertex presentation, cervical dilation 3 to
5 cm, pain score ⬎50 mm (of 100 mm), and requesting labor analgesia were enrolled in this
study. Patients received 2.5 mg of intrathecal bupivacaine in 2 mL using a combined
spinal-epidural technique. Analgesia duration was the time from intrathecal injection until the
first request for additional analgesia. The duration of analgesia was analyzed by visual inspection
of the data, application of smoothing functions (Supersmoother; LOWESS and LOESS [locally
weighted scatterplot smoothing functions]), analysis of variance, Cosinor (Chronos-Fit), Excel,
and NONMEM (nonlinear mixed effect modeling). Confidence intervals (CIs) were determined by
bootstrap analysis (1000 replications with replacement) using PLT Tools.
RESULTS: Eighty-two women were included in the study. Examination of the raw data using 3
smoothing functions revealed a bimodal pattern, with a peak at approximately 0630 and a
subsequent peak in the afternoon or evening, depending on the smoother. Analysis of variance did
not identify any statistically significant difference between the duration of analgesia when intrathecal
injection was given from midnight to 0600 compared with the duration of analgesia after intrathecal
injection at other times. Chronos-Fit, Excel, and NONMEM produced identical results, with a mean
duration of analgesia of 38.4 minutes (95% CI: 35.4 – 41.6 minutes), an 8-hour periodic waveform
with an amplitude of 5.8 minutes (95% CI: 2.1–10.7 minutes), and a phase offset of 6.5 hours (95%
CI: 5.4 – 8.0 hours) relative to midnight. The 8-hour periodic model did not reach statistical
significance in 40% of bootstrap analyses, implying that statistical significance of the 8-hour periodic
model was dependent on a subset of the data. Two data points before the change of shift at 0700
contributed most strongly to the statistical significance of the periodic waveform. Without these data
points, there was no evidence of an 8-hour periodic waveform for intrathecal bupivacaine analgesia.
CONCLUSION: Chronobiology includes the influence of external daily rhythms in the environment
(e.g., nursing shifts) as well as human biological rhythms. We were able to distinguish the influence
of an external rhythm by combining several novel analyses: (1) graphical presentation superimposing
the raw data, external rhythms (e.g., nursing and anesthesia provider shifts), and smoothing
functions; (2) graphical display of the contribution of each data point to the statistical significance;
and (3) bootstrap analysis to identify whether the statistical significance was highly dependent on a
data subset. These approaches suggested that 2 data points were likely artifacts of the change in nursing
and anesthesia shifts. When these points were removed, there was no suggestion of biological rhythm in
the duration of intrathecal bupivacaine analgesia. (Anesth Analg 2010;111:980–5)

T he duration of epidural analgesia, and the hypnotic

effect of many drugs used for anesthesia or sedation,
have been shown to vary in a periodic pattern
through the day.1 The transcutaneous passage of lidocaine
Authors’ affiliations are listed at the end of the article.
Accepted for publication March 10, 2010.
Supported solely by institutional sources.
Supplemental digital content is available for this article. Direct URL citations
appear in the printed text and are provided in the HTML and PDF versions
of this article on the journal’s Web site (www.anesthesia-analgesia.org).
Address correspondence and reprint requests to Dominique Chassard, MD,
PhD, Department of Anesthesiology and Intensive Care, Hôpital Mère Enfant,
Bron 69500, France. Address e-mail to dominique.chassard @ chu-lyon.fr.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181dd22d4
has been investigated in the morning or in the evening;
significantly higher plasma lidocaine levels were obtained
after evening dosing in children.2 It has also been shown
under conditions of daily dental practice that the duration
of local anesthesia after subcutaneous mepivacaine or arti-
caine is not constant throughout the day.3 These studies
suggest a longer duration of anesthesia in the afternoon at
approximately 1500 for a variety of local anesthetic drugs.4
We have demonstrated a significant effect of the time of
administration on the duration of epidural ropivacaine
analgesia in parturients, with a 30% longer duration of
analgesia when the block is placed in the afternoon.5
Similarly, spinal opioids such as fentanyl or sufentanil do
not have a constant duration of analgesia throughout a

980 www.anesthesia-analgesia.org October 2010 • Volume 111 • Number 4

24-hour period.6,7 These studies indicate that epidurally or
spinally injected anesthetic drugs exhibit a circadian pat-
tern. Because there are no such data on intrathecal local
anesthetics, we tested the hypothesis that the duration of
action of intrathecal local anesthetics has a circadian
rhythm in parturients during the first stage of labor. In
exploring this hypothesis, we developed several novel
approaches to assess the influence of external rhythms on
chronobiological models.
METHODS
After obtaining agreement from the Ethical Committee of
the University of Lyon (France) and written informed
consent, we enrolled pregnant women, ASA physical status
I, parity 0 and 1, with a singleton vertex pregnancy of at
least 36 weeks of gestation. Each woman was in the first
stage of spontaneous labor with a cervical dilation 2 to 4 cm
when requesting labor analgesia. Only patients with un-
complicated pregnancy and normal fetal heart rate tracings
requesting analgesia were enrolled. We excluded patients
receiving antihypertensive drugs and patients who had
already received opioids during labor.
Each patient received an intrathecal injection of 2.5 mg of
plain bupivacaine in 2 mL (a mixture of 1 mL 0.25% bupiva-
caine and 1 mL normal saline) using a combined spinal-
epidural technique with the Braun Escopan威 combination
needle (Braun Medical SA, Boulogne, France). After insertion
of a 27-gauge pencil-point spinal needle and intrathecal
bupivacaine injection, a 20-gauge multiorifice catheter was
inserted in the epidural space. The epidural catheter was left
in position, but no drug was delivered epidurally until the
end of the intrathecal bupivacaine effect.
Pain was assessed using a visual analog scale (VAS) of
pain intensity at 15, 30, and 45 minutes after the block, and
when secondary analgesia was requested. Each patient was
presented with a 100-mm line, labeled as no pain (left end)
and worst pain imaginable (right end). Patients were asked
to mark the line to indicate the intensity of their pain at the
peak of a contraction. Patients could request additional
analgesia if pain relief was unsatisfactory by 15 minutes
after injection of the intrathecal bupivacaine. If a patient
requested relief within 15 minutes after injection, this was
considered a technical failure; these patients were excluded
from further data collection. When additional analgesia
was requested, the study protocol and data collection were
terminated, and analgesia duration was recorded. Analge-
sia duration was the time from intrathecal bupivacaine
administration to the first request for additional analgesia.
Data are presented as mean ⫾ SD. Circadian rhythms
were explored graphically using the 3 smoothing functions,
Supersmoother,* LOWESS,8 and LOESS,9 as implemented
in the R statistical programming language (R Project for
Statistical Computing: http://www.r-project.org). Super-
smoother is a variable span scatterplot smoothing function,
whereas LOWESS and LOESS are locally weighted scatterplot
smoothing functions. To prevent edge effects, the smoothed
value for each time point was calculated by rescaling the X
data to place the point in the center of the data. This was done
*Friedman JH. A variable span scatterplot smoother. Laboratory for Com-
putational Statistics, Stanford University Technical Report No. 5, 1984.
October 2010 • Volume 111 • Number 4
by subtracting 24 hours from all data points ⬎12 hours after
the point, and adding 24 hours for each data point ⬎12 hours
before the data point. This required calculating a new
smoother function for every data point, rather than just once
for across all data. However, for this data set, the compu-
tational time was negligible. The R code for Figure 2
appears in Appendix 1 (see Supplemental Digital Con-
tent1, http://links.lww.com/AA/A123).
The influence of parity and time of day (day [0800 –2000]
versus night [2000 – 0800]) on the VAS score before intra-
thecal injection was analyzed using Mann-Whitney tests.
Analgesia durations were divided into 4 periods based on
the time of intrathecal injection: midnight to 0600, 0600 to
1200, 1200 to 1800, and 1800 to midnight. Times exactly on
the border of 2 periods (e.g., 0600) were assigned to the
later period. Differences in analgesia duration among peri-
ods were analyzed using analysis of variance.
Parametric analysis estimated the duration of analgesia
as a cosine function of time, based on the equation:
analgesia duration ⫽ mesor ⫹ 冘冉
i
1
amplitudei
⫻ cos 冋 2␲
periodi
(t ⫺ acrophasei) 册冊 (1)
where t is the time of initiation of intrathecal analgesia,
mesor is the mean of the fitted curve (i.e., average duration
of anesthesia), amplitude is half of the magnitude of the
circadian variation, and acrophase is the phase offset
relative to midnight (time 0). These parameters were calcu-
lated for periods of 6, 8, 12, and 24 hours.
Three fitting programs were used: Chronos-Fit,†10 Excel
2007 (Microsoft Corp., Redmond, WA) (using the Solver
function to minimize ⫺2 log likelihood), and nonlinear mixed
effect modeling (NONMEM) (version 6.2, Icon Development
Solutions, Ellicott City, MD) (which also minimized ⫺2 log
likelihood). Results were considered significant if P values
were ⬍0.05. No adjustment was made for multiple tests (i.e.,
we considered 4 different periodic signals but did not adjust
our P value to reflect the multiple tests).
The contribution of each subject to the final model was
estimated using NONMEM, comparing the individual ⫺2
log likelihood of a null model (analgesia duration ⫽ mesor
[e.g., mean]) with the final model (Eq. 1, period ⫽ 8 hours).
Ninety-five percent confidence intervals for the estimated
parameters were calculated using a NONMEM bootstrap
analysis of 1000 random bootstrap samples with replace-
ment. The dependency of the statistical significance on a
subset of the data was estimated by NONMEM using
bootstrap analysis to calculate the difference in ⫺2 log
likelihood between the null (no period) and final (8-hour
period) models in 1000 random samples with replacement.
NONMEM was run using the PLT Tools platform (PLTsoft,
San Francisco, CA; www.pltsoft.com).
No formal power analysis was done. The sample size
†Zuther P, Lemmer B. Chronos-Fit, version 1.05, 2004. Available at:
http://www.ma.uni-heidelberg.de/inst/phar/forschungLemmer.html. Ac-
cessed June 6, 2008.
www.anesthesia-analgesia.org 981
Pitfalls in Chronobiological Studies
Table 1. Demographic Data (Mean ⴞ SD)
Age (y) 29 ⫾ 4
Height (cm) 164 ⫾ 5
Weight (kg) 72 ⫾ 9
Dilation (cm) 3 (2–4)
Parity 0/1 (n) 38/44
Figure 1. Analgesia duration versus the time of intrathecal injection.
Black horizontal line ⫽ median analgesic duration; red line ⫽
Supersmoother; blue line ⫽ LOWESS smoother; green line ⫽ LOESS
smoother. Arrows denote the peaks for each smoother of the same
color. Orange horizontal bar ⫽ daytime certified registered nurse
anesthetist and midwife shift; purple horizontal bar ⫽ daytime
anesthesia fellow and anesthesia attending shift.
was chosen based simply on the experience of the investi-
gators. Power analysis for chronobiological studies is re-
viewed in more detail in the Discussion section.
The data files for this analysis, the R programming code,
the Excel spreadsheet for the nonlinear regression, and
NONMEM control and output files for chronobiology
analysis using these data as an example are available as a
Web supplement to this article.
RESULTS
Ninety-one parturients were enrolled in this study from
June 2005 through October 2005. Nine were excluded for
spinal analgesia failure, protocol violation, or emergency
cesarean delivery before a request for analgesia. Patient
characteristics (n ⫽ 82) are provided in Table 1.
The mean VAS score before intrathecal injection was
70 ⫾ 17 mm (median, 70 mm; range, 40 –100 mm). VAS scores
before intrathecal injection were not significantly affected by
the period of day (74 ⫾ 13 mm from 0800 to 2000 vs 69 ⫾ 12
mm from 2000 to 0800; P ⫽ 0.08) or parity (72 ⫾ 16 mm for
parity 0 vs 68 ⫾ 10 mm for parity 1; P ⫽ 0.17). Visual
inspection of the baseline VAS data over the 24-hour time
course did not suggest any periodic signal. Fifteen minutes
after intrathecal bupivacaine, the mean VAS score decreased
to a median of 20 mm (range, 0 –70 mm).
Figure 1 shows analgesia duration versus the time of
intrathecal injection. The black horizontal line is the aver-
age analgesic duration. The smoothers (red ⫽ Super-
smoother, blue ⫽ LOWESS, and green ⫽ LOESS) all reveal
a bimodal pattern with peaks (colored arrows) at 0624 and
982 www.anesthesia-analgesia.org
Table 2. Analgesia Duration by Period
Period Analgesia duration (min ⴞ SD)
1 (0:01 to 6:00) 33.1 ⫾ 10.4
2 (6:01 to 12:00) 41.2 ⫾ 17.8
3 (12:01 to 18:00) 40.0 ⫾ 14.5
4 (18:01 to midnight) 40.0 ⫾ 14.2
Table 3. Periodic Model
Parameter Typical value 95% CI
Mean (min) 38.4 35.4–41.6
Period (h) 8 na
Amplitude (min) 5.8 2.1–10.7
Offset (h) 6.5 5.4–8.0
CI ⫽ confidence interval; na ⫽ not applicable.
2349 (Supersmoother), 0609 and 2318 (LOWESS smoother),
and 0722 and 1812 (LOESS smoother). The orange horizon-
tal bar shows the change in the certified registered nurse
anesthetist (CRNA) and midwife shift, whereas the purple
horizontal bar shows the change in the shift for the anes-
thesia fellow and anesthesia attending.
The duration of analgesia was modestly decreased dur-
ing the period from midnight to 0600 compared with the
other periods (Table 2). This trend did not reach statistical
significance by analysis of variance (P ⫽ 0.42).
Chronos-Fit, Excel, and NONMEM produced identical
results. Excel and NONMEM returned identical ⫺2 log
likelihood estimates. There was just 1 statistically signifi-
cant (P ⫽ 0.04) periodic waveform, with an 8-hour period,
amplitude of 5.8 minutes, acrophase (phase offset) of 6.5
hours, and a mean duration of analgesia (mesor) of 38.4
minutes (Table 3 and Fig. 2). The periodic waveform
captured the peak seen in the original data at approxi-
mately 0630, and approximately captured the peaks at 18
hours (LOESS) and 24 hours (Supersmoother, LOWESS).
In Figure 2, the contribution of each data point can be
inferred from the size and color of the dot. Black data
points represent observations better fit by the model with
the 8-hour period than by the null model, which only
modeled the mesor (average analgesic duration). Red data
points represent observations worse fit by the model with
the 8-hour period than by the null model. The size of the
data point represents the increase (black) or decrease (red)
in ⫺2 log likelihood of the 8-hour periodic model over the
null model.
As shown in Figure 2, the 2 points that contributed most to
the improvement in ⫺2 log likelihood were from the 2
intrathecal injections performed shortly before the change in
nursing shift at 0700. One was placed at 0624 and lasted 81
minutes, and the other was placed at 0653 and lasted 90
minutes.
Figure 3 shows the result of the bootstrap analysis based
on 1000 samples of the data set with replacement. The
x-axis shows the difference in ⫺2 log likelihood, and the
y-axis shows the cumulative fraction. The vertical lines
mark differences in ⫺2 log likelihood associated with P ⫽
0.05, 0.01, 0.001, and 0.0001. If a model was highly statisti-
cally significant, regardless of which subset of the data was
ANESTHESIA & ANALGESIA
Figure 2. Same result as Figure 1, with the 8-hour periodic waveform
shown in gold. The size of the points reflects the magnitude of the
contribution of each point toward (black) or against (red) statistical
significance of the final model.
Figure 3. The cumulative distribution of the improvement in ⫺2 log
likelihood comparing the null model with the final model in 1000
bootstraps. In 40% of the bootstrap analyses, the difference was not
statistically significant, suggesting that the statistical significance
depended on the inclusion of a subset of the data points in the
analysis.
considered, then the vast majority of replicates in a boot-
strap analysis would show a P value ⬍0.05. That is not the
case. Statistical significance was not achieved in 40% of the
replicates, suggesting that statistical significance is highly
dependent on the subset of data included. Given that the 2
data points that contribute most to the model occur shortly
before the change in nurse and anesthesia provider, we
investigated the performance of the model with those
outliers excluded.
Figure 4 shows the data with the 2 suspected outlying data
points removed. When those 2 data points were excluded,
there was no longer a statistically significant periodic signal
according to our reanalysis with Chronos-Fit, Excel, and
NONMEM. Only the LOWESS smoother (blue line) offered
any evidence of periodicity. The Supersmoother and LOESS
smoother functions were essentially flat.
DISCUSSION
The mean duration of action of intrathecal bupivacaine
reported in our study, 39 ⫾ 15 minutes, is similar to that
October 2010 • Volume 111 • Number 4
Figure 4. The same analysis as in Figure 1, but with the 2 outlying
data points at 0624 and 0653 removed. Only the LOWESS smoother
(blue line) continues to show evidence of periodicity. The Super-
smoother and LOESS smoother are essentially flat.
obtained by Wong et al.,11 39 ⫾ 25 minutes. Stocks et al.,12
using a technique of up-down sequential allocation, dem-
onstrated that the minimum local analgesic dose of intra-
thecal bupivacaine was 1.99 mg, and that the duration of
analgesia with this dose was 43 ⫾ 19 minutes. However,
using the same dosage, Lim et al.13 reported a longer
duration of action, 76.3 ⫾ 5.9 minutes. The results could
have been influenced by several factors other than the time
of the day: for example, the method to calculate analgesia
duration, VAS scores, cervical dilation at the time of
inclusion, parity, and the use of oxytocin. These factors
must be incorporated in future studies of chronobiology in
parturients, as well as the observation that the perception of
labor pain varies with time of day.14
This article was initially submitted to Anesthesia &
Analgesia as a demonstration of the chronobiology of anal-
gesia after intrathecal bupivacaine. Over the course of peer
review and additional analysis, this article has metamor-
phosed into a cautionary tale of the influence of external
rhythms on chronobiological analysis, and an exploration
of data analysis methods to facilitate detection of artifacts.
Our presentation of the methods and results in this article
reflects the way we think this type of analysis should
proceed. The actual analysis was much more convoluted.
The most critical element in data analysis is visualizing
the data. Figure 1 is our proposal for visualization of the
raw data in chronobiological analyses. The figure shows
the raw data and adds 3 smoothing functions to draw the
eye toward the underlying signal. Because smoothers are
prone to edge effects, the smoothing functions have been
calculated by centering each observation in a 24-hour time
window, permitting the smoothing function to properly
“wrap around” midnight. The smoothers show that there
can be, at most, 2 peaks. Thus, the data themselves suggest
that there is little reason to explore a periodic signal with a
period shorter than 12 hours.
In addition, Figure 1 shows the primary external
rhythms likely to influence the data: the start and end of the
daytime nursing and anesthesia shifts. It is immediately
obvious that the spike in analgesia duration for intrathecal
www.anesthesia-analgesia.org 983
Pitfalls in Chronobiological Studies
injections before the change in CRNA shift might be
explained because the CRNAs, and subsequently the anes-
thesiologists, are starting their daytime shift, delaying
assessment of patients’ pain. We suggest that future chro-
nobiological analyses include a figure similar to Figure 1 to
identify the basic patterns in the data, assess the likely
period of any chronobiological signal by using smoothing
functions, and assess the relationship, if any, to external
rhythms. Of course, it is entirely possible that rhythms
unknown to the investigators (e.g., lunch breaks, the timing
of the change in bed sheets, and the regular 4 am test of the
backup generators) might be present. If that was the case,
isolated spikes would appear at given times, but an under-
lying sine rhythm would not be evident.
Our second step was to divide the data into 4 time periods
that approximately divided the 24-hour day into “night”
(midnight to 0600), “morning” (0600 to 1200), “afternoon”
(1200 to 1800), and “evening” (1800 to midnight). We did not
see any statistically significant differences.
Our third step was modeling the data. The “model” is a
simple cosine function. We chose to model periods of 24,
12, 8, and 6 hours. However, the only period that makes a
priori sense biologically is the 24-hour period. Of course,
there could be 2 peaks in some biological function, but why
should they necessarily be 12 hours apart? If the day is
divided into a rhythmic pattern with multiple peaks, why
should it be sinusoidal rather than (for example) 16 hours
alternating with 8 hours, as in the awake/asleep cycle? It is
even more mysterious why the day should be divided into
sinusoidal 6- or 8-hour periods. Interestingly, even though
we only saw 2 peaks with the smoothing functions, the
12-hour rhythm was not statistically significant. The only
rhythm that was statistically significant was a cosine signal
with a period of 8 hours, shown in Figure 2. Figure 2 also
introduces another new graphic element: showing the
extent to which each observation contributes for, or against,
the statistical significance of the model.
Inspection of Figure 2 reveals a problem with the model.
It captures the observed peak at 0630, but the other 2 peaks
in the sinusoidal model do not correspond with peaks in
the data identified by the smoothing functions. Examining
the individual points, the 2 largest black dots are the
observations that contribute the most to the statistical
significance of the 8-hour periodic waveform. Looking at
the external rhythms (horizontal orange and purple bars),
these are the very observations that are most likely to
reflect the influence of the change in nursing and anesthesia
shifts.
Before discarding the fit entirely, we turned to a boot-
strap analysis to address whether the statistical significance
of the model was highly dependent on the specific data
sampled. If the data broadly supported the statistical
significance of the more complex model, then statistical
significance should be evident from any subset of the data
analyzed. However, if statistical significance depends on
just a few points, then a significant fraction of the bootstrap
analyses will fail to be statistically significant (the boot-
straps that do not include the critical observations),
whereas some bootstraps will have greatly increased sta-
tistical significance (bootstraps that have ⬎1 copy of the
984 www.anesthesia-analgesia.org
critical observations). Because 40% of the bootstrap analy-
ses did not reach statistical significance (Fig. 3), a small
number of data points are driving the statistical result.
Supported by the evidence in Figures 1, 2, and 3 that the
statistical significance was driven by an artifact, the obser-
vations at 0624 (81 minutes of analgesia) and 0653 (90
minutes of analgesia) were removed from the data. Figure
4 shows that there is still a peak in duration at the time of
the nursing shift change, but that the peak is significantly
attenuated by the loss of these 2 points. Indeed, Super-
smoother and the LOESS smoother were essentially flat.
Similarly, no periodic signal was statistically significant
once these points were removed.
Our lives are governed by powerful external rhythms.
Most obviously, we are synchronized to the rotation of the
earth, and the resulting periods of light and darkness.
Other rhythms infuse our lives. These can affect studies of
chronobiology in 2 ways. First, they may induce real
physiological changes. For example, arterial blood pressure
might increase twice daily with the commute to and from
work. However, external rhythms might produce artifacts
that have nothing to do with biology, such as delayed
patient assessment during the changing of provider shifts.
We suggest 6 steps to assess the influence of external
artifacts on chronobiological analyses:
1. Graph the raw data over 24 hours, along with 1 or
more smoothing functions. To avoid edge effects, the
function needs to wrap around midnight. We elimi-
nated edge effects by calculating the value of the
smoothing function for each observation with the
time centered in a 24-hour window. If the smoothing
functions do not suggest a periodic waveform, then
no further analysis should be undertaken.
2. If a periodic pattern is evident in the smoothing
functions, compare the smoothed functions with ex-
ternal rhythms, such as shift changes, that might
affect the observations.
3. If a periodic pattern is evident in the smoothing
functions, fit the data with a regression program,
such as Chronos-Fit, Excel, or NONMEM. If there is
only 1 point per subject, then it makes no difference
what program is used. In this analysis, all 3 programs
returned results that were identical in all reported
digits. If there are multiple observations per subject,
only NONMEM can separate interindividual vari-
ability from intraindividual variability.
4. Represent the data points to show the contribution
for, or against, the final model. This will identify
whether the model is being driven by a subset of the
data.
5. Undertake a bootstrap analysis to confirm whether
statistical significance is driven by a small subset of
the data points.
6. If data artifacts are driving the result, remove them
and reassess the model.
Removing data that appear to be artifacts may bias study
results in favor of the investigator’s expectation. For example,
we do not know for certain that the observations at 0624 and
0653 are artifacts. These data points could be completely valid
observations. However, this analysis demonstrates that even
ANESTHESIA & ANALGESIA
if these points are valid (which we doubt), the lack of
correlation between the data and the other 2 peaks in the
“statistically significant” 8-hour periodic signal preclude our
having any confidence in the model.
We have only explored rhythms that are synchronized
among all patients. If each patient had a strong individual
rhythm, but a random phase relationship to every other
patient, then the underlying chronobiology would be
missed in this type of analysis. Individual rhythms with
variable phase offset could only be detected by analyzing
multiple observations in each individual, and using a
mixed effects model with intersubject variability in the
phase offset (“acrophase”).
Conventional power analyses do not readily handle a
data analysis plan that involves comparing 1 model to a
second model, as in this case in which a straight line is
being compared with a sine wave. In our view, the best way
to approach power analysis is to perform a Monte Carlo
simulation, in which multiple synthetic data sets are gen-
erated for a range of sample sizes, and the percent of
successful trials is calculated. The study size is based on the
sample size in the Monte Carlo simulation that produces
the desired percent of models correctly identified.
In conclusion, circadian rhythms are an important as-
pect of human biology. On the basis of these data, there is
no important chronobiological rhythm in spinal bupiva-
caine analgesia during labor. External daily rhythms may
contaminate the data when assessing biological rhythms.
We suggest that the approach used in this study may help
to separate true biological signals from external artifact in
future studies.
AUTHORS’ AFFILIATIONS
From the *Department of Anesthesiology, Columbia University,
New York, New York; †Institute of Pharmacology & Toxicology,
University of Heidelberg, Heidelberg, Germany; ‡Department of
Anesthesiology and Intensive Care, Hôtel-Dieu Hospital; and
§University Claude Bernard-Lyon 1, Lyon, France.
RECUSE NOTE
Steven L. Shafer is Editor-in-Chief of the Journal. The manu-
script was handled by James G. Bovill, Guest Editor-in-Chief,
and Dr. Shafer was not involved in any way with the editorial
process or decision.
DISCLOSURE
Dr. Shafer is codeveloper of PLT Tools and has a financial
interest in the software.
October 2010 • Volume 111 • Number 4
ACKNOWLEDGMENTS
The authors thank Peter Tucker (American Translator, Univer-
sity of Aix-Marseille, France) for assistance in the preparation
of the manuscript. The authors also acknowledge the many
insightful suggestions of Dr. Dennis Fisher to this work.
REFERENCES
1. Chassard D, Bruguerolle B. Chronobiology and anesthesia.
Anesthesiology 2004;100:413–27
2. Bruguerolle B, Giaufre E, Prat M. Temporal variations in
transcutaneous passage of drugs: the example of lidocaine in
children and in rats. Chronobiol Int 1991;8:277– 82
3. Lemmer B, Wiemers R. Circadian changes in stimulus thresh-
old and in the effect of a local anaesthetic drug in human teeth:
studies with an electronic pulptester. Chronobiol Int 1989;
6:157– 62
4. Reinberg A, Reinberg MA. Circadian changes of the duration
of action of local anaesthetic agents. Naunyn Schmiedebergs
Arch Pharmacol 1977;297:149 –52
5. Debon R, Chassard D, Duflo F, Boselli E, Bryssine B, Allaouch-
iche B. Chronobiology of epidural ropivacaine. Anesthesiology
2002;96:542–5
6. Debon R, Boselli E, Guyot R, Allaouchiche B, Lemmer B,
Chassard D. Chronopharmacology of intrathecal sufentanil for
labor analgesia: daily variations in duration of action. Anes-
thesiology 2004;101:978 – 82
7. Pan PH, Lee S, Harris L. Chronobiology of subarachnoid
fentanyl for labor analgesia. Anesthesiology 2005;103:595–9
8. Cleveland WS. Robust locally weighted regression and
smoothing scatterplots. J Am Stat Assoc 1979;74:829 –36
9. Cleveland WS, Grosse E, Shyu WM. Local regression models.
In: Chambers JM, Hastie TJ, eds. Statistical Models in S. Pacific
Grove, CA: Wadsworth & Brooks/Cole, 1992:309 –76
10. Zuther P, Witte K, Lemmer B. ABPM-FIT and CV-SORT: an
easy-to-use software package for detailed analysis of data from
ambulatory blood pressure monitoring. Blood Press Monit
1996;1:347–54
11. Wong CA, Scavone BM, Loffredi M, Wang WY, Peaceman AM,
Ganchiff JN. The dose-response of intrathecal sufentanil added
to bupivacaine for labor analgesia. Anesthesiology 2000;
92:1553– 8
12. Stocks GM, Hallworth SP, Fernando R, Engl AJ, Columb MO,
Lyons G. Minimum local analgesic dose of intrathecal bupiv-
acaine in labor and the effect of intrathecal fentanyl. Anesthe-
siology 2001;94:593– 8
13. Lim Y, Ocampo CE, Sia AT. A comparison of duration of
analgesia of intrathecal 2.5 mg of bupivacaine, ropivacaine,
and levobupivacaine in combined spinal epidural analgesia for
patients in labor. Anesth Analg 2004;98:235–9
14. Aya AG, Vialles N, Mangin R, Robert C, Ferrer JM, Ripart J, de
La Coussaye JE. Chronobiology of labour pain perception: an
observational study. Br J Anaesth 2004;93:451–3
www.anesthesia-analgesia.org 985
The Influence of Time of Day of Administration on
Duration of Opioid Labor Analgesia
Barbara M. Scavone, MD, Robert J. McCarthy, PharmD, Cynthia A. Wong, MD, and John T. Sullivan, MD

BACKGROUND: Medications administered into the epidural or intrathecal space for labor
analgesia may demonstrate variable effects dependent on time of day, and this may affect
clinical research trials investigating the pharmacology of specific drugs. In this retrospective
study, we evaluated the effect of time of day of administration of intrathecal fentanyl and
systemic hydromorphone labor analgesia from data collected as part of a randomized clinical trial
examining the influence of analgesia method on labor outcome.
METHODS: Six hundred ninety-two healthy parturients were randomized early in labor to receive
combined spinal-epidural (intrathecal fentanyl 25 ␮g followed by a lidocaine and epinephrine
containing epidural test dose) versus systemic (hydromorphone 1 mg IV and 1 mg IM) labor
analgesia at first analgesia request. No further analgesics were administered until the patient
requested additional analgesia (second analgesia request). Subjects were assigned to the
daytime group (DAY) if initial analgesia (neuraxial or systemic) was administered between the
hours of 07:01 and 23:00 and to the nighttime group (NIGHT) if it was administered between
23:01 and 07:00. Within each mode of analgesia study arm (neuraxial or systemic), the DAY and
NIGHT groups were compared. The primary outcome variable was analgesia duration, defined as
the time interval from administration of labor analgesia until the second analgesia request.
Cervical dilation at first and second analgesia requests, pain score at first analgesia request, and
average amount of pain between analgesia administration and second analgesia request were
also compared between DAY and NIGHT groups. Rhythm analyses for duration of analgesia,
cervical dilation, and pain scores were performed.
RESULTS: There was no difference in the median duration of either neuraxial or systemic
analgesia in DAY versus NIGHT subjects, and no harmonic variation was observed for analgesia
duration. Rhythm analysis demonstrated a 24-h harmonic cycle for cervical dilation at first
analgesia request with maximum values occurring near 17:00 and minimum values near 05:00,
but the amplitude of the difference was very small. Rhythm analysis demonstrated a 24-h
harmonic cycle with maximum values occurring near 22:00 and minimum values near 10:00 for
the average amount of pain between analgesia administration and second analgesia request in
neuraxial group patients, but amplitude was small.
CONCLUSIONS: Time of day of administration did not seem to influence combined spinal-
epidural or systemic labor analgesia duration under these study conditions. (Anesth Analg
2010;111:986 –91)

C hronobiology is the study of the effect of time,

especially biologic rhythms, on living organisms;
chronopharmacology is the study of the relation
between time of administration and the effects of drugs.
Chronopharmacologic effects can be either chronopharmaco-
dynamic (time-dependent differences in organism re-
sponse to drugs) or chronopharmacokinetic (time-
dependent differences in pharmacokinetic parameters,
such as absorption, distribution, and clearance).1
Several authors have noted time-dependent effects of
medications administered into the epidural or intrathecal
From the Department of Anesthesiology, Northwestern University Feinberg
School of Medicine, Chicago, Illinois.
Accepted for publication September 7, 2009.
Supported by Departmental funds.
Cynthia A. Wong is Section Editor of Obstetric Anesthesiology for the
Journal. This manuscript was handled by Steve Shafer, Editor-in-Chief, and
Dr. Wong was not involved in any way with the editorial process or
decision.
Reprints will not be available from the author.
Address correspondence to Barbara M. Scavone, MD, 251 E. Huron St.,
F5-704, Chicago, IL 60611. Address e-mail to bimscavone@aol.com.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181c29390
space for labor analgesia; specifically, differences in analgesia
duration after administration of epidural ropivacaine and
intrathecal sufentanil and fentanyl depending on time of day
of administration.2– 4 This has prompted concern regarding
the impact of chronopharmacology on clinical research trials
investigating the pharmacology of specific drugs.1 However,
these trials involved small numbers of subjects, especially
during the night hours, and did not always control for
external factors that may indirectly influence duration of
analgesia (e.g., differences in labor unit staffing, visiting
hours, and sleep deprivation).
The purpose of this retrospective study was to evaluate
possible effects time of administration may have on intra-
thecal fentanyl and systemic hydromorphone labor analge-
sia from data collected as part of a randomized clinical trial
examining the influence of analgesia method on labor
outcome.5 In the aforementioned study, duration of labor
analgesia was a secondary outcome, but the influence of
time of day was not examined. This study is a secondary
analysis of the previously obtained data, examining the
influence of time of administration of analgesia on duration
of labor analgesia. The null hypothesis for this analysis was
that duration of labor analgesia would not vary with time
of day.

986 www.anesthesia-analgesia.org October 2010 • Volume 111 • Number 4

METHODS
The study was approved by the Northwestern University
Office for the Protection of Research Subjects IRB. Details of
the study design and protocol have been published.5
Briefly, written informed consent was obtained from 884
healthy, opioid naïve, nulliparous women. Additional in-
clusion criteria included uncomplicated, term, singleton
pregnancy with vertex position, and spontaneous labor or
spontaneous rupture of membranes. Parturients were re-
cruited shortly after admission to the Labor and Delivery
Unit by anesthesia personnel 24 h per day, 7 days per wk
over a 38-mo period commencing October 2000. Subjects
were enrolled (n ⫽ 750) in the study if the cervix was
dilated ⬍4 cm at the first request for analgesia and were
randomized to receive neuraxial versus systemic opioid
analgesia. At the first analgesia request, subjects in the
neuraxial group received a combined spinal-epidural via a
needle through needle technique. They received intrathecal
fentanyl 25 ␮g and an epidural test dose of lidocaine 45 mg
and epinephrine 15 ␮g in 3 mL total volume. Subjects in the
systemic analgesia group received 1 mg IV and 1 mg IM
hydromorphone. No further analgesics were administered
until the patient requested additional analgesia (second
analgesia request). At the second analgesia request, a
cervical examination was performed and the subject re-
ceived additional medications per study protocol.
Subjects rated peak contraction pain using a verbal rating
scale for pain: a 0–10 scale in which 0 ⫽ no pain and 10 ⫽ worst
imaginable pain. Pain scores were obtained at the first analgesia
request (pain score1st request). In addition, at the second anal-
gesia request, when the effects of analgesia were diminish-
ing, the patient was asked to rate the “average amount of
pain” experienced between receiving intrathecal fentanyl
or systemic hydromorphone and the second analgesia
request (pain scoreaverage).
In this secondary analysis, subjects were assigned to the
daytime group (DAY) if initial analgesia (neuraxial or
systemic) was administered between 07:01 and 23:00 h, and
to the nighttime group (NIGHT) if it was administered
between 23:01 and 07:00 h.2 Within each mode of analgesia
study arm (neuraxial or systemic), the DAY and NIGHT
groups were compared. Duration of analgesia was the
primary outcome variable and was defined as the time
interval between the administration of intrathecal fentanyl
or systemic hydromorphone and the second analgesia
request. Age, height, weight, gestational age, oxytocin
administration, time interval from analgesia administration
to delivery, infant weight, mode of delivery, cervical dila-
tion, and pain scores were also compared in DAY versus
NIGHT group subjects.
Previous studies examining the chronopharmacologic
effect of intrathecal opioids have demonstrated a 30-min
difference between day and night administration in the
duration of analgesia after administration of intrathecal
fentanyl 20 ␮g in 77 subjects,4 and a 23-min difference after
administration of intrathecal sufentanil 10 ␮g in 91 sub-
jects.3 A sample of 345 subjects would achieve 96% power
to detect a difference of 7 min between day and night
duration of analgesia assuming an average duration of 90
min with an estimated standard deviation of 25 min at an ␣
equal to 0.05 using a 2-sided Mann-Whitney test.
October 2010 • Volume 111 • Number 4
DAY and NIGHT groups within each mode of analgesia
arm (neuraxial or systemic) were compared using the Mann-
Whitney U-test (age, height, weight, gestational age, time to
delivery, analgesia duration, cervical dilation, and pain scores) or
␹2 test (oxytocin administration and mode of delivery). The time
data were plotted against the data for duration of analgesia, as
well as cervical dilation and pain scores at first and second
request and average pain score between analgesia interven-
tions and evaluated using weighted least squares regression
(lowess), weighted quadratic least squares regression (loess),
and nonparametric smoothing (the “super smoother” of
Friedman).*6 Curve fitting trend lines were determined using
the R functions “lowess,” “loess,” and “supsmu”). To im-
prove smoothness at the beginning and end of the 24-h cycle,
data were concatenated from 25 h before and after the 24-h
cycle. Lowess and loess curves were estimated at a smoother
span of 0.15. Data were analyzed using R version 2.9.1
(http://www.r-project.org).
Rhythm analyses for duration of analgesia, cervical dila-
tion, and pain scores were performed using a nonlinear
WIN-ABPM-FIT program.† This analysis calculates the mesor
(rhythm-adjusted mean), amplitude (half of the peak to
trough of the rhythm-adjusted harmonic), and acrophase
(time of the occurrence of the rhythm-adjusted harmonic).
Data were evaluated for 24-, 12-, 8-, and 6-h harmonics. The
solver add in for Microsoft Excel 2007 was used to calculate
the r2, ⫺2 log likelihood, and P value for the harmonics
identified. P ⬍ 0.05 was required to reject the null hypothesis.
RESULTS
There were 366 patients assigned to the neuraxial analgesia
arm and 362 to the systemic analgesia arm of the original
study. Data from 343 of the neuraxial group subjects and
349 of the systemic group subjects are included in the
present analysis (the remainder of subjects had protocol
violations or missing data). Comparison of demographic
data between the DAY versus NIGHT groups is presented
in Table 1. A significantly larger proportion of DAY sub-
jects than NIGHT subjects were receiving oxytocin at the
time of analgesia in both the neuraxial and systemic arms of
the study. DAY subjects in both analgesic arms of the study
had a shorter analgesia to delivery time interval than
NIGHT subjects.
Outcome data are presented in Table 2, and scatter plots
and smoother curves for duration of analgesia, cervical
dilation, and pain scores are shown in Figures 1– 4. There
was no difference in the median duration of either
neuraxial or systemic analgesia in DAY versus NIGHT
subjects (Table 2). No harmonic variation was apparent for
either neuraxial or systemic analgesia duration (Fig. 1).
Despite a similar median value, there was a difference in
distribution of cervical dilation at first analgesia request
between the DAY and NIGHT subjects who received
neuraxial analgesia. Subjects who received systemic anal-
gesia did not demonstrate the same variation. Rhythm
analysis of all subjects demonstrated a 24-h harmonic cycle
for cervical dilation at first analgesia request (Figs. 2 and 5).
*Friedman JH. A variable span smoother: technical report LC55. Palo Alto,
CA: Department of Statistics, Stanford University, 1984.
†Available at: http://www.ma.uni-heidelberg.de/inst/phar/forschungLemmer.
html. 2004. Accessed February 8, 2008.
www.anesthesia-analgesia.org 987
Opioid Labor Analgesia and Time of Day

Table 1. Subject Demographic Characteristics

Neuraxial analgesia Systemic analgesia
Time of analgesia administration Time of analgesia administration
07:01–23:00 h 23:01–07:00 h P 07:01–23:00 h 23:01–07:00 h P
Age (yr) 32 (29–35) 31 (28–35) 0.34 31 (28–35) 32 (30–35) 0.22
Height (cm) 165 (160–170) 163 (160–169) 0.58 165 (160–170) 163 (160–170) 0.61
Weight (kg) 77 (69–88) 77 (70–85) 0.78 79 (71–88) 76 (69–87) 0.09
Gestational age (wk) 39 (38–40) 40 (39–40) 0.37 40 (39–40) 39.5 (39–40) 0.63
Oxytocin administered at 180/227 (79) 66/110 (60) ⬍0.001 196/240 (82) 57/99 (58) ⬍0.001
time of first
analgesia, n (%)
Time to delivery (min) 419 (295–562) 491 (314–651) 0.03 505 (345–653) 559 (276–575) 0.03
Infant weight (g) 3490 (3168–3768) 3415 (3136–3679) 0.42 3410 (3120–3665) 3450 (3185–3732) 0.17
Vaginal delivery (%) 83 82 0.89 80 76 0.40

Table 2. Outcome Variables

Neuraxial analgesia Systemic analgesia
Time of analgesia administration Time of analgesia administration
07:01–23:00 h 23:01–07:00 h P 07:01–23:00 h 23:01–07:00 h P
Duration of analgesia (min) 95 (73–117) 92 (75–125) 0.36 108 (80–143) 105 (80–145) 0.85
Cervical dilation at 1st analgesia 2 (1.5–3) 2 (1–3) 0.001 2 (1–3) 2 (1–3) 0.25
request (cm)
Distribution, n (%)
ⱕ1.5 cm 65 (27) 48 (39) 97 (38) 55 (50)
⬎1.5 to ⬍3.0 cm 80 (33) 50 (49) 83 (33) 28 (26)
ⱖ3.0 cm 97 (40) 26 (21) 73 (29) 26 (24)
Pain score1st request 8 (7–9) 8 (7–9) 0.02 8 (7–9) 8 (7–9) 0.36
Distribution, n (%)
0–3 3 (2) 2 (2) 7 (3) 5 (5)
4–7 93 (39) 32 (26) 92 (38) 34 (31)
8–10 140 (59) 88 (72) 142 (59) 69 (64)
Cervical dilation at 2nd 4 (3–5) 4 (3–5) 0.05 4 (3–5) 4 (2–5) 0.36
analgesia request (cm)
Pain scoreaverage 2 (1–3) 2 (0–3) 0.33 6 (4–7) 5 (4–7) 0.11

There was no difference observed between DAY and
NIGHT subjects regarding cervical dilation at second anal-
gesia request, and no harmonic variation was observed for
this variable (Fig. 3).
The distribution of pain score1st request was different
during the night period compared with the day in subjects
who received neuraxial analgesia despite a similar median
value. Subjects who received systemic analgesia did not
demonstrate a difference. Pain score1st request did not exhibit
any harmonic variation (Fig. 2). Median pain scoreaverage did
not differ between DAY and NIGHT subjects in either the
neuraxial or systemic arms of the study; however, a 24-h
harmonic variation was observed for pain scoreaverage in the
neuraxial group (Figs. 4 and 5); no harmonic variation was
observed among subjects in the systemic group (Fig. 4).
DISCUSSION
Time of day of administration of analgesia did not seem to
influence combined spinal-epidural or systemic labor anal-
gesia duration under the conditions of this study. We did
observe a harmonic cycle in the cervical dilation at which
labor analgesia was first requested; however, the small
amplitude associated with the harmonic variation and the
small r2 value imply that any clinical implications would be
minimal. We also observed harmonic variations in the
average amount of pain experienced between intrathecal
fentanyl administration and second analgesia request,
again, with small amplitude and r2 value.
Biologic rhythms are both genetically determined and
influenced by environmental factors known as synchroniz-
ers, such as the light-dark cycle, eating, and fasting periods.
In humans, the main circadian (referring to a 24-h cycle)
pacemaker is located in the hypothalamus. It is modulated
by the external environment, receiving inputs from the
retina, as well as nonphotic synchronizers, such as those
involved with locomotor activity and eating. Certain drugs
also influence the pacemaker. The output from this pace-
maker in turn influences a myriad of biologic activities,
such as melatonin production, cortisol secretion, adreno-
corticotropin releasing hormone secretion, and sensitivity
of end organs to adrenocorticotropin releasing hormone.
Organs such as the heart and liver, and even isolated cells,
demonstrate their own circadian rhythms independent of
the central circadian pacemaker.1
It can be difficult to separate true time of day effects
from external factors that may influence circadian human
behaviors. Our study subjects demonstrated a minimum
cervical dilation at first analgesia request in the very early

988 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA


Figure 1. Analgesia duration versus the time of administration of
analgesia (top panel ⫽ neuraxial group; bottom panel ⫽ systemic
group). Black horizontal line ⫽ mean analgesic duration; blue line ⫽
lowess smoother; green line ⫽ loess smoother; and red line ⫽
supersmoother. Diamonds denote the maximum and minimum
values for each smoother of the same color.
morning hours, perhaps indicating increased early labor
pain perception during that time interval. Also, patients in
the neuraxial arm of the study had higher pain scores at
first analgesia request between the hours of 23:01 and 07:00
than between 07:01 and 23:00. These findings are consistent
with a previous study in which nulliparous women in
spontaneous labor with cervical dilation between 3 and 5
cm and ruptured membranes reported lower pain scores
between 07:00 and 13:00 h than at other times of the day.7
This circadian rhythmicity to pain perception could be
either attributable to true time of day effects or influenced
by external factors. For instance, it could be explained by
the circadian rhythm associated with ␤-endorphin secre-
tion (highest in the morning).8 However, it also could be
explained by the negative effects that factors such as sleep
deprivation may have on pain perception. Alternatively,
subjects who experience labor pain during the day may be
more likely to go to the hospital when they are in less
distress than patients who experience that pain at night.
Although we did not observe any time-related differences
in analgesia duration, we did find some harmonic variation
in effectiveness of intrathecal opioid analgesia as measured
by pain scoreaverage, which showed a peak in the late
evening and a nadir in the morning. Once again, our study
October 2010 • Volume 111 • Number 4
Figure 2. Cervical dilation (top panel) and pain score at first request
for analgesia (bottom panel) versus time of administration of anal-
gesia. Black horizontal line ⫽ mean cervical dilation/pain score; blue
line ⫽ lowess smoother; green line ⫽ loess smoother; and red
line ⫽ supersmoother. Diamonds denote the maximum and mini-
mum values for each smoother of the same color.
design does not distinguish between true time of day
explanations for this versus external factors that influence
human behavior.
To our knowledge, there have been no previous studies
examining the effect of time of administration on systemic
opioid labor analgesia. Several authors have demonstrated
time-related effects for neuraxial labor analgesia. Epidural
ropivacaine had a longer duration between 07:00 and 13:00
h compared with other times of the day.2 Interestingly, this
is the same time interval associated with lowest labor pain
scores.7 Intrathecal opioids given for labor analgesia also
exhibited chronobiologic characteristics. The duration of
intrathecal fentanyl 20 ␮g was longer between 12:00 and
18:00 h than between 20:00 and 02:00 h.4 The duration of
intrathecal sufentanil 10 ␮g followed a 12-h cycle, with
peaks at 12:00 and 00:00 h.3
In contrast to these previous studies, we did not observe
any harmonic cycle in the duration of labor analgesia. One
possible explanation for the difference in our results from
previous studies is that our patients were recruited in early
labor, with a median cervical dilation of 2 cm, whereas
those in other studies were in more advanced labor. It is
possible that chronopharmacologic analgesic effects are
more likely to be observed as labor progresses and pain
worsens. Another explanation may lie in the fact that the
analgesic doses differed. Not only did subjects in this study
www.anesthesia-analgesia.org 989
Opioid Labor Analgesia and Time of Day
Figure 3. Cervical dilation at second request for analgesia (top
panel ⫽ neuraxial group; bottom panel ⫽ systemic group). Black
horizontal line ⫽ mean cervical dilation; blue line ⫽ lowess
smoother; green line ⫽ loess smoother; and red line ⫽ super-
smoother. Diamonds denote the maximum and minimum values for
each smoother of the same color.
receive a relatively high dose of opioid but they also
received a lidocaine/epinephrine epidural test dose. No
one has studied the time of day effect of a neuraxial mixture
of opioid and local anesthetic. It is possible that pharmaco-
logic rhythms are less likely to be observed with higher
doses of drugs or mixtures of drugs. Our results are
consistent with a study examining chronic neuropathic
pain, which demonstrated chronobiology to pain percep-
tion but not to analgesic efficacy.9
The number of subjects in previous studies was small
(largest n ⫽ 194) compared with this study. Additionally, in
previous studies, explaining the study objective to the
subject may have influenced the subject’s perception of
analgesic efficacy. It is well established that a significant
placebo effect can be measured in studies involving anal-
gesia administration.10 Pain scores in the previous studies
were assessed at regular intervals after the initial analgesic
dose, and the presence of an investigator assessing pain
may have influenced when the subject requested additional
analgesia. Subjects in previous studies were recruited over
a relatively brief interval (months) compared with our
study (years), and it is possible that this may also have
influenced results.
990 www.anesthesia-analgesia.org
Figure 4. Average pain score between first and second analgesia
administration (top panel ⫽ neuraxial group; bottom panel ⫽ sys-
temic group). Black horizontal line ⫽ mean pain score; blue line ⫽
lowess smoother; green line ⫽ loess smoother; and red line ⫽
supersmoother. Diamonds denote the maximum and minimum
values for each smoother of the same color.
Finally, none of the studies, including ours, controlled
for other factors that may influence pain and analgesia,
including presence or absence of medical personnel, lay
support people or family members, room lighting, sleep
deprivation before enrollment in the study, time since last
sleep, sleep during the study period, the subjects’ normal
activity and sleep cycles, and other unknown factors. We
manage a high volume of patients (approximately 10,000
deliveries per year) on our delivery unit, and therefore, the
light and noise levels at night may differ from those of
smaller units. It would be desirable to control variables
such as these in a study specifically investigating the
influence of chronobiology on labor pain and analgesia.
There are other limitations to this study. Variables such
as the administration of oxytocin were not controlled
resulting in a larger fraction of subjects during the day
receiving oxytocin at the initiation of analgesia. It is pos-
sible that both nurses and obstetricians are less likely to
augment labor at night because of inconvenience or other
factors; alternatively, patients who come to the hospital at
night may be a class of patients in more active labor (in
more pain even at low cervical dilations) and therefore in
less need of oxytocin augmentation. It is possible that the
increased use of oxytocin during the day masked time-
related effects that would have been evident had this
ANESTHESIA & ANALGESIA

Figure 5. Harmonic cycles (a) identified by fitting the equation (b).
Upper panel: pain scoreaverage, neuraxial group. Orange line ⫽
predicted values from function with periodi ⫽ 24 h, mesor ⫽ 2,
amplitudei ⫽ 0.6, acrophasei ⫽ 21.3 h (r2 ⫽ 0.012, P ⫽ 0.005),
with maximum values occurring near 22:00 h and minimum values
near 10:00 h. Lower panel: cervical dilation at first request for
analgesia. Orange line ⫽ predicted values from function with pe-
riodi ⫽ 24 h, mesor ⫽ 2 cm, amplitudei ⫽ 0.1 cm, acrophasei ⫽ 16.4 h
(r2 ⫽ 0.011, P ⫽ 0.02), with maximum values occurring near 17:00 h,
and minimum values near 05:00 h.
variable been controlled. In addition, more subjects were
recruited during the day than at night, and this may have
confounded results. Patients at night may be different than
patients during the day (for instance, a group in more pain
and therefore more “motivated” to travel to the hospital at
night).
October 2010 • Volume 111 • Number 4
Our aim in this analysis was not to study the chronop-
harmacology of systemic hydromorphone and intrathecal
fentanyl per se, but to determine whether time-related
effects confounded our original study results, as some
authors have suggested they might. Time-related variables
do not seem to have played a major role in our previous
study. There were no time-related differences in analgesia
duration, and time-related differences that we did observe,
such as differences in cervical dilation and pain scores,
were minimal. Furthermore, our study design does not
allow us to draw conclusions regarding the cause for these
observed differences; they may be related to true chrono-
biology or to environmental factors that influence human
behavior. Chronobiology as a possible confounding factor
in labor analgesia trials requires further examination.
ACKNOWLEDGMENTS
The authors thank an anonymous reviewer for his insight into
the analysis and presentation of the data in this article.
REFERENCES
1. Chassard D, Bruguerolle B. Chronobiology and anesthesia.
Anesthesiology 2004;100:413–27
2. Debon R, Chassard D, Duflo F, Boselli E, Bryssine B,
Allaouchiche B. Chronobiology of epidural ropivacaine:
variations in the duration of action related to the hour of
administration. Anesthesiology 2002;96:542–5
3. Debon R, Boselli E, Guyot R, Allaouchiche B, Lemmer B,
Chassard D. Chronopharmacology of intrathecal sufentanil
for labor analgesia: daily variations in duration of action.
Anesthesiology 2004;101:978 – 82
4. Pan PH, Lee S, Harris L. Chronobiology of subarachnoid
fentanyl for labor analgesia. Anesthesiology 2005;103:595–9
5. Wong CA, Scavone BM, Peaceman AM, McCarthy RJ, Sullivan
JT, Diaz NT, Yaghmour E, Marcus RJ, Sherwani SS, Sproviero
MT, Yilmaz M, Patel R, Robles C, Grouper S. The risk of
cesarean delivery with neuraxial analgesia given early versus
late in labor. N Engl J Med 2005;352:655– 65
6. Cleveland WS, Devlin SJ. Locally weighted regression: an ap-
proach to regression analysis by local fitting. JASA 1988;83:596–610
7. Aya AG, Vialles N, Mangin R, Robert C, Ferrer JM, Ripart J, de
La Coussaye JE. Chronobiology of labour pain perception: an
observational study. Br J Anaesth 2004;93:451–3
8. Lindow SW, Newham A, Hendricks MS, Thompson JW, van der
Spuy ZM. The 24-hour rhythm of oxytocin and beta-endorphin
secretion in human pregnancy. Clin Endocrinol (Oxf) 1996;45:443–6
9. Odrcich M, Bailey JM, Cahill CM, Gilron I. Chronobiological
characteristics of painful diabetic neuropathy and postherpetic
neuralgia: diurnal pain variation and effects of analgesic
therapy. Pain 2006;120:207–12
10. Zubieta JK, Yau WY, Scott DJ, Stohler CS. Belief or need?
Accounting for individual variations in the neurochemistry of
the placebo effect. Brain Behav Immun 2006;20:15–26
www.anesthesia-analgesia.org 991
 CASE REPORT

Epidural Hematoma Nine Days After Removal of a

Labor Epidural Catheter
Patrick J. Guffey, MD, Warren R. McKay, MD, and Rachel Eshima McKay, MD

Timely recognition and surgical decompression are crucial to minimize risk of permanent
neurologic deficit from epidural hematoma. We present the case of a patient who developed
acute back pain, sensory deficit, and ascending weakness 9 days after removal of a labor
epidural catheter. Magnetic resonance imaging revealed a heterogeneous fluid collection
extending from C6-7 through the lumbar region, with cord deformity at T9-11. Decompression
laminectomy was performed within 4 hours of symptom onset. Twelve hours later, her motor
function had fully recovered. Subsequent anatomic and hematologic workup was inconclu-
sive. This presentation is atypical given the delayed presentation of symptoms after epidural
placement. (Anesth Analg 2010;111:992–5)

E pidural hematoma related to neuraxial anesthesia is

a rare but potentially devastating complication; pub-
lished reports in patients undergoing epidural cath-
eter placement estimate an incidence of about 1 case in
150,000 anesthetics.1–3 Most often, symptoms arise within a
few hours after placement or removal of the catheter.3
Cases of spontaneous epidural hematoma occurring in
pregnancy are exceedingly rare; only 6 have been re-
ported in the English-language literature, only 1 of which
occurred after delivery.3–7 We found no previous case
reports of epidural hematoma presenting more than 1
week after catheter removal. We report a case of an
epidural hematoma in a previously healthy woman who
presented 9 days after labor and vaginal delivery with
epidural analgesia. Written informed consent was ob-
tained from the patient before preparation and submis-
sion of the manuscript.
CASE DESCRIPTION
A 32-year-old, G1P0 woman with body mass index of 27
kg/m2 underwent epidural catheter placement during la-
bor. Localization of the epidural space required 2 passes in
the midline of the L2-3 interspace with an 18-gauge Tuohy
needle. After the epidural space was identified by loss of
resistance to saline at 7 cm from the skin, a flexible,
wire-embedded catheter (FlexTip Plus®; Arrow Interna-
tional, Reading, PA) was passed without resistance, pares-
thesia, or discomfort to a distance 6 cm beyond the tip of
the needle. The catheter was secured at 13 cm at the skin.
Aspiration through the catheter with a 3-mL syringe failed
to yield blood or fluid. Lidocaine 45 mg and epinephrine 15
␮g were administered without change in heart rate or
initiation of motor or sensory loss. Subsequent analgesia
From the Department of Anesthesia and Perioperative Care, University of
California San Francisco, San Francisco, California.
Accepted for publication June 3, 2010.
Disclosure: The authors report no conflicts of interest.
Reprints will not be available from the author.
Address correspondence to Rachel Eshima McKay, MD, Department of
Anesthesia and Perioperative Care, University of California San Francisco,
521 Parnassus Ave., C450, San Francisco, CA 94143-0648. Address e-mail to
eshimar@anesthesia.ucsf.edu.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181effd8f
was effective, and the remainder of labor and delivery (5
hours) progressed without complication. Shortly after de-
livery, the patient developed profuse vaginal bleeding and
was taken to the operating room for cervical dilation and
curettage; surgical anesthesia was successfully adminis-
tered via the in situ epidural catheter. The cause of the
bleeding was not clearly attributed to retained placental
fragments or cervical tear. Total estimated blood loss was
750 mL, and 2.5 L lactated Ringer solution was adminis-
tered. After the procedure, the epidural catheter was re-
moved, and the patient was discharged home 2 days later
in excellent condition without complaints.
Nine days later, she presented to the emergency depart-
ment with severe low back pain and acute, ascending lower
extremity paralysis that began 2 hours before arrival.
Immediate evaluation by obstetric and anesthesia physi-
cians revealed arterial blood pressure of 190/100 mm Hg
and lower extremity paralysis and sensory deficit ascend-
ing to the umbilicus. Laboratory values included hemato-
crit 30%, platelet count 358 ⫻ 109/␮L, prothrombin time
(PT) 14.0 seconds (normal, 12.5–16.0 seconds; international
normalized ratio, 1.0), and fibrinogen 236 mg/dL (normal,
202– 430 mg/dL). Dexamethasone 12 mg was administered
IV and urgent neurosurgical consultation was requested.
Magnetic resonance imaging revealed a large, cylindrical
heterogeneous fluid collection around the spine extending
from C6-7 through the lumbar region, with significant mass
effect, cord deformity, and T2 signal prolongation at T9-11;
the majority of the blood had collected between T6 and T12
(Fig. 1). Within 2 hours of her arrival at the hospital, she
was taken to the operating room to undergo decompression
surgery. Baseline laboratory values revealed a hematocrit
of 28%, and platelet count, PT, and partial thromboplastin
time within normal limits. Neither thromboelastography
nor platelet function assay was available in our institution.
Because of ongoing blood loss and oozing in the field, 1
pheresis pack of platelets and 4 U fresh frozen plasma were
administered, resulting in satisfactory hemostasis. Three
units of packed red blood cells were administered during
surgery, and at the conclusion of the 4-hour case, blood loss
was estimated to be 1.5 L. The paralysis resolved com-
pletely over the next 12 hours, and the sensory examination
returned to normal over the next week with the exception
of an area of paresthesia along the medial aspect of her
lower right medial leg.

992 www.anesthesia-analgesia.org October 2010 • Volume 111 • Number 4

Hematoma After Labor Epidural Removal

Figure 1. A, Sagittal T1-weighted magnetic resonance image showing presence of extensive hematoma around the spinal cord, visible here
from C7 to T12 and below. Blood here is visible anterior to the cord, and compression is visible most prominently between T9 to T12. B, Axial
T2-weighted magnetic resonance image showing hematoma surrounding the spinal cord at the level of T12. C, Sagittal T2-weighted magnetic
resonance image demonstrating extension of blood throughout the thoracic and lumbar region (T12 to sacrum shown).

Neuroangiographic evaluation for spinal arteriovascular

malformation was negative. A subsequent detailed history,
performed in conjunction with the hematology service, did
not reveal personal or family history of abnormal bleeding.
The patient was noted to have used ibuprofen 800 mg every
8 hours for 9 days after delivery for low back and general-
ized musculoskeletal pain. Further laboratory workup con-
ducted 2 weeks after hematoma evacuation included a
ristocetin cofactor assay of 88% (reference range, 42%–200%),
fibrinogen of 301 mg/dL (normal range, 202– 430 mg/dL),
and factor VIII activity of 162% (normal, 43%–168%). These
values were all within 2 standard deviations of mean
values at our institution for a patient who is not pregnant.
At 2-month follow-up, the patient had made a complete
neurologic recovery.

DISCUSSION
Spinal epidural hematoma is a rare but potentially devas-
tating event, classified as spontaneous (occurring without
apparent cause or with delayed onset after minor injury), or
secondary to identifiable cause, including spinal or epi-
dural anesthesia. Patients at higher risk for hematoma after
epidural anesthesia include those with advanced age, spi-
nal stenosis, coagulopathy (longstanding or acute), platelet
inhibition, arteriovenous malformation, or multiple punc-
tures.8 Overall, hematomas related to neuraxial anesthesia
seem to be less frequent in the obstetrical compared with
the elderly surgical population.2 There are also rare reports
of spontaneous epidural hematoma in parturients that
occur in the absence of neuraxial block, and delay in
recognition and management has resulted in devastating
consequences.2,9 This particular case was unusual because
it occurred more than 1 week after epidural catheter
removal, beyond the timeframe previously described after
an inciting event.10
A systematic review of 613 reported cases of spinal and
epidural hematoma demonstrates that trauma, neuraxial
anesthesia, coagulopathy, arteriovenous malformation, or a
combination of these factors, are frequently associated with
hematoma; 30% were found to have no identifiable cause.3
In all, 63 cases were associated with neuraxial anesthesia.
Of these 63 cases, coagulation status was known in 49; 34 of
Figure 2. Graph showing hours between initiation of neuraxial
procedure (lumbar puncture or spinal/epidural anesthesia) and
symptoms of epidural hematoma. Of the 63 cases of epidural
hematoma associated with spinal or epidural anesthesia among
613 reported, 49 had documented evidence confirming or refuting
presence of antiplatelet or anticoagulant medication use, coagulopa-
thy, or platelet abnormality. (The graph was constructed from data
from Kreppel et al.3)
49 had confirmed coagulopathy or were receiving concur-
rent anticoagulation therapy; and 41 of these 49 patients
developed symptoms within 72 hours of the neuraxial
procedure (Fig. 2).3 Five pregnant patients were among this
group of 63; 1 had an elevated PT from hepatic disease
related to the pregnancy, another was classified as hyper-
tensive, and the remaining 3 had no identifiable risk
factors.
What may have put our patient at risk? The presence of
postpartum hemorrhage may have been related to numer-
ous factors other than coagulation disorder, but could also
have been related to a subtle disorder of coagulation that
was not detected during her workup after surgery.11,12
Although the values of her PT, fibrinogen, and subsequent
ristocetin cofactor assay and factor VIII activity were within

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 993

 CASE REPORT
the normal range for a nonpregnant patient, these may
have been falsely normal given the typical changes in
coagulation state that accompany pregnancy and the early
postpartum period, when circulating levels of clotting
factors may be 20% to 200% above nonpregnant levels and
do not fully return to nonpregnancy baseline values until 8
weeks postpartum.13
Additionally, we questioned whether the hematoma
could have been related to the patient’s nonsteroidal anti-
inflammatory drug (NSAID) use, or to NSAIDs in combi-
nation with a subtle impairment in coagulation function.
Antiplatelet drugs, including nonselective NSAIDs and
aspirin, taken within 1 week of surgery, have been associ-
ated with increased risk of hematoma after craniotomy and
hip arthroplasty.14,15 However, there is consensus among
experts that NSAID use does not increase risk of epidural
hematoma. This consensus is based on prospective studies
of hundreds of NSAID users undergoing neuraxial anes-
thesia8 and epidural steroid injection.11 Use of throm-
boelastography or other platelet function assays at the time
of presentation might have established whether NSAID use
contributed to the development of this epidural hematoma.
A more thorough hematologic evaluation, performed fur-
ther out from the peripartum period, would be needed to
exclude an underlying bleeding disorder; to date, the
patient has declined further workup.
In either case, it is possible that inadequate hemostasis
caused gradual expansion of a small collection of blood that
was initiated by epidural placement or removal. Over days,
this small asymptomatic hematoma may have slowly con-
tinued to bleed and expand, eventually reaching a critical
size sufficient to cause dramatic symptoms. The predomi-
nantly thoracic location of the hemorrhage, higher than the
site of epidural puncture, does not argue against the
epidural catheter as the primary cause. First, the catheter
tip may have extended a considerable distance in a cepha-
lad direction, and catheter removal may have initiated
trauma at that location. Second, blood from a hematoma
secondary to needle trauma in the lumbar region may have
collected preferentially in the thoracic epidural space be-
cause of anatomic factors such as the relatively narrower
dimension of the spinal cord, convexity of the thoracic
spine, the relatively lower intrathoracic pressure, and
greater capacity of the thoracic epidural space.16,17
Given the prolonged period of time between catheter
manipulation and presentation of symptoms, it is possible
that this patient’s condition resulted from factors other than
epidural catheter placement. However, we think it more
likely that the event was related to epidural catheter
placement/removal, with possible contributing factors in-
cluding elevated venous pressure related to the pregnancy,
platelet inhibition from persistent NSAID use, and hyper-
tension. The relative contribution of these various factors, if
any, however, remains undetermined.
Pregnancy is characterized by a relatively hypercoagu-
lable state, effectively reducing the risk of epidural hema-
toma. However, the epidural space houses an extensive
venous system containing tributaries from the spinal cord
and vertebral bodies that drain into the external vertebral
venous plexus and become more engorged during preg-
nancy. Because this venous system does not contain valves,
994 www.anesthesia-analgesia.org
and exists within a low-pressure environment, any pres-
sure exerted proximally because of a Valsalva maneuver,
vomiting, or mechanical compression of the vena cava can
dramatically increase the transmural venous pressure, leav-
ing the veins vulnerable to injury. It has been postulated
that spontaneous epidural hematoma may occur from
preexisting faults in the venous wall exacerbated by well-
described mechanical and hyperdynamic conditions exist-
ing in the peripartum period.4 –7,18
Numerous studies have demonstrated the most favor-
able outcomes when a symptomatic hematoma is decom-
pressed within 36 hours, and some authors suggest even
faster intervention, within 6 hours.19,20 In this case, inter-
vention occurred within 4 hours of the onset of symptoms.
Through emergent, definitive treatment, the patient made a
full recovery, illustrating the importance of prompt recog-
nition and treatment of epidural hematoma to maximize
the patient’s chance of a favorable outcome.
AUTHOR CONTRIBUTIONS
PJG helped with patient consent, data collection, and manu-
script preparation; WRM helped with manuscript preparation;
and REM helped with data collection, manuscript preparation,
and construction of figures.
REFERENCES
1. Loo CC, Dahlgren G, Irestedt L. Neurological complications in
obstetric regional anaesthesia. Int Obstet Anesth 2000;9:99 –124
2. Kopp SL, Horlocker TT. Anticoagulation in pregnancy and
neuraxial blocks. Anesthesiol Clin 2008;26:1–22
3. Kreppel D, Antoniadis G, Seeling W. Spinal hematoma: a
literature survey with meta-analysis of 613 patients. Neurosurg
Rev 2003;26:1– 49
4. Bidzinski J. Spontaneous spinal epidural hematoma during
pregnancy: case report. J Neurosurg 1966;24:1017
5. Yonekawa Y, Mehdorn HM, Nishikawa M. Spontaneous spinal
epidural hematoma during pregnancy. Surg Neurol 1975;3:
327– 8
6. Carroll SG. Spontaneous spinal extradural hematoma during
pregnancy. J Matern Fetal Med 1997;6:218 –9
7. Bose S, Ali Z, Rath P, Prabhakar H. Spontaneous spinal
haematoma: a rare cause of quadriplegia in the post-partum
period. Br J Anaesth 2007;99:855–7
8. Horlocker TT, Wedel DJ, Schroeder DR, Rose SH, Elliot BA,
McGregor DG, Wong GY. Preoperative anti-platelet therapy
does not increase the risk of spinal hematoma associated with
regional anesthesia. Anesth Analg 1995;80:303–9
9. Doblar DD, Schumacher SD. Spontaneous acute thoracic epi-
dural hematoma causing paraplegia in a patient with severe
preeclampsia in early labor. Int J Obstet Anesth 2005;14:256 – 60
10. Pear BL. Spinal epidural hematoma. Am J Roentgenol Radium
Ther Nucl Med 1972;155:155– 64
11. Horlocker TT, Bajwa ZH, Ashraf Z, Khan S, Wilson JL, Sami N,
Peeters-Asdourian C, Powers CA, Schroeder DR, Decker PA,
Warfield CA. Risk assessment of hemorrhagic complications
associated with nonsteroidal anti-inflammatory medications in
ambulatory pain clinic patients undergoing epidural steroid
injection. Anesth Analg 2002;95:1691–7
12. Kadir RA, Kingmam CEC, Chi C, Lee CA, Economides DL. Is
primary postpartum haemorrhage a good predictor of inher-
ited bleeding disorders? Haemophilia 2007;13:178 – 81
13. Bremme KA. Haemostatic changes in pregnancy. Best Pract
Res Clin Haematol 2003;16:153– 68
14. Palmer JD, Sparrow OC, Ianotti F. Postoperative hematoma: a
5-year survey and identification of risk factors. Neurosurgery
1994;35:1061–5
15. Robinson CM, Christie J, Malcolm-Smith N. Nonsteroidal
anti-inflammatory drugs, perioperative blood loss, and trans-
fusion requirements in elective hip arthroplasty. J Arthroplasty
1993;8:607–10
ANESTHESIA & ANALGESIA
Hematoma After Labor Epidural Removal
16. Visser WA, Liem TH, van Egmond J, Gielen MJ. Extension of
sensory blockade after thoracic epidural administration of a
test dose of lidocaine at three different levels. Anesth Analg
1998;86:332–5
17. Igarashi T, Hirabayashi Y, Shimizu R, Saitoh K, Fukuda H.
Thoracic and lumbar extradural structure examined by extra-
duroscope. Br J Anaesth 1998;81:121–5
18. Beatty RM, Winston KR. Spontaneous cervical epidural
haematoma: a consideration of etiology. J Neurosurg 1984;61:
143– 8
October 2010 • Volume 111 • Number 4
19. Groen RJ, van Alphen HA. Operative treatment of sponta-
neous spinal epidural hematomas: a study of the factors
determining postoperative outcome. Neurosurgery 1996;39:
494 –508
20. Lawton MT, Porter RW, Heiserman JE, Jacobowitz R, Sonntag
VK, Dickman CA. Surgical management of spinal epidural
hematoma: relationship between surgical timing and neuro-
logical outcome. J Neurosurg 1995;83:1–7
www.anesthesia-analgesia.org 995
Society for Pediatric Anesthesia
Section Editor: Peter J. Davis

Ipsilateral Transversus Abdominis Plane Block

Provides Effective Analgesia After Appendectomy in
Children: A Randomized Controlled Trial
John Carney, MB,*† Olivia Finnerty, MB, FCARCSI,*† Jassim Rauf, MB,† Gerard Curley, MB, FCARCSI,*†
John G. McDonnell, MB, FCARCSI,*† and John G. Laffey, MD, MA, BSc, FCARCSI*†‡

BACKGROUND: The transversus abdominis plane (TAP) block provides effective postoperative
analgesia in adults undergoing major abdominal surgery. Its efficacy in children remains unclear,
with no randomized clinical trials in this population. In this study, we evaluated its analgesic
efficacy over the first 48 postoperative hours after appendectomy performed through an open
abdominal incision, in a randomized, controlled, double-blind clinical trial.
METHODS: Forty children undergoing appendectomy were randomized to undergo unilateral TAP
block with ropivacaine (n ⫽ 19) versus placebo (n ⫽ 21) in addition to standard postoperative
analgesia comprising IV morphine analgesia and regular diclofenac and acetaminophen. All
patients received a standard general anesthetic, and after induction of anesthesia, a TAP block
was performed using the landmark technique with 2.5 mg 䡠 kg⫺1 ropivacaine 0.75% or an equal
volume (0.3 mL 䡠 kg⫺1) of saline on the ipsilateral side to the incision.
RESULTS: The TAP block with ropivacaine reduced mean (⫾SD) morphine requirements in the
first 48 postoperative hours (10.3 ⫾ 12.7 vs 22.3 ⫾ 14.7 mg; P ⬍ 0.01) compared with placebo
block. The TAP block also reduced postoperative visual analog scale pain scores at rest and on
movement compared with placebo. Interval morphine consumption was reduced over the first 24
postoperative hours. There were no between-group differences in the incidence of sedation or
nausea and vomiting. There were no complications attributable to the TAP block.
CONCLUSIONS: Unilateral TAP block, as a component of a multimodal analgesic regimen,
provided superior analgesia compared with placebo in the first 48 postoperative hours after
appendectomy in children. (Anesth Analg 2010;111:998 –1003)

A ppendectomy is one of the most frequently per-

formed surgical procedures in children and is asso-
ciated with significant postoperative discomfort and
pain.1 Multimodal approaches to the provision of postopera-
tive analgesia often incorporate blockade of the abdominal
wall, such as ilioinguinal blockade2 or wound infiltration.3 How-
ever, the efficacy of these approaches is unclear.3,4 The optimal
approach to the blockade of the abdominal wall in children
undergoing appendectomy remains to be determined.
An important component of the pain experienced by
children after appendectomy derives from the abdominal
wall incision. The nerves that supply the abdominal wall
course through the neurofascial transversus abdominis
From the *Department of Anaesthesia, Clinical Sciences Institute, National
University of Ireland, Galway; and †Department of Anaesthesia and Inten-
sive Care Medicine, and ‡Clinical Research Facility, Galway University
Hospitals, Galway, Ireland.
Accepted for publication May 26, 2010.
Supplemental digital content is available for this article. Direct URL citations
appear in the printed text and are provided in the HTML and PDF versions
of this article on the journal’s Web site (www.anesthesia-analgesia.org).
Funded by departmental resources.
Disclosure: The authors report no conflicts of interest.
Reprints will not be available from the author.
Address correspondence to John G. Laffey, MD, MA, BSc, FCARCSI,
Department of Anaesthesia, Clinical Sciences Institute, National University
of Ireland, Galway, Ireland. Address e-mail to john.laffey@nuigalway.ie.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181ee7bba
plane (TAP) between the internal oblique and the transver-
sus abdominis muscles.5 Using anatomic studies, our group
identified the lumbar triangle of Petit as an access point for
introducing local anesthetic drugs into the TAP.6 This
triangle is bounded posteriorly by the latissimus dorsi
muscle, anteriorly by the external oblique, and inferiorly by
the iliac crest.7 The floor of the triangle, from superficial to
deep, is composed of subcutaneous tissue, and the fascial
borders of the external oblique, the internal oblique, and
the transversus abdominis muscles, respectively.
The efficacy of the TAP block in providing postoperative
analgesia has been shown in adults undergoing bowel
surgery,8 cesarean delivery,9 and total abdominal hysterec-
tomy.9 Most recently, an ultrasound-guided approach to
the TAP block has been described in a series of 8 children
undergoing inguinal hernia repair10 and has shown anal-
gesic efficacy in a randomized controlled trial for appen-
dicectomy in adults.11 However, there are no randomized
controlled clinical trials demonstrating the efficacy of the
TAP block in children. We hypothesized that, compared
with a placebo block, the TAP block, as part of a multimo-
dal analgesic regimen, would result in decreased opioid
consumption and improved analgesia in the first 48 hours
for children undergoing appendectomy.
METHODS
After obtaining approval by the Hospital Ethics Committee,
and written informed parental consent, we studied children

998 www.anesthesia-analgesia.org October 2010 • Volume 111 • Number 4

younger than 16 years, ASA grade I to III, scheduled for
emergency open appendectomy, in a randomized, double-
blind, controlled clinical trial. Patients were excluded if
there was a history of relevant drug allergy, if they were
receiving medical therapies considered to result in toler-
ance to opioids, or if they were deemed to be unable to
independently assess their pain.
Patients were randomly allocated to undergo TAP block
with 2.5 mg 䡠 kg⫺1 ropivacaine 0.75% (to a maximal dose of
150 mg) on the right side or TAP block with an equal
volume (0.3 mL 䡠 kg⫺1) of 0.9% saline. The allocation se-
quence was generated by a random number table, and
group allocation was concealed in sealed, opaque enve-
lopes, which were not opened until patient consent had
been obtained. The patients, their anesthetists, and staff
providing postoperative care were blinded to group assign-
ment. All patients received a standardized rapid sequence
induction of anesthesia. After application of standard
monitoring and oxygen administration for 3 minutes, an-
esthesia was induced with propofol (2–3 mg 䡠 kg⫺1), cricoid
pressure was applied and muscle relaxation was produced
with succinylcholine (1–1.5 mg 䡠 kg⫺1), and the trachea was
intubated. Anesthesia was maintained using 1 to 1.5 mini-
mum alveolar concentration sevoflurane in oxygen and air.
All patients also received IV morphine 0.15 mg 䡠 kg⫺1,
rectal diclofenac 1 mg 䡠 kg⫺1, and rectal acetaminophen 20
mg 䡠 kg⫺1 immediately before surgical incision. Prophylac-
tic antiemetics were not administered.
The TAP block was performed after induction of anes-
thesia but before surgical incision by 1 of 2 investigators
(JC, OF) using the following technique.8 A 22-gauge 25-mm
or 50-mm blunted regional anesthesia needle (Plexufix威; B.
Braun, Melsungen AG, Germany) was attached with flex-
ible tubing to a syringe filled with the study solution. With
the patient in a supine position and the investigator stand-
ing on the contralateral side, the iliac crest was palpated
from anterior to posterior until the latissimus dorsi muscle
insertion was appreciated. The triangle of Petit was pal-
pated between the anterior border of latissimus dorsi
muscle, the posterior border of the external oblique muscle,
and the iliac crest. The skin over the triangle of Petit was
pierced with the needle held at right angles to the coronal
plane. The needle was stabilized and advanced at right
angles to the skin in a coronal plane until resistance was
encountered. This first resistance indicated that the needle
tip was abutting the fascial extension of the external
oblique muscle. Further gentle advancement of the needle
resulted in a loss of resistance, or “pop” sensation, as the
needle entered the plane between the external and internal
oblique fascial layers. Further gentle advancement resulted
in the appreciation of a second increased resistance and its
loss indicated entry into the TAP. After careful aspiration to
exclude vascular puncture, a test dose of 1 mL was injected.
The presence of substantial resistance to this injection
indicates the needle is not between fascial planes and the
needle should be repositioned. The study solution (0.3
mL 䡠 kg⫺1 ropivacaine 7.5 mg 䡠 mL⫺1 or saline 0.9%) was
injected through the needle in increments while observing
closely for signs of toxicity.
October 2010 • Volume 111 • Number 4
After completion of the surgical procedure, patients
were transferred to the postanesthesia care unit. All pa-
tients received oral acetaminophen 20 mg 䡠 kg⫺1 every 6
hours and rectal diclofenac 1 mg 䡠 kg⫺1 every 12 hours
postoperatively. Children older than 8 years were given IV
morphine patient-controlled analgesia (PCA) (bolus dose
20 ␮g 䡠 kg⫺1; lockout 6 minutes). Children younger than 8
years were given nurse-administered IV morphine (20
␮g 䡠 kg⫺1 bolus) on demand. The presence and severity of
pain, nausea, and sedation were assessed systematically by
an investigator blinded to group allocation. These assess-
ments were performed in the postanesthesia care unit and
at 2, 4, 6, 12, 24, 36, and 48 hours after TAP blockade. All
patients were asked to give scores for their pain at rest and
on movement (knee flexion) and for the degree of nausea at
each time point. Pain severity was measured using a visual
analog scale (VAS) with superimposed faces pain severity
scale (10-cm line in which 0 cm ⫽ no pain and10 cm ⫽
worst pain imaginable) and a categorical pain scoring
system (none ⫽ 0, mild ⫽ 1, moderate ⫽ 2, and severe ⫽ 3).
Nausea was measured using a categorical scoring system
(none ⫽ 0, mild ⫽ 1, moderate ⫽ 2, and severe ⫽ 3). The
patient was deemed to have been nauseated if they had a
nausea score ⬎0 at any postoperative time point. Sedation
scores were assigned by the investigator using a sedation
scale (awake and alert ⫽ 0, quietly awake ⫽ 1, asleep but
easily roused ⫽ 2, and deep sleep ⫽ 3). The patient was
deemed to have been sedated if they had a sedation score
⬎0 at any postoperative time point. Rescue antiemetics
were offered to any patient who complained of nausea or
vomiting. The study ended 48 hours after TAP blockade.
The primary outcome measure in this study was 48-hour
morphine consumption. Secondary outcome measures in-
cluded time to first request for morphine, VAS scores, and
side effects associated with morphine consumption. A pilot
study of children undergoing open appendectomy found a
mean 48-hour morphine requirement of 24 mg, with a
standard deviation of 8 mg in the control group. We
intended to be able to detect a minimum 33% reduction in
morphine requirement in the patients receiving TAP block-
ade. Based on these projections, we calculated that at least
17 patients would be required per group for an experimen-
tal design incorporating 2 groups, with ␣ ⫽ 0.05 and ␤ ⫽
0.2. We therefore planned to recruit 40 patients into the
study.
Statistical analyses were performed using a standard
statistical program (SigmaStat 3.5; Systat Software, San
Jose, CA). Demographic data were analyzed using Student
t, ␹2, or Fisher exact tests as appropriate. The data were
tested for normality using the Kolmogorov-Smirnov nor-
mality test. Repeated measurements (pain scores, nausea
scores) were analyzed by repeated-measures analysis of
variance where normally distributed, with further paired
comparisons at each time interval performed using the t
test. For non-normally distributed data, between-group
comparisons at each time point were made using Wilcoxon
rank sum test. Categorical data were analyzed using the ␹2
analysis or Fisher exact test. The time to first request for
morphine was analyzed using the log rank test. Normally
distributed data are presented as mean ⫾ SD, non-normally
distributed data are presented as median (interquartile
www.anesthesia-analgesia.org 999
Transversus Abdominis Plane Block in Children

undergo TAP blockade with normal saline. Groups were

Table 1. Baseline Patient Characteristics comparable in terms of age, which ranged from 4 to 16
Group years in the TAP group and 5 to 16 years in the control
Control TAP block group. There were no differences between the groups
Characteristic (n ⴝ 21) (n ⴝ 19) regarding weight and height, history of prior abdominal
Weight (kg) 47.5 ⫾ 18.6 37.7 ⫾ 15.2 surgery, method of postoperative morphine administra-
Height (m) 1.5 ⫾ 0.2 1.4 ⫾ 0.2 tion, or surgical technique (Tables 1 and 2). A similar
Body mass index (kg/m2) 20.2 ⫾ 4.9 18.7 ⫾ 3.2
Prior abdominal surgery 关n (%)兴 0 (0) 0 (0) number of patients in both groups had appendicitis, with
Histologic diagnosis 14 (74%) in the TAP group versus 14 (67%) in the control
Appendicitis 14 14 group having a histologic diagnosis of appendicitis with or
Appendicitis with perforation 5 4 without perforation (Table 1). Five patients in the TAP
of appendix
group and 4 patients in the control group had histologic
TAP ⫽ transversus abdominis plane. evidence of perforation with peritonitis. In all patients, the
Continuous data are presented as mean ⫾ SD. Categorical variables are
presented as number and proportion.
triangle of Petit was located easily on palpation, the TAP
There were no significant differences between groups. was localized after 1 or 2 attempts, and the block performed
without complication.
Children undergoing unilateral TAP block with ropiva-
Table 2. Postoperative Analgesia, Nausea, caine had reduced 48-hour morphine requirements (Fig. 1)
and Sedation and increased time to first requirement for morphine (Fig.
Group 2). The total amount of morphine required in the first 48
Control TAP block
postoperative hours was 10.3 ⫾ 12.7 mg in the TAP group
(n ⴝ 21) (n ⴝ 19) compared with 22.3 ⫾ 14.7 mg (P ⬍ 0.01) in the control
Patient-controlled morphine 17 14 group. The median (interquartile range) time to first re-
analgesia quirement for morphine was 55 (30 –300) minutes in chil-
Nurse-controlled morphine 4 5 dren who received a TAP block compared with 16 (7–30)
analgesia minutes in the control group (P ⬍ 0.001). The TAP block
Interval morphine requirement
(␮g/kg)
with ropivacaine reduced cumulative postoperative mor-
0–6 h 117 (60–203) 55 (10–150)* phine consumption compared with placebo block at all
6–12 h 40 (0–115) 0 (0–17)* time points (Fig. 1). Interval morphine consumption was
12–24 h 60 (19–175) 0 (0–159)* also significantly lower at 6, 12, and 24 hours but not at the
24–36 h 0 (0–40) 0 (0–0)
later time points in the patients who had TAP blockade
36–48 h 0 (0–0) 0 (0–0)
Postoperative nausea scores (Table 2). There were no differences within either the TAP
0–6 h 0 (0–0) 0 (0–0) or control group regarding the amounts of morphine
6–12 h 0 (0–0) 0 (0–0) required between patients in whom the morphine was
12–24 h 0 (0–0) 0 (0–0) nurse administered versus those who received morphine
24–36 h 0 (0–0) 0 (0–0)
36–48 h 0 (0–0) 0 (0–0)
via PCA. In addition, morphine consumption within each
Postoperative sedation scores group did not differ significantly depending on whether or
2h 1 (1–1) 1 (0.5–1) not the patient had histologic evidence of appendicitis.
4h 1 (0.25–1) 1 (0–1) Postoperative VAS pain scores at rest and on movement
6h 1 (0–1) 1 (0–1)
were reduced after TAP block at all time points assessed
12 h 1 (0–2) 1 (0–1)
24 h 1 (0–1) 0 (0–0) (Figs. 3 and 4). Categorical pain scores were significantly
36 h 0 (0–0) 0 (0–0) lower in patients who received the TAP block at some but
48 h 0 (0–0) 0 (0–0) not all postoperative time points (data not shown). Pain
TAP ⫽ transversus abdominis plane. severity was low in both groups after the first 24 hours, and
The data are presented as medians (interquartile range). several patients were discharged in both groups between
*P ⱕ 0.05 (control versus TAP block). 24 and 48 hours.
There was no significant difference in the incidence of
range), and categorical data are presented as raw data and nausea or distribution of nausea scores between groups at
frequencies. The ␣ level for all analyses was set at P ⬍ 0.05, any time point (Table 2). The limited nausea experienced
and the Bonferroni correction for multiple comparisons resulted in low scores in both groups; the median and
was used where appropriate. interquartile scores are shown in Table 2. Six patients (27%)
in the control group and 4 patients (16%) in the TAP
RESULTS blockade group developed postoperative nausea. There
Forty-two children participated in this study. No patients was no significant difference in the incidence of sedation or
were excluded on the basis of the exclusion criteria (Ap- distribution of sedation scores between groups at any time
pendix Figure; see Supplemental Digital Content 1, point (Table 2).
http://links.lww.com/AA/A161). Two patients, 1 from
each group, were excluded after enrollment because of
postoperative analgesic protocol violations. Of the remain- DISCUSSION
ing 40 patients, 19 were randomized to undergo TAP Open appendectomy is one of the most frequently per-
blockade with ropivacaine and 21 were randomized to formed surgical procedures in the pediatric population

1000 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

Figure 1. A box plot of postoperative cumulative
morphine consumption in each group in the first 48
postoperative hours. The middle line in each box
represents the median value, the outer margins of
the box represent the interquartile range, and the
whiskers represent the 10th and 90th centile for
each time point. *Significantly (P ⬍ 0.05, Wilcoxon
rank sum test) higher morphine consumption com-
pared with the transversus abdominis plane (TAP)
block group.

Figure 2. A Kaplan-Meier graph of the proportion of

patients in each group over time that did not require
supplemental morphine (P ⫽ 0.004, log rank test).

worldwide and is a cause of significant pain in the postop-
erative period.1 The optimal analgesic regimen should
provide safe, effective analgesia, with minimal side effects
for the child. A multimodal analgesic regimen is most likely
to achieve these goals; however, the optimal components
remain to be determined. The TAP block provides blockade
of nociception from the abdominal wall; however, there is
also nociceptive input from the abdominal organs and the
onset of the block is not immediate. Therefore, the block is
used as part of a multimodal approach. This trial demon-
strated that an ipsilateral TAP block provides effective
analgesia in children undergoing open appendectomy.
A TAP block on the side of the surgical incision reduced
overall postoperative morphine requirements by approxi-
mately 50% and the interval morphine requirements for up
to 24 hours after surgery. Thereafter, morphine require-
ments were very low in both groups, with no morphine
consumed over the second 24 postoperative hours in any
patient studied. The TAP block delayed the time to first
request for supplemental opioid analgesia and reduced
pain scores at rest and on movement. Because the surgery
was 1 sided, the dose of local anesthetic used was limited to
2.5 mg 䡠 kg⫺1 on the ipsilateral side, which is within the
recommended safe dose range. Alternative abdominal tech-
niques, such as the ilioinguinal iliohypogastric nerve block,
have been performed in children at a dose of 5 mg 䡠 kg⫺1
ropivacaine without central nervous system or systemic
toxicity.12 The reduction in opioid use, coupled with the
reduction in postoperative pain, highlights the potential of
the TAP block in children undergoing appendectomy.

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 1001

Transversus Abdominis Plane Block in Children
These findings, when taken in context with the demon-
strated efficacy of the TAP block in adults,6,8 –10 attest that
the TAP block may provide effective postoperative analge-
sia for a wide variety of abdominal procedures in both
children and adults.
We used the landmark-based technique for the TAP
block, which was performed without difficulty in the
children in this study. Alternative approaches to the TAP
block using ultrasound guidance have recently been de-
scribed in a case series of children undergoing inguinal
hernia repair,10 in adults for appendicectomy,12 and an
adult study for laparoscopic cholecystectomy.11 The opti-
mal approach remains to be demonstrated. Other regional
anesthesia– based approaches for analgesia after appendec-
tomy include ilioinguinal/iliohypogastric nerve blockade
and local wound infiltration. However, in the case of
ilioinguinal/iliohypogastric block, the landmark technique
1002 www.anesthesia-analgesia.org
Figure 3. Box plots of postoperative visual analog
scale (VAS) pain scores at rest in each group over
the first 48 postoperative hours. The middle line in
each box represents the median value, the outer
margins of the box represent the interquartile
range, and the whiskers represent the 10th and
90th centile for each time point. *Significantly (P ⬍
0.05, Wilcoxon rank sum test) higher VAS pain
scores compared with the transversus abdominis
plane (TAP) block group.
Figure 4. Box plots of postoperative visual analog
scale (VAS) pain scores on movement in each group
over the first 48 postoperative hours. The middle
line in each box represents the median value, the
outer margins of the box represent the interquartile
range, and the whiskers represent the 10th and
90th centile for each time point. *Significantly (P ⬍
0.05, Wilcoxon rank sum test) higher VAS pain
scores compared with the transversus abdominis
plane (TAP) block group.
has been shown to be relatively unreliable, producing
deposition of local anesthetic solution in close proximity to
the nerves in as few as 14% of children.4 This may be
improved by the use of ultrasound to guide needle posi-
tion.4 Potentially serious complications, such as bowel
wall hematoma,13 have been reported after the use of
ilioinguinal/iliohypogastric blocks in children. In our ex-
perience, performing the TAP block in a pediatric popula-
tion has been technically easier than in adults because the
degree of obesity is usually less than that of an adult
population, there is a lesser degree of muscle laxity, and the
1-in. Plexufix needle that is available allows easier percep-
tion of the loss of resistance as described. Local infiltration
of the appendectomy wound with local anesthetic drugs is
also widely practiced. However, the efficacy of this ap-
proach is also unclear, with studies reporting no demon-
strable reduction in morphine consumption compared with
ANESTHESIA & ANALGESIA
standard care.3 Indeed, the technique has demonstrated to
be inconsistent with a lack of evidence after surgery on the
abdominal wall apart from herniorrhaphy.14
There are now a variety of techniques for the TAP block
and the analgesic merit of each is being elucidated in
ongoing studies. Although it is possible to ultrasonically
visualize the 3 muscle layers of the abdominal wall, there is
variation in these muscle layers that can restrict the use of
ultrasound over the triangle of Petit.15 As a result, the
needle insertion point as described in the ultrasound stud-
ies, which is dependent on the adequate identification of
the 3 muscle layers, can vary. This will alter the location of
the injectate as will the angle of the needle insertion to skin,
which contrasts to the landmark approach’s description.
Anteriorly placed injectate may not spread throughout the
TAP and does not according to ongoing work at this
institution. Although there is an ever-increasing access to
ultrasound, it is far from universal and there is a continuing
interest in landmark techniques.
There are a number of limitations to this study. First,
there are difficulties in adequately blinding these types of
studies because of the loss of sensation of the abdominal
wall. Although patients and the investigator conducting the
postoperative assessments were technically blinded to
group allocation, true blinding may not have been possible.
Second, 2 methods of postoperative administration of mor-
phine were used: patients younger than 8 years received
nurse-administered morphine whereas children older than
8 years received PCA morphine. However, there were no
differences in the proportions of each method between the
groups, so these differences do not seem to have contrib-
uted to the reported findings. Third, rates of appendicitis in
our study (66%–74%) are somewhat lower than those ob-
tained internationally.16 This likely reflects a lower threshold
for operative intervention in this pediatric population, as
evidenced by lower rates of appendiceal perforation in our
patients compared with those reported internationally.16
Fourth, the study was not large enough to independently
assess safety. There is a risk of inadvertent peritoneal
puncture with this block as with other abdominal
blocks.17,18 Although the incidence is not known, if the
block is performed as described, the risk of peritoneal
puncture is likely to be low. This institution has not
encountered complications relating to peritoneal puncture.
Finally, a dose-response study has not yet been performed
to determine whether effective postoperative analgesia
could be produced with a lower dose of ropivacaine.
In conclusion, the TAP block holds considerable promise
as part of a multimodal analgesic regimen for postappen-
dectomy analgesia. With experience, the TAP block is easy
to perform, has provided reliable and effective analgesia in
this study, and no complications from the TAP block were
detected.
ACKNOWLEDGMENTS
The authors thank the nursing staff on the pediatric ward for
their assistance with this study.
October 2010 • Volume 111 • Number 4
REFERENCES
1. Hale DA, Molloy M, Pearl RH, Schutt DC, Jaques DP. Appen-
dectomy: a contemporary appraisal. Ann Surg 1997;225:252– 61
2. Willschke H, Marhofer P, Bösenberg A, Johnston S, Wanzel O,
Cox SG, Sitzwohl C, Kapral S. Ultrasonography for
ilioinguinal/iliohypogastric nerve blocks in children. Br J
Anaesth 2005;95:226 –30
3. Jensen SI, Andersen M, Nielsen J, Qvist N. Incisional local
anaesthesia versus placebo for pain relief after appendectomy
in children: a double-blinded controlled randomised trial. Eur
J Pediatr Surg 2004;14:410 –3
4. Weintraud M, Marhofer P, Bösenberg A, Kapral S, Willschke
H, Felfernig M, Kettner S. Ilioinguinal/iliohypogastric blocks
in children: where do we administer the local anesthetic
without direct visualization? Anesth Analg 2008;106:89 –93
5. Netter FH. Back and spinal cord. In: Netter FH, ed. Atlas of
Human Anatomy. Summit, NJ: Ciba-Geigy Corporation,
1989:145–55
6. McDonnell JG, O’Donnell BD, Farrell T, Gough N, Tuite D,
Power C, Laffey JG. Transversus abdominis plane block: a
cadaveric and radiological evaluation. Reg Anesth Pain Med
2007;32:399 – 404
7. Netter FH. Abdomen posterolateral abdominal wall. In: Netter
FH, ed. Atlas of Human Anatomy. Summit, NJ: Ciba-Geigy
Corporation, 1989:230 – 40
8. McDonnell JG, O’Donnell B, Curley G, Heffernan A, Power C,
Laffey JG. The analgesic efficacy of transversus abdominis
plane block after abdominal surgery: a prospective random-
ized controlled trial. Anesth Analg 2007;104:193–7
9. Carney J, McDonnell JG, Ochana A, Bhinder R, Laffey JG. The
transversus abdominis plane block provides effective postop-
erative analgesia in patients undergoing total abdominal hys-
terectomy. Anesth Analg 2008;107:2056 – 60
10. Fredrickson M, Seal P, Houghton J. Early experience with the
transversus abdominis plane block in children. Paediatr An-
aesth 2008;18:891–2
11. Niraj G, Searle A, Mathews M, Misra V, Baban M, Kiani S,
Wong M. Analgesic efficacy of ultrasound-guided transversus
abdominis plane block in patients undergoing open appen-
dicectomy. Br J Anaesth 2009;103:601–5
12. Dalens B, Ecoffey C, Joly A, Giaufre E, Gustafsson U, Huledal
G, Larsson LE. Pharmacokinetics and analgesic effect of ropi-
vacaine following ilioinguinal/iliohypogastric nerve block in
children. Paediatr Anaesth 2001;11:415–20
13. Frigon C, Mai R, Valois-Gomez T, Desparmet J. Bowel hema-
toma following an iliohypogastric-ilioinguinal nerve block.
Paediatr Anaesth 2006;16:993– 6
14. Moiniche S, Mikkelsen S, Wetterslev J, Dahl JB. A qualitative
systematic review of incisional local anaesthesia for postopera-
tive pain relief after abdominal operations. Br J Anaesth
1998;81:377– 83
15. Loukas M, Tubbs RS, El-Sedfy A, Jester A, Polepalli S, Kinsela
C, Wu S. The clinical anatomy of the triangle of Petit. Hernia
2007;11:441– 4
16. Newman K, Ponsky T, Kittle K, Dyk L, Throop C, Gieseker K,
Sills M, Gilbert J. Appendicitis 2000: variability in practice,
outcomes, and resource utilization at thirty pediatric hospitals.
J Pediatr Surg 2003;38:372–9
17. Farooq M, Carey M. A case of liver trauma with a blunt
regional anesthesia needle while performing transversus ab-
dominis plane block. Reg Anesth Pain Med 2008;33:274 –5
18. Frigon C, Mai R, Valois-Gomez T, Desparmet J. Bowel hema-
toma following an iliohypogastric-ilioinguinal nerve block.
Paediatr Anaesth 2006;16:993– 6
www.anesthesia-analgesia.org 1003
Dexmedetomidine Infusion for Analgesia and
Prevention of Emergence Agitation in Children with
Obstructive Sleep Apnea Syndrome Undergoing
Tonsillectomy and Adenoidectomy
Anuradha Patel, MD, FRCA,* Melissa Davidson, MD,* Minh C. J. Tran, MD, MPH,*
Huma Quraishi, MD,† Catherine Schoenberg, BSN,* Manasee Sant, MD,* Albert Lin, MD,*
and Xiuru Sun, MS*

BACKGROUND: Dexmedetomidine, a specific ␣2 agonist, has an analgesic-sparing effect and

reduces emergence agitation. We compared an intraoperative dexmedetomidine infusion with
bolus fentanyl to reduce perioperative opioid use and decrease emergence agitation in children
with obstructive sleep apnea syndrome undergoing adenotonsillectomy (T&A).
METHODS: One hundred twenty-two patients with obstructive sleep apnea syndrome undergoing
T&A, ages 2 to 10 years, completed this prospective, randomized, U.S. Food and Drug
Administration–approved study. After mask induction with sevoflurane, group D received IV
dexmedetomidine 2 ␮g 䡠 kg–1 over 10 minutes, followed by 0.7 ␮g 䡠 kg–1 䡠 h–1, and group F
received IV fentanyl bolus 1 ␮g 䡠 kg–1. Anesthesia was maintained with sevoflurane, oxygen, and
nitrous oxide. Fentanyl 0.5 to 1 ␮g 䡠 kg–1 was given to subjects in both groups for an increase
in heart rate or systolic blood pressure 30% above preincision values that continued for 5
minutes. Observers in the postanesthesia care unit (PACU) were blinded to treatment groups.
Pain was evaluated using the objective pain score in the PACU on arrival, at 5 minutes, at 15
minutes, then every 15 minutes for 120 minutes. Emergence agitation was evaluated at the
same intervals by 2 scales: the Pediatric Anesthesia Emergence Delirium scale and a 5-point
scale described by Cole. Morphine (0.05 to 0.1 mg 䡠 kg–1) was given for pain (score ⬎4) or severe
agitation (score 4 or 5) lasting more than 5 minutes.
RESULTS: In group D, 9.8% patients needed intraoperative rescue fentanyl in comparison with
36% in group F (P ⫽ 0.001). Mean systolic blood pressure and heart rate were significantly lower
in group D (P ⬍ 0.05). Minimum alveolar concentration values were significantly different
between the 2 groups (P ⫽ 0.015). The median objective pain score was 3 for group D and 5 for
group F (P ⫽ 0.001). In group D, 10 (16.3%) patients required rescue morphine, in comparison
with 29 (47.5%) in group F (P ⫽ 0.002). The frequency of severe emergence agitation on arrival
in the PACU was 18% in group D and 45.9% in group F (P ⫽ 0.004); at 5 minutes and at 15
minutes, it was lower in group D (P ⫽ 0.028). The duration of agitation on the Cole scale was
statistically lower in group D (P ⫽ 0.004). In group D, 18% of patients and 40.9% in group F had
an episode of SPO2 below 95% (P ⫽ 0.01).
CONCLUSIONS: An intraoperative infusion of dexmedetomidine combined with inhalation anes-
thetics provided satisfactory intraoperative conditions for T&A without adverse hemodynamic
effects. Postoperative opioid requirements were significantly reduced, and the incidence and
duration of severe emergence agitation was lower with fewer patients having desaturation
episodes. (Anesth Analg 2010;111:1004 –10)

A denotonsillectomy (T&A) is one of the most com-

mon surgical procedures performed in children.
The presence of obstructive symptoms is replacing
recurrent tonsillitis as the primary indication for T&A. The
prevalence of obstructive sleep apnea syndrome (OSAS) in
children, also referred to as sleep disordered breathing, is
From the *Department of Anesthesiology and Perioperative Medicine and
the †Department of Otolaryngology, University of Medicine and Dentistry
of New Jersey, New Jersey Medical School, Newark, New Jersey.
Accepted for publication June 10, 2010.
Financial support was provided by an Institutional Cost of Drug support
grant from Hospira Worldwide, Lake Forest, Illinois.
Address correspondence and reprint requests to Dr. Anuradha Patel,
Associate Professor, Department of Anesthesiology and Perioperative Medi-
cine, New Jersey Medical School, University of Medicine and Dentistry of
New Jersey, Medical Science Building E-581, 185 South Orange Avenue,
Newark, NJ 07101. Address e-mail to patelan@umdnj.edu.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181ee82fa
estimated to be 1%–3%.1 Postoperative pain can be severe
after T&A, and providing effective and safe perioperative
analgesia in this group of patients is challenging. Not only
are children with OSAS undergoing T&A at significant risk
of respiratory and cardiovascular complications,2 they also
have enhanced analgesic sensitivity to opiates and reduced
morphine requirements after T&A.3 A high incidence of
emergence agitation (EA) in patients having otolarygologic
procedures adds another challenge.4 Dexmedetomidine
(Dex) (Precedex, Hospira Worldwide, Lake Forest, Illinois),
a specific ␣ 2-adrenergic receptor agonist, has sedative,
anxiolytic, and analgesic properties5 and is very effective in
prevention of EA in children.6,7 An intraoperative infusion
of Dex used as a substitute for fentanyl has been shown to
reduce opiate use in the postoperative period in adult
patients undergoing bariatric surgery,8 but clinical data on
the analgesic-sparing effect of Dex in children are conflict-
ing. The present study was performed to evaluate whether

1004 www.anesthesia-analgesia.org October 2010 • Volume 111 • Number 4

an intraoperative infusion of Dex combined with general
anesthesia would be a safe and effective substitute to
opiates intraoperatively, reduce opiate requirements post-
operatively, and also be effective in reducing the incidence
and severity of EA in children with OSAS undergoing
T&A.
METHODS
An Investigational New Drug number (76,041) was ob-
tained from the U.S. Food and Drug Administration. The
study was registered at www.clinicaltrials.gov (registration
number NCT00468052) and approved by the IRB of the
University of Medicine and Dentistry of New Jersey. One
hundred thirty-seven children ages 2 to 10 years, ASA
physical status II–III, undergoing elective T&A, were en-
rolled in this investigator-initiated, prospective, random-
ized, blinded, controlled study. Informed, written consent
to participate in the study was obtained from the parent or
legal guardian and assent from children older than 7 years
of age. All patients had OSAS on the basis of clinical
symptoms or diagnostic polysomnography. Clinical grad-
ing of OSAS was done by the surgeon on the basis of
severity of symptoms such as restless sleep, severe snoring,
apnea witnessed by the parents, nocturnal enuresis, stertor,
hyperactivity, or failure to thrive. Exclusion criteria were
known allergy to ␣2 agonists, developmental delay, cardiac
and craniofacial abnormalities, anxiety disorder, chronic
disabilities or pain syndrome, and use of psychotherapeutic
medications, ␤ blockers, digoxin, cimetidine, ␣2 agonists,
anticonvulsants, or psychotropic medications. A random
number table was used to assign subjects into 1 of 2
treatment groups: Dex infusion (group D) or IV fentanyl
(group F). The anesthesiologists and data collectors in the
operating room (OR) were not blinded; the subjects, their
parents, and observers in the postanesthesia care unit
(PACU) were blinded to treatment group.
No premedication was given. Monitoring included
pulse oximetry, electrocardiogram, noninvasive arterial
blood pressure (NIBP), end-tidal CO2 (etco2), and a depth
of anesthesia monitor, the Bispectral Index (BIS; Aspect
Medical Systems, Natick, Massachusetts). Anesthesia was
induced with 8% inspired sevoflurane and 60% nitrous
oxide (N2O) in oxygen by facemask. Group D received IV
Dex (2 ␮g 䡠 kg–1 over 10 minutes, followed by 0.7 ␮g 䡠 kg–1 䡠 h–1
until 5 minutes before the end of the surgery), and
group F received IV fentanyl (1 ␮g 䡠 kg–1) as a bolus, as soon
as IV access was obtained. A balanced salt solution was
administered according to standard fluid administration
guidelines. Rocuronium 0.6 mg 䡠 kg–1 was used to facilitate
tracheal intubation. End-tidal sevoflurane concentration
was maintained at 1 minimum alveolar concentration
(MAC) with 60% N2O as long as the BIS remained below 60
during surgery. If the BIS reached 60 or more, the sevoflu-
rane concentration was increased to reduce the BIS below
60. All patients received IV dexamethasone 0.5 mg 䡠 kg–1
(maximum 10 mg) and rectal acetaminophen 30 to 40 mg 䡠
kg–1 up to a maximum of 1000 mg before the start of
surgery. The data collector recorded the heart rate (HR),
systolic and diastolic blood pressures (NIBP), hemoglobin
oxygen saturation (Spo2), etco2 tension, MAC, and BIS
every 5 minutes during the anesthetic. The values in the
October 2010 • Volume 111 • Number 4
holding area for HR and systolic blood pressure were used
as baseline. Both groups received fentanyl 0.5 to 1 ␮g 䡠 kg–1
for an increase in HR or systolic NIBP 30% above the value
before start of surgery and sustained for 5 minutes. Lac-
tated Ringer’s solution 15 mL/kg was administered as a
fluid bolus for a 30% decrease of systolic blood pressure
from baseline, which continued for 2 readings and glyco-
pyrrolate 0.01 mg 䡠 kg–1 for a 30% decrease in HR. Sevoflu-
rane was discontinued once hemostasis was achieved and
muscle relaxation was reversed with neostigmine 0.05 mg 䡠
kg–1 and glycopyrrolate 0.01 mg 䡠 kg–1. The time to awak-
ening (TA), defined as spontaneous eye opening or on
command from end of surgery, and the time to extubation
(TE), defined as time from end of surgery to tracheal
extubation, were recorded. All patients were observed
continuously in the PACU for 2 hours by observers blinded
to study group. Pain was evaluated using the objective pain
score (OPS)9 in the PACU on arrival and at 5 minutes, at 15
minutes, and then every 15 minutes for 120 minutes. EA
was evaluated at the same intervals by 2 scales: the
Pediatric Anesthesia Emergence Delirium (PAED) scale10
and a 5-point agitation scale described by Cole.11 Duration
of severe EA was noted on the Cole scale. Morphine (0.05 to
0.1 mg 䡠 kg–1) was given for pain (score ⬎4) or severe
agitation (score 4 or 5) lasting more than 5 minutes. HR,
systolic and diastolic NIBP, respiratory rate (RR), and Spo2
were recorded in the PACU every 5 minutes for the first 15
minutes, then at 15-minute intervals for the next 2 hours.
Any desaturation episode with Spo2 below 95% was noted.
Statistical Methods
A power analysis indicated that 60 subjects were required
per group to show that the number of patients needing
intraoperative rescue fentanyl and rescue morphine in the
PACU would be 50% lower in the subjects receiving Dex.
Sixty subjects were also required per group to determine
that treatment with Dex would decrease the incidence of
severe EA after surgery by 50% with 80% power (␣ ⫽ 0.05)
in comparison with the control group.
Data were analyzed using SPSS software (version 16,
Chicago, Illinois), and are presented as number (n) or
percentage (%), mean ⫾ sd, or median as appropriate.
Student t test was used to compare the mean value of
quantitative data between the 2 groups. Two-way
repeated-measures analysis of variance (ANOVA) was
used for NIBP, HR, Spo2, MAC, and BIS. Student t test was
used for the comparisons of intragroup values of intraop-
erative and postoperative systolic blood pressure and HR.
Nonparametric data such as pain score, PAED score, and
EA score on the Cole scale were compared between groups
with Mann–Whitney U test. Fischer exact test was used for
comparision of gender; percentage of patients in each
group with a preoperative diagnosis of mild, moderate, or
severe OSAS; and number of patients rescued with fentanyl
or morphine and those with episodes of severe EA. P value
of 0.05 or less was considered statistically significant.
RESULTS
Results are presented for 122 patients. One hundred thirty-
seven subjects were enrolled in this study; 15 subjects were
eliminated from data analysis for the following reasons:
www.anesthesia-analgesia.org 1005
Dexmedetomidine Infusion for Adenotonsillectomy

Table 1. Demographic Data

Group F Group D
(n ⴝ 61) (n ⴝ 61) P value
Age (years) 3.8 ⫾ 1.5 4.2 ⫾ 2.1 0.16
2–3 years old (%) 26 (42.6) 26 (42.6) 1
Gender (M/F) 35/26 35/26 1
Weight (kg) 18.3 ⫾ 5.7 20.4 ⫾ 8.6 0.12
Baseline HR (beats/ 105 ⫾ 18 104 ⫾ 15 0.77
minute)
Baseline systolic 101 ⫾ 13.7 104 ⫾ 12.6 0.29
NIBP (mm Hg)
OSAS (% patients)
Mild 30 26
Moderate 50 60
Severe 20 14 0.55
Data are expressed as n (%) and mean ⫾ SD.
Group D ⫽ dexmedetomidine group; group F ⫽ fentanyl group; HR ⫽ heart
rate; NIBP ⫽ noninvasive arterial blood pressure; OSAS ⫽ obstructive sleep
apnea syndrome.

Table 2. Intraoperative Data

Group F Group D
(n ⴝ 61) (n ⴝ 61) P value
Rescue by fentanyl, 22 (36.1) 6 (9.8) 0.001*
n (%)
Fentanyl rescue 1.04 ⫾ 0.67 0.73 ⫾ 0.25 0.312
dosage (␮g/kg) Figure 1. A, Heart rate and B, systolic blood pressure during the first
Time of rescue 10.82 ⫾ 12.5 17.6 ⫾ 6.77 0.256 60 minutes of the procedure. Both variables were statistically lower
(minutes) in the dexmedetomidine group (P ⬍ 0.05).
Acetaminophen 31.51 ⫾ 4.96 28.30 ⫾ 6.59 0.02*
dosage (mg/kg)
Dexamethasone 0.30 ⫾ 0.12 0.30 ⫾ 0.14 0.847
dosage (mg/kg) different between the 2 groups (P ⫽ 0.015); MAC was lower
Duration of surgery 43.33 ⫾ 17.36 37.54 ⫾ 13.33 0.041* in group D, ranging from 5.7% to 41.6%. There was a
(minutes) statistical difference in TA and TE, both lower in group D
Duration of 75.08 ⫾ 24.73 69.80 ⫾ 16.82 0.175 than in group F (P ⬍ 0.05). Duration of surgery was
anesthesia
statistically lower in group D (P ⫽ 0.041). There was no
(minutes)
Time to awake 8.75 ⫾ 4.06 7.18 ⫾ 4.05 0.03* difference in the average dose of intraoperative fentanyl
(minutes) and dexamethasone between the 2 groups. The dose of
Time to extubate 10.44 ⫾ 4.15 8.59 ⫾ 4.51 0.02* acetaminophen was lower in group D. None of the subjects
(minutes) needed glycopyrrolate for bradycardia or fluid bolus for
Data are expressed as n, mean ⫾ SD, and percentage. hypotension in the OR.
Group D ⫽ dexmedetomidine group; group F ⫽ fentanyl group; HR ⫽ heart The variables measured in the PACU are shown in Table
rate; NIBP ⫽ noninvasive arterial blood pressure.
3. In group D 10 (16.3%) patients required rescue morphine,
* P ⬍ 0.05.
in comparison with 29 (47.5%) in group F (P ⫽ 0.002). The
median of the maximum OPS was 3 for group D and 5 for
surgery was cancelled for 2 patients, 1 refused to participate group F (P ⫽ 0.001). The percentage of patients with an
after enrolling, and 1 patient had an intraoperative complica- OPS score of 4 and above (Fig. 2A) from arrival (P ⫽ 0.001)
tion. Eleven subjects who completed the study had deviation and at 5 and 15 minutes was statistically lower in group D
from this strictly controlled protocol or incomplete data and (P ⬍ 0.05). On the Cole scale (5-point scale), severe EA was
were also removed before data analysis. defined as a score of 4 to 5. The frequency of severe EA is
The 2 groups were comparable in age, gender, baseline shown in Figure 2B. On arrival in the PACU it was
HR, systolic NIBP, and diagnosis of OSAS (Table 1). The statistically lower, 18% in group D and 45.9% in group F
age range of patients in the study was 2 to 10 years, 90% of (P ⫽ 0.004). At 5 and 15 minutes it was statistically lower in
patients were 6 years or younger, and 26 patients (46.2%) in group D (P ⫽ 0.028). At 30 minutes none of the patients had
each group were 2 to 3 years old. severe EA in group D, and 1.6% of patients in group F had
Intraoperative data are presented in Table 2. In group D, severe EA. The duration of agitation on the Cole scale
6 patients (9.8%) needed rescue fentanyl in comparison showed statistical significance; it was 6.59 ⫾ 7.4 minutes
with 22 (36%) in group F (P ⫽ 0.001). Mean HR (P ⫽ 0.001) (mean ⫾ sd) for group D and 11.85 ⫾ 12.0 minutes (mean ⫾
(Fig. 1A) and mean systolic NIBP (Fig. 1B) were signifi- sd) for group F (P ⫽ 0.004). There was a statistical differ-
cantly lower in group D during the first 60 minutes (P ⫽ ence in the median of the highest score on the Cole scale, 3
0.019). Mean diastolic NIBP was not statistically different in for group D and 4 for group F (P ⫽ 0.001). The percentage
the 2 groups (P ⫽ 0.29). During the first 60 minutes of the of patients with a score of 10 and above for the PAED (Fig.
anesthetic, MAC values of sevoflurane were significantly 2C) was statistically lower in group D at arrival (P ⫽ 0.004)

1006 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

 Table 3. Postanesthesia Recovery Unit Data
Group F Group D
(n ⴝ 61) (n ⴝ 61) P value
OPS maximum 5 (0–10) 3 (0–10) 0.001*
(range)
EA score 4 (1–5) 3 (1–5) 0.001*
maximum
(range)
Duration of 11.85 ⫾ 12.02 6.59 ⫾ 7.42 0.004*
severe EA
(minutes)
PAED score 14 (0–20) 10 (0–20) 0.051
maximum
(range)
Rescue by 29 (48) 10 (17) 0.0003*
morphine,
n (%)
Morphine dosage 0.073 ⫾ 0.033 0.074 ⫾ 0.033 0.928
(mg/kg)
SpO2 below 25 (41) 11 (18) 0.01*
95%, n (%)
OPS, PAED, and EA (Cole scale) are expressed as median values of the
maximum score.
Other data are expressed as n (%) and mean ⫾ SD.
Group D ⫽ dexmedetomidine group; group F ⫽ fentanyl group; OPS ⫽
objective pain score; EA ⫽ emergence agitation; PAED ⫽ pediatric anesthesia
emergence delirium.
* P ⬍ 0.05.

and at 5 and 15 minutes (P ⬍ 0.05). The median of the

highest score on the PAED scale did not show a statistical
difference, 10 for group D and 14 for group F (P ⫽ 0.051).
On arrival in the PACU until 90 minutes, HR was
statistically lower in group D (P ⫽ 0.001) than in group F.
There was no statistical difference in mean systolic NIBP,
respiratory rate, and Spo2 in the PACU. There was a
statistically significant difference in the number of patients
with Spo2 below 95% between the 2 groups, 11 (18%) in
group D and 25 (40.9%) in group F (P ⫽ 0.01).
DISCUSSION
In this study of children undergoing T&A, an intraopera-
tive initial loading dose of 2 ␮g 䡠 kg–1 Dex followed by an
infusion at 0.7 ␮g 䡠 kg–1 䡠 h–1 decreased intraoperative
opiate and anesthetic requirements and decreased opiate
requirements in the PACU, in comparison with a control
group receiving intraoperative IV fentanyl. Additionally,
there was a significantly lower incidence and duration of
severe EA in children who received Dex.
We are reporting on the use of a high initial loading dose
of Dex (2 ␮g 䡠 kg–1) followed by a relatively high dose
continuous intraoperative infusion, because T&A is a pro-
cedure with an intense surgical stimulus, and surgery starts
immediately with no preparation time. An analgesic-
sparing effect of Dex has been shown,5 and when combined
with N2O there is an additive interaction to enhance
analgesia.12 We aimed to use the analgesic-sparing effect of
Dex and evaluated whether a continuous infusion could be
used as a substitute for bolus fentanyl. In our study, 90% of
patients in group D did not require any other intraopera-
tive analgesics. Because there are no studies using a high-
dose continuous infusion of Dex combined with inhalation
anesthetics, the U.S. Food and Drug Administration man-
dated reporting hemodynamic changes as a safety concern.
Figure 2. A, Percentage of patients with an objective pain score
(OPS) of 4 and above. Score 4 and above lasting more than 5
minutes was treated. Statistically lower in group D (dexmedetomi-
dine) at arrival (P ⫽ 0.001), at 5 minutes (P ⫽ 0.028), and at 15
minutes (P ⫽ 0.011). B, Percentage of patients with severe emer-
gence agitation (EA), defined as a score of 4 or 5 on the 5-point
scale. Lower in group D at arrival (P ⫽ 0.001), at 5 minutes (P ⫽
0.028), and at 15 minutes (P ⫽ 0.028). C, Percentage of patients
with Pediatric Anesthesia Emergence Delirium (PAED) score of 10
and above. Statistically lower in group D at arrival (P ⫽ 0.001), at 5
minutes (P ⫽ 0.028), and at 15 minutes (P ⫽ 0.011).
Children in group D had statistically lower systolic blood
pressure and HR, almost the entire duration of the anes-
thetic (Fig. 1), but none of the patients needed intervention
for bradycardia or hypotension on the basis of study
criteria. Hemodynamic data are consistent with reports by
other investigators. Mason et al.13 used higher doses of Dex
(2 to 3 ␮g 䡠 kg–1 loading dose and infusion of 1.5 to 2 ␮g 䡠
kg–1 䡠 h–1) as the sole drug for sedation in children and
observed a decrease in HR and blood pressure, which were
within 20% of awake normal range. In children anesthe-
tized with one MAC sevoflurane or desflurane and given a
single, lower dose of Dex (0.5 ␮g 䡠 kg–1), Deutsch et al.14
found a significant decrease in HR, but neither the systolic
nor diastolic blood pressure was statistically lower. The
biphasic response usually seen in adults, with an initial
increase in systolic blood pressure and a reflex decrease in

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 1007

Dexmedetomidine Infusion for Adenotonsillectomy
HR followed by stabilization of these variables below
baseline, is not observed in children.15
On the basis of routine clinical practice, fentanyl was
given in a dose of 1 ␮g 䡠 kg–1 as a bolus to the control group.
This lower dose is based on the enhanced analgesic sensi-
tivity to opiates in children with OSAS.3 It is noteworthy
that in the control group only 36% of patients needed
rescue fentanyl, indicating that our technique of low-dose
fentanyl is effective in almost two thirds of patients. HR
and systolic blood pressure increase was used as the trigger
for rescue fentanyl in both groups in response to surgical
stimulation. The BIS monitor was used to ensure that
patients in group D had an adequate depth of anesthesia
because they may not display hemodynamic changes due
to the inherent sympatholytic properties of Dex. In an
attempt to maintain equivalent depth of anesthesia in both
groups, the sevoflurane concentration was titrated to main-
tain a BIS value below 60. Consistent with studies in adult
patients, the concentration of sevoflurane required to main-
tain the BIS below 60 was smaller in patients receiving Dex
(MAC in group D was 5.7% to 41.6% lower). Tufanoguallari
et al.8 found reductions in the average end-tidal desflurane
concentration of 19%–22%, depending on the rate of Dex
infusion, which ranged from 0.2 to 0.8 ␮g 䡠 kg–1 䡠 h–1. The
anesthetic-sparing effect of Dex appears to have an added
advantage in facilitating earlier awakening and tracheal
extubation. In the present study, TA and TE were statisti-
cally lower in group D, despite the high dose used. Most
investigators using Dex as a low-dose intraoperative infu-
sion or as a single bolus reported no difference in TA and
TE in comparison with placebo.6,16 Only 1 study reported
that a single dose of 0.5 ␮g 䡠 kg–1 Dex, 5 minutes before the
end of surgery significantly prolonged TA and TE in
comparison with placebo in patients having T&A.7
Evaluation of postoperative pain is complicated by the
difficulty in assessing pain in younger children and by the
occurrence of EA. It is often difficult to distinguish between
pain and EA because of the overlapping clinical picture,
and pain itself can be the source of agitation.17 Most
investigators have used different assessment tools to try
and separate the two, but there is generally overlap in the
scales, because a child who is restless or thrashing will
score high on both scales. We did find a positive correlation
between agitation and pain; group F had higher pain and
EA scores than did group D. Results on the OPS, Cole scale,
and PAED showed a very similar trend in both groups;
scores were highest on arrival in the PACU and decreased
over time (Fig. 2, A–C). A significantly smaller number of
patients needed rescue morphine in group D, 18% in
comparison with 44% in group F. Because it is difficult to
separate pain and EA, and the fact that the rescue drug for
both agitation and pain in our study was morphine, it is not
possible to determine whether the morphine was given for
pain or for agitation. On the basis of the effectiveness of
smaller doses of intraoperative Dex in adult patients for
reducing postoperative morphine consumption for 24
hours,8,18 we could assume that an analgesic effect would
be present in our study patients in the immediate postop-
erative period. In children, Guler et al.7 found that 23%
patients who received a single dose of 0.5 ␮g/kg Dex
before the end of the procedure (T&A) required opoiods for
1008 www.anesthesia-analgesia.org
analgesia in the PACU in comparison with 53% in the
placebo group. Erdil et al.19 compared a single dose of 0.5
␮g 䡠 kg–1 Dex with 2.5 ␮g 䡠 kg–1 fentanyl in patients
undergoing adenoidectomy and concluded that Dex pro-
vided residual analgesia similar to that of fentanyl.
Pain can be severe after T&A, and it is commonly treated
with opioids, despite a known sensitivity of patients with
OSAS and recurrent hypoxemia to opiates. Brown et al.3
reported enhanced analgesic morphine sensitivity in chil-
dren with OSAS during T&A and reduced morphine
requirements after T&A. Therefore, several nonopioid an-
algesics such as ketorolac, ketamine, and tramadol have
been evaluated for pain management after T&A,20 –22 but
none have gained widespread use or acceptance because of
concerns with side effects or inadequate analgesia. A
morphine-sparing effect of acetaminophen has been dem-
onstrated in pediatric day-case surgery,23 and dexametha-
sone also reduces post-tonsillectomy pain.24 In the present
study, all patients were given 30 to 40 mg 䡠 kg⫺1 of acetamin-
ophen rectally before start of surgery and intraoperative IV
dexamethasone. A multimodal, opioid-sparing, analgesic ap-
proach including Dex, such as the one used in our study, is
worth considering in this patient population with a high
potential for adverse respiratory events. The incidence of
nausea or vomiting was extremely low in this study. Only 1
patient needed an antiemetic in the PACU, probably because
of the antiemetic effect of dexamethasone.
EA is a complex phenomenon, the etiology of which is
multifactorial. The wide variability in the incidence of
agitation in the different studies on EA may be due to the
criteria used to define this phenomenon and the time in the
PACU when EA was measured.17 We did repeated mea-
surements at frequent time intervals, because a single
measurement may not reflect the true incidence of EA.11
Group D had a statistically lower frequency of severe EA
than did group F until 30 minutes (Fig. 2B). At 30 minutes
there was no incidence of severe EA in group D, and in
group F it was 1.6%. Severe EA lasting more than 5 minutes
was treated. The incidence of severe EA on arrival in the
PACU in group D (18%) was similar to that reported by
Guler et al.7 (17%), who used a single dose of Dex 5 minutes
before the end of the procedure in children undergoing
T&A. The occurrence of EA in younger patients and
otolaryngologic procedures is reported to be high, although
the exact reason for this is not known.4 Ninety percent of
patients in our study were 6 years old or younger, and 26
patients (46.2%) in each group were 2 to 3 years old.
Hyperactivity and attention deficit disorder are frequently
seen in children with OSAS, possibly explaining or contrib-
uting to a high incidence of EA in our T&A patients.
Dexmedetomidine has been used successfully as an infu-
sion (0.2 ␮g 䡠 kg–1 䡠 h–1) continued into the postoperative
period for 15 minutes or single dose at the end of surgery
(0.5 ␮g 䡠 kg–1) to prevent or reduce emergence delirium in
children.6,7,16 It must be noted that these studies compared
Dex with placebo, whereas our control group received
fentanyl 1 ␮g 䡠 kg–1, which also reduces EA. However, a
higher dose is reported to be effective in patients having
painful procedures.25 From our study and others, it re-
mains difficult to discern whether the analgesic or sedative
effects of ␣2 agonists are responsible for reducing EA in
ANESTHESIA & ANALGESIA
children; regardless of the mechanism, Dex appears to be
effective in a wide range of doses. The half-life of Dex is
reported to be 1.8 hours in children,15 but there are no data
on duration of sedative or analgesic effects after discontinu-
ation of Dex infusion. The HRs were significantly slower in
group D until 90 minutes in the PACU. The residual effects
on HR of an intraoperative Dex infusion and the potential
for an attenuated response to postoperative bleeding in
T&A patients may be a concern and a disadvantage of
using a Dex infusion.
The risk of respiratory morbidity after T&A in children
with OSAS is reported to be about 20%.2 Sanders et al.26
reported that although the patients with OSAS were more
likely to require supplemental oxygen, oral airway use, or
assisted ventilation on emergence, severe complications
such as laryngospasm and bronchospasm were uncom-
mon. In the present study, there were no instances of
laryngospasm or bronchospasm after extubation. One pa-
tient developed intraoperative pulmonary edema and was
excluded from the study because she remained intubated
overnight. Although not a study variable, we noted that
extubation was much smoother with less coughing and
breath-holding in patients given Dex. All patients were
observed continuously in the PACU for 2 hours, and the
observers were asked to record the lowest Spo2 during this
period. There was a statistically significant difference in the
number of patients with Spo2 below 95% in the PACU
between the 2 groups, 11 in group D and 25 in group F. This
could be related to the smaller requirement for opiates in
the PACU in group D or to the lower incidence and
duration of severe EA in group D. The goal of having a
child who was settled, comfortable, and less restless, with
application of monitors and administration of supplemen-
tal oxygen in the PACU, was easier to achieve in patients
who received Dex.
A few methological considerations of this study need to
be mentioned. The anesthesiologist and the data recorder in
the OR were not blinded to the study group. We believe
that knowledge of study group assignment did not bias the
conduct of the anesthetic, because the study protocol was
tightly controlled, with specific criteria regarding intraop-
erative rescue fentanyl, sevoflurane concentration, the time
to discontinue sevoflurane, extubation criteria, and use of
rescue morphine in the PACU.
The PAED is the only validated rating scale for emer-
gence delirium.10 The investigators who developed the
PAED scale rated emergence behavior 10 minutes after the
child awakened and remained awake (did not fall back to
sleep). Early in the present study, we found this to be a
potential problem because children who were asleep were
receiving ratings of 4 on the first 3 items of the scale
because they could not make eye contact, their actions were
not purposeful, and they were not aware of their surround-
ings. Therefore we had to modify the scoring on the scale
and rate these items as zero. Clearly, the children were not
agitated if they were sleeping. Because we used a modified
version of the PAED, we used a second scale (Cole) to run
concomitantly to support the findings with the modified
version of the PAED. The 1 to 5 scale described by Cole et
al.11 has been used in several studies of EA. It is not a
October 2010 • Volume 111 • Number 4
validated scale, but is easy to use, and defining the catego-
ries of mild or severe is clear.
The OPS is not a validated scale, but this scale or some
modification of it has been used in several studies in
children. Although 2 other studies on EA6,19 have used the
OPS, it is perhaps not the best scale to use in a study on EA
because of considerable overlap on the items being scored.
We did not follow patients once they were discharged
from the PACU. A future study with overnight pulse
oximetry data and use of postoperative analgesics would
be worthwhile to perform.
CONCLUSION
In children undergoing T&A, the goal is to minimize
respiratory and airway compromise and have an awake,
settled, comfortable child after the surgery. An opioid-
sparing technique is particularly appealing in children with
OSAS, when airway obstruction is known to preexist and
may persist on the night after surgery. An intraoperative
infusion of Dex combined with sevoflurane and N2O
provided satisfactory intraoperative conditions for T&A
without adverse hemodynamic effects. TA and TE were
shorter than they were for the patients receiving fentanyl.
Postoperative opoiod requirements were significantly re-
duced, and the incidence and duration of severe EA was
lower, resulting in a smooth recovery. We have described a
practical, effective, and safe technique for using Dex infu-
sion. A multimodal, opioid-sparing, analgesic approach
including Dex, such as the one used in our study, can be
useful in children with OSAS undergoing other surgical
procedures besides T&A, wherein the advantages of de-
creased perioperative opioid requirements and a reduced
occurrence of EA will be beneficial.
REFERENCES
1. Brown KA. What we don’t know about childhood obstructive
sleep apnea. Paediatr Anaesth 2001;11:385–9
2. McColley SA, April MM, Carroll JL, Naclerio RM, Loughlin
GM. Respiratory compromise after adenotonsillectomy in chil-
dren with obstructive sleep apnea. Arch Otolaryngol Head
Neck Surg 1992;118(9):940 –3
3. Brown KA, Laferriere A, Lakheeram I, Moss IR. Recurrent
hypoxemia in children is associated with increased analgesic
sensitivity to opiates. Anesthesiology 2006;105(4):645–7
4. Voepel-Lewis T, Malviya S, Tait AR. A prospective cohort
study of emergence agitation in the pediatric postanesthesia
care unit. Anesth Analg 2003;96(6):1625–30
5. Hall JE, Uhrich TD, Barney JA, Arain SR, Ebert TJ. Sedative,
amnestic, and analgesic properties of small-dose dexmedeto-
midine infusions. Anesth Analg 2000;90:699 –705
6. Shukry M, Clyde MC, Kalarickal PL, Ramadhyani U. Does
dexmedetomidine prevent emergence delirium in children
after sevoflurane-based general anesthesia? Paediatr Anaesth
2005;15:1098 –104
7. Guler G, Akin A, Tosun Z, Ors S, Esmaoglu A, Boyaci A.
Single-dose dexmedetomidine reduces agitation and provides
smooth extubation after pediatric adenotonsillectomy. Paediatr
Anaesth 2005;15(9):762– 6
8. Tufanoguallari B, White PF, Peixoto MP, Kianpour D, Lacour
T, Griffin J, Skrivanek G, Macaluso A, Shah M, Provost DA.
Dexmedetomidine infusion during laparoscopic bariatric sur-
gery: the effect on recovery outcome variables. Anesth Analg
2008;106(6):1741– 8
9. Hannallah RS, Broadman LM, Belman AB, Abramowitz MD,
Epstein BS. Comparison of caudal and ilioinguinal/
iliohypogastric nerve blocks for control of post-orchiopexy pain
in pediatric ambulatory surgery. Anesthesiology 1987;66(6):832– 4
www.anesthesia-analgesia.org 1009
Dexmedetomidine Infusion for Adenotonsillectomy
10. Sikich N, Lerman J. Development and psychometric evaluation
of the pediatric anesthesia emergence delirium scale. Anesthe-
siology 2004;100(5):1138 – 45
11. Cole JW, Murray DJ, McAllister JD, Hirshberg GE. Emergence
behaviour in children: defining the incidence of excitement
and agitation following anaesthesia. Paediatr Anaesth 2002;
12:442–7
12. Dawson C, Ma D, Chow A, Maze M. Dexmedetomidine
enhances analgesic action of nitrous oxide: mechanisms of
action. Anesthesiology 2004;100(4):894 –904
13. Mason KP, Zurakowski D, Zgleszewski SE, Robson CD, Car-
rier M, Hickey PR, Dinardo JA. High dose dexmedetomidine
as the sole sedative for pediatric MRI. Paediatr Anaesth
2008;18(5):403–11
14. Deutsch E, Tobias JD. Hemodynamic and respiratory changes
following dexmedetomidine administration during general
anesthesia: sevoflurane vs desflurane. Paediatr Anaesth
2007;17:438 – 44
15. Petroz GC, Sikich N, James M, van Dyk H, Shafer SL, Schily M,
Lerman J. A phase I, two center study of the pharmacokinetics
and pharmacodynamics of dexmedtomidine in children.
Anesthesiology 2006;105:1098 –110
16. Ibacache ME, Munoz HR, Brandes V, Morales AL. Single-dose
dexmedetomidine reduces agitation after sevoflurane anesthe-
sia in children. Anesth Analg 2004;98(1):60 –3
17. Vlajkovic GP, Sindjelic RP. Emergence delirium in children:
many questions, few answers. Anesth Analg 2007;104:84 –91
18. Gurbet A, Basagan-Mogol E, Turker G, Ugun F, Kaya FN,
Ozcan B. Intraoperative infusion of dexmedetomidine reduces
perioperative analgesic requirements. Can J Anaesth 2006;
53(7):646 –52
1010 www.anesthesia-analgesia.org
19. Erdil F, Demirbilek S, Begec Z, Ozturk E, Ulger MH, Ersoy MO.
The effects of dexmedetomidine and fentanyl on emergence
characteristics after adenoidectomy in children. Anaesth Inten-
sive Care 2009;37(4):571– 6
20. Marret E, Flahault A, Samama CM, Bonnet F. Effects of
postoperative, nonsteroidal, antiinflammatory drugs on bleed-
ing risk after tonsillectomy: meta-analysis of randomized,
controlled trials. Anesthesiology 2003;98(6):1497–502
21. Erhan OL, Göksu H, Alpay C, Beçstaçs A. Ketamine in
post-tonsillectomy pain. Int J Pediatr Otorhinolaryngol 2007;
71(5):735–9
22. Hullett BJ, Chambers NA, Pascoe EM, Johnson C. Tramadol vs
morphine during adenotonsillectomy for obstructive sleep
apnea in children. Paediatr Anaesth 2006;16(6):648 –53
23. Korpela R, Korvenoja P, Meretoja OA. Morphine-sparing effect
of acetaminophen in pediatric day-case surgery. Anesthesiol-
ogy 1999;91:442–7
24. Afman CE, Welge JA, Steward DL. Steroids for posttonsillec-
tomy pain reduction: meta-analysis of randomized controlled
trials. Otolaryngol Head Neck Surg 2006;134:181– 6
25. Cohen IT, Finkel JC, Hannallah RS, Hummer KA, Patel KM.
The effect of fentanyl on the emergence characteristics after
desflurane or sevoflurane anaesthesia. Paediatr Anaesth
2002;12:442–7
26. Sanders JC, King MA, Mitchell RB, Kelly JP. Perioperative
complications of adenotonsillectomy in children with obstruc-
tive sleep apnea syndrome. Anesth Analg 2006;103(5):1115–21
ANESTHESIA & ANALGESIA
Epidemiology of Ambulatory Anesthesia for Children in
the United States: 2006 and 1996
Jennifer A. Rabbitts, MB, ChB,* Cornelius B. Groenewald, MB, ChB,* James P. Moriarty, MSc,†
and Randall Flick, MD, MPH*

BACKGROUND: There are few data that describe the frequency, anesthetic type, provider, or
disposition of children requiring outpatient anesthesia in the United States (US). Since the early
1980s, the frequency of ambulatory surgery has increased dramatically because of advances in
medical technology and changes in payment arrangements. Our primary aim in this study was to
quantify the number of ambulatory anesthetics for children that occur annually and to study the
change in utilization of pediatric anesthetic care over a decade.
METHODS: The US National Center for Health Statistics performed the National Survey of
Ambulatory Surgery in 1994 through 1996 and again in 2006. The survey is based on data
abstracted from a national sample of ambulatory surgery centers and provides data on visits for
surgical and nonsurgical procedures for patients of all ages. We abstracted data for children
who had general anesthesia, regional anesthesia, or monitored anesthesia care during the
ambulatory visit. We obtained the information from the 2006 and 1996 databases and used
population census data to estimate the annual utilization of ambulatory anesthesia per 1000
children in the US.
RESULTS: In 2006, an estimated 2.3 million ambulatory anesthesia episodes of care were
provided in the US to children younger than 15 years (38 of 1000 children). This amount
compares with 26 per 1000 children of the same age group in 1996. In most cases, an
anesthesiologist was involved in both time periods (74% in 2006 and 85% in 1996). Of the
children, 14,200 were admitted to the hospital postoperatively, a rate of 6 per 1000 ambulatory
anesthesia episodes.
CONCLUSION: The number and rate of ambulatory anesthesia episodes for US children
increased dramatically over a decade. This study provides an example of how databases can
provide useful information to health care policy makers and educators on the utilization of
ambulatory surgical centers by children. (Anesth Analg 2010;111:1011–5)

T he introduction of the first freestanding ambulatory

surgery centers (ASCs) in the 1970s resulted in a
rapid increase in the proportion of operations per-
formed on an outpatient basis, from ⬍10% in 1979 to 50% in
1990.1 The number of ASCs continues to increase, with a
150% increase per 100,000 population reported in metro-
politan areas from 1993 to 2001.2 The number of Medicare-
certified ASCs increased 64% between 2000 and 2007, from
3028 to 4964.3 Improvements in surgical and anesthetic
techniques have increased the proportion of procedures
performed on an outpatient basis to ⬎70% of the total
surgical interventions currently performed in the United
States (US).1
No quantification has been made of the pediatric proce-
dures occurring on an outpatient basis in the US. As the
country enters an era of health care reform, epidemiologic
data on the utilization of medical resources may be helpful
to policy makers as health care expenditures are analyzed.
For example, current Medicare payments to freestanding
From the *Department of Anesthesiology, and the †Division of Health Care
Policy & Research, Mayo Clinic, Rochester, Minnesota.
Accepted for publication June 10, 2010.
Supported by the Department of Anesthesiology, Mayo Clinic, Rochester, MN.
Disclosure: The authors report no conflicts of interest.
Address correspondence and reprint requests to Randall Flick, MD, MPH,
Department of Anesthesiology, Mayo Clinic, 200 First St. SW, Rochester, MN
55905. Address e-mail to flick.randall@mayo.edu.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181ee8479
ASCs are less than for corresponding services in hospital-
based outpatient departments. In addition, copayments
and charges to patients are generally less at ASCs than at
hospitals. Almost 90% of all US freestanding ASCs are
wholly or partially owned by physicians and 96% are
for-profit institutions.4
The purpose of this study was to describe, for the first
time, the utilization of freestanding and hospital-based
ASCs in regard to their care of children. We quantified the
number of ambulatory anesthesia episodes occurring an-
nually for children in accordance with age group, anes-
thetic type, and anesthesia provider and described the
change in utilization over a decade. Secondary analyses
examined the distribution of perioperative time and dispo-
sition and used unplanned admission as an end point.
METHODS
The National Survey of Ambulatory Surgery (NSAS) is the
only US national study of ambulatory surgery in hospital-
based and freestanding ASCs.5 We abstracted the data for
ambulatory anesthesia of children from this public data-
base for 1996 and 2006. National census data were used to
estimate utilization rates.
The NSAS Database
The NSAS was performed by the National Center for
Health Statistics on a nationally representative sample of
surgery centers that perform ambulatory procedures. The
complete sampling and survey methods have been de-
scribed5 and select data have been published for patients of

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 1011

Ambulatory Anesthesia in Children
all ages who had both surgical and nonsurgical proce-
dures.6,7 In summary, eligible hospital-based facilities were
identified from the SMG Marketing Group, Inc., Hospital
Market Database5 and included all short-stay or general
(medical, surgical, or children’s) noninstitutional, nonfed-
eral hospitals in the 50 states and the District of Columbia
with 6 or more beds staffed for patient use. Eligible
freestanding facilities were identified from the SMG Free-
Standing Outpatient Surgery Center Database and the
Health Care Financing Administration Provider of Services
Public Use File.
Hospital-based and freestanding ASCs consisted of hos-
pitals that were state regulated or certified for Medicare
that performed at least 50 ambulatory procedures in the
previous year and excluded dental, podiatry, pain, abor-
tion, family planning, and birthing centers. The included
procedures were both surgical and nonsurgical (e.g., lum-
bar puncture, computed tomographic scanning) procedures
performed on an ambulatory basis in general operating
rooms, dedicated ambulatory surgery rooms, and other
specialized rooms, including endoscopy units and cardiac
catheterization laboratories.
A multistage probability design was used, in which
independent samples of hospitals and freestanding ASCs
were selected at the first or second stages and visits to these
facilities were selected at the final sampling stages.5 An
NSAS medical abstract form (Appendix) was used to
collect data for each sampled visit during which ⬎1 proce-
dure may have been performed. Data were abstracted from
the medical record by facility staff in 30% of cases and by
US Census Bureau personnel in 70% of cases. Data ab-
stracted for the NSAS database included patient character-
istics, payment information, surgical and nonsurgical
procedures, surgical visit information (e.g., perioperative
times, anesthesia provider, type of anesthesia), and patient
disposition.
In 2006, data were collected for approximately 52,000
ASC visits at 437 centers (142 hospital-based and 295
freestanding centers), with an overall response rate of 74%
of sampled centers.6 Survey responses were received from
75% of sampled hospital-based ASCs and 74% of sampled
freestanding ASCs. In 1996, data were collected for 125,000
ASC visits to 488 centers, with an overall response rate of
81% of sampled centers.7 Survey responses were received
from 91% of sampled hospital-based centers and 70% of
sampled freestanding ASCs.
Data Abstraction from the NSAS Database
We abstracted data pertaining to type of anesthetic admin-
istered, anesthesia provider present, procedure time vari-
ables, primary procedure, gender, source of payment, and
discharge status. We combined the data of patients younger
than 15 years with data from the population census to
estimate the rate of visits to an ASC for ambulatory
procedures with anesthesia for US pediatric patients. Age
categories were ⬍1 year, 1 to 4 years, and 5 to 14 years
based on available census data. All statistical analyses were
conducted with Stata/SE 10.1 software (StataCorp LP,
College Station, TX). Where data were missing, we catego-
rized the result as “not specified” (e.g., for the anesthesia
provider category in 1996 data).
1012 www.anesthesia-analgesia.org
Figure 1. Rate of ambulatory anesthesia for children in 1996 and
2006. Rate increased from 26 per 1000 children younger than 15
years in 1996 to 38 per 1000 children of this age group in 2006.
Data Abstraction from the National Hospital
Discharge Survey
To help interpret the trends observed in the utilization of
ambulatory surgery facilities, we abstracted a limited
amount of information from the National Hospital Dis-
charge Survey. The survey is a national database of inpa-
tient medical and surgical care that is similar to the NSAS
database.8,9 It does not include information on the admin-
istration of anesthesia during procedures performed at
inpatient facilities, and therefore, we could not differentiate
the procedures performed with anesthesia from the nonin-
vasive procedures (including imaging studies) or proce-
dures performed without anesthesia. To help interpret the
change in rate of utilization of ASCs for surgical proce-
dures, we abstracted the number of inpatient visits in both
1996 and 2006 for which tonsillectomy or adenoidectomy,
or both, was listed as the first procedure. We combined the
data on inpatients younger than 15 years with population
census data to estimate the rate of these procedures.
RESULTS
Utilization of ASCs for Children, 2006
In 2006, 2,300,651 (standard error [SE], 315,651) ambulatory
anesthesia episodes of care were performed for patients
younger than 15 years in the US, which is a rate of 38
ambulatory anesthetic procedures per 1000 children (Fig.
1). Among these cases, anesthetics were given to 1,329,976
(SE, 160,647) boys and 1,071,650 (SE, 168,697) girls, or rates
of 43 (SE, 5.2) per 1000 boys and 36 (SE, 5.7) per 1000 girls.
Data by age group and type of anesthesia are provided in
Table 1.
The 3 most frequently performed procedures were ton-
sillectomy, adenoidectomy, and myringotomy with ear
tube.6 Data regarding the provider of anesthesia are dis-
played in Table 2.
Perioperative Data
The breakdown of perioperative times is displayed in
Figure 2. Of the children who received anesthetics, 12,030
were admitted postoperatively to an inpatient facility (data
on those patients readmitted after discharge were not
available), for a rate of 6 (SE, 1.3) inpatient admissions per
1000 ambulatory anesthetics. An estimated 2,193,686 (SE,
311,507) of the 2,401,626 children receiving ambulatory
ANESTHESIA & ANALGESIA
 Table 1. Ambulatory Anesthesia Sessions for Monitored Anesthesia Care or Regional or General
Anesthetics Only by Age Group, 2006 and 1996
Overall rate
Age, y MAC (SE) Regional (SE) General (SE) Total (SE) per 1000 children
2006
⬍15 44,462 (10,149)
a
26,484 a(7036) 2,241,985 (313,649) 2,300,651 (315,651) 38
⬍1 — —
196,991 (36,173) 202,412 (36,363) 49
—a —a
1–4 963,733 (141,654) 974,915 (141,977) 60
5–14 38,215 (9823) 11,156 (3741) 1,081,261 (145,287) 1,123,295 (147,728) 28
1996b
⬍15 53,943 14,776 1,490,686 1,522,883 26
a a
⬍1 — —
138,661 140,639 37
1–4 12,283 3177 633,454 640,424 41
5–14 39,351 9338 718,571 741,820 19
MAC ⫽ monitored anesthesia care; SE ⫽ standard error.
a
Sample size was too small or SE was too large.
b
1996 Data did not contain some of the survey sampling variables needed to accurately estimate the SEs and thus the SEs are not reported.

Table 2. Anesthesia Provider Involved During Admission to Ambulatory Center When Anesthesia Was
Provided by an Anesthesiologist or CRNA Only, 2006 and 1996
Both anesthesiologist
Age, y Anesthesiologist only (SE) CRNA only (SE) and CRNA (SE)
2006
⬍15 1,389,393 (209,784) 603,695 (158,713) 292,630 (52,055)
⬍1 130,681 (23,159) 52,145a 18,375a
1–4 577,712 (89,577) 256,924a 135,772 (25,799)
5–14 681,000 (104,418) 294,626 (69,382) 138,483 (26,847)
1996b
⬍15 936,944 219,716 314,919
⬍1 95,883 17,738 24,910
1–4 387,108 93,878 137,596
5–14 453,953 108,100 152,413
CRNA ⫽ certified registered nurse anesthetist; SE ⫽ standard error.
a
Sample size too small or SE too large.
b
Data of 1996 did not contain some of the survey sampling variables needed to accurately estimate the SEs and thus the SEs are not reported.

insurance or through self-pay. For the other visits, the cost

of 816,185 visits was paid through public forms of funding
(e.g., Medicaid, TRICARE). Of the visits for which funding
was known, the cost for 65% of visits was paid from a
private or commercial source and for 35% of visits from a
government source.

Utilization of ASCs for Children, 1996

Figure 2. Mean perioperative times for children younger than 15
years. Postoperative time accounted for the largest portion of the
perioperative period during pediatric visits to ambulatory surgery
centers for surgical procedures. Room time ⫽ the difference be-
tween total operating room time (from entrance into until exit out of
the operating room, or 45 [2] minutes) and surgical time (from the
operation’s start to its finish, or 26 [1] minutes); postoperative
time ⫽ from entrance into until exit from the recovery room, or 71 (3)
minutes; perioperative time ⫽ from entrance into the operating room
until exit from the recovery room.
anesthesia were recorded as having routine discharge (913
of 1000 ambulatory anesthetics; SE, 138).
Payment Information, 2006
In 2006, the cost of 1,547,744 visits to ASCs for children
younger than 15 years was paid by private or commercial
In 1996, an estimated 1,522,883 ASC visits included anes-
thesia administration, which is a rate of 26 ambulatory
anesthetic procedures per 1000 children younger than 15
years. Data by age group and type of anesthetic are
provided in Table 1. Data regarding the provider of anes-
thesia are displayed in Table 2.
Payment Information, 1996
In 1996, most (1,142,481) of the ASC visits for children were
funded through private or commercial insurance or self-
pay; 494,665 (30%) were funded through public sources
(including Medicaid and TRICARE). Of the visits for which
funding was known, 70% of visits were paid from a private
or commercial source and 30% from a government source.
Rate of Inpatient and Ambulatory Tonsillectomy
and Adenoidectomy, 1996 and 2006
The rate of inpatient tonsillectomy or adenoidectomy, or
both, in 1996 was 0.39 (SE, 0.08) per 1000 children younger
than 15 years. In 2006, it was 0.18 (SE, 0.04) per 1000

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 1013

Ambulatory Anesthesia in Children
Table 3. Rates of Tonsillectomy or
Adenoidectomy, or Both, per 1000 Children
Performed on an Ambulatory Basisa and an
Inpatient Basis,b 2006 and 1996
Rate per 1000 children (SE)
Age, y Ambulatoryc Inpatient
2006
⬍15 9.7 (2.0) 0.18 (0.04)
—d
⬍1
1–4 13.2 (2.8)
5–14 9.2 (2.0)
1996
⬍15 5.3 0.39 (0.08)
d d
⬍1 — —
1–4 6.2
5–14 5.5
SE ⫽ standard error.
a
From National Survey of Ambulatory Surgery data.
b
From National Hospital Discharge Survey data.
c
Ambulatory data from 1996 did not contain some of the survey sampling
variables needed to accurately estimate SEs and thus SEs are not reported.
d
Sample size was too small or SE too large.
children of that age. By comparison, the rate of ambulatory
tonsillectomy or adenoidectomy, or both, in 1996 was 5.3
per 1000 children younger than 15 years; in 2006, it was 9.7
(SE, 2.0) per 1000 children of that age. Information by age is
provided in Table 3.
DISCUSSION
Over the 10 years between 1996 and 2006, pediatric visits to
ASCs during which anesthesia was administered increased
almost 50%, from approximately 1.6 million in 1996 to 2.3
million in 2006. During that period, the population of
pediatric patients increased only 5.3%, suggesting that the
increase in ASC visits requiring anesthesia was the result of
a change in overall utilization or a shift in practice from
inpatient to outpatient, or both. Overall utilization in-
creased from 26 to 38 ASC visits per 1000 children, repre-
senting an almost 40% increase.
Whether this increase in rate of ambulatory anesthesia is
attributable to an increase in surgical procedures or a shift
of procedures from inpatient to outpatient settings has
important implications for health care spending. No data
are available that permit a direct comparison of inpatient
and outpatient utilization rates for procedures requiring
anesthesia.
Therefore, we abstracted the rate of either tonsillectomy
or adenoidectomy and of both procedures from the NSAS
database and the National Hospital Discharge Survey da-
tabase, because tonsillectomy and adenoidectomy are com-
mon pediatric procedures that may be performed in an
inpatient or an outpatient setting and always require anes-
thesia. The rate of these procedures as an inpatient opera-
tion decreased approximately 54% from 1996 to 2006
whereas the rate for the ambulatory setting increased 82%.
This change suggests that there may have been a shift of
procedures from the inpatient, short-stay hospitals to the
hospital-based and freestanding ASCs during these 10
years. This shift is consistent with data from the Medicare
Online Survey Certification and Reporting System and the
1014 www.anesthesia-analgesia.org
—d
—d
—d
Figure 3. Provider of ambulatory anesthesia for children in 2006 and
—d
1996. An anesthesiologist was involved in most anesthesia epi-
—d
sodes for ambulatory surgery in both time periods (61% in 2006 and
64% in 1996). CRNA ⫽ certified registered nurse anesthetist.
American Hospital Association Annual Surveys of Hospi-
tals, which showed a 28% increase in hospital-based out-
patient surgery and a 4.5% decrease in inpatient surgery
from 1993 to 2001.2 However, these data must be inter-
preted with caution because there may be a different
explanation for this change. For example, surgeons may
schedule tonsillectomies as outpatient procedures in chil-
dren who stay overnight for payment reasons.
During both 1996 and 2006, the highest rate of ASC visits
with general anesthesia administration was in the 1 to 4
years age group and the lowest rate was in the 5 to 14 years
age group. Most of the ambulatory pediatric anesthesia was
delivered by an anesthesiologist in both time periods (74%
in 2006 and 85% in 1996). However, with the increased use
of ambulatory anesthesia, the proportion of anesthetics
provided by a certified registered nurse anesthetist alone
increased whereas the proportion of anesthetics provided
by a certified registered nurse anesthetist working with an
anesthesiologist decreased (Fig. 3). Nongovernmental
groups (private and commercial insurance and self-pay)
were the funding source for most visits in both 1996 and
2006.
Economic and Educational Implications
This study is an example of how a database can be used to
abstract data useful to health care policy makers, adminis-
trators, and educators and to provide important informa-
tion when changes have to be made in health care systems.
The increase in ambulatory anesthesia itself may be
interpreted as an increase in health care spending. How-
ever, it may be associated with a decrease in inpatient
anesthesia, which could decrease health care expenditures.4
If this trend continues, further savings may occur.
The dramatic increase in pediatric ambulatory surgery
has direct implications for residency and fellowship train-
ing, and this effect may be the most important impact of
this trend. Currently, programs are based at inpatient
medical centers, and training at ambulatory anesthesia
centers may be limited. As pediatric anesthesia shifts to
outpatient and ambulatory centers, education for residents
and fellows may need to be adapted to adequately prepare
anesthesiologists to manage the unique challenges of am-
bulatory anesthesia in children.10,11
ANESTHESIA & ANALGESIA
Limitations
The main limitations of this study are those inherent to the
NSAS database and the medical charts that were reviewed
for it, because our study was reliant on data collected by the
National Center for Health Statistics for the NSAS database.
There was an average response rate of 74% by sampled
hospitals in 2006 and 81% in 1996. Data were extracted from
the medical records of sampled patients by nonmedical
personnel after training,5 and it is possible that the medical
abstract form (Appendix) was not uniformly interpreted.
This process was also limited by the data that were
available and retrievable from the medical records. Infor-
mation was missing for some cases; specifically, the source
of funding was unknown for a large portion of the pediatric
ambulatory visits in 2006.
The statistical software we used could abstract data only
for specific visits and the primary procedure during the
visit. These visits potentially could have included multiple
procedures and anesthetics that were counted as 1 visit.
Sample size was limited in the pediatric population and,
therefore, further data could not be reported because of
unacceptable standard errors. Also, the 1996 and 2006
NSAS medical abstracts were not identical. For example,
the “not-specified” field used in 2006 was not used in 1996,
and thus “not specified” in 1996 was defined as no other
field filled. Options for payment source were slightly
different in the 2 time periods, and therefore, comparisons
cannot be made for this category.
In addition, sampling variables were not available for
the 1996 NSAS database and thus accurate standard errors
could not be calculated for 1996 data. This lack of sampling
variables limited the comparisons that we could make
between the 2 time periods. Percentages do not add up to
100% because all data represent estimates based on sam-
pling rates and population size.
CONCLUSIONS
The rate of ambulatory anesthesia for children in the US
increased by ⬎40% over a decade, partly because of a shift
in procedures from an inpatient to an outpatient setting.
These databases are useful to health care policy makers,
October 2010 • Volume 111 • Number 4
educators, and administrators, as well as other parties
involved in health care organization and provision. This
type of information is currently of particular importance in
this era of health care reform when, to make decisions
regarding health care spending and reform, data on utili-
zation of all aspects of health care are needed from all
groups.
APPENDIX
Medical Abstract Form of the National Survey of Ambula-
tory Surgery, NSAS-5 (2-1-2006). (Adapted from US Census
Bureau and US Department of Commerce. Available at:
http://www.cdc.gov/nchs/data/hdasd/nsas_participant/
nsas5.pdf.)
REFERENCES
1. Pregler JL, Kapur PA. The development of ambulatory anes-
thesia and future challenges. Anesthesiol Clin North Am
2003;21:207–28
2. Bian J, Morrisey MA. Free-standing ambulatory surgery cen-
ters and hospital surgery volume. Inquiry 2007;44:200 –10
3. Medicare Payment Advisory Commission (MedPAC). June 2008
Healthcare Spending and the Medicare Program: A Data Book.
Available at: http://www.medpac.gov/documents/Jun08Data
Book_Entire_report.pdf. Accessed February 16, 2010
4. Medicare Payment Advisory Commission (MedPAC). Report
to the Congress: Medicare Payment Policy. Available at:
http://www.medpac.gov/documents/Mar09_EntireReport.
pdf. Accessed February 16, 2010
5. McLemore T, Lawrence L. Plan and operation of the National
Survey of Ambulatory Surgery. Vital Health Stat 1
1997;37:I–IV, 1–124
6. Cullen KA, Hall MJ, Golosinskiy A. Ambulatory surgery in the
United States, 2006. Natl Health Stat Report 2009;11:1–25
7. Hall MJ, Lawrence L. Ambulatory surgery in the United States,
1995. Adv Data 1997;296:1–15
8. DeFrances CJ, Lucas CA, Buie VC, Golosinskiy A. 2006 Na-
tional Hospital Discharge Survey. Natl Health Stat Report
2008;5:1–20
9. Graves EJ, Owings MF. 1996 summary: National Hospital
Discharge Survey. Adv Data 1998;301:1–12
10. Emhardt JD, Saysana C, Sirichotvithyakorn P. Anesthetic con-
siderations for pediatric outpatient surgery. Semin Pediatr
Surg 2004;13:210 –21
11. Twersky RS. Educational protocols in ambulatory anesthesia.
Ambul Surg 1997;5:117–9
www.anesthesia-analgesia.org 1015
 REVIEW ARTICLE

CME

The Anesthetic Considerations of Tracheobronchial

Foreign Bodies in Children: A Literature Review of
12,979 Cases
Christina W. Fidkowski, MD,* Hui Zheng, PhD,† and Paul G. Firth, MBChB*‡

Asphyxiation by an inhaled foreign body is a leading cause of accidental death among children
younger than 4 years. We analyzed the recent epidemiology of foreign body aspiration and
reviewed the current trends in diagnosis and management. In this article, we discuss anesthetic
management of bronchoscopy to remove objects. The reviewed articles total 12,979 pediatric
bronchoscopies. Most aspirated foreign bodies are organic materials (81%, confidence interval
[CI] ⫽ 77%– 86%), nuts and seeds being the most common. The majority of foreign bodies (88%,
CI ⫽ 85%–91%) lodge in the bronchial tree, with the remainder catching in the larynx or trachea.
The incidence of right-sided foreign bodies (52%, CI ⫽ 48%–55%) is higher than that of left-sided
foreign bodies (33%, CI ⫽ 30%–37%). A small number of objects fragment and lodge in different
parts of the airways. Only 11% (CI ⫽ 8%–16%) of the foreign bodies were radio-opaque on
radiograph, with chest radiographs being normal in 17% of children (CI ⫽ 13%–22%). Although
rigid bronchoscopy is the traditional diagnostic “gold standard,” the use of computerized
tomography, virtual bronchoscopy, and flexible bronchoscopy is increasing. Reported mortality
during bronchoscopy is 0.42%. Although asphyxia at presentation or initial emergency bronchos-
copy causes some deaths, hypoxic cardiac arrest during retrieval of the object, bronchial rupture,
and unspecified intraoperative complications in previously stable patients constitute the majority
of in-hospital fatalities. Major complications include severe laryngeal edema or bronchospasm
requiring tracheotomy or reintubation, pneumothorax, pneumomediastinum, cardiac arrest,
tracheal or bronchial laceration, and hypoxic brain damage (0.96%). Aspiration of gastric
contents is not reported. Preoperative assessment should determine where the aspirated foreign
body has lodged, what was aspirated, and when the aspiration occurred (“what, where, when”).
The choices of inhaled or IV induction, spontaneous or controlled ventilation, and inhaled or IV
maintenance may be individualized to the circumstances. Although several anesthetic tech-
niques are effective for managing children with foreign body aspiration, there is no consensus
from the literature as to which technique is optimal. An induction that maintains spontaneous
ventilation is commonly practiced to minimize the risk of converting a partial proximal obstruction
to a complete obstruction. Controlled ventilation combined with IV drugs and paralysis allows for
suitable rigid bronchoscopy conditions and a consistent level of anesthesia. Close communica-
tion between the anesthesiologist, bronchoscopist, and assistants is essential. (Anesth Analg
2010;111:1016 –25)

A spiration of foreign bodies by children is a common

problem around the world. Asphyxiation from
inhaled foreign bodies is a leading cause of acci-
dental death among children younger than 4 years. During
the 19th century, treatment of foreign body aspiration by
From the *Department of Anesthesia, Critical Care and Pain Medicine, and
the †Biostatistics Center, Massachusetts General Hospital, Boston; and the
‡Department of Anesthesia, Massachusetts Eye and Ear Infirmary, Boston.
Christina W. Fidkowski is now affiliated with the Department of Anesthe-
siology, Henry Ford Hospital, Detroit, Michigan.
Accepted for publication June 10, 2010.
Statistical and administrative support funded by the Department of Anes-
thesia, Critical Care and Pain Medicine, Massachusetts General Hospital,
Boston, and the Department of Anesthesia, Massachusetts Eye and Ear
Infirmary, Boston.
Supplemental digital content is available for this article. Direct URL citations
appear in the printed text and are provided in the HTML and PDF versions
of this article on the journal’s Web site (www.anesthesia-analgesia.org).
Address correspondence and reprint requests to Paul G. Firth, Department
of Anesthesia, Critical Care and Pain Medicine, Massachusetts General
Hospital, Boston, MA 02114. Address e-mail to pfirth@partners.org.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181ef3e9c
purges, bleeding, and emetics were largely ineffective.
Mortality was estimated at 23%. This rate plummeted with
the development of bronchoscopic techniques for the re-
moval of these foreign bodies.1 In 1897, Gustav Killian, a
German otolaryngologist, performed the first bronchos-
copy using a rigid esophagoscope to successfully remove a
pig bone from a farmer’s right main bronchus.1,2 Algernon
Coolidge performed the first successful removal of a tracheal
foreign body in the United States at the Massachusetts Gen-
eral Hospital in 1898.1 Shortly thereafter, Chevalier Jackson
developed the lighted bronchoscope and several special-
ized instruments for the removal of foreign bodies.3 He
pioneered developments in the field and saved the lives of
hundreds of children who had aspirated objects.4 While
early clinicians used topical anesthesia, general anesthesia
became commonplace for the removal of aspirated objects
with increased experience with the rigid bronchoscope and
advances in anesthetic delivery. The flexible bronchoscope
was introduced by Shigeto Ikeda in 1966,5 and the removal
of an airway foreign body using this instrument was
reported in the 1970s.6

1016 www.anesthesia-analgesia.org October 2010 • Volume 111 • Number 4

Anesthetic Considerations of Tracheobronchial Foreign Bodies

Over the years, the steady accumulation of clinical

reports has provided greater insight into the management Table 1. Sensitivity, Specificity, Positive
Predictive Value (PPV), and Negative Predictive
of foreign body aspiration in children. In part as a result,
Value (NPV) of a Witnessed Aspiration Effect for
morbidity and mortality from foreign body aspiration have Foreign Body Aspiration
drastically diminished. This article analyzes the recent
Sensitivity Specificity PPV NPV
epidemiology of foreign body aspiration, reviews the cur-
Aydogan et al. 93.2 45.1 86.5 63.6
rent trends in diagnosis and management, and discusses (1887, 1493)7
anesthetic management for bronchoscopy. Ciftci et al. 91.1 46.0 90.5 47.9
(663, 563)11
Tomaske et al. 74.7 53.7 70.5 58.8
(370, 221)35
CURRENT TRENDS IN FOREIGN BODY ASPIRATION Ayed et al. 81.6 37.9 90.3 22.4
A preliminary Medline search of the literature from 1950 to (235, 206)8
2009 yielded nearly 20,000 cases of foreign body aspiration Tokar et al. 84.9 87.1 94.2 70.1
in children. Before 2000, however, most case series were (214, 152)34
small, included both aspirated and ingested objects, Skoulakis et al. 91.5 56.3 77.3 80.4
mingled adult and pediatric patients, or were reported in (210, 130)31
Kiyan et al. 37.3 96.3 96.6 35.1
various styles in differing specialist journals that precluded
(207, 153)25
meta-analysis. In addition, anesthetic and surgical tech- Erikci et al. 58.3 87.1 90.2 50.5
niques have altered considerably in recent years, making a (189, 127)16
detailed review of older series less relevant. However, Heyer et al. 75.4 92.1 96.8 53.8
numerous large case series have been published recently (160, 122)20
that collectively allow for a clearer representation of the Cohen et al. 83.6 32.1 48.1 72.2
(142, 61)12
problem of pediatric aspiration, as well as current trends in
management. A Medline search using the keywords foreign Values are percentages.
Data were available from 10 of the 30 studies that were reviewed. Study size
body aspiration with limits of the year 2000 to present was is denoted (n, n) to represent the total number of patients and the number of
performed on October 1, 2009. Analysis was limited to patients with an aspirated foreign body, respectively.
studies (1) containing only patients with suspected or
proven foreign body aspiration, (2) with ⬎100 patients, (3)
Most (81%, CI ⫽ 77%– 86%) of the aspirated foreign
containing only children ages 18 years and younger, and (4)
bodies are organic materials.7,9 –11,19 –21,23,24,26 –36 Nuts (es-
written in English. Of the 698 articles obtained, 33 met the
pecially peanuts) and seeds (mainly sunflower and water-
inclusion criteria, of which 3 were excluded for containing
melon) are the most commonly aspirated foreign bodies
duplicate patient data. The 30 articles reviewed report
reported in almost all studies. The exception to this finding
12,979 children with suspected foreign body aspiration, of
is an Italian series that found teeth to be the most com-
whom 11,145 were found to have aspirated an object.7–36
monly aspirated objects (33/121).13 In adolescent Turkish
Twenty-seven of these studies are retrospective analyses,
females, headscarf pins are commonly aspirated.24,34 As
and 3 studies are prospective analyses of all children with
was reported in 24 studies, the majority of foreign bodies
suspected or actual foreign body aspiration.12,13,19 The
(88%, CI ⫽ 85%–91%) lodge in the bronchial tree, with the
cases in these series occurred within the last 20 years in 21 remainder catching in the larynx or trachea.7,9,10,12–19,22,25–36 A
studies, 8,9,11,12,14 –18,20,21,23–26,28 –31,33,34 8 covered a larger higher incidence of right-sided foreign bodies (52%, CI ⫽
time span dating back to the 1980s,10,13,19,22,27,32,35,36 and 1 48%–55%) in comparison with left-sided foreign bodies (33%,
study covered a 30-year time frame starting in 1973.7 CI ⫽ 30%–37%) was reported in all of these studies, with the
To obtain a robust estimate of the various rates (true exception being a small series in Israel.12 A small number of
positive, gender, foreign body type and location, and objects fragment and lodge in different parts of the airways.
radiographic outcome distributions), we applied a meta- A history of a witnessed choking event is highly sug-
analysis to the published data to account for the number of gestive of an acute aspiration. Data were available to
patients, the number of foreign body cases, and the number determine the sensitivity, specificity, positive predictive
of outcomes reported in these 30 articles.7–36 These meta- value, and negative predictive value of a witnessed event in
analyses use a Bayesian model to consider variations in 10 studies (Table 1).7,8,11,12,16,20,25,31,34,35 Children with as-
study design, inclusion and exclusion criteria, and the pirated foreign bodies typically present with the symptoms
study population among the different reported studies. of coughing, dyspnea, wheezing, cyanosis, or stridor. Data
Most patients with aspirated foreign bodies are children were available in 10 studies to determine the sensitivity and
younger than 3 years. Four series reported the median age, the specificity of these presenting signs and symptoms
and 7 series reported the mean age of children with (Table 2).8,12,16,20,23,25,29,31,34,35 A history of cough is highly
aspirated foreign bodies. The median and the mean age sensitive for foreign body aspiration but is not very specific.
ranged from 1 to 2 years12,18,21,33 and from 2.1 to 3.8 On the other hand, a history of cyanosis or stridor is very
years,11,12,14,26,28,29,32 respectively. With the exception of 2 specific for foreign body aspiration but is not very sensitive.
Turkish studies with a high incidence of adolescent girls Signs and symptoms typical in delayed presentations in-
aspirating headscarf pins,24,34 boys account for 61% (confi- clude unilateral decreased breath sounds and rhonchi,
dence interval [CI] ⫽ 59%– 63%) of children with foreign persistent cough or wheezing, recurrent or nonresolving
body aspiration.7,9,17,21,22,26 –29,32,33 pneumonia, or rarely, pneumothorax.

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 1017

 REVIEW ARTICLE

Table 2. Sensitivity (Sens) and Specificity (Spec) of Symptoms for Foreign Body Aspiration
Cough Dyspnea Wheeze Cyanosis Stridor
Sens Spec Sens Spec Sens Spec Sens Spec Sens Spec
Tomaske et al. (370, 221)35 87.8 45.0 57.9 73.2 39.4 74.5
Ayed et al. (235, 206)8 80.1 34.5 30.1 65.5 16.5 65.5
Tokar et al. (214, 152)34 94.1 32.3 27.6 66.1
Skoulakis et al. (210, 130)31 82.3 53.8 24.6 85.0 5.4 100 11.5 98.8
Kiyan et al. (207, 153)25 67.3 20.4 16.3 74.1 79.1 27.8 7.2 98.1
Erikci et al. (189, 127)16 51.2 83.9 4.7 93.5 18.9 93.5
Shivakumar et al. (165, 105)29 92.4 8.3 61.9 66.7 64.8 0 12.4 100 4.8 100
Heyer et al. (160, 122)20 41.0 55.3 33.6 68.4
Kadmon et al. (150, 80)23 51.3 12.9 18.8 72.9
Cohen et al. (142, 61)67 93.4 28.4 14.8 92.6
Values are percentages.
Data were available from 10 of the 30 studies that were reviewed to determine the sensitivity (Sens) and specificity (Spec) of the symptoms of cough, dyspnea,
wheeze, cyanosis, and stridor for foreign body aspiration. Study size is denoted (n, n) to represent the total number of patients and the number of patients with
an aspirated foreign body, respectively.

Table 3. Sensitivity (Sens) and Specificity (Spec) of Radiographic Findings for Foreign Body Aspiration
Air trapping Atelectasis Mediastinal shift Infiltrate
Sens Spec Sens Spec Sens Spec Sens Spec
Tokar et al. (214, 152)34 41.7 91.9 12.6 71.0 11.9 74.2
Skoulakis et al. (210, 130)31 39.2 91.6 9.2 88.8 0 76.3
Kiyan et al. (207, 153)25 63.8 79.6 8.0 94.4 4.4 94.4
Shivakumar et al. (165, 105)29 49.5 80.0 22.9 83.3 3.8 41.7
Heyer et al. (160, 122)20 62.3 97.4 8.2 97.4 20.5 97.4 18.9 84.2
Kadmon et al. (150, 80)23 50.0 90.0 38.8 97.1
Cohen et al. (142, 61)67 49.2 86.4 6.6 96.3 13.1 100 14.8 79.0
Values are percentages.
Data were available from 8 of the 30 studies that were reviewed to determine the sensitivity (Sens) and specificity (Spec) of the radiographic findings of localized
air trapping, atelectasis, mediastinal shift, and infiltrate for foreign body aspiration. Study size is denoted (n, n) to represent the total the number of patients and
the number of patients with an aspirated foreign body, respectively.

Most stable patients had chest radiographs. As was re-
ported in 20 studies, only 11% (CI ⫽ 8%–16%) of the foreign
bodies were radio-opaque.7,9,12–17,19,20,23–25,27–29,32,34 –36 Chest
radiographs were normal in 17% (CI ⫽ 13%–22%) of children
with aspirated objects as were reported in 14 studies.
9,11,12,17,22,24,25,27–29,32–35
The common radiographic abnor-
malities included localized emphysema and air trapping,
atelectasis, infiltrate, and mediastinal shift. Data were avail-
able in 8 studies to calculate the sensitivity and specificity of
these radiographic findings (Table 3).11,12,20,23,25,29,31,34 Pneu-
mothorax and pneumomediastinum are less common find-
ings on chest radiograph (range: 0.1%–3.7%), as was reported
in 7 studies.14,15,22,28,32,33
While rigid bronchoscopy was used solely for the re-
moval of foreign bodies in most studies, both flexible and
rigid bronchoscopies were used in 4 series.12,13,20,33 A
minority of foreign bodies were removed by flexible bron-
choscopy in 3 of these studies (range: 4.1%–10.7%),12,13,20
whereas Tang et al33. reported successful removal by
flexible bronchoscopy in 91.3% of children with foreign
body aspiration. For this study, local anesthesia with seda-
tion was used during bronchoscopy. For children undergo-
ing rigid bronchoscopy, general anesthesia was used in all
studies, and details regarding the anesthetic technique
were provided in 12 studies. Both inhaled7,15,31,36 and
IV13,19 inductions were reported. Similarly, anesthesia was
maintained with either inhaled15,19,31,36 or IV20,22,25 drugs
or a balanced anesthetic.13 Five studies reported the use of
neuromuscular blockers.7,9,15,19,22 Bittencourt et al.9 and
Hasdiraz et al.19 used paralysis as needed during the
procedure and attempted to maintain spontaneous ventila-
tion when possible. On the other hand, Divisi et al.13
commented that spontaneous ventilation is not suitable for
rigid bronchoscopy because of resultant oxygen desatura-
tion and used a balance anesthetic with sevoflurane and
remifentanil. Shivakumar et al.29 used jet ventilation to
prevent oxygen desaturation. None of the authors com-
mented on using drying drugs such as glycopyrolate before
bronchoscopy. Seven studies commented on using steroids
for laryngeal edema, with the majority of those authors
favoring steroid use only as needed,7,13,19,22,30 as opposed
to routine administration.25,36 In 4 studies, antibiotics were
given routinely preoperatively,19,25 postoperatively,36 or as
a 5-day course,30 whereas authors in 3 studies favored
antibiotic administration only as needed for infection.7,13,24
Major iatrogenic complications were specified in 21
studies with 9437 children with aspirated foreign bodies.
The other 9 studies did not provide details or rates of
complications. These complications included severe laryn-
geal edema or bronchospasm requiring tracheotomy or
reintubation, pneumothorax, pneumomediastinum, cardiac
arrest, tracheal or bronchial laceration, and hypoxic brain
damage. These major complications occurred in 91 of these
9437 children (0.96%) (Table 4). Of the 11 cardiac arrests
that were reported, 1 occurred after induction of anesthesia
in a child who was hypoxic on admission, 5 occurred
during bronchoscopy because of hypoxia (3) or bleeding
(2), and the remaining 5 were not specified. Other reported

1018 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

Anesthetic Considerations of Tracheobronchial Foreign Bodies
Table 4. Morbidity Associated with Bronchoscopy
for the Removal of Tracheobronchial
Foreign Bodies
Complication
Major nonfatal complications (n ⫽ 91)
Severe laryngeal edema or bronchospasm
requiring tracheotomy or
reintubation15,19,22,25,27,30,32
Pneumothorax or
pneumomediastinum7,11,15,18,19,22,30,32,33
Cardiac arrest11,15,25,27
Hypoxic brain damage20,21
Tracheal or bronchial laceration requiring
repair11,15,27
Other serious complications (n ⫽ 136)
Infection13,19,21,26,32
Failed bronchoscopy requiring thoracotomy
(27)7,8,11,13,15,19,36 or tracheotomy (10)7,15
Bleeding15,19,27
Thoracotomy (5)8,15,19 or tracheotomy
(7)7,32—not specified
Major iatrogenic complications, as were specified in 21 studies, occurred in
91 of the 9437 children with aspirated foreign bodies. Other serious
complications occurred in 136 of these 9437 children.
Table 5. Mortality Associated with Bronchoscopy
for the Removal of Tracheobronchial
Foreign Bodies
Cause of death (n ⴝ 43)
Cardiac/respiratory arrest
Hypoxic arrest at presentation7,9,11,22,27,32,36
Arrest due to tracheal foreign body19,26
Cardiac arrest during bronchoscopy, not
specified10,11
Postoperative arrest19,29
Hypoxic arrest due to shifting foreign body22
Rupture of puss under pressure behind
foreign body19
Respiratory arrest due to inhaled cement
powder19
Not specified15
Bronchial rupture15
Severe bronchospasm15
Postoperative infection11
Multiorgan failure21
Not specified18
Deaths, as were specified in 26 studies, occurred in 43 of the 10,236
children with aspirated foreign bodies.
serious complications included infection, bleeding, and
failed bronchoscopic removal that necessitated thora-
cotomy or tracheotomy to remove the object (Table 4).
Mortality data were obtained from 26 articles with 43
deaths among 10,236 children (0.42%) with aspirated for-
eign bodies (Table 5). The remaining 4 articles did not
provide details of death rates. Twenty-five deaths occurred
in the 5 largest series with 5927 children (0.42%).7,15,19,22,33
In 2003, Eren et al.15 reported 10 deaths in 1068 children
(0.94%) undergoing rigid bronchoscopy for foreign body
removal under general anesthesia in Turkey. Seven died of
hypoxic arrest during bronchoscopy, 2 of bronchial rup-
ture, and 1 of intractable bronchospasm. Shortly thereafter,
their countrymen Aydogen et al. reported 4 deaths in 1493
children (0.27%) with foreign body aspiration undergoing
rigid bronchoscopy over a 31-year period.7 All 4 fatalities
October 2010 • Volume 111 • Number 4
were in children presenting with acute cyanosis and respi-
ratory distress. Hasdiraz et al.19 reported 8 deaths in 911
Turkish children (0.88%) with foreign body aspiration
undergoing rigid bronchoscopy. Three children developed
Total n
cardiac arrest from total tracheal obstruction, 2 had heart
failure and bronchopneumonia at the time of bronchoscopy
43
and developed cardiac arrest postoperatively, 1 developed
respiratory arrest due to inhalation of cement powder, 1
27 developed sepsis and respiratory failure after explosive
release of purulent discharge from behind the foreign body,
11 and 1 developed a respiratory arrest after a negative
5
bronchoscopy and was found to have a tracheal foreign
5
body at autopsy. In 2008, Hui et al.22 reported 3 deaths
among 1428 children (0.21%) undergoing rigid bronchos-
58 copy over a 22-year period in China. Two died after foreign
37 body displacement during bronchoscopy, and 1 died of
asphyxia during a delay before bronchoscopy. In 2009,
29
Tang et al.33 reported no deaths among 1027 children in
12
China undergoing bronchoscopy for foreign body removal.
In that series, 938 foreign bodies were removed by flexible
bronchoscopy, and 89 foreign bodies were removed by
rigid bronchoscopy. Of the remaining 18 deaths in the 21
other reports, 10 were due to irreversible cardiac arrest on
admission.9,11,27,32,36
DIAGNOSIS AND MANAGEMENT
A suggestive history is important in diagnosing an aspi-
n
rated object, because it is often difficult to make a definitive
37
15
diagnosis on the basis of an abnormal physical examination
5 or radiological studies alone. The work-up of the stable
3 patient should include a chest radiograph to assess for
other potential causes of symptoms, to identify a radio-
3 opaque foreign body, or to detect the position of a foreign
2
1 body on the basis of localized emphysema and air-trapping,
atelectasis, infiltrate, or mediastinal shift.37 The common ab-
1 normality of unilateral hyperinflation seen on the chest radio-
graph due to air trapping behind the foreign body is best
7 viewed at end expiration (Fig. 1). (Video 1; see Supplemental
2
1 Digital Content 1, http://links.lww.com/AA/A169; see the
1 Appendix for video legends). Although a decubitous view has
1 been suggested to look for air trapping in the dependent lung
1 of small children who cannot cooperate with expiratory films,
one study found this to be an unreliable technique.38
Neck radiographs may be helpful in managing upper
aerodigestive tract foreign bodies. The alignment of flat
objects, such as coins, may suggest the location of an object
(Fig. 2A, 2B). Tracheal objects tend to align in the sagittal
plane, whereas esophageal objects tend to align in the
anterior plane. An object that overlaps the boundaries of
the airway on an anterior–posterior view is unlikely to be
inside the airway. Lateral radiographs may show soft tissue
swelling, loss of cervical lordosis, or an object posterior to
the trachea. Proximal esophageal objects can be removed
with a forceps under direct vision, with the laryngoscope
blade inserted into the esophagus to visualize the body and
protect the airway during removal of the object.
Thoracic computed tomography (CT) and virtual
bronchoscopy—a reformatted 3-dimensional CT image that
generates intraluminal views of the airway to the sixth and
seventh generation bronchi—are emerging as new modali-
ties to diagnose tracheobronchial foreign bodies in children
www.anesthesia-analgesia.org 1019
 REVIEW ARTICLE

Figure 1. A, Chest radiograph on end inspiration of a patient with a delayed presentation of an aspirated foreign body aspiration. B, Chest
radiograph on end expiration. Delayed emptying of the left lung suggests local air trapping. The foreign body was in the left bronchus. C, The
offending object seen on rigid bronchoscopy. The airway edema (white-gray) can be seen around the black foreign body, with bubbles reflecting
delayed air release during expiration. A Fogarty catheter is passed beyond the object in preparation for dislodgement. Images courtesy of Dr.
Dan Doody, Department of Surgery, Massachusetts General Hospital, Boston, Massachusetts.

Figure 2. A, The sagittal orientation of a

proximal aerodigestive foreign body sug-
gests an esophageal location. B, A lat-
eral view can demonstrate a position
posterior to the esophagus. Images cour-
tesy of Dr. Allan Goldstein, Department of
Surgery, Massachusetts General Hospi-
tal, Boston, Massachusetts.

Figure 3. A, B, Computerized tomography scan of an aspirated soda can top, using a low-resolution pediatric protocol to minimize radiation
exposure. The object was not seen on initial chest radiograph. A small aluminum object, although metal, has insufficient radiopacity for a plain
chest radiograph, and the object did not produce major obstruction leading to overt pulmonary changes. A computed tomography (CT) scan has
a greater range of sensitivity. C, The offending object in the bronchus intermedius. (Images courtesy of Dr. Pallavi Sagar, Department of
Radiology, and Dr. David Lawlor, Department of Surgery, Massachusetts General Hospital, Boston, MA.)

(Fig. 3).39,40 CT and virtual bronchoscopy are more sensi- false-positive findings. In a retrospective analysis, spiral CT
tive diagnostic modalities for foreign body aspiration in correctly identified all 42 children with aspirated foreign
comparison with conventional chest radiography.41,42 Se- bodies.41 In that study, 3 children had false-positive CT
cretions, tumors, or other obstructive lesions can produce images due to excess bronchial secretions, and 6 children

1020 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

Anesthetic Considerations of Tracheobronchial Foreign Bodies
had true negative scans. In 2 retrospective studies, virtual
bronchoscopy correctly identified 11of 11 and 15 of 23
children, respectively, with aspirated foreign bodies.42,43
No false-positive virtual bronchoscopies were reported in
those studies. The diagnostic utility of virtual bronchos-
copy has also been shown prospectively.44,45 Haliloglu et
al.45 demonstrated that virtual bronchoscopy findings cor-
related with those of conventional bronchoscopy in 23
children, of whom 7 had foreign body aspiration and 16 did
not have foreign body aspiration. In a prospective study of
37 children with suspected foreign body aspiration, 16 had
a positive virtual bronchoscopy, of whom 13 had a foreign
body found with conventional bronchoscopy, and 3 had
either mucous plugs or a schwannoma found with conven-
tional bronchoscopy.44 The remaining 21 patients had a
negative bronchoscopy and were observed with improve-
ment in their symptoms.44 These studies demonstrate that
CT and virtual bronchoscopy correctly identified all cases
of foreign body aspiration. Therefore, some authors sug-
gested that children with a negative CT and virtual bron-
choscopy may not require conventional bronchoscopy as a
definitive work-up.44
A drawback of CT and virtual bronchoscopy is the
potential for excessive radiation exposure. A chest radio-
graph exposes the child to 0.1 mSv of radiation, equivalent
to several days of background environmental radiation.
Although a high-resolution pediatric chest CT can involve
up to 7 mSv of radiation, a lower-resolution scan protocol
using 1.5 mSv is usually sufficient to diagnose a foreign
body. Adequate 3-dimensional views can be subsequently
formatted from this level of detail.
Further limitations include the cost and limited avail-
ability of equipment and radiologists. In addition, CT
examination is limited to stable and cooperative children,
because anesthesia in a remote location for a child with an
unstable object that can potentially acutely obstruct the
airway poses significant risks.
Although rigid bronchoscopy has traditionally been the
definitive method to diagnose and remove tracheobron-
chial foreign bodies, a diagnostic flexible bronchoscopy
under local anesthesia may be indicated for patients with-
out a clear history or findings of aspiration.23,46 – 48 In a
prospective study, children with convincing evidence of
foreign body aspiration were examined with rigid bron-
choscopy under general anesthesia, whereas others with
less-suggestive findings underwent flexible bronchoscopy
with local anesthesia.47 Of the 28 children who underwent
rigid bronchoscopy, 23 (82%) had a foreign body aspiration.
Of the 55 children who underwent flexible bronchoscopy,
only 17 (34%) had a foreign body aspiration. Another
prospective study found that 43 (84%) of 51 children who
underwent rigid bronchoscopy and only 7 (37%) of 19
children undergoing flexible bronchoscopy had positive
studies for foreign body aspiration.48 Both studies found a
significant association of aspirated foreign bodies with
unilateral decreased breath sounds, localized wheezing,
and obstructive emphysema on chest radiograph.47,48
These authors recommended that children undergo rigid
bronchoscopy only if they have acute asphyxiation, a
radio-opaque foreign body, unilateral pulmonary signs, or
obstructive emphysema. All other children should undergo
October 2010 • Volume 111 • Number 4
a diagnostic flexible bronchoscopy. When this algorithm
was applied retrospectively, the negative finding rate of
rigid bronchoscopy decreased from 18% to 4% and from
16% to 6%, respectively.47,48 No adverse events were re-
ported with flexible bronchoscopy.47,48 Therefore, diagnos-
tic flexible bronchoscopy in selected children minimizes the
potential complications of rigid bronchoscopies. More re-
cently, Kadmon et al.23 proposed a computer model based
on history, physical examination, and radiographic find-
ings to calculate a score that predicts the likelihood of
foreign body aspiration in children. They further suggested
an algorithm to observe a child, perform diagnostic flexible
bronchoscopy, or perform therapeutic rigid bronchoscopy
on the basis of the calculated score. A prospective study is
warranted to determine the utility of this model.
In addition to aiding in the diagnosis of aspirated foreign
bodies, flexible bronchoscopy is becoming more popular for
the removal of foreign bodies.33,49 –52 In a large retrospective
study, a foreign body was successfully removed by flexible
bronchoscopy in 938 (91.3%) children.33 Flexible bronchos-
copy is better suited for removing foreign bodies from distal
airways and upper lobe bronchi, because of the smaller
diameter and greater flexibility in comparison with the rigid
bronchoscope. Fewer instruments, however, are available for
use with the flexible bronchoscope to remove the foreign
bodies. Rigid bronchoscopy continues to be used to remove
aspirated foreign bodies because multiple extraction instru-
ments are available and because it provides good visualiza-
tion, controls the airway, and allows ventilation.
ANESTHETIC MANAGEMENT FOR
BRONCHOSCOPY
Anesthetic considerations encompass preoperative assess-
ment, management techniques for flexible or rigid bron-
choscopy, and postbronchoscopic disposition.
Preoperative Assessment
The preoperative assessment should determine where the
aspirated foreign body has lodged, what was aspirated,
and when the aspiration occurred. If the foreign body is
located in the trachea, the child is at risk for complete
airway obstruction and should be taken urgently to the
operating room. Conversely, the risk of complete airway
obstruction is less if the object is firmly lodged beyond the
carina. It is important to determine the type of foreign
body: Organic materials can absorb fluid and swell, oils
from nuts cause localized inflammation, and sharp objects
can pierce the airway. The time since the aspiration should
be established because airway edema, granulation tissue,
and infection may make retrieval more difficult with de-
layed presentations. A recently aspirated object may move
to a different position with coughing.
The time of the last meal should be established to assess
the risk of aspiration. There are no reports of aspiration of
gastric contents in the literature surveyed, although fatal
progression of obstruction has been reported.7,9,11,22,27,32,36
In acute cases, therefore, the dangers of delayed removal
appear to outweigh the risk of a full stomach in a well-
conducted anesthetic. In urgent cases, the stomach can be
suctioned through a large-bore gastric tube after induction but
before the bronchoscope is inserted to minimize the risk of
www.anesthesia-analgesia.org 1021
 REVIEW ARTICLE
gastric aspiration. In delayed presentations in which bron-
choscopy is not urgent, a preanesthetic fast is appropriate.
The airway patency should be assessed. If the patient is
in severe distress, urgent bronchoscopy should be per-
formed. If the patient is stable, however, some authors
suggest that bronchoscopy may be performed during nor-
mal daytime operating hours to ensure optimal conditions
with an experienced bronchoscopist and anesthesiologist.53
These authors found no increase in morbidity in stable
patients by delaying bronchoscopy for a suspected foreign
body until the next available elective daytime slot.53
Anesthetic Considerations for Rigid
Bronchoscopy
Because surgeon and anesthesiologist share management of a
potentially obstructed airway, clear communication and good
cooperation are essential. Before induction, a detailed anes-
thetic and operative plan should be discussed. The 3 main
anesthetic issues involve the methods of induction, ventilation
during bronchoscopy, and maintenance of anesthesia.
The choice of induction is dominated by the consider-
ation of converting a proximal partial obstruction into a
complete obstruction. The conversion from spontaneous
negative pressure breathing to positive pressure ventilation
theoretically risks dislodging an unstable proximal body,
causing complete obstruction.54 Although hypoxic arrest
during the initial stages of bronchoscopy is a recognized
cause of death,10,11,19 the relative contributions of obstruction
on initial presentation, during the induction of anesthesia, and
from dislodgement during bronchoscopy, are unclear from
published accounts. A survey of 838 pediatric anesthesiolo-
gists found that the majority preferred an inhaled induction
when foreign bodies were present in the tracheobronchial
tree.55 A cautious IV induction that maintains spontaneous
ventilation is also possible, although this was not an option in
that particular survey study. While the optimal method of
induction is not definitively established, maintaining sponta-
neous ventilation during the induction of a patient with a
proximal foreign body is commonly practiced.
After induction of general anesthesia, the rigid broncho-
scope is inserted through the glottic opening. The anesthesia
circuit is connected to the sideport of the bronchoscope to
allow ventilation. Both spontaneous ventilation and con-
trolled ventilation are feasible for removal of foreign bodies.
Spontaneous ventilation around the bronchoscope may be
more suitable for removal of proximal bodies, during which
leakage around the scope may make effective positive pres-
sure ventilation difficult. Manually closing the mouth and
nose can diminish a large leak around the scope and improve
ventilation. Positive pressure ventilation down the broncho-
scope, with intermittent apnea while manipulating the object,
may be more suitable for distal retrieval. The use of optical
forceps allows for positive pressure ventilation to be main-
tained while the foreign body is being manipulated so that
periods of apnea can be minimized (Video 2; see Supplemen-
tal Digital Content 2, http://links.lww.com/AA/A170; see
the Appendix for video legends). Because airway trauma and
rupture are significant and potentially fatal complications, it is
essential to avoid coughing and bucking secondary to the
intense stimulation from a rigid bronchoscope deep in the
1022 www.anesthesia-analgesia.org
bronchial tree. Movement can be prevented with neuromus-
cular blocking drugs9,54,56,57 or with a deep level of anesthesia.
One study suggests that topicalization of the tracheobronchial
mucosal using a rigid bronchoscope coated with local anes-
thetic gel improves surgical conditions and more effectively
maintains spontaneous ventilation while decreasing the doses
of anesthetics.58 Although the risk of positive pressure venti-
lation causing distal air trapping by a ball-valve effect has
been suggested,59,60 there is no clear clinical evidence in the
literature surveyed to support this as a practical concern.
A retrospective review of 94 children with aspirated
foreign bodies detected no difference in adverse events on
the basis of the type of ventilation.61 However, 5 of 18
children who were maintained on assisted ventilation and
11 of 26 who were maintained on spontaneous ventilation
were switched to controlled ventilation. A prospective
study of 36 children with aspirated foreign bodies found
that controlled ventilation is more effective than is sponta-
neous ventilation.54 All children in the spontaneous venti-
lation group were switched to either assisted or controlled
ventilation because of coughing and bucking. It is possible,
however, that the necessity of switching from spontaneous
to either assisted or controlled ventilation was due to an
inadequate depth of anesthesia with inhaled drugs rather
than an inherent problem with spontaneous ventilation.62
Larger prospective studies, with both inhaled and IV
maintenance techniques, are necessary to further evaluate
whether spontaneous or controlled ventilation is more
advantageous. In a nonrandomized observational study,
manual jet ventilation was shown to decrease the incidence
of intraoperative hypoxemia in comparison with manual
controlled ventilation and spontaneous ventilation.63
Manual jet ventilation may better allow oxygenation and
ventilation of the unobstructed lung during manipulation
of the foreign body because the jet ventilation catheter was
inserted separately from the bronchoscope.63
Halothane and sevoflurane are 2 volatile anesthetics that
are widely used in pediatric practice. Meretoja et al.64
compared sevoflurane with halothane in 120 children
undergoing bronchoscopy, gastroscopy, or combined
procedures. They reported a higher incidence of cardiac
arrhythmias (nodal rhythm, bigeminy or ventricular
ectopy) in the halothane group (18/60 vs. 4/60). Batra et
al.65 compared the 2 drugs in 44 children undergoing
bronchoscopy specifically for foreign body removal and
found a higher incidence of cardiac arrhythmias in the
halothane group (7/22 vs. 2/22). When comparing halo-
thane and sevoflurane for 62 pediatric bronchoscopies,
Davidson66 found no differences in cord closure, desatura-
tions, breath holding, or coughing.
Although inhaled drugs have traditionally been used for
the maintenance of anesthesia,54,59,61,67 total IV techniques
are becoming more popular in the pediatric popula-
tion.56,62,68,69 A total IV anesthetic with propofol (200 to 400
␮g 䡠 kg⫺1 䡠 min⫺1) and remifentanil (0.05 to 0.2 ␮g 䡠 kg⫺1 䡠
min⫺1) infusions in combination with vocal cord topical-
ization with lidocaine (1 mg 䡠 kg⫺1) allows for spontaneous
ventilation.62 Children younger than 3 years of age can
tolerate higher doses of remifentanil and still maintain
spontaneous ventilation in comparison with older chil-
dren.70 An advantage of an IV anesthetic is that it provides
ANESTHESIA & ANALGESIA
Anesthetic Considerations of Tracheobronchial Foreign Bodies
a constant level of anesthesia irrespective of ventilation. By
contrast, hypoventilation and leaks around the rigid bron-
choscope may produce an inadequate depth of inhaled
anesthesia. Pollution of the operating room, due to the
combination of leaks around the rigid bronchoscope and
high gas flows needed for ventilation, are additional draw-
backs of inhalation anesthetics. Chen et al.63 showed that a
total IV technique with spontaneous ventilation was asso-
ciated with a higher incidence of body movement, breath
holding, and laryngospasm in comparison with an inhaled
technique. However, the doses of IV propofol (100 to 150
␮g 䡠 kg⫺1 䡠 min⫺1) and remifentanil (0.1 ␮g 䡠 kg⫺1 䡠 min⫺1)
were less than those previously described to provide anes-
thesia and maintain spontaneous ventilation.
Dropping the foreign body during retrieval is a poten-
tially life-threatening complication.71,72 The vocal cords
should be well relaxed, either by residual topicalization,
paralysis, or an adequate depth of anesthesia, before re-
moval of the foreign body through the larynx. Dropping
the foreign body has a higher correlation with the experi-
ence level of the bronchoscopist than with the mode of
ventilation.72 If the object is dropped in the proximal
airway and cannot immediately be removed, pushing it
back into a bronchus can eliminate an obstruction. If a
bronchial body falls into the other bronchus, there is
potential for complete airway obstruction due to edema
and inflammation at the original site.71 In the setting of a
marginal airway, optimization of other components of
ventilation is essential. Ventilation may be impaired not
only by the object, but also proximally by upper-airway soft
tissue or cord closure, and distally by atelectasis after
prolonged intraoperative hypoventilation. Optimal head
position, open cords, reinflation of atelectatic segments,
and slow prolonged breaths with adequate pressure can
provide ventilation past a partial obstruction. If ventilation
is impossible, emergent efforts must be made to extract or
move the object. In severe cases of cardiopulmonary failure
due to foreign body obstruction, extracorporeal membrane
oxygenation may facilitate foreign body removal and car-
diopulmonary recovery.73
After the extraction of the foreign body and the removal
of the rigid bronchoscope, the choice of ventilation during
emergence is influenced by pulmonary gas exchange and
the degree of airway edema. For uncomplicated cases,
spontaneous ventilation assisted by mask ventilation as
needed may be adequate. Intubation during emergence
may be indicated for a marginal airway, pulmonary com-
promise, or residual neuromuscular blockade.
Anesthetic Considerations for
Flexible Bronchoscopy
Flexible bronchoscopy can be performed with local anes-
thetic topicalization and sedation in both children and
adults.47– 49,74 –77 IM meperidine and oral diazepam,75 IV
midazolam or fentanyl,75 and atropine and diazepam49,76
or sublingual codeine74 have been successfully used to
sedate adolescents and adults. Topical lidocaine to the
nasopharynx and larynx was combined with 0.1 to 0.3
mg/kg rectal midazolam for 19 younger children.48 Aero-
solized lidocaine in combination with an IM dose of either
atropine (0.01 to 0.02 mg/kg) and diazepam (0.1 to 0.2
October 2010 • Volume 111 • Number 4
mg/kg) or midazolam (0.1 to 0.15 mg/kg) was used for
sedation in 938 young children who had a foreign body
removed by flexible bronchoscopy.33 In that series, the
flexible bronchoscope was inserted intranasally unless na-
sal stenosis was present.33
In smaller children who are unable to cooperate, several
techniques of general anesthesia have been reported. A
balanced anesthetic using IV propofol and sevoflurane with
topical lidocaine and oxymetazoline was used for 23 chil-
dren ages 9 months to 16 years.50 The fiberoptic broncho-
scope was then inserted through a T-piece on the child’s
facemask and advanced transnasally. In a series of 6
children ages 1.2 to 5 years spontaneously breathing under
sevoflurane anesthesia, the bronchoscope was inserted
through a swivel adapter on a laryngeal mask airway.52
The foreign bodies were removed en bloc with the laryn-
geal mask with no adverse events. Flexible bronchoscopy
through endotracheal tubes under general anesthesia is
also described in which the foreign body, bronchoscope,
and endotracheal tube are removed en bloc.51 A standard
pediatric bronchoscope (3.6 mm outer diameter) can be
used with a size 4.5 or larger endotracheal tube, whereas
standard adult bronchoscopes (4.9 mm diameter) will fit
into size 2 or larger laryngeal mask.
Postoperative Considerations
Early discharge after uncomplicated bronchoscopy is rea-
sonable. In one study, 187 (65%) children were discharged
home within 4 hours after rigid bronchoscopy.78 In another
study, 82 (60.7%) children had a hospital stay ⬍1 day.32
Prolonged pulmonary recovery may prevent early dis-
charge. Predictive factors of prolonged recovery included
evidence of inflammation on preoperative radiographs,
aggravation of pulmonary lesions on postoperative films,
and a prolonged duration of bronchoscopy.28,79 Ciftci et
al.11 found bronchoscopy time (57 ⫾ 2.9 minutes vs. 23 ⫾
1.2 minutes) to be prolonged in children with postoperative
complications in comparison with those without complica-
tions. Chen et al.63 found that postoperative hypoxemia
was associated with prolonged emergence from anesthesia
and with foreign bodies that were plant seeds.
CONCLUSIONS
Aspiration of a foreign body is a potentially lethal event.
Although many deaths occur before arrival at the hospital,
anesthesia and bronchoscopy to remove the offending item
are associated with considerable mortality and morbidity.
Outcomes have improved over the years because of ad-
vances in anesthesia and bronchoscopy. Although several
anesthetic techniques are effective for managing children
with foreign body aspiration, there is no consensus from
the literature as to which technique is optimal. An induc-
tion that maintains spontaneous ventilation is commonly
practiced to minimize the risk of converting a partial
proximal obstruction to a complete obstruction. Controlled
ventilation combined with IV drugs and paralysis allows
for suitable rigid bronchoscopy conditions and a consistent
level of anesthesia. The use of CT and virtual bronchoscopy
to diagnose foreign body aspiration and the use of flexible
bronchoscopy for the diagnosis and removal of foreign
bodies may decrease the necessity for rigid bronchoscopy
www.anesthesia-analgesia.org 1023
 REVIEW ARTICLE
under general anesthesia in patients with suspected foreign
body aspiration. As a result, morbidity and mortality in
these children may further decrease. Regardless of the
management strategy, close cooperation within a skilled
surgical and anesthetic team is essential to avoid the
potential hazards of foreign body aspiration.
ACKNOWLEDGMENTS
We wish to thank the following for assistance in preparation of
this manuscript: Daniel Doody, MD, Gennadiy Fuzaylov, MD,
Allan Goldstein, MD, Kenan Haver, MD, David Lawlor, MD,
and Pallavi Sagar, MD, all of the Massachusetts General
Hospital; Cory Collins, DO, and Christopher Hartnick, MD,
both of the Massachusetts Eye and Ear Infirmary; and all of
Harvard Medical School, Boston, Massachusetts.
APPENDIX: VIDEO CAPTIONS
Video 1. Rigid bronchoscopy down the left mainstem bronchus.
Bubbles formed by release of trapped air can be seen during
spontaneous breathing.
Video 2. An optical forceps is used to grasp and remove the object
via rigid bronchoscopy.
REFERENCES
1. Clerf LH. Historical aspects of foreign bodies in the air and
food passages. South Med J 1975;68:1449 –54
2. Zaytoun GM, Rouadi PW, Baki DH. Endoscopic manage-
ment of foreign bodies in the tracheobronchial tree: predic-
tive factors for complications. Otolaryngol Head Neck Surg
2000;123:311– 6
3. Jackson C. The Life of Chevalier Jackson: An Autobiography.
New York: Macmillan, 1938
4. Jackson C. Foreign Bodies in the Air and Food Passages.
Charted Experiences in Cases from no. 631 to no. 1155 at the
Bronchoscopic Clinic. New Bedford: Reynolds the Printer, 1923
5. Miyazawa T, Miyazu Y, Kanoh K, Iwamoto Y. Flexible bron-
choscopy historical perspective. In: edited by Beamis J, Mathur
P, Mehta A, eds. Interventional Pulmonary Medicine (1st ed.,
chap. 3). New York: M. Dekker, 2004:33– 48
6. Dikensoy O, Usalan C, Filiz A. Foreign body aspiration: clinical
utility of flexible bronchoscopy. Postgrad Med J 2002;78:399 – 403
7. Aydogan LB, Tuncer U, Soylu L, Kiroglu M, Ozsahinoglu C.
Rigid bronchoscopy for the suspicion of foreign body in the
airway. Int J Pediatr Otorhinolaryngol 2006;70:823– 8
8. Ayed AK, Jafar AM, Owayed A. Foreign body aspiration in
children: diagnosis and treatment. Pediatr Surg Int 2003;19:485– 8
9. Bittencourt PF, Camargos PA, Scheinmann P, de Blic J. Foreign
body aspiration: clinical, radiological findings and factors
associated with its late removal. Int J Pediatr Otorhinolaryngol
2006;70:879 – 84
10. Brkic F, Delibegovic-Dedic S, Hajdarovic D. Bronchoscopic
removal of foreign bodies from children in Bosnia and Herze-
govina: experience with 230 patients. Int J Pediatr Otorhino-
laryngol 2001;60:193– 6
11. Ciftci AO, Bingol-Kologlu M, Senocak ME, Tanyel FC, Buyuk-
pamukcu N. Bronchoscopy for evaluation of foreign body
aspiration in children. J Pediatr Surg 2003;38:1170 – 6
12. Cohen S, Avital A, Godfrey S, Gross M, Kerem E, Springer C.
Suspected foreign body inhalation in children: what are the
indications for bronchoscopy? J Pediatr 2009;155:276 – 80
13. Divisi D, Di Tommaso S, Garramone M, Di Francescantonio W,
Crisci RM, Costa AM, Gravina GL, Crisci R. Foreign bodies
aspirated in children: role of bronchoscopy. Thorac Cardiovasc
Surg 2007;55:249 –52
14. Emir H, Tekant G, Besik C, Elicevik M, Senyuz OF, Buyukunal
C, Sarimurat N, Yeker D. Bronchoscopic removal of tracheo-
broncheal foreign bodies: value of patient history and timing.
Pediatr Surg Int 2001;17:85–7
15. Eren S, Balci AE, Dikici B, Doblan M, Eren MN. Foreign body
aspiration in children: experience of 1160 cases. Ann Trop
Paediatr 2003;23:31–7
1024 www.anesthesia-analgesia.org
16. Erikçi V, Karaçay S, Arikan A. Foreign body aspiration: a
four-years experience. Ulus Travma Acil Cerrahi Derg
2003;9:45–9
17. Girardi G, Contador AM, Castro-Rodríguez JA. Two new
radiological findings to improve the diagnosis of bronchial
foreign-body aspiration in children. Pediatr Pulmonol 2004;38:
261– 4
18. Gregori D, Salerni L, Scarinzi C, Morra B, Berchialla P, Snidero
S, Corradetti R, Passali D, Klaus A, Isidor H, Gernot S, Jan B,
Bernard B, Karchev T, Tzolov T, Ranko M, Lana K, Ivo S, Mirko
T, Caye-Thomasen P, Anne P, Volker J, Onder G, Simasko N,
Matilda C, Christopoulos I, Passà li GC, Passà li F, Damiani V,
MieczysÅ,aw C, Dorin S, Gheorghe DC, Janka J, Miha Z, Ales
G, Ales M, Lorenzo R, Javier C, Pontus S, Philippe P, Ahmed C,
Metin OT, Ozden CA, Riza D, John G, Peter R, Rupert O.
Foreign bodies in the upper airways causing complications
and requiring hospitalization in children aged 0 –14 years:
results from the ESFBI study. Eur Arch Otorhinolaryngol
2008;265:971– 8
19. Hasdiraz L, Oguzkaya F, Bilgin M, Bicer C. Complications of
bronchoscopy for foreign body removal: experience in 1,035
cases. Ann Saudi Med 2006;26:283–7
20. Heyer CM, Bollmeier ME, Rossler L, Nuesslein TG, Stephan V,
Bauer TT, Rieger CH. Evaluation of clinical, radiologic, and
laboratory prebronchoscopy findings in children with sus-
pected foreign body aspiration. J Pediatr Surg 2006;41:1882– 8
21. Higuchi O, Adachi Y, Ichimaru T, Asai M, Kawasaki K. Foreign
body aspiration in children: a nationwide survey in Japan. Int
J Pediatr Otorhinolaryngol 2009;73:659 – 61
22. Hui H, Na L, Zhijun CJ, Fugao ZG, Yan S, Niankai ZK, Jingjing
CJ. Therapeutic experience from 1428 patients with pediatric
tracheobronchial foreign body. J Pediatr Surg 2008;43:718 –21
23. Kadmon G, Stern Y, Bron-Harlev E, Nahum E, Battat E,
Schonfeld T. Computerized scoring system for the diagnosis of
foreign body aspiration in children. Ann Otol Rhinol Laryngol
2008;117:839 – 43
24. Karakoc F, Karadag B, Akbenlioglu C, Ersu R, Yildizeli B,
Yuksel M, Dagli E. Foreign body aspiration: what is the
outcome? Pediatr Pulmonol 2002;34:30 – 6
25. Kiyan G, Gocmen B, Tugtepe H, Karakoc F, Dagli E, Dagli TE.
Foreign body aspiration in children: the value of diagnostic
criteria. Int J Pediatr Otorhinolaryngol 2009;73:963–7
26. Mahafza T, Khader Y. Aspirated tracheobronchial foreign bodies:
a Jordanian experience. Ear Nose Throat J 2007;86:107–10
27. Roda J, Nobre S, Pires J, Estêvão MH, Félix M. Foreign bodies
in the airway: a quarter of a century’s experience. Rev Port
Pneumol 2008;14:787– 802
28. Roh JL, Hong SJ. Lung recovery after rigid bronchoscopic
removal of tracheobronchial foreign bodies in children. Int
J Pediatr Otorhinolaryngol 2008;72:635– 41
29. Shivakumar AM, Naik AS, Prashanth KB, Shetty KD, Praveen
DS. Tracheobronchial foreign bodies. Indian J Pediatr 2003;70:
793–7
30. Shubha AM, Das K. Tracheobronchial foreign bodies in infants.
Int J Pediatr Otorhinolaryngol 2009;73:1385–9
31. Skoulakis CE, Doxas PG, Papadakis CE, Proimos E,
Christodoulou P, Bizakis JG, Velegrakis GA, Mamoulakis D,
Helidonis ES. Bronchoscopy for foreign body removal in
children. A review and analysis of 210 cases. Int J Pediatr
Otorhinolaryngol 2000;53:143– 8
32. Tan HK, Brown K, McGill T, Kenna MA, Lund DP, Healy GB.
Airway foreign bodies (FB): a 10-year review. Int J Pediatr
Otorhinolaryngol 2000;56:91–9
33. Tang LF, Xu YC, Wang YS, Wang CF, Zhu GH, Bao XE, Lu MP,
Chen LX, Chen ZM. Airway foreign body removal by flexible
bronchoscopy: experience with 1027 children during 2000 –2008.
World J Pediatr 2009;5:191–5
34. Tokar B, Ozkan R, Ilhan H. Tracheobronchial foreign bodies in
children: importance of accurate history and plain chest radi-
ography in delayed presentation. Clin Radiol 2004;59:609 –15
35. Tomaske M, Gerber AC, Stocker S, Weiss M. Tracheobronchial
foreign body aspiration in children— diagnostic value of
symptoms and signs. Swiss Med Wkly 2006;136:533– 8
ANESTHESIA & ANALGESIA
Anesthetic Considerations of Tracheobronchial Foreign Bodies
36. Yadav SP, Singh J, Aggarwal N, Goel A. Airway foreign bodies
in children: experience of 132 cases. Singapore Med J 2007;48:
850 –3
37. Pinto A, Scaglione M, Pinto F, Guidi G, Pepe M, Del Prato B,
Grassi R, Romano L. Tracheobronchial aspiration of foreign
bodies: current indications for emergency plain chest radiog-
raphy. Radiol Med 2006;1112:497–506
38. Assefa D, Amin N, Stringel G, Dozor AJ. Use of decubitus
radiographs in the diagnosis of foreign body aspiration in
young children. Pediatr Emerg Care 2007;23:154 –7
39. Sodhi KS, Saxena AK, Singh M, Rao KL, Khandelwal N. CT
virtual bronchoscopy: new non invasive tool in pediatric
patients with foreign body aspiration. Indian J Pediatr
2008;75:511–3
40. Veras TN, Hornburg G, Schner AM, Pinto LA. Use of virtual
bronchoscopy in children with suspected foreign body aspira-
tion. J Bras Pneumol 2009;35:937– 41
41. Hong SJ, Goo HW, Roh JL. Utility of spiral and cine CT scans
in pediatric patients suspected of aspirating radiolucent for-
eign bodies. Otolaryngol Head Neck Surg 2008;138:576 – 80
42. Kosucu P, Ahmetoglu A, Koramaz I, Orhan F, Ozdemir O,
Dinc H, Okten A, Gumele HR. Low-dose MDCT and virtual
bronchoscopy in pediatric patients with foreign body aspira-
tion. AJR Am J Roentgenol 2004;183:1771–7
43. Huang HJ, Fang HY, Chen HC, Wu CY, Cheng CY, Chang CL.
Three-dimensional computed tomography for detection of
tracheobronchial foreign body aspiration in children. Pediatr
Surg Int 2008;24:157– 60
44. Adaletli I, Kurugoglu S, Ulus S, Ozer H, Elicevik M, Kantarci F,
Mihmanli I, Akman C. Utilization of low-dose multidetector
CT and virtual bronchoscopy in children with suspected
foreign body aspiration. Pediatr Radiol 2007;37:33– 40
45. Haliloglu M, Ciftci AO, Oto A, Gumus B, Tanyel FC, Senocak
ME, Buyukpamukcu N, Besim A. CT virtual bronchoscopy in
the evaluation of children with suspected foreign body aspi-
ration. Eur J Radiol 2003;48:188 –92
46. Even L, Heno N, Talmon Y, Samet E, Zonis Z, Kugelman A.
Diagnostic evaluation of foreign body aspiration in children: a
prospective study. J Pediatr Surg 2005;40:1122–7
47. Martinot A, Closset M, Marquette CH, Hue V, Deschildre A,
Ramon P, Remy J, Leclerc F. Indications for flexible versus
rigid bronchoscopy in children with suspected foreign-body
aspiration. Am J Respir Crit Care Med 1997;155:1676 –9
48. Righini CA, Morel N, Karkas A, Reyt E, Ferretti K, Pin I,
Schmerber S. What is the diagnostic value of flexible bronchos-
copy in the initial investigation of children with suspected
foreign body aspiration? Int J Pediatr Otorhinolaryngol
2007;71:1383–90
49. Gencer M, Ceylan E, Koksal N. Extraction of pins from the
airway with flexible bronchoscopy. Respiration 2007;74:674 –9
50. Ramirez-Figueroa JL, Gochicoa-Rangel LG, Ramirez-San Juan
DH, Vargas MH. Foreign body removal by flexible fiberoptic
bronchoscopy in infants and children. Pediatr Pulmonol
2005;40:392–7
51. Swanson KL, Prakash UB, Midthun DE, Edell ES, Utz JP, McDou-
gall JC, Brutinel WM. Flexible bronchoscopic management of
airway foreign bodies in children. Chest 2002;121:1695–1700
52. Yazbeck-Karam VG, Aouad MT, Baraka AS. Laryngeal mask
airway for ventilation during diagnostic and interventional fibre-
optic bronchoscopy in children. Paediatr Anaesth 2003;13:691– 4
53. Mani N, Soma M, Massey S, Albert D, Bailey CM. Removal of
inhaled foreign bodies—middle of the night or the next morn-
ing? Int J Pediatr Otorhinolaryngol 2009;73:1085–9
54. Soodan A, Pawar D, Subramanium R. Anesthesia for removal
of inhaled foreign bodies in children. Paediatr Anaesth
2004;14:947–52
55. Kain ZN, O’Connor TZ, Berde CB. Management of tracheo-
bronchial and esophageal foreign bodies in children: a survey
study. J Clin Anesth 1994;6:28 –32
56. Sersar SI, Rizk WH, Bilal M, El Diasty MM, Eltantawy TA,
Abdelhakam BB, Elgamal AM, Bieh AA. Inhaled foreign
bodies: presentation, management and value of history and
plain chest radiography in delayed presentation. Otolaryngol
Head Neck Surg 2006;134:92–9
October 2010 • Volume 111 • Number 4
57. Soysal O, Kuzucu A, Ulutas H. Tracheobronchial foreign body
aspiration: a continuing challenge. Otolaryngol Head Neck
Surg 2006;135:223– 6
58. Yu H, Yang XY, Liu B. EMLA cream coated on the rigid
bronchoscope for tracheobronchial foreign body removal in
children. Laryngoscope 2009;119:158 – 61
59. Baraka A. Bronchoscopic removal of inhaled foreign bodies in
children. Br J Anaesth 1974;46:124 – 6
60. Farrell PT. Rigid bronchoscopy for foreign body removal:
anaesthesia and ventilation. Paediatr Anaesth 2004;14:84 –9
61. Litman RS, Ponnuri J, Trogan I. Anesthesia for tracheal or
bronchial foreign body removal in children: an analysis of
ninety-four cases. Anesth Analg 2000;91:1389 –91
62. Buu NT, Ansermino M. Anesthesia for removal of inhaled
foreign bodies in children. Paediatr Anaesth 2005;15:533
63. Chen LH, Zhang X, Li SQ, Liu YQ, Zhang TY, Wu JZ. The risk
factors for hypoxemia in children younger than 5 years old
undergoing rigid bronchoscopy for foreign body removal.
Anesth Analg 2009;109:1079 – 84
64. Meretoja OA, Taivainen T, Raiha L, Korpela R, Wirtavuori
K. Sevoflurane-nitrous oxide or halothane-nitrous oxide for
paediatric bronchoscopy and gastroscopy. Br J Anaesth
1996;76:767–71
65. Batra YK, Mahajan R, Bangalia SK, Chari P, Rao KL.
A comparison of halothane and sevoflurane for broncho-
scopic removal of foreign bodies in children. Ann Card
Anaesth 2004;7:137– 43
66. Davidson A. The correlation between bispectral index and
airway reflexes with sevoflurane and halothane anaesthesia.
Paediatr Anaesth 2004;14:241– 6
67. Cohen SR, Geller KA. Anesthesia and pediatric endoscopy: the
surgeon’s view. Otolaryngol Clin North Am 1981;14:705–13
68. Maddali MM, Badur RS, Fernando MS, Alsajwani MJ. Total
contralateral atelectasis following rigid bronchoscopy in a
child with scarf pin aspiration. Paediatr Anaesth 2006;16:1095–7
69. Perrin G, Colt HG, Martin C, Mak MA, Dumon JF, Gouin F.
Safety of interventional rigid bronchoscopy using intravenous
anesthesia and spontaneous assisted ventilation. a prospective
study. Chest 1992;102:1526 –30
70. Barker N, Lim J, Amari E, Malherbe S, Ansermino JM. Rela-
tionship between age and spontaneous ventilation during
intravenous anesthesia in children. Paediatr Anaesth 2007;17:
948 –55
71. Kumar S, Saxena AK, Kumar M, Rautela RS, Gupta N, Goyal
A. Anesthetic management during bronchoscopic removal of a
unique, friable foreign body. Anesth Analg 2006;103:1596 –7
72. Pawar DK. Dislodgement of bronchial foreign body during
retrieval in children. Paediatr Anaesth 2000;10:333–5
73. Brown KL, Shefler A, Cohen G, DeMunter C, Pigott N,
Goldman AP. Near-fatal grape aspiration with complicating
acute lung injury successfully treated with extracorporeal
membrane oxygenation. Pediatr Crit Care Med 2003;4:243–5
74. Mise K, Jurcev Savicevic A, Pavlov N, Jankovic S. Removal of
tracheobronchial foreign bodies in adults using flexible bron-
choscopy: experience 1995–2006. Surg Endosc 2009;23:1360 – 4
75. Tariq SM, George J, Srinivasan S. Inhaled foreign bodies in
adolescents and adults. Monaldi Arch Chest Dis 2005;63:193– 8
76. Uskul TB, Turker H, Arslan S, Slevi A, Kant A. Use of
fiberoptic bronchoscopy in endobronchial foreign body re-
moval in adults. Turkish Resp J 2007;8:39 – 42
77. Wong KS, Lai SH, Lien R, Hsia SH. Retrieval of bronchial
foreign body with central lumen using a flexible bronchoscope.
Int J Pediatr Otorhinolaryngol 2002;62:253– 6
78. Tomaske M, Gerber AC, Weiss M. Anesthesia and periinter-
ventional morbidity of rigid bronchoscopy for tracheobron-
chial foreign body diagnosis and removal. Paediatr Anaesth
2006;16:123–9
79. Chung MK, Jeong HS, Ahn KM, Park SH, Cho JK, Son YI, Baek
CH. Pulmonary recovery after rigid bronchoscopic retrieval of
airway foreign body. Laryngoscope 2007;117:303–7
www.anesthesia-analgesia.org 1025
Economics, Education, and Policy
Section Editor: Franklin Dexter
SPECIAL ARTICLE

An Optimistic Prognosis for the Clinical Utility of

Laboratory Test Data
Ming Zheng, PhD,* Palanikumar Ravindran, PhD,† Jianmei Wang, PhD,† Richard H. Epstein, MD,‡
David P. Chen, PhD,* Atul J. Butte, MD, PhD,* and Gary Peltz, MD, PhD*

It is hoped that anesthesiologists and other clinicians will be able to increasingly rely upon
laboratory test data to improve the perioperative care of patients. However, it has been
suggested that in order for a laboratory test to have clinically useful diagnostic performance
characteristics (sensitivity and specificity), its performance must be considerably better than
those that have been evaluated in most etiologic or epidemiologic studies. This pessimism
about the clinical utility of laboratory tests is based upon the untested assumption that
laboratory data are normally distributed within case and control populations.
We evaluated the data distribution for 700 commonly ordered laboratory tests, and found
that the vast majority (99%) do not have a normal distribution. The deviation from normal was
most pronounced at extreme values, which had a large quantitative effect on laboratory test
performance. At the sensitivity and specificity values required for diagnostic utility, the
minimum required odds ratios for laboratory tests with a nonnormal data distribution were
significantly smaller (by orders of magnitude) than for tests with a normal distribution.
By evaluating the effect that the data distribution has on laboratory test performance, we
have arrived at the more optimistic outlook that it is feasible to produce laboratory tests with
diagnostically useful performance characteristics. We also show that moderate errors in the
classification of outcome variables (e.g., death vs. survival at a specified end point) have a
small impact on test performance, which is of importance for outcomes research that uses
anesthesia information management systems. Because these analyses typically seek to identify
factors associated with an undesirable outcome, the data distributions of the independent
variables need to be considered when interpreting the odds ratios obtained from such
investigations. (Anesth Analg 2010;111:1026 –35)

I n contemporary clinical practice, a large number of

laboratory tests are performed to facilitate diagnosis, to
assess disease progression or the effect of a therapy, and
for prognostication. With the advent of genomic science, an
even larger number of variables (mRNAs, proteins, and
metabolites) can be measured in blood or tissues in a
cost-effective way. For anesthesiologists, it is likely that
many different metabolites, proteins, or neural variables
will be measured in the perioperative setting that could
provide data useful for patient assessment and treatment.
Genetic risk factors can be measured to identify individuals
at increased risk for disease or perioperative complications,
or for predicting surgical outcome. Also, analyzing joined
From *Stanford University, Stanford, California; †Roche Palo Alto LLC, Palo
Alto, CA; and ‡Jefferson Medical College, Philadelphia, Pennsylvania.
Accepted for publication June 4, 2010.
Gary Peltz was partially supported by funding (7R56 GMO68885 to 05) from
the NIGMS. Funding was provided by NLM T15 LM007033 (to David P.
Chen) and by the Lucile Packard Foundation for Children’s Health (to Atul
J. Butte).
Supplemental digital content is available for this article. Direct URL citations
appear in the printed text and are provided in the HTML and PDF versions
of this article on the journal’s Web site (www.anesthesia-analgesia.org).
Address correspondence to Gary Peltz, MD, PhD, Stanford University, 800
Welch Road, Room 213, Palo Alto, CA 94304. Address e-mail to
gpeltz@stanford.edu.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181efff0c
datasets from multiple institutions’ anesthesia information
management systems (AIMS) through data-mining tech-
niques could identify threshold parameters associated with
undesirable outcomes or to assess risk.
There is reason to believe that additional data and
genetic risk factor measurements will improve our ability
to diagnose, stratify, and optimize the perioperative care of
our patients. For example, a quantitative simulation dem-
onstrated how the use of pharmacogenetic information to
individualize drug dosage has the potential to significantly
improve treatment outcome.1
Nevertheless, several publications have suggested that
newly discovered laboratory tests would have limited
diagnostic utility for individual patients. The argument
advanced was that in order for a laboratory measurement
to have clinically useful performance characteristics (sensi-
tivity and specificity), the magnitude of the odds ratio (OR)
for the test (Text Box) must be considerably higher than
those seen in most etiologic or epidemiologic studies.2– 4
For example, a hypothetical analysis by Ware2 indicated
that an OR of 228 would be required for a test result to have
sufficient diagnostic or predictive utility (80% specificity
and 80% sensitivity).
If this level of certainty is truly required, it would be
virtually impossible to identify a genetic test with accept-
able performance for the vast majority of clinical situations.
This is because ORs for most of the genetic risk factors for

1026 www.anesthesia-analgesia.org October 2010 • Volume 111 • Number 4

An Optimistic Prognosis for Clinical Utility of Lab Test
quantitative traits or susceptibility to common diseases that
have been identified are usually much lower than the
proposed threshold. For example, analysis of the 1967
identified human single nucleotide polymorphisms with a
reported OR for a studied traita indicated that 67% have
ORs ⬍5, and 95% have ORs ⬍30.
In a similar negative vein, Pepe et al.4 investigated the
relationship between the OR and classification accuracy.
They analyzed hypothetical data with a normal distribution
and concluded that an OR of 74 was required to obtain
clinically useful performance characteristics (79% sensitiv-
ity and 79% specificity). Taken at face value, these 2
analyses provide a pessimistic outlook for the utility of
laboratory tests, because they imply that the performance
of most laboratory tests will not be sufficient to have much
diagnostic or predictive utility.
However, these analyses2– 4 are based on a fundamental
assumption that laboratory data are normally distributed
within case and control populations. They also assume
that currently used disease definitions and classification
schema, developed before the discovery of genetic risk
factors or novel laboratory tests, will continue to be
applied.
In this article, we demonstrate that the clinical utility of
laboratory data has a much better prognosis than that
suggested by Ware and by Pepe et al. The vast majority of
laboratory results do not have a normal distribution within
either control or disease populations. As a consequence, the
performance characteristics (i.e., sensitivity and specificity
for disease diagnosis) of laboratory tests are substantially
improved over Gaussian assumptions. Thus, laboratory
data whose performance characteristics are within the
usual range observed in epidemiologic or etiologic studies
can have substantial diagnostic utility. These consider-
ations lead to a more optimistic assessment of the utility of
laboratory tests in improving patient care.
METHODS
Analysis of Clinical Laboratory Test Data
All available data for 700 clinical laboratory tests per-
formed between 2000 and 2006 were retrieved from the
Stanford Translational Research Integrated Database
(STRIDE) according to a protocol that was approved by the
IRB. These tests were routinely performed at the Stanford
and Lucile Packard Children’s Hospitals, and included
patients between 0 and 108 years of age. At least 1000
measurements were available for each test (minimum 1001,
maximum 2,090,227, median 3466). The Jarque–Bera test7
was used to evaluate the normality of the data distribu-
tion in each test. The test was performed on the original
data and on the logarithm-transformed data, which is a
transformation commonly used to analyze data with a
significant rightward skewing. The larger of the 2 P
values obtained was reported. The Benjamini–Hochberg
adjustment method was then used to adjust for multiple
testing.9 Because the null hypothesis was that the distri-
bution was indeed normal, the smaller the adjusted P
value, the greater the deviation from normal. When the
a
http://www.genome.gov/gwastudies, accessed on April 10, 2010.
October 2010 • Volume 111 • Number 4
adjusted P value was smaller than 0.05, the null hypoth-
esis was rejected at the 5% significance level (i.e., the data
were not normally distributed).
Biomarker Data Analysis
For the 3 types of laboratory data studied in detail, deiden-
tified data were obtained from the STRIDE. The Interna-
tional Classification of Diseases, Clinical Modification
(ICD-9), codes for each individual with an available labo-
ratory value were evaluated to identify the control and
disease populations by using the STRIDE Anonymous
Patient Cohort Discovery Tool that was developed by the
Stanford Center for Clinical Informatics.10 The hemoglobin
A1C (immunoassay), CD19⫹ cell counts (flow immunocy-
tometry), and protein S activity (automated latex immuno-
assay) were measured using standard protocols in the
clinical laboratory at Stanford University.
The following ICD-9 codes were used to classify indi-
viduals with available laboratory data: diabetes (250), lym-
phoma (200,201), and coagulopathy (286). There were
33,958 A1C measurements from 20,590 nondiabetic indi-
viduals and 41,541 A1C measurements from 10,677 diabetic
individuals from the database. CD19⫹ cell counts included
17,706 measurements from 4498 individuals not diagnosed
as having a lymphoma, and 1861 measurements from 541
individuals with a lymphoma. Protein S activities included
3701 measurements from 3385 individuals without a diag-
nosed coagulopathy and 519 measurements from 420 indi-
viduals with a diagnosed coagulopathy. When a test was
performed more than once on an individual, the value
obtained on the initial visit was used. (However, results
were insensitive to the substitution of lab values from
subsequent visits.)
The A1C lab values were extensively right skewed;
therefore, a log-transformation was applied to make the
data distribution more symmetric. Because values of 0 were
obtained for some CD19 and the protein S values, 1 was
added to the original value before log transformation,
because log(0) is undefined. All subsequent analyses were
performed on the transformed data.
Additional information about the OR determinations for
different data distributions and for A1C and other lab test
data, and for the other simulation studies, is provided in
the online Supplementary Methods (see Supplemental
Digital Content 1, http://links.lww.com/AA/A172) for
this journal.
RESULTS
Clinical Laboratory Data Usually Are not
Normally Distributed
Even if most of the population data appear to follow the
shape of a normal distribution, the values at the tails
(within the population extremes) may diverge markedly
from that predicted by the normal distribution. For in-
stance, the measured serum concentration of a protein
could have a nonnormal distribution within the extremes
of a population because of a limitation in synthetic
capacity, a biological feedback mechanism that limits its
maximum concentration, or a threshold level of stimula-
tion that may be required to initiate the synthesis of a
www.anesthesia-analgesia.org 1027
 SPECIAL ARTICLE

Text Box: OR Definitions and Data Distribution.

The OR compares the measured value for a risk factor within a population with a disease to that in a control population
without the disease. Traditionally, the relationship between a binary test result and a binary outcome is represented in
a two-way table. For example, the Bispectral Index (BIS) is used to titrate the administration of drugs during general
anesthesia to minimize the risk of intraoperative awareness.5 This measure can be used to illustrate how a quantitative
measure can be converted into a binary parameter (“BIS status”). Suppose analysis of an AIMS database that includes
adverse outcome data leads to a putative test to predict the occurrence of intraoperative awareness if the BIS were more
than 70 for more than 5 min during the interval from intubation to the end of surgery. Such a test might be useful to
guide the rapidity with which BIS increases should be treated as well as to identify patients who need postoperative
follow-up to assess and possibly treat the consequences of intraoperative awareness. Cases where the BIS was elevated
as such are characterized as BIS⫹ and those where there were no such epochs as BIS-. Then, we can construct the
following two-way table:
Awareness Amnesia
BIS- n11 n12
BISⴙ n21 n22
We have a total of n individuals (n ⫽ n11⫹n12⫹n21⫹n22), where n11 of them are BIS- and have awareness during the
surgery and n12, n21 and n22 are similarly defined according to the row and column labels for the corresponding cells
in the table. Then, the odd of having awareness versus amnesia during the surgery in the BIS negative group is n11/n12.
The odd of having awareness versus amnesia in the BIS positive group is n21/n22. The OR is then calculated as the ratio
of these two odds: OR ⫽ (n11/n12)/(n21/n22) ⫽ n11n22/n21n12. If the criterion of the BIS being more than 70 for ⬎5 min
were a good indicator of outcome, then n11 and n22 would be large, and n12 and n21 small, relative to n11 and n22. As
a consequence, a large OR is obtained. In contrast, a small OR (close to 1) indicates that the test result has a small effect
on outcome probability, indicating that it has a poor performance.
Several other important measurements are frequently used to assess test performance. If we denote the individuals
having amnesia during the surgery as cases, and those with awareness as controls, then n11, n12, n21 and n22 are known
as true negative (TN), false negative (FN), false positive (FP) and true positive (TP), respectively. Specificity is the
probability that a control individual is correctly classified; and sensitivity is the probability that a case individual is
correctly classified. Additionally, the positive predictive value and negative predictive value are the chances that an
individual predicted as case (or control) is actually a case or control, respectively. The following table illustrates how
these values are calculated:

Awareness Amnesia
BIS- n11 (TN) n12 (FN) NPVⴝTN/(TNⴙFN)
BISⴙ n21 (FP) n22 (TP) PPVⴝTP/(TPⴙFP)
Specificity ⴝ Sensitivity ⴝ
TN/(TN ⴙ FP) TP/(TPⴙFN)
However, by focusing on different case and control populations, different methods can be used to calculate the OR,
which can markedly affect its value. In the analysis by Ware,2 the OR is defined as the ratio of the frequency of an event
occurring within the group with the disease (cases) whose risk factor value places them at the 90th percentile (Figure
T1, left panel, arrow labeled Group 2) relative to the frequency of events within the control group whose risk factor
value is at the 10th percentile (Figure T1, left panel, arrow labeled Group 1). The difference in the mean frequencies
between the case and control groups is used to calculate the OR for a risk factor. However, more conventionally, the
OR is defined as the ratio of the odds of an event occurring in one group relative to the odds of it occurring in another
group where all values are at or above a given value (90th percentile) in one group (Figure T1, right panel, orange area),
and the values that are at or below a given value (10th percentile) in the second group (Figure T1, right panel, blue area).
The conventional method for OR calculation can be more easily and accurately determined.
The data distribution within a population also affects laboratory data performance. The data distribution for
populations with a normal (Blue), a Double-Exponential (green) or a Cauchy (Red) distribution are shown in Figure T2.
Note the differences at the extremes of the population distribution, where the Cauchy curve and the Double-
Exponential curve do not tail-off as rapidly as does the normal curve. The maximum likelihood method is used to fit
the actual data to different types of distributions. The fitted curve can be overlaid to the histogram. Visual inspection
provides an estimate of the goodness-of-fit of the fitted distribution and statistical tests (Kolmogorov-Smirnov test6 for
general distributions, and the Jarque-Bera7 or Shapiro-Wilk8 tests for normal distribution) can be used to rigorously
assess the fit.

1028 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

An Optimistic Prognosis for Clinical Utility of Lab Test

Figure T1: Definition of odds ratio using the Ware (Left) and Conventional Definitions (Right). According to the Ware Definition, (2) the odds
ratio is defined as the ratio of the frequency of an event occurring within the group with the disease (cases) whose risk factor value places them
at the 90th percentile (arrow labeled in Group 2) relative to the frequency of events within the control group whose risk factor value is at the
10th percentile (arrow labeled in Group 1). The difference in the mean frequencies between the case and control groups is used to calculate
the odds ratio for a risk factor. According to the Conventional Definition, the odds ratio is defined as the ratio of the odds of an event occurring
in one group relative to the odds of it occurring in another group where all values are at or above a given value (90th percentile) in one group
(orange area), and the values that are at or below a given value (10th percentile) in the second group (blue area).

Figure T2: The data distribution for popu-

lations with normal (Blue), Double-
Exponential (green) or Cauchy (Red)
distributions.

disease-associated protein. Any of these effects would
flatten the data distribution curve at the extremes of
control or diseased populations. This is important be-
cause calculation of ORs involves assessments at the tails
of the distribution, whether one follows the calculation
method described by Ware or the more conventional
approach (Text Box).
To assess the possibility that the prior estimates of
ORs required for adequate test performance are overly
conservative because of deviations from the normal
distribution at the tails, we evaluated the data distribu-
tion of all 700 clinical laboratory tests. A detailed ex-
ample follows.
The A1C test (HbA1c, glycated hemoglobin, or glycosy-
lated hemoglobin) is a commonly used laboratory test that
reflects the effectiveness of blood glucose regulation.11
Histograms of A1C lab values for 20,590 control (nondia-
betic) individuals and 10,677 diabetic individuals reveal a
deviation of these data from a normal distribution in either
population (Fig. 1). Although a superficial visual inspection
of the shape of the data distribution might seem to indicate
a normal distribution, this is not a rigorous method for such
determinations. Therefore, we used the Jarque–Bera test to
assess the normality of this data. The resulting P values are
nearly zero for the case and control populations, indicating
that these data have a highly nonnormal distribution,
which is consistent with the shape of the histograms. The
same data are also graphed as quantile– quantile plots,
where the A1C lab values are plotted in relation to the
percentile for the theoretical normal distribution and de-
partures from linearity indicate where the data do not have
a normal distribution. These plots demonstrate that the

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 1029

 SPECIAL ARTICLE
distribution of A1C lab values at the extremes deviates
significantly from the normal distribution, especially in the
control population (Fig. 1).
We analyzed the raw data for all 700 laboratory
tests with the same methodology, and found that 699
(99.9%) tests were not normally distributed. After log-
transformation, the data in 694 (99.1%) tests remained non-
normally distributed. The serum transferrin level was the only
test whose data had a normal distribution; only 5 other tests
(B-type natriuretic peptide, 1-hour glucose tolerance test,
glucostatin, hematocrit, and total protein) were normally
distributed after log-transformation. Thus, the data for nearly
all laboratory tests do not have a normal distribution.
A Nonnormal Distribution Significantly Alters
Laboratory Data Performance
Because the ORs and laboratory data performance are
assessed with data obtained from the extremes of a popu-
lation (Text Box), the distribution of the data within the
population extremes has a large effect on its utility for
disease diagnosis. For example, if laboratory data were
more flatly distributed at the extremes, the data distribu-
tion would resemble a double-exponential12 or a Cauchy
distribution13 (Text Box). These two types of data distribu-
tion may better fit the rate of decay of the probability
density for the tails. Consequently, these distributions
better fit the actual data distribution at the extremes than
does the normal distribution.
To investigate the potential implications of this effect,
we plotted the OR as a function of the sensitivity and
specificity for lab test data with a normal or a Cauchy
1030 www.anesthesia-analgesia.org
Figure 1. A1C lab values are not normally
distributed in control (nondiabetic) or dia-
betic populations. Top panel, The histo-
gram (left) and the quantile– quantile (QQ)
plot (right) of the (log-transformed) A1C lab
values in control populations. Bottom pan-
els, The histogram (left) and the QQ plot
(right) of the (log-transformed) A1C lab
values in diabetic populations. In the
histograms, the red curve represents
the fitted normal curve. In the QQ plots,
the red lines indicate data with a normal
distribution. The P values for the null
hypothesis that the data are normally
distributed are 0 for both the control
and diabetic populations, indicating
that the data have a highly nonnormal
distribution.
distribution (Supplemental Figs. 1 [see Supplemental Digital
Content 2, http://links.lww.com/AA/A173 ] and 2 [see Supple-
mental Digital Content 2, http://links.lww.com/AA/A174 ]; see
Supplementary Methods section for figure legends,
http://links.lww.com/AA/A172). The dramatically different
shapes of these curves demonstrate that laboratory test per-
formance (sensitivity and specificity) at a specified OR is
markedly altered if the data are not normally distributed.
Furthermore, the effect of a nonnormal data distribution
is especially pronounced under conditions in which high
sensitivity and specificity are required. To illustrate this, we
prepared a table showing the values of ORs calculated for
clinically useful levels of sensitivity and specificity for data
with a normal, a double-exponential or a Cauchy distribu-
tion (Tables 1 and 2). The OR definition applied clearly
impacts laboratory test performance. In general, use of the
Ware definition increased the ORs that were required for a
laboratory test to have clinically useful performance
(80%–90% sensitivity and 80%–90% specificity) by ⬎10-fold
in relation to the conventional OR definition.
However, independent of which OR definition was
used, the data distribution within the extremes of a popu-
lation had a very significant effect on test utility and
performance characteristics. A laboratory test with 80%
sensitivity and 90% specificity require an OR (Ware defini-
tion) of 231 for normally distributed data. However, using
the same assumptions as Ware (the data distribution in the
case and control populations has the same shape), an OR of
15 or 25 can provide the same performance for data with a
Cauchy distribution or a double-exponential distribution,
ANESTHESIA & ANALGESIA
An Optimistic Prognosis for Clinical Utility of Lab Test

Table 1. Comparison of Odds Ratio Values Table 3. Performance Characteristics (Specificity

Required for Laboratory Data with a Normal, and Sensitivity) of A1C Laboratory Test Data
Double-Exponential, or Cauchy Distribution to No. correctly
Achieve the Indicated Performance Characteristics identified Specificity
(Sensitivity and Specificity) Control population
Odds ratio Actual 17,319 84.1%
(Ware definition) Normal distribution 13,880 67.4%
Improvement ⫹3439 13.7%
Double No. correctly
Sensitivity Specificity Normal exponential Cauchy identified Sensitivity
80% 80% 75 25 11 Diabetic population
80% 90% 231 25 7 Actual 7532 70.5%
90% 80% 231 25 5 Normal distribution 7198 67.4%
90% 90% 713 25 4 Improvement ⫹334 3.1%
The odds ratios were calculated as described in the METHODS section using Using 2.63 as the diagnostic cutoff, which corresponds to an odds ratio of
the Ware (2) definition. Of note, regardless of the odds ratio definition 24.3, the number and percentage of individuals who were correctly classified
used, the data distribution has a large impact on the required odds ratio for among the 20,590 control (nondiabetic) and 10,677 diabetic individuals are
certain performance characteristics: the odds ratio required for normally shown. This analysis was repeated using data with a normal distribution that
distributed data may be 1 or 2 magnitudes higher than that required for data had the same number of individuals and odds ratios. The number of
with the double-exponential or Cauchy distribution. individuals who were correctly classified using the actual data was compared
with that in the normally distributed data (improvement).

Table 2. Comparison of Odds Ratio Values

Required for Laboratory Data with a Normal,
Double-Exponential, or Cauchy Distribution to
Achieve the Indicated Performance Characteristics
(Sensitivity and Specificity)
Odds ratio
(conventional definition)
Double
Sensitivity Specificity Normal exponential Cauchy
80% 80% 433 38 8 of performance could be achieved, which enabled the
80% 90% 2415 100 11
90% 80% 2415 100 15
90% 90% 14,910 225 18
The odds ratios were calculated as described in the METHODS section using
the conventional definition. Of note, regardless of the odds ratio definition
used, the data distribution has a large impact on the required odds ratio for
certain performance characteristics: the odds ratio required for normally
distributed data may be 1 or 2 magnitudes higher than that required for data
with the double-exponential or Cauchy distribution.
respectively. The same trend was noted when the conven-
tional OR definition was applied. More dramatically, an OR
of 14,910 is required to achieve 90% sensitivity and 90%
specificity for data with a normal distribution, and an OR of
225 or 6 is required for data that has a double-exponential
or Cauchy distribution, respectively (Table 2).
At the sensitivity and specificity values required for
clinical utility, the minimum required OR values for data
with a Cauchy distribution are substantially smaller than
those for data with a normal distribution. The same results
were consistently obtained when the shape of the data
distribution in the case and control groups were different
(Supplementary Table 1; see Supplementary Methods,
http://links.lww.com/AA/A172).
We further investigated whether the distribution of
actual laboratory data affects the OR that is required for a
laboratory test to achieve a certain level of diagnostic
performance. Using a cutoff of 2.63 for the A1C lab values,
this test would achieve 84% specificity and 71% sensitivity
for the diagnosis of diabetes in the populations evaluated.
Using the actual A1C data, the calculated OR was 24. If
AIC data were normally distributed, the OR required to
achieve this performance would have been ⬎10-fold
(271) higher.
To evaluate the impact that the data distribution had on
the required OR, we artificially shifted the distribution
curve for the A1C data in the case population to achieve the
desired performance characteristics. This enabled us to
directly compare the calculated ORs from the actual A1C
data (after the hypothetical right shift) with those for data
with a normal distribution. By varying the extent of the
rightward shift in the diabetic population, any desired level
impact of the shape of the distribution curve (especially at
the 2 tails) on the OR (conventional definition) to be
characterized (Table 3).
The required OR for the A1C test to achieve clinically
useful performance characteristics is significantly smaller
than if the data had a normal distribution. This analysis
also indicated that a double-exponential distribution could
reasonably approximate the properties of the A1C data
distribution at the 2 tails. Thus, the nonnormal data distri-
bution at the extremes had a very significant effect on the
performance of a laboratory test; the ORs required to
achieve clinically useful performance characteristics were
significantly less than expected.
Nonnormal Data Distribution Improves
A1C Performance
We next evaluated the effect that the data distribution had
on the performance of A1C laboratory data. To do this, we
determined the number of individuals who would be
correctly classified as control (nondiabetic) or diabetic on
the basis of actual A1C laboratory data, and compared that
with the expected results if the A1C data had a normal
distribution (Table 3). Using the indicated cutoff of 2.63
(which produced an OR of 24.3) for the A1C data, we
correctly identified 3439 more control (nondiabetic) indi-
viduals by this result than if the data had a normal
distribution. In addition, 334 more diabetic individuals
were correctly identified than if the A1C data were nor-
mally distributed (Table 3). In other words, the nonnormal
distribution of the A1C laboratory data resulted in 13.7%
fewer FP and 3.1% fewer FN classifications than if the data
had followed a normal distribution.

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 1031

 SPECIAL ARTICLE

Figure 2. Receiver operating characteristic (ROC) curves for (A) A1C, (B) CD19, (C) protein S actual, and (D) shifted protein S laboratory data.
The ROC curve for the data distribution is shown in red, and the ROC curve for corresponding data with a normal distribution with the same
odds ratio is shown in green. A, The areas under the ROC curve for the A1C and for the normal distribution data are 0.84 and 0.74, respectively.
The specificity and sensitivity for the 2 points (triangle) are shown in Table 2. B, For CD-19 data, the actual odds ratio is 12.6, and the area
under the curve (AUC) is 0.77. The corresponding AUC for normally distributed data with the same odds ratio is only 0.69. C, For protein S data,
the actual odds ratio is 2.5, and the AUC is 0.6. The corresponding AUC for normally distributed data with the same odds ratio is 0.57. D, For
protein S data, the data for the case group were artificially shifted to the right to achieve the performance of 80% specificity and 80% sensitivity.
The odds ratio for the transformed data is 38, and the AUC is 0.86. The corresponding AUC for normally distributed data with the same odds
ratio is 0.77.

Generation of a receiver operating characteristics (ROC)
curve is a useful procedure for assessing the overall per-
formance of a test. By varying the cutoff for A1C data, we
obtain different pairs of specificity and sensitivity. Instead
of focusing on a specific pair, we can plot all such pairs in
a scatter plot. The resulting curve is an ROC curve, showing
the change of sensitivity with respect to the change in
specificity (Fig. 2A). An ROC curve close to the ideal point
of 100% specificity and 100% sensitivity indicates nice
performance, and a curve close to the diagonal line from 0%
specificity and 100% sensitivity to 100% specificity and 0%
sensitivity indicates a poor performance. The total area
under the ROC curve is also an indicator of the perfor-
mance, with a value of 1 indicating perfect classification
power and a value of 0.5 demonstrating that the variable
used for classification is irrelevant to the actual outcome.

1032 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

An Optimistic Prognosis for Clinical Utility of Lab Test

Table 4. Odds Ratios Required to Achieve the Table 5. Results of a Simulation Examining the
Indicated Performance Characteristics (Sensitivity Impact of Misclassified Samples (% Mislabeled or
and Specificity) Intercenter Differences in Data Distribution
Odds ratio Parameters [Intercenter ⌬] on Biomarker
(conventional definition) Sensitivity [at 80% Specificity])
Sensitivity Specificity A1C CD-19 Protein S Sensitivity
80% 80% 50 44 38 Actual Normal
80% 90% 165 139 87 % Mislabeled
90% 80% 165 71 61 0 74.5% 74.5%
90% 90% 332 235 129 1% 73.7%⫾0.1% 73.6%⫾0.5%
The odds ratios were calculated using the conventional definition of the odds 3% 72.0%⫾0.2% 71.9%⫾0.5%
ratio and under the assumption that the data distribution had the same shape 5% 70.4%⫾0.2% 70.1%⫾0.5%
as that of A1C, CD-19⫹ cell count, or protein S activity. These distributions 7.5% 68.4%⫾0.3% 67.9%⫾0.5%
were shifted in order to obtain the desired performance characteristics, as is 10% 66.3%⫾0.3% 65.6%⫾0.5%
described in the METHODS section. Intercenter ⌬
0 74.5% 74.5%
The improved performance of the actual A1C data in 10% 74.1%⫾0.2% 74.3%⫾0.2%
relation to that of the corresponding normal distribution 30% 71.9%⫾1.1% 72.6%⫾0.9%
50% 68.8%⫾2.1% 69.7%⫾2.0%
is demonstrated by comparison of their ROC curves (Fig. 75% 64.3%⫾3.3% 64.8%⫾3.6%
2A); the area under the ROC curve for the A1C data 100% 60.4%⫾4.0% 59.9%⫾4.7%
(0.84) is ⬎10% more than that of the normal distribution
The results are shown for simulations using the actual A1C data or when it
(0.74). was adjusted to have a normal distribution.
The performances of 2 other laboratory tests were simi-
larly evaluated. The number of CD19⫹ cells is a potential
biomarker for the diagnosis of lymphoma, and protein S distribution had on laboratory test performance when the
activity is a potential biomarker for coagulopathy. The case and control populations were better separated by
histograms and the quantile– quantile plots indicate that the test. In this case, there was a 10% gain in the AUC in the
the data for both of these biomarkers have a nonnormal shifted data in relation to that of the normally distributed
distribution (Supplementary Figs. 3 [see Supplementary data (Fig. 2D). These results further demonstrate that the
Digital Content 4, http://links.lww.com/AA/A175] data distribution has a large impact on the performance of
and 4 [see Supplementary Digital Content 5, laboratory test data.
http://links.lww.com/AA/A176]; see Supplementary Methods
for figure legends, http://links.lww.com/AA/A172). The Additional Factors Affecting the Laboratory
ORs required for these 2 tests to achieve a useful level of Data Performance
diagnostic performance were also significantly smaller than In addition to the data distribution, other factors can also
if they had a normal distribution (Table 4). The required significantly impact the performance of laboratory data.
ORs were comparable in magnitude to those for data For example, lactic dehydrogenase (LDH) is a prognostic
following a double-exponential distribution, and were biomarker for survival in adult leukemia–lymphoma
much larger than those for data following a Cauchy distri- cases.14 However, the response variable (survival at 5
bution. Therefore, the 2 tails of the distribution curves for years) could have an erroneous value in a small percentage
the data in the control and disease groups play an impor- of patients if death were due to a nonleukemia-associated
tant role in determining the relationship between biomar- cause (e.g., an automobile accident) or because of a failure
ker performance and OR, and this data distribution was in mortality reporting. In the OR example presented in the
better approximated by a double-exponential distribution. introductory section of this article, such misclassification
The ROC curves of CD19 and protein S were also would represent errors due to patients having awareness
compared with the ROC curves of a normally distributed under anesthesia that was not discovered by the inves-
case and control population with the same OR. For the tigators, or patients claiming to have had awareness but
CD19 data, there was a ⬎10% gain in the area under the the events recalled did not take place during a general
curve (AUC) for the actual CD19 data in relation to that of anesthetic.
the data with a normal distribution and the same OR (Fig. In a simulation, we evaluated the impact of such mis-
2B). The performance of the actual data was superior to that classification of the response variable on laboratory data
of the normally distributed data over most of the range. The performance. At 80% specificity, a 1% (or 10%) misclas-
protein S data did not have good diagnostic utility because sification incidence decreased the average sensitivity
the AUC of the ROC curve was close to 0.5. However, the from 74.5% to 73.7% (or 66.3%) (Table 5). The same
ROC curve of the actual protein S data had a gain in AUC decrease occurred when the underlying data distribution
in relation to that of the normally distributed data (Fig. 2C). was assumed to be normal. This simulation indicates that
The distributions of the protein S data in the case and laboratory data performance decreases as the measure-
control populations were close to each other, which made ment error in the binary dependent variable increases,
the impact of data distribution hard to compare. Therefore, but the decrease is not very large if the measurement
the protein S data in the case group was artificially shifted error is small.
to the right to create a performance of 80% specificity and In many perioperative medicine studies, data are often
80% sensitivity, to investigate the effect that the data collected from multiple centers to achieve a desired sample

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 1033

 SPECIAL ARTICLE
size. However, laboratories at different centers could pro-
duce data with different statistical distribution parameters,
which introduces a degree of heterogeneity into the pooled
data. Therefore, we evaluated the impact of intercenter
differences on a simulation study examining A1C data
distributions obtained from 10 hypothetical centers. At 80%
specificity, the sensitivity of the A1C data only decreased
from 74.5% (if all data were from a single center) to 74.1%
if the SD in the distribution parameter across centers was
10% of the control A1C data (Table 5); and the shape of the
data distribution curve did not affect the size of the
decrease in the sensitivity. Thus, combining data from
multiple centers had a small effect on laboratory data
performance if the tests achieved a reasonable level of
standardization across the centers. If care is taken to
minimize the variation among different data centers, the
benefit from a larger sample size outweighs any decrease in
laboratory data performance resulting from a multicenter
study.
DISCUSSION
The effect that a nonnormal data distribution has on
laboratory test performance may at first appear to be a topic
that is of interest to statisticians rather than physicians.
However, this finding has significant implications for sci-
entists who are discovering and developing new diagnostic
markers, and subsequently for physicians who will use the
results provided by the next generation of diagnostic tests
to care for their patients. This information is also relevant to
researchers using retrospective AIMS data to develop pre-
dictors of postoperative outcomes on the basis of quantita-
tive data recorded during anesthesia.
First, this analysis demonstrates that a nonnormal dis-
tribution of laboratory data enables laboratory tests with
moderate ORs (6 –30) to provide useful diagnostic tools.
Second, these ORs are within the range observed for
laboratory data identified in etiologic and epidemiologic
studies. This implies that contemporary genomic studies
have the potential to produce clinically useful diagnostic
tools. Third, when evaluated within the larger context
of populations that are affected by common diseases
(prevalence ⬎1% of the population), the improvement in
laboratory test performance due to the nonnormal data
distribution assumes substantial significance. Finally, the
evaluation of the usefulness of a marker (i.e., the indepen-
dent variable) in predicting an outcome (i.e., the dependent
variable) requires a determination of the probability distri-
bution of that marker in the population. It is insufficient to
make assumptions that the marker follows a normal distri-
bution without evaluating this formally, because the failure
to do this may result in a useful test being inappropriately
discarded. This caveat applies equally to quantitative fac-
tors identified retrospectively from data-mining analyses of
AIMS databases to be associated with an undesirable
outcome.
For example, it is estimated that 18.2 million people
(6.3% of the population) in the United States had diabetes
in 2002, and 5.2 million of these were undiagnosed cases.15
Let us imagine that a scientist has just discovered that
measurement of A1C could be a potential diagnostic
marker that could be used for long-term monitoring of
1034 www.anesthesia-analgesia.org
patients with diabetes. The scientist has just received a
large number of serum samples that were obtained from
control and diabetic individuals. After the A1C values were
measured, the test developer must establish a threshold
value that will enable physicians to determine whether an
individual has an abnormal test result. As is shown here,
the nonnormal distribution of A1C laboratory values re-
sulted in a 13.7% increase in specificity and a 3.1% increase
in sensitivity for the diagnosis of diabetes on the basis of
A1C test results. Thus, for every 1 million diabetic indi-
viduals that were evaluated by this newly developed test, if
the analyst used threshold values that were based upon a
double-exponential distribution of A1C laboratory data in
the population, 31,000 more test results would be correctly
classified as abnormal. Similarly, when the test results from
a population of 1 million control individuals were ana-
lyzed, the use of these cutoff values would avoid misclas-
sifying 137,000 individuals as having an abnormal test
result.
However, diagnoses are not based solely upon a labo-
ratory test result; clinical context and judgment will always
be essential for the proper use of any risk stratification tool.
The prevalence of a particular condition and the level of
pretest suspicion within the tested population (pretest
probability) will impact laboratory test performance and
the interpretation of test results. For many genomic tests
that an anesthesiologist might use, different clinical sce-
narios might lead to requirements for different test perfor-
mance characteristics (sensitivity and specificity).
For example, suppose that there was a genomic screen-
ing test that predicted susceptibility to malignant hyper-
thermia (MH). For this test, we would want a sensitivity of
⬎99%, because this condition can be fatal and we do not
want to miss anyone. However, we could accept a specific-
ity of only 50% (for every 2 patients who are resistant to
MH, only 1 will be predicted correctly), because there are
effective alternative methods of providing anesthesia using
non-MH triggering drugs. Alternatively, suppose another
genetic test predicted whether a patient is likely to experi-
ence prolonged sedation when given midazolam. For this
test, we do not need an extremely high sensitivity, but want
a higher level of specificity than for MH. We do not want to
deprive too many patients of the anxiolytic benefits of this
drug, and we can effectively reverse the effect of midazo-
lam with an antagonist (flumazenil), if necessary. However,
knowing whether a patient is at risk for prolonged sedation
would be helpful in assessing a patient who is slow to
awaken after an anesthetic, and it could improve outcome.
If the patient were at risk for excessive sedation, the
practitioner might more quickly reach a decision to admin-
ister flumazenil than if the patient were not at risk. Avoid-
ing the administration of flumazenil in situations in which
slow emergence is not likely due to midazolam is desirable,
because provoking a seizure is a potential complication of
reversal. These 2 different scenarios indicate how very
different test performance characteristics can be required to
address different clinical situations.
The results from this study are especially important for
organizations such as the Anesthesia Quality Institute
ANESTHESIA & ANALGESIA
An Optimistic Prognosis for Clinical Utility of Lab Test
(AQI)b and the Multicenter Perioperative Outcomes group
(MPOG),c both of which have recently initiated efforts to
improve the quality of anesthesia care through the retro-
spective analysis of anesthesia data. The finding that up to
a 10% error in survival classification has a small impact on
test performance is fortunate. Typically, the cause of death
cannot be determined from mortality databases that are
publically available, and there is also incomplete reporting
to these databases. If small reporting errors adversely
affected test performance, misclassification of the cause of
death would call into question the interpretation of such
studies. Also encouraging for AIMS research involving
collaborative databases is that statistical variation in the
distributions of reported parameters among various con-
tributing centers are not likely to have a major impact on
the analysis. However, AQI and MPOG need to consider
the implications of the data distribution of identified inde-
pendent variables and report on these in their published
reports. The data distributions of the independent variables
will also affect power analysis calculations (which usually
are based on assumptions that the data are normally
distributed) for subsequent prospective studies, because
group sizes necessary to demonstrate a specified change in
outcome will be influenced.
The prognosis for laboratory data is also further im-
proved when viewed through a future-oriented lens. First,
although this analysis evaluated commonly available labo-
ratory data, it has significant implications for newly emerging
diagnostic tests identified using contemporary transcriptional,
proteomic, or metabolomic technologies. It is likely that the
data for these newly discovered tests would also have a
similar (nonnormal) distribution, and we also have reason to
be more optimistic about their diagnostic performance. Sec-
ond, this analysis evaluated the performance of a single
laboratory test. However, it is likely that many laboratory tests
will be identified by contemporary genomic analyses.
As has been noted for disease-associated gene expres-
sion signatures,16 it is likely that a combination of genetic
risk factors and laboratory test data will produce better
clinical stratification tools. Fortunately, many statistical
learning methods and modeling tools,17 including nonlin-
ear mixed-effects modeling,18 have been developed to
identify optimal combinations for clinical stratification.
Also, this analysis evaluated test performance within the
context of existing systems for disease classification. Cur-
rently, very large and diverse groups of individuals are
included within a disease grouping that is largely based
upon a collection of symptoms, which often are disparate,
and sometimes incorporating various laboratory variables.
However, genetic risk factors and laboratory data have the
potential to transform our diagnostic and stratification
practices. In the not too distant future, patient groupings
b
http://www.aqihq.org/resources.aspx, last accessed June 2, 2010.
c
http://mpog.med.umich.edu/, last accessed June 2, 2010.
October 2010 • Volume 111 • Number 4
may be redefined using genetic risk factors and laboratory
test measurements. When emerging laboratory test results
and genetic data are incorporated into disease classification
criteria, patients with common underlying predispositions
and pathogenesis will be similarly classified. The sensitiv-
ity and specificity of the laboratory results that are used for
diagnosis and prognosis will then improve.
ACKNOWLEDGMENTS
We thank Dr. Gomathi Krishnan for retrieving the biomarker
data, and Dr. Bob Lewis for helpful discussions.
REFERENCES
1. Guo Y, Shafer S, Weller P, Usuka J, Peltz G. Pharmacogenomics
and drug development. Pharmacogenomics 2005;6:857– 64
2. Ware JH. The limitations of risk factors as prognostic tools.
New Engl J Med 2006;355:2615–7
3. Wald NJ, Hackshaw AK, Frost CD. When can a risk factor be
used as a worthwhile screening test? BMJ 1999;319:1562–5
4. Pepe MS, Janes H, Longton G, Leisenring W, Newcomb P.
Limitations of the odds ratio in gauging the performance of a
diagnostic, prognostic, or screening marker. Am J Epidemiol
2004;159:882–90
5. Sigl JC, Chamoun NG. An introduction to bispectral analysis
for the electroencephalogram. J. Clin Monit 1994;10:392– 404
6. Conover WJ. Practical Nonparametric Statistics. New York:
John Wiley & Sons, 1971
7. Jarque C, Bera A. Efficient tests for normality, homoscedastic-
ity and serial independence of regression residuals. Econ Lett
1980;6:255–9
8. Royston P. An extension of Shapiro and Wilk’s W test for
normalit to large samples. Appl Stat 1982;31:115–24
9. Benjamini Y, Hochberg Y. Controlling the false discovery rate:
a practical and powerful approach to multiple testing. J R Stat
Soc B 1995;57:289 –300
10. Lowe HJ, Ferris TA, Hernandez PM, Weber SC. STRIDE—An
integrated standards-based translational research informatics
platform. AMIA Ann Symp Proc 2009;2009:391–5
11. Larsen ML, Horder M, Mogensen EF. Effect of long-term
monitoring of glycosylated hemoglobin levels in insulin-
dependent diabetes mellitus. N Engl J Med 1990;323:1021–5
12. Norton RM. The double exponential distribution: using calcu-
lus to find a maximum likelihood estimator. Am Stat
1984;38:135– 6
13. Cassela G, Berger R. Statistical Inference. Pacific Grove, CA:
Duxbury Press, 2002
14. Tsukasaki K, Hermine O, Bazarbachi A, Ratner L, Ramos JC,
Harrington W Jr, O’Mahony D, Janik JE, Bittencourt AL, Taylor
GP, Yamaguchi K, Utsunomiya A, Tobinai K, Watanabe T.
Definition, prognostic factors, treatment, and response criteria
of adult T-cell leukemia–lymphoma: a proposal from an inter-
national consensus meeting. J Clin Oncol 2009;27:453–9
15. Centers for Disease Control and Prevention. National Diabetes
Fact Sheet: National Estimates on Diabetes. Atlanta, GA:
Centers for Disease Control and Prevention, 2005
16. van de Vijver MJ, He YD, van’t Veer LJ, Dai H, Hart AA,
Voskuil DW, Schreiber GJ, Peterse JL, Roberts C, Marton MJ,
Parrish M, Atsma D, Witteveen A, Glas A, Delahaye L, van der
Velde T, Bartelink H, Rodenhuis S, Rutgers ET, Friend SH,
Bernards R. A gene-expression signature as a predictor of
survival in breast cancer. N Engl J Med 2002;347:1999 –2009
17. Hastie T, Tibshirani R, Friedman J. The Elements of Statistical
Learning. New York: Springer, 2001
18. Lindstrom MJ, Bates DM. Nonlinear mixed effects models for
repeated measures data. Biometrics 1990;46:673– 87
www.anesthesia-analgesia.org 1035
General Article

A Comparison of Liver Function After Hepatectomy

with Inflow Occlusion Between Sevoflurane and
Propofol Anesthesia
J. C. Song, MD,* Y. M. Sun, MD,* L. Q. Yang, MD,* M. Z. Zhang, MD,† Z. J. Lu, MD,*
and W. F. Yu, MD*

BACKGROUND: In this study, we compared liver function tests after hepatectomy with inflow
occlusion as a function of propofol versus sevoflurane anesthesia.
METHODS: One hundred patients undergoing elective liver resection with inflow occlusion were
randomized into a sevoflurane group or a propofol group. General anesthesia was induced with
3 ␮g/kg fentanyl, 0.2 mg/kg cisatracurium, and target-controlled infusion of propofol, set at a
plasma target concentration of 4 to 6 ␮g/mL, or sevoflurane initially started at 8%. Anesthesia
was maintained with target-controlled infusion of propofol (2– 4 ␮g/mL) or sevoflurane
(1.5%–2.5%). The primary end point was postoperative liver injury assessed by peak values of
liver transaminases.
RESULTS: Transaminase levels peaked between the first and the third postoperative day. Peak
alanine aminotransferase was 504 and 571 U/L in the sevoflurane group and the propofol group,
respectively. Peak aspartate aminotransferase was 435 U/L after sevoflurane and 581 U/L in
the propofol group. There were no significant differences in peak alanine aminotransferase or
peak aspartate aminotransferase between groups. Other liver function tests including bilirubin
and alkaline phosphatase, and peak values of white blood cell counts and creatinine, were also
not different between groups.
CONCLUSIONS: Sevoflurane and propofol anesthetics resulted in similar patterns of liver
function tests after hepatectomy with inflow occlusion. These data suggest that the 2
anesthetics are equivalent in this clinical context. (Anesth Analg 2010;111:1036 –41)

I nflow occlusion by clamping of the portal triad (Pringle

maneuver) is routinely used in many centers1–3 to
prevent blood loss during liver transsection.4,5 How-
ever, the Pringle maneuver induces ischemic injury in the
remnant liver, which is associated with increased morbid-
ity and mortality.6 Diseased livers such as steatotic or
fibrotic livers may be the most vulnerable to temporary
interruption of blood flow.7–9
Although volatile anesthetics and propofol have been
studied and compared in ischemia/reperfusion injuries in
many organ systems, few studies have compared volatile
anesthetics and propofol for liver resection with inflow
occlusion. In this study, we compared sevoflurane- and
propofol-based anesthetics for hepatectomy with inflow
occlusion, using indices of postoperative liver function as
the major end point for comparison.
METHODS
One hundred consecutive ASA physical status I/II/III
patients undergoing elective liver resection with inflow
From the *Department of Anesthesiology, Eastern Hepatobiliary Surgery
Hospital, Second Military Medical University; and †Department of Anes-
thesiology, Shanghai Children’s Medical Center, Shanghai Jiao Tong Uni-
versity School of Medicine, Shanghai, China.
Accepted for publication June 23, 2010.
Supported by a grant from the National Natural Science Foundation of
China (No. 30901394).
Disclosure: The authors report no conflicts of interest.
Address correspondence and reprint requests to W.F. Yu, MD, Department
of Anesthesiology, Eastern Hepatobiliary Surgery Hospital, Second Military
Medical University, Changhai Rd., No. 225, Shanghai, China. Address e-mail
to ywf606@sohu.com.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181effda8
occlusion between July 2009 and December 2009 were
assessed for study eligibility. Exclusion criteria were age
⬍18 years, additional ablation therapies (cryosurgery or
radiofrequency ablation), prior liver resection for donation,
or scheduled resection not requiring inflow occlusion.
Enrolled patients were randomized at the beginning of the
operation into a sevoflurane group (inhaled anesthesia
with sevoflurane) or a propofol group (target-controlled
infusion of propofol). All other anesthetic and surgical
management was the same. The randomization sequence
without any stratification was generated by computer and
sealed with consecutively numbered envelopes providing
concealment of random allocation. The study was ap-
proved by our local institutional research ethics committee.
Written informed patient consent was obtained from all
participants.
All patients received oral midazolam (5.0 mg) and IM
atropine (0.5 mg) as a premedication. Electrocardiogram,
radial arterial blood pressure, arterial oxygen saturation,
and bispectral index were monitored routinely. Epidural
catheters were placed at T7-10 interspaces before anesthe-
sia, and a test dose of lidocaine was injected. A bolus of 7 to
10 mL ropivacaine 0.75% was administered epidurally after
induction of anesthesia. In the propofol group, general
anesthesia was induced with 3 ␮g/kg fentanyl, followed by
a target-controlled infusion of propofol, set at a plasma
target concentration of 4 to 6 ␮g/mL, and 0.2 mg/kg
cisatracurium. In the sevoflurane group, anesthesia was
induced with 3 ␮g/kg fentanyl, sevoflurane initially started
at 8%, and 0.2 mg/kg cisatracurium. After tracheal intuba-
tion, anesthesia was maintained with a target-controlled
infusion of propofol (in the propofol group, as above) or

1036 www.anesthesia-analgesia.org October 2010 • Volume 111 • Number 4

sevoflurane (1.5%–2.5%), fentanyl 1 to 2 ␮g/kg, and cisa-
tracurium 5- to 10-mg boluses according to clinical needs. Table 1. Patient Characteristics and Preoperative
To keep arterial mean blood pressure at a target of 60 mm
Laboratory Values
Hg, we administered 2 to 5 ␮g 䡠 kg⫺1 䡠 min⫺1 dopamine as Sevoflurane group Propofol group
(n ⴝ 50) (n ⴝ 50)
indicated. Depth of anesthesia was determined with the
Age (y) 48.5 (8.9) 51.4 (7.8)
bispectral index with a target range between 35 and 45 Body height (cm) 170.3 (3.7) 167.5 (10.3)
during surgery. Epidural catheters were removed after the Weight (kg) 70.3 (9.8) 68.9 (12.1)
operation. Intravenous analgesia (continuous IV infusion of Gender, male/female 40/10 36/14
1000 mg tramadol and 100 ␮g sufentanil over 48 hours) was Baseline ALT (U/L) 40 (13) 43 (26)
Baseline AST (U/L) 36 (11) 46 (29)
continued in all patients postoperatively. Baseline bilirubin (␮mol/L) 14.6 (5.2) 13.5 (5.9)
Surgical procedures were performed in a standardized Baseline WBC count (⫻103/mL) 5.9 (1.8) 6.1 (2.1)
manner under the supervision of 2 experienced hepatobili- Baseline creatinine (␮mol/L) 74.6 (10.1) 70.3 (12.3)
ary surgeons. After mobilization of the liver, inflow occlu- Baseline ALP (U/L) 92 (41) 103 (57)
MELD scorea 6.85 (1.01) 6.94 (0.82)
sion was achieved by the tourniquet technique around the
Cirrhosis, yes/nob 33/17 27/23
portal triad with a 4-mm Mersilene tape. During resections, Malignant/benign disease 45/5 43/7
a low central venous pressure (0 –5 mm Hg) was main- Hepatocellular carcinomas 36 36
tained. The length of time for continuous inflow occlusion Intrahepatic 9 7
was determined by the surgeons. All patients received the cholangiocarcinoma
Intrahepatic bile duct stone 2 3
same chemotherapy regimen before and after surgery.
Angioleiomyolipoma of liver 3 4
Each patient was followed for the entire hospitalization.
Data are expressed as mean (SD).
The primary end point was postoperative hepatocyte injury
ALT ⫽ alanine aminotransferase; AST ⫽ aspartate aminotransferase; WBC ⫽
defined by peak alanine aminotransferase (ALT) and aspar- white blood cell; ALP ⫽ alkaline phosphatase.
tate aminotransferase (AST) levels over 6 postoperative days. a
MELD, the Model for End-Stage Liver Disease, consists of serum bilirubin
Additional end points were peak values of white blood cells, and creatinine levels, international normalized ratio (INR) for prothrombin
time, and etiology of liver disease. The formula for the MELD score is 3.8 ⫻
bilirubin, alkaline phosphatase, creatinine levels, and length loge(bilirubin 关mg/dL兴) ⫹ 11.2 ⫻ loge(INR) ⫹ 9.6 ⫻ loge(creatinine
of hospital stay. All outcome variables were measured before 关mg/dL兴) ⫹ 6.4 ⫻ (etiology: 0 if cholestatic or alcoholic, 1 otherwise).
b
and 1, 3, and 6 days after surgery. Additionally, cirrhosis Cirrhosis (yes/no) was defined by histologic evaluation.
(yes/no) was defined by histologic evaluation.
Group sample size was calculated based on differences Table 2. Intraoperative Data from Patients
in postoperative peak ALT concentration in a pilot study of Undergoing Hepatectomy with Inflow Occlusion
patients who received propofol anesthesia (685 ⫾ 392 U/L) Sevoflurane group Propofol group
and sevoflurane anesthesia (487 ⫾ 308 U/L). The following (n ⴝ 50) (n ⴝ 50)
formula: n ⫽ 15.7/ES2 ⫹ 1, where ES ⫽ effect size ⫽ Pringle time (min) 21.4 (8.5) 18.4 (6.3)
(difference between groups)/(mean of the SD between Operation time (min) 136.0 (38.4) 124.3 (29.1)
Blood loss (mL) 302 (269) 291 (187)
groups), with ␣ ⫽ 0.05 and power ⫽ 0.8 was used to
Bispectral index 40.6 (5.2) 39.1 (6.1)
determine that the study would be adequately powered Size of excised liver (cm3) 474.1 (450.4) 527.7 (398.9)
with n ⫽ 50 per group.10 Major/minor resection
We compared peak transaminases (primary outcome) Major resection ⱖ3 22 23
between groups using a linear regression model with peak segments
Minor resection ⬍3 28 27
transaminases as the dependent and group allocation as the
segments
independent variable (corresponding to a 2-sample t test).
In addition, we adjusted these comparisons for baseline Data are mean (SD).

transaminases and bilirubin levels, ischemic (Pringle) time,

size of excised liver (cm3), and blood loss in multivariable Table 3. Intraoperative Hemodynamic Data
linear regression analyses (analysis of covariance). The other Mean arterial blood
measurement data were also compared using a 2-sample t pressure (mm Hg) Heart rate (bpm)
test, and count data were compared using ␹2. Hepatic Sevoflurane Propofol Sevoflurane Propofol
function in patients with cirrhosis might be more sensitive group group group group
to inflow occlusion than without cirrhosis. Therefore, we (n ⴝ 50) (n ⴝ 50) (n ⴝ 50) (n ⴝ 50)
conducted a limited number of subgroup analyses to assess T0 94.9 (9.9) 97.1 (13.2) 86.0 (18.7) 83.7 (14.1)
the effect of cirrhosis (yes/no) on postoperative hepatocyte T1 76.7 (10.2) 80.3 (11.4) 74.6 (13.6) 74.2 (9.4)
T2 76.2 (10.9) 79.9 (10.9) 72.6 (9.1) 69.5 (8.6)
injury (ALT/AST) using a 2-sample t test. All analyses were T3 82.7 (11.8) 80.7 (10.9) 74.9 (10.8) 69.7 (11.5)
conducted using SPSS 16.0 (SPSS Inc., Chicago, IL). Results T4 80.0 (9.9) 84.6 (9.7) 70.3 (9.6) 66.2 (8.9)
were expressed as the mean ⫾ SD. A P value ⬍0.05 was
Data are mean (SD).
considered to represent statistical significance. T0 ⫽ baseline; T1 ⫽ 5th min after anesthetic induction; T2 ⫽ 5th min before
Pringle; T3 ⫽ middle moment of Pringle; T4 ⫽ 5th min after Pringle.

RESULTS
Fifty patients were included in each group. Table 1 shows
the patient characteristics and baseline values of the out-
come variables. Table 2 shows a summary of the important
intraoperative data. In both data sets (Tables 1 and 2), there
were no significant differences. Table 3 shows the hemo-
dynamic data during surgery. Epidural anesthesia was not
attempted in 2 patients in each group for a platelet count
⬍80 ⫻ 109/L.

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 1037

Liver Function Tests are Similar After Hepatectomy with Propofol or Sevoflurane

Table 4. Postoperative Laboratory Data and Length of Hospital Stay

95% confidence interval
of the difference
Sevoflurane Propofol (equal variances assumed)
group group
(n ⴝ 50) (n ⴝ 50) Lower Upper
Peak ALT (U/L) 504 (295) 571 (460) ⫺159 293
Peak AST (U/L) 435 (275) 581 (494) ⫺92 383
Peak bilirubin (␮mol/L) 25.2 (9.3) 33.3 (28.7) ⫺4.3 20.6
Peak ALP (U/L) 121 (35) 144 (83) ⫺16 61
Peak WBC (⫻103/mL) 13.1 (2.7) 14.6 (4.6) ⫺0.6 3.7
Peak creatinine (␮mol/L) 70.3 (11.0) 66.8 (11.7) ⫺9.7 2.6
Hospital stay (d) 16.1 (4.8) 14.0 (2.9) ⫺4.6 0.4
Data are mean (SD).
ALT ⫽ alanine aminotransferase; AST ⫽ aspartate aminotransferase; ALP ⫽ alkaline phosphatase; WBC ⫽ white blood cell.

Table 5. The Peak Values of ALT/AST of the Subgroups over 6 Postoperative Days
Sevoflurane group Propofol group
Cirrhosis Noncirrhosis Cirrhosis Noncirrhosis
(n ⴝ 33) (n ⴝ 17) P value (n ⴝ 27) (n ⴝ 23) P value
Peak ALT (U/L) 558 (310) 398 (245) 0.215 675 (554) 449 (309) 0.231
Peak AST (U/L) 482 (289) 342 (232) 0.218 672 (581) 474 (310) 0.326
Data are mean (SD).
ALT ⫽ alanine aminotransferase; AST ⫽ aspartate aminotransferase.

No patient died in this study. Major complications
included sepsis (2 patients in each group), bleeding (2
patients in each group), and biloma (1 patient in the
sevoflurane group). The mean hospital stay was 2 days
shorter in the propofol group (14 vs 16 days) but without
statistical significance. The degree of ischemia and reper-
fusion injury of the liver was assessed by postoperative
peak serum ALT and AST levels. Transaminase levels
peaked between the first and the third postoperative day.
The sevoflurane group had slightly lower peak ALT and
AST levels than the propofol group, but these were not
statistically different (P ⬎ 0.05) (Table 4). Unadjusted and
adjusted results in multivariable linear regression analyses
(analysis of covariance) were almost identical. Unadjusted
results are presented in Table 4. Other liver function tests
such as bilirubin and alkaline phosphatase, as well as peak
white blood cell levels and creatinine, were not different
between groups (P ⬎ 0.05). The results of subgroup analy-
ses comparing cirrhotic and noncirrhotic patients are pre-
sented in Table 5. Although the serum levels of ALT/AST
were higher in cirrhotic patients than in noncirrhotic pa-
tients, there were no significant differences between these
subgroups (P ⬎ 0.05). Figures 1 and 2 show ALT and AST
levels over 6 postoperative days, respectively.
DISCUSSION
We compared the effect of volatile anesthetics with propo-
fol anesthetics on liver function in patients undergoing
liver resection with inflow occlusion. There were no signifi-
cant differences in postoperative liver function as measured
by serial transaminase levels, or in clinical outcomes in the
2 groups.
Numerous strategies have been designed to reduce
ischemia/reperfusion injury after liver resection. Two pro-
tective strategies to prevent ischemic-reperfusion injury
have been clinically accepted: ischemic preconditioning4,9
and intermittent clamping11of the portal triad. Both proce-
dures require a surgical intervention and prolong the overall
time of the surgical procedure. Hence, a pharmacological ap-
proach not requiring additional surgical procedures may be
a more attractive alternative than the established surgical
strategies. In this study, we wanted to address whether the
anesthetic affects postoperative hepatic function in patients
undergoing elective liver resection with inflow occlusion.
Many factors contribute to hepatic injury and outcomes
after liver resection. Among these, patient characteristics
such as the severity of liver disease and surgical complica-
tions are common contributors to poor postoperative out-
come. In this study, the patient characteristics, baseline
values of the outcome variables, and the data of surgery-
related events showed that the 2 patient groups were truly
equivalent (Tables 1 and 2). In addition, surgery-related
factors (Pringle time, size of excised liver, and blood loss)
and baseline transaminases and bilirubin levels were con-
sidered in multivariable linear regression analyses.
The effect of sevoflurane on hepatic function has been
examined in several studies. The study conducted by Ebert
and Arain12 suggested that sevoflurane, but not propofol,
was associated with increased liver injury in patients
undergoing hepatectomy without inflow occlusion, but the
liver function tests in sevoflurane-exposed patients were in
the upper limits of normal. The study by Beck-Schimmer et
al.,13 however, suggested that sevoflurane preconditioning
was protective against ischemia/reperfusion injury during
liver resection. Sevoflurane preconditioning was shown to
prevent hepatic injury, defined by transaminase levels, and
improve clinical outcome. In the volatile preconditioning
group, the expression of inducible nitric oxide synthase
upon reperfusion significantly increased compared with
the baseline value, which points to a possible protective
role of nitric oxide in pharmacological preconditioning. In a
rat hepatic ischemia/reperfusion injury model, clinically

1038 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

Figure 1. Alanine aminotransferase (ALT) levels
over 6 postoperative days. T0 ⫽ baseline; T1 ⫽ first
postoperative day; T2 ⫽ third postoperative day;
T3 ⫽ sixth postoperative day; sevoflur ⫽ sevoflu-
rane; CI ⫽ confidence interval. Each bar represents
the mean ⫾ SD.

Figure 2. Aspartate aminotransferase (AST) levels

over 6 postoperative days. T0 ⫽ baseline; T1 ⫽ first
postoperative day; T2 ⫽ third postoperative day;
T3 ⫽ sixth postoperative day; sevoflur ⫽ sevoflu-
rane; CI ⫽ confidence interval. Each bar represents
the mean ⫾ SD.

relevant concentrations of sevoflurane given before, dur-
ing, and after hepatic ischemia protected the liver against
ischemia/reperfusion injury. Increased hepatic adenosine
triphosphate and energy levels decreased hepatocyte in-
jury, and the hepatic tissue blood flow almost completely
recovered after ischemia/reperfusion in the sevoflurane
group.14 The study by Imai et al.15 suggested that volatile
anesthetics may protect the fasted liver from early,
neutrophil-independent, ischemia/reperfusion injury by
acting during the reperfusion phase. Sevoflurane reduced
hepatic oxygen consumption and attenuated lactate dehydro-
genase release during reperfusion. Sevoflurane precondition-
ing may provide a new and easily applicable therapeutic
option to protect the liver in hepatectomy.

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 1039

Liver Function Tests are Similar After Hepatectomy with Propofol or Sevoflurane
Propofol is an IV sedative-hypnotic drug frequently
used in anesthesia and intensive care that not only inhibits
lipid peroxidation, but also enhances the cellular antioxi-
dant defense system in several tissues including brain,
liver, and heart.16,17 Propofol pretreatment attenuates
ischemia/reperfusion–induced intestinal epithelial apopto-
sis, which might be attributable to its antioxidant property
modulating the ceramide pathway.18 Propofol improves
survival of liver cells in rat hepatocyte suspensions exposed
to oxidant injury by a free radical generator 2,2⬘-azobis
(2-amidinopropane) dihydrochloride (AAPH).19 Propofol
protects hepatic L02 cells from hydrogen peroxide–induced
apoptosis, partly through activating the MEK-ERK path-
way and further suppressing Bad and Bax expression.20
The study by Shimono et al., however, suggested that
propofol did not have a protective effect against
hypoxia/reoxygenation injury in rat liver slices.21 The
effects of pharmacological preconditioning with propo-
fol in patients undergoing liver resection with inflow
occlusion remain to be determined.
In our study, patients with biopsy-proven cirrhosis did
not have worse postoperative liver dysfunction than those
without cirrhosis, likely because the cirrhotic patients had
mild disease, indicated by the Model for End-Stage Liver
Disease score. In addition, the period of ischemia was short;
longer ischemic stress may have unmasked a difference
between cirrhotic and noncirrhotic patients. Previous stud-
ies suggest that patients with early-stage cirrhosis can
tolerate as much as 60 to 75 minutes of inflow occlusion
during hepatectomy without serious postoperative liver
decompensation.22–24
Our study had several limitations. It is not possible to
determine from our results whether sevoflurane and
propofol are protective in the setting of hepatic
ischemia/reperfusion injury. Second, patients were not
systematically assessed for hepatic fibrosis or steatosis,
both risk factors for liver dysfunction after liver ischemia.
In conclusion, our data suggest that sevoflurane or propo-
fol can be used safely in patients undergoing hepatectomy
with inflow occlusion, and that there is no advantage of one
drug over the other.
AUTHOR CONTRIBUTIONS
JCS helped design and conduct the study, analyze the data,
and write the manuscript. This author has seen the original
study data, reviewed the analysis of the data, approved the
final manuscript, and is the author responsible for archiv-
ing the study files. YMS helped design and conduct the
study, analyze the data, and write the manuscript. This
author has seen the original study data, reviewed the
analysis of the data, and approved the final manuscript.
LQY helped conduct the study and analyze the data. This
author has seen the original study data, reviewed the
analysis of the data, and approved the final manuscript.
MZZ helped conduct the study. This author has seen the
original study data and approved the final manuscript. ZJL
helped conduct the study. This author has seen the original
study data and approved the final manuscript. WFY helped
design the study. This author has seen the original study
data, reviewed the analysis of the data, and approved the
final manuscript.
1040 www.anesthesia-analgesia.org
REFERENCES
1. Gozzetti G, Mazziotti A, Grazi GL, Jovine E, Gallucci A,
Gruttadauria S, Frena A, Morganti M, Ercolani G, Masetti M,
Pierangeli F. Liver resection without blood transfusion. Br J
Surg 1995;82:1105–10
2. Kooby DA, Stockman J, Ben-Porat L, Gonen M, Jarnagin WR,
Dematteo RP, Tuorto S, Wuest D, Blumgart LH, Fong Y.
Influence of transfusions on perioperative and long-term out-
come in patients following hepatic resection for colorectal
metastases. Ann Surg 2003;237:860 –70
3. van der Bilt JD, Livestro DP, Borren A, van Hillegersberg R,
Borel Rinkes IH. European survey on the application of
vascular clamping in liver surgery. Dig Surg 2007;24:423–35
4. Clavien PA, Yadav S, Sindram D, Bentley RC. Protective
effects of ischemic preconditioning for liver resection per-
formed under inflow occlusion in humans. Ann Surg
2000;232:155– 62
5. Lesurtel M, Selzner M, Petrowsky H, McCormack L, Clavien
PA. How should transection of the liver be performed? A
prospective randomized study in 100 consecutive patients:
comparing four different transection strategies. Ann Surg
2005;242:814 –23
6. Clavien PA, Petrowsky H, DeOliveira ML, Graf R. Strategies
for safer liver surgery and partial liver transplantation. N Engl
J Med 2007;356:1545–59
7. Wei AC, Tung-Ping Poon R, Fan ST, Wong J. Risk factors for
perioperative morbidity and mortality after extended hepatec-
tomy for hepatocellular carcinoma. Br J Surg 2003;90:33– 41
8. Coelho JC, Claus CM, Machuca TN, Sobottka WH, Gonçalves
CG. Liver resection: 10-year experience from a single institu-
tion. Arq Gastroenterol 2004;41:229 –33
9. Clavien PA, Selzner M, Rüdiger HA, Graf R, Kadry Z, Rousson
V, Jochum W. A prospective randomized study in 100 consecu-
tive patients undergoing major liver resection with versus
without ischemic preconditioning. Ann Surg 2003;238:843–52
10. Lerman J. Study design in clinical research: sample size esti-
mation and power analysis. Can J Anaesth 1996;43:184 –91
11. Petrowsky H, McCormack L, Trujillo M, Selzner M, Jochum W,
Clavien PA. A prospective, randomized, controlled trial com-
paring intermittent portal triad clamping versus ischemic
preconditioning with continuous clamping for major liver
resection. Ann Surg 2006;244:921–30
12. Ebert TJ, Arain SR. Renal responses to low-flow desflurane,
sevoflurane, and propofol in patients. Anesthesiology 2000;
93:1401– 6
13. Beck-Schimmer B, Breitenstein S, Urech S, De Conno E, Wit-
tlinger M, Puhan M, Jochum W, Spahn DR, Graf R, Clavien PA.
A randomized controlled trial on pharmacological precondi-
tioning in liver surgery using a volatile anesthetic. Ann Surg
2008;248:909 –18
14. Bedirli N, Ofluoglu E, Kerem M, Utebey G, Alper M, Yilmazer
D, Bedirli A, Ozlu O, Pasaoglu H. Hepatic energy metabolism
and the differential protective effects of sevoflurane and isoflu-
rane anesthesia in a rat hepatic ischemia-reperfusion injury
model. Anesth Analg 2008;106:830 –7
15. Imai M, Kon S, Inaba H. Effects of halothane, isoflurane and
sevoflurane on ischemia-reperfusion injury in the perfused
liver of fasted rats. Acta Anasthesiol Scand 1996;40:1242– 8
16. De La Cruz JP, Sedeño G, Carmona JA, Sánchez de la Cuesta F.
The in vitro effects of propofol on tissular oxidative stress in
the rat. Anesth Analg 1998;87:1141– 6
17. Kharasch ED, Armstrong AS, Gunn K, Artru A, Cox K, Karol
MD. Clinical sevoflurane metabolism and disposition. II. The
role of cytochrome P450 2E1 in fluoride and hexafluoroisopro-
panol formation. Anesthesiology 1995;82:1379 – 88
18. Liu KX, Chen SQ, Huang WQ, Li YS, Irwin MG, Xia Z.
Propofol pretreatment reduces ceramide production and
attenuates intestinal mucosal apoptosis induced by intesti-
nal ischemia/reperfusion in rats. Anesth Analg 2008;107:
1884 –91
19. Navapurkar VU, Skepper JN, Jones JG, Menon DK. Propofol
preserves the viability of isolated rat hepatocyte suspensions
under an oxidant stress. Anesth Analg 1998;87:1152–7
ANESTHESIA & ANALGESIA
20. Wang H, Xue Z, Wang Q, Feng X, Shen Z. Propofol protects 23. Kim YI, Nakashima K, Tada I, Kawano K, Kobayashi M.
hepatic L02 cells from hydrogen peroxide-induced apoptosis Prolonged normothermic ischaemia of human cirrhotic liver
via activation of extracellular signal-regulated kinases path- during hepatectomy: a preliminary report. Br J Surg
way. Anesth Analg 2008;107:534 – 40 1993;80:1566 –70
21. Shimono H, Goromaru T, Kadota Y, Tsurumaru T, Kanmura Y. 24. Jeppsson B. Liver resection: prolonged inflow occlusion in
Propofol displays no protective effect against hypoxia/ human cirrhotic livers. HPB Surg 1996;10:123–5
reoxygenation injury in rat liver slices. Anesth Analg
2003;97:442– 8
22. Kim YI, Kobayashi M, Aramaki M, Nakashima K, Mitarai Y,
Yoshida T. “Early-stage” cirrhotic liver can withstand 75
minutes of inflow occlusion during resection. Hepatogastroen-
terology 1994;41:355– 8

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 1041

Pain Medicine
Section Editor: Spencer S. Liu

The Costs and Benefits of Extending the Role of the

Acute Pain Service on Clinical Outcomes After Major
Elective Surgery
Anna Lee, PhD, Simon K. C. Chan, MBBS, Phoon Ping Chen, MBBS, Tony Gin, MD,
Angel S. C. Lau, MPhil, and Chun Hung Chiu, MPhil

BACKGROUND: Acute pain services have received widespread acceptance and formal support
from institutions and organizations, but available evidence on their costs and benefits is scarce.
Although there is good agreement on the provision of acute pain services after many major
surgical procedures, there are other procedures for which the benefits are unclear. Data are
required to justify any expansion of acute pain services. In this randomized, controlled clinical
trial we compared the costs and effects of acute pain service care on clinical outcomes with
conventional pain management on the ward. Patients included in the trial were considered by
their anesthesiologist to have either arm be suitable for the procedure.
METHODS: Four hundred twenty-three patients undergoing major elective surgery were random-
ized either to an anesthesiologist-led, nurse-based acute pain service group with patient-
controlled analgesia or to a control group with IM or IV boluses of opioid analgesia. Both groups
were treated with medications to treat opioid-related adverse effects and received the usual care
from health professionals assigned to the ward. The main outcome measures were quality of
recovery scores, pain intensity measures, global measure of treatment effectiveness, and overall
pain treatment cost. Cost-effectiveness acceptability curves were drawn to detect a difference in
the joint cost-effect relationship between groups.
RESULTS: There was no difference in quality of recovery score on postoperative day 1 between
treatment and control groups (mean difference, 0; 95% confidence interval [CI], ⫺0.7 to 0.7; P ⫽
0.94) or in the rate of improvement in quality of recovery score (mean difference, ⫺0.1; 95% CI,
⫺0.4 to 0.1; P ⫽ 0.34). The proportion of patients with 1 or more days of highly effective pain
management was higher in the acute pain service group than in the control group (86% vs. 75%;
P ⬍ 0.01). Costs were higher in the acute pain service group (mean difference, US$46; 95% CI,
$44 to $48 per patient; P ⬍ 0.001). A cost-effectiveness acceptability curve showed that the
acute pain service was more cost effective than was control for providing highly effective pain
management if the decision maker was willing to pay more than US$546 per patient per 1 day
with highly effective treatment.
CONCLUSION: In extending the role of the acute pain service to a specific group of major surgical
procedures, the acute pain service was likely to be cost effective. (Anesth Analg 2010;111:
1042–50)

A pproximately 234 million major surgical procedures

are performed every year worldwide, and most of
these will require some form of pain management.1
Although patient-controlled analgesia (PCA) and epidural
analgesia provide better postoperative analgesia than does
intermittent IM analgesia,2 these techniques require special
care and monitoring from acute pain service (APS) teams. In
Hong Kong, it is routine practice for APS,3 rather than ward
nurses, to oversee PCA because of a lack of clinical nurse
specialists.
From the Department of Anaesthesia and Intensive Care, The Chinese
University of Hong Kong, Prince of Wales Hospital, Shatin, NT, Hong Kong.
Accepted for publication May 28, 2010.
The work described in this paper was fully supported by a grant from the
Central Policy Unit of the Government of HKSAR and the Research Grants
Council of the HKSAR, China (project reference: CUHK4004-PPR20051).
Address correspondence to Anna Lee, PhD, Department of Anaesthesia and
Intensive Care, The Chinese University of Hong Kong, Prince of Wales
Hospital, Shatin, NT, Hong Kong. Address e-mail to annalee@cuhk.edu.hk.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181ed1317
In some countries, APS is considered to be a standard of
care,4 – 6 but little is known about its economic benefits. Our
systematic review of 10 economic evaluations of APS
programs (involving 14,774 patients) also showed that
there was insufficient evidence to draw conclusions about
the cost effectiveness and cost– benefit ratio of APS.7 The
overall quality of published economic evaluations was poor
because most studies were limited to partial economic
analyses.7
A limitation of the latest guidelines, “Acute Pain Man-
agement: Scientific Evidence,”8 is that they generalized the
evidence and did not present data on a specific procedure.9
There is usually good agreement on the provision of APS
after many major surgical procedures (e.g., with upper
abdominal laparotomy). One problem with assessing the
costs and benefits of an established APS program is that it
would be unreasonable to revert to standard postoperative
ward care for these patients, making a randomized trial of
APS or no APS difficult. However, there are some surgical
procedures for which the benefits of APS are unclear.4 This

1042 www.anesthesia-analgesia.org October 2010 • Volume 111 • Number 4

may be because of new surgical techniques that may
decrease postoperative pain (laparoscopic assisted proce-
dures), or patients expected to have significant postopera-
tive pain but had not been offered APS in the past (e.g.,
cardiac surgery). Although all anesthesiologists would con-
sider using APS in these patients, mixed views were held
by staff on the desirability and necessity of providing
APS for them. Thus, data are required to justify any
expansion of acute pain services to these patients. Be-
cause they had not traditionally been receiving APS, it
was possible to randomize these patients to APS or
standard ward care.
Therefore, we performed prospective cost-effectiveness
analyses alongside a randomized controlled trial of APS
care versus conventional pain management on the ward in
patients undergoing major elective surgery, for whom the
attending anesthesiologist was uncertain about the benefits
of APS care. Our objective was to compare the costs and
effects of APS care on clinical outcomes with conventional
pain management on the ward in patients undergoing
major elective surgery from the perspective of the Hospital
Authority (a government body funding public health ser-
vices in Hong Kong).
METHODS
The study was conducted at the Prince of Wales Hospital in
Hong Kong, a large university teaching hospital. The study
was approved by the local Clinical Research Ethics Commit-
tee. The study was registered in the Centre for Clinical Trials
Clinical Registry of The Chinese University of Hong Kong
(trial no. CUHK_CCT00105) on February 28, 2006, available at
http://www.cct.cuhk.edu.hk/registry/publictriallist.aspx. After
giving written informed consent, adult patients were en-
rolled from April 8, 2006, to February 12, 2009. The APS is
staffed by 2.2 anesthesiologists and 1 nurse full-time
equivalent to provide the service continuously 24 h/day,
serving an average of 17 patients per day (unpublished
data for 2009).
Patients
Groups of patients were identified that could possibly
benefit from APS, but for whom APS care was not often
used in the past. Inclusion criteria were adults, ages 18
years or older and undergoing major surgery (such as
laparoscopic hemicolectomy, open reduction and internal
fixation, total abdominal hysterectomy) or complex major
surgery (e.g., coronary artery bypass graft, discectomy) as
is defined by the Hong Kong Government Gazette.10 Pa-
tients recruited to this study were only those identified by
the attending anesthesiologist as being candidates who
may or may not benefit from APS, in comparison with
conventional ward pain service (CWPS). All patients had
general anesthesia with no additional regional anesthesia.
Patients were excluded if they were younger than 18 years
of age, undergoing emergency or obstetric surgery, had a
history of cognitive impairment or preoperative opioid use,
or were unable to give consent.
The principal investigator (Anna Lee) generated the
random allocation sequence using a computer and was not
involved in the data collection process. Two investigators
(Angel S. C. Lau, and Chun Hung Chiu) were responsible
October 2010 • Volume 111 • Number 4
for randomizing eligible patients within 24 hours of meet-
ing the study criteria to either APS or CWPS care, by using
a sealed opaque envelope containing a computer-generated
random treatment allocation.
Treatment Procedures
Patients randomized to the APS group received IV mor-
phine PCA with or without supplementary oral analgesics,
and medications to treat opioid-related adverse effects.
They were seen by an APS nurse or an anesthesiologist or
both once daily (normal practice). The APS team was
informed if any of the following occurred: inadequate pain
control (persistent pain score ⱖ3 of 10), oxygen desatura-
tion (Spo2 ⬍90%), bradypnoea (respiratory rate ⬍10/min),
hypotension (systolic blood pressure ⬍90 mm Hg), uncon-
trolled nausea and vomiting with parenteral antiemetic,
severe pruritus if uncontrolled with chlorpheniramine 5 mg
q8h prn IM/IV, or difficulty awakening the patient. Al-
though APS did offer acute pain techniques such as periph-
eral nerve blocks and epidural PCA, these techniques were
not used in this study. Patients randomized to the CWPS
group received opioid (pethidine, morphine, tramadol), non-
opioid (diclofenac, paracetamol/phenyltoloxamine, paraceta-
mol) analgesics by IM, IV boluses and/or oral routes, and
medications to treat opioid-related adverse effects prescribed
by surgeons. All patients received the usual care from sur-
geons and nursing professionals assigned to the surgical ward
(and intensive care unit for postoperative care if undergoing
coronary artery bypass graft). The patient, attending health
professionals, and research staff were not blinded to the
treatment assignment.
Data Collection
All surgical patients were admitted as inpatients on the day
before surgery. Patients were interviewed by an investiga-
tor (Angel S. C. Lau or Chun Hung Chiu) on 3 consecutive
days after their surgery with a standardized questionnaire.
We collected data on patient’s demographics, ASA Physical
Status, and length of hospital stay. We measured pain
intensity (worse pain, average pain in the last 24 hours, and
current pain), pain at rest, pain during movement and pain
interference with daily activities (walking ability, mood,
sleep, relations with others, and ability to concentrate), and
ratings on a numeric rating scale (NRS) using the modified
Brief Pain Inventory questionnaire11 on days 1 to 3 after
surgery. A global measure of treatment effectiveness was
measured by asking patients, “How effective do you think
the treatment for pain was?” using a 5-point Likert scale
(0 ⫽ poor, 1 ⫽ fair, 2 ⫽ good, 3 ⫽ very good, 4 ⫽ excellent).12
The 9-item Quality of Recovery (QOR) score13 was col-
lected to measure the patient’s health-related quality of life
after anesthesia and surgery on a daily basis with a score
from 0 to 18. The frequency, severity, and distress of
opioid-related side effects (nausea, vomiting, difficulty in
concentrating, drowsiness or difficulty staying awake, feel-
ing confused, and feeling of general fatigue or weakness)
were totaled daily into an adverse effect score (from 0 to
60). These specific symptoms from the opioid-related
symptom distress scale14 were chosen because they were
thought to have a negative effect on patient’s daily activi-
ties and recovery after surgery.
www.anesthesia-analgesia.org 1043
Cost Effectiveness of Acute Pain Service
All calculated direct costs related to postoperative pain
management were based on the first 3 days after surgery.
From the patient’s drug chart, we recorded the type, dose,
and frequency of analgesic drugs and the drugs used to
treat opioid-related side effects. The medication costs were
estimated from the unit costs of the hospital pharmacy. The
PCA costs, obtained from the hospital administration,
included the cost for infusion pump, IV tubing sets, car-
tridges, catheters, batteries, syringes, needles, swabs, dress-
ings, saline, and morphine. From the patient’s APS record,
the staff cost was calculated using the total nursing and
anesthesiologist time spent for each patient. The nursing
and anesthesiologists’ staff salaries, obtained from the
hospital administration, were based on the midpoint of the
relevant pay scale. The ward nursing cost for APS and
CWPS groups were assumed to be the same as a previous
study at our hospital,15 showing that there was no signifi-
cant difference in total ward nursing time (communication,
documentation, administration of drug, and observations)
for patients receiving PCA or IM opioid injection. The total
postoperative pain management cost was a total of the costs
for analgesic drugs, drugs to treat opioid-related side
effects, PCA, and APS staffing. At the time of reporting the
study results (October 31, 2009), 1 US$ ⫽ HK$7.75. All costs
are reported in U.S. dollars.
Outcome Measures
The primary outcome measure was QOR scores. Pain
intensity (mean pain ratings for worst pain, average pain in
the last 24 hours, and current pain), pain intensity at rest,
pain intensity during movement, global measure of treat-
ment effectiveness, and overall pain treatment cost out-
comes were also measured. These outcome measures were
used in defining the incremental cost-effectiveness ratios
and incremental net benefits, the appropriate measures
of reporting results from a cost-effectiveness analysis.16 Spe-
cifically, the effectiveness of the intervention for cost-
effectiveness analysis purposes was expressed as the number
of pain-free days at rest,17 pain-free days with movement,17
and days with highly effective treatment. A pain-free day
was defined as having a NRS ⱕ3 on a 0 to 10 scale. The
analgesic effectiveness was 3 if the patient had 3 pain-free
days, and 0 if the patient did not experience NRS ⱕ3 at all.
The number of days with highly effective treatment was 3
if the patient rated his or her global measure of effective-
ness as excellent or very good on all 3 days, and 0 if the
patient did not have days with excellent or very good ratings.
Other secondary end points included pain interference
during daily activities, adverse effect score, in-hospital mor-
tality, and length of hospital stay. A pain-free interference day
was defined as having a mean NRS of 0 with pain interference
on daily activities scales ranging from 0 to 10. The
interference-free effect was 3 if the patient had 3 pain-free
interference days, and 0 if the patient experienced pain
that interfered with daily activities on all 3 days. The
length of stay was not included in the cost-effectiveness
analysis because we believed that it was a weak outcome
measure. The delay in patient discharge from hospital
was often not due to pain or analgesia-related side effects
but due to postoperative rehabilitation plans, level of
1044 www.anesthesia-analgesia.org
social support, and surgeon’s postoperative treatment
preferences.
Statistical Analysis
We calculated the sample size using QOR as the primary
outcome because this was a more patient-centered outcome
than were pain ratings and cost. We calculated that a
sample size of 522 would provide 80% power to detect
a small to moderate effect size (0.25) between groups using
a 2-sample t test (EAST 5.2, Cytel Software Corporation,
Cambridge, Massachusetts), allowing for interim analyses.
Two interim analyses were planned after 174 and 348
patients had completed their participation in the study
using the O’Brien–Fleming stopping rules, with ad priori
boundaries of P ⱕ 0.0002 and P ⱕ 0.0121 to reject the null
hypothesis (efficacy boundary, if large treatment differ-
ences appear before the end of the study), and P ⱖ 0.9659
and P ⱖ 0.3444 to accept the null hypothesis (futility
boundary, if there is little chance of finding a significant
difference between groups).
The primary analyses were performed on a modified
intention-to-treat basis (i.e., patients were analyzed accord-
ing to their randomized allocated groups but were ex-
cluded from the analysis if they did not adequately adhere
to the protocol after randomization.). We used the 2-sample
t test, ␹2 test, Fisher exact test, and Mann–Whitney U test to
compare baseline characteristics. The mean difference is
defined as the APS outcome measure minus CWPS outcome
measure. For the QOR, pain intensity and interference out-
comes, global measure of effectiveness, and adverse effect
score, we used multilevel regression models18,19 to assess the
intervention effect on the change between the measure-
ments taken on the first to third day after surgery. An
advantage of using a multilevel regression model over a
repeated-measure analysis of variance is that it can account
for complex covariance structure and accommodate incom-
plete data.19
Given the expected large variability of cost data, the
study was underpowered to test the economic hypothesis
that APS would be more cost effective than would CWPS.
However, in the absence of sufficient power to test the
economic hypotheses, there have been methodological ad-
vances in examining sampling uncertainty for incremental
cost-effectiveness ratios, with emphasis on the likelihood
that the intervention represents good value for the cost
rather than on economic hypothesis testing.20,21
We assumed that APS was cost effective if the extra cost
of an extra gain in effect was less than the decision maker’s
willingness to pay (WTP) for it.22 For example, if the
maximum WTP was set at $200, APS would be cost
effective if the incremental cost-effectiveness ratio (ratio of
the extra cost to extra benefit, i.e., ⌬C/⌬E) was less than
$200. The 95% confidence interval (95% CI) around the
incremental cost-effectiveness ratio was estimated using
the Fieller method. Because there is uncertainty on the WTP
value and the true estimate of the incremental cost-
effectiveness ratio, the cost-effectiveness acceptability curve
was constructed from net benefit (NB) regressions.23 Under
the NB framework, each subject’s NB is computed from the
observed data as WTP ⫻ effecti ⫺ costi, where effecti and
costi are the data for the ith person’s effect and cost,
ANESTHESIA & ANALGESIA
respectively, and WTP is a willingness-to-pay number that
must be specified.24 In its simplest form, the NB regression
involves fitting the following linear regression model:

NBi ⴝ ␤ 0 ⴙ ␤ TX TXi ⴙ ␧ i,

where TXi is the ith person’s treatment indicator (TXi ⫽ 1 for

APS and 0 for CWPS) and ␧i is a stochastic error term.24 The
equation is fitted several times, each time with a different
value of WTP value.24 To generate a cost-effectiveness accept-
ability curve, we used the probability that ␤TX ⬎0 for the y
axis and WTP for the x axis.24 In other words, the cost-
acceptability curves showed the probability that APS was
more cost effective than was CWPS for a range of values that
decision makers might be willing to pay for 1 day gained of
beneficial effect. All analyses were performed with STATA
software version 10.0 (StataCorp, College Station, Texas), and
cost-effectiveness analysis was performed using the macro
“iprogs” available from the University of Pennsylvania
(www.uphs.upenn.edu/dgimhsr/stat-cicer.htm; accessed April
28, 2010). A 2-sided P ⬍ 0.05 was considered statistically
significant. We declared the trial to be positive if more than 1
of the outcomes, but not all, were significant after correcting
for multiplicity using the Holm stepwise approach (corrected
overall critical P value was 0.0167).25

RESULTS
The trial was stopped at the second interim analysis, after 402
patients had completed the study, on the basis of slower-than-
anticipated accrual rate and a prespecified futility stopping
rule. On the basis of 398 patients who had complete day 1
QOR data (195 in the APS group and 203 in the CWPS group),
comparison of the day 1 QOR via a 2-sample t test (P ⫽ 0.43)
crossed the a priori futility boundary for early stopping with
acceptance of the null hypothesis of no difference between
groups. At interim analysis, it was calculated that if the study
had continued to the planned enrollment of 522, the probabil-
ity of demonstrating a difference in day 1 QOR between
treatment groups was ⬍1% under the alternative hypothesis
on the basis of the observed unadjusted day 1 QOR treatment
group differences.
Study Population
Of the 470 surgical patients screened for the study, 422 met
the study criteria and were randomized. Two hundred nine
patients were allocated to the APS group, and 213 to the
CWPS group (Fig. 1). Ten patients from each group with-
drew from the study after randomization. Although the
gender and type of surgery distributions in the withdrawal
group were similar to those of patients who completed the
study (Fisher exact test P ⫽ 1.00 and ␹2 test P ⫽ 0.16,
respectively), the patients who withdrew were older than
those who completed the study (mean [SD], 58 [8] for those
who withdrew and 52 [12] for those who completed;
2-sample t test P ⫽ 0.02). The baseline characteristics at
enrollment were similar for age, gender, type and magni-
tude of surgery, ASA Physical Status, and length of stay in
the intensive care unit between APS and CWPS patients
(Table 1). Patients in the APS group had longer duration of
anesthesia than did those in the CWPS group (mean
difference, 26 minutes; 95% confidence interval [CI], 10 to
Figure 1. Patient flow through clinical trial. APS ⫽ acute pain service;
CWPS ⫽ conventional ward pain service.
42, 2-sample t test P ⬍ 0.01). Although the proportion of
admissions to the intensive care unit after surgery was
similar (␹2 test P ⫽ 0.12) between the 2 groups, the
predicted risk of death from the Acute Physiologic and
Chronic Health Evaluation II score26 was higher in the APS
group (Mann–Whitney U test, P ⬍ 0.01). The median [IQR]
duration of PCA with morphine was 29.5 [18 to 43] hours.
The median [IQR] time that anesthesiologists (including
PCA set-up time in the recovery room) and pain nurses
spent caring for the patients in the APS group were 31 [23
to 39] and 16 [8 to 16] minutes, respectively.
Outcomes
The point estimates on the first day after surgery and the
mean change over 3 days for QOR, pain intensity, interfer-
ence with daily activity from pain, global measure of
treatment effectiveness, and adverse effect outcomes are
shown in Table 2. There were no differences between
groups for outcomes on the first day of surgery: QOR (P ⫽
0.94), pain intensity (P ⫽ 0.31), and pain on movement (P ⫽
0.17). However, APS patients had lower pain scores at rest,
less interference with daily activities because of pain, and
better treatment effectiveness than did CWPS patients on
the first day after surgery (Table 2). The rate of improve-
ment in QOR scores (P ⫽ 0.34), daily rate reductions in pain
intensity (P ⫽ 0.20), and pain during movement (P ⫽ 0.07)
between the 2 groups were similar. The APS group had
significantly smaller daily reductions in pain scores at rest

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 1045

Cost Effectiveness of Acute Pain Service

Table 1. Patient Characteristics at Enrollment

Acute pain service group Conventional ward pain service group
Characteristic (N ⴝ 209) (N ⴝ 213) P value
Age, mean (SD), years 53 (12.3) 52 (11.3) 0.41
Women, no. (%) 106 (50.7) 108 (50.7) 1.00
Type of surgery, no. (%)a
Orthopedic 32 (15.3) 36 (15.9)
Gynecology 47 (22.5) 50 (23.5) 0.64
Cardiothoracic 80 (38.3) 69 (32.4)
General/other 50 (23.9) 58 (27.2)
Magnitude of surgery, no. (%)
Major 123 (58.9) 140 (65.7) 0.13
Ultramajor 86 (41.1) 73 (34.3)
ASA physical status grade, no. (%)b
I 62 (29.8) 78 (40.0)
II 78 (37.5) 81 (38.4) 0.37
III 61 (29.3) 45 (21.3)
IV 7 (3.4) 7 (3.3)
Duration of anesthesia, mean 225 (88.1) 199 (79.6) ⬍0.01
(SD), minutes
Admission to ICU, no. (%) 64 (30.6) 51 (23.9) 0.12
APACHE II score, median (IQR)c 12 (10–16) 11 (8–13) ⬍0.01
Length of ICU stay, median (IQR), 22 (19–23) 21 (18–22) 0.17
hours
SD, standard deviation; ASA, American Society of Anesthesiologists; ICU, intensive care unit; APACHE II, Acute Physiology and Chronic Health Evaluation II; IQR,
interquartile range.
a
Because of rounding, percentages may not all total 100.
b
American Society of Anesthesiologists’ physical status could not be determined for 3 patients (1 in acute pain service group and 2 in conventional ward pain
service group).
c
APACHE II score ranges from 0 to 71, with higher scores indicating higher probability of death.

Table 2. Point Estimate on First Day After Surgery and Mean Change Over the First 3 Days After
Surgery by Group, and Mean Difference Between Groups (95% Confidence Intervals) for Primary
and Secondary Outcomes
Acute pain service Conventional ward pain service Mean difference between groups

Day 1 Mean change over Day 1 Mean change over Day 1 Mean change over
estimate 3 days estimate 3 days estimate 3 days
Primary outcome
QORa 11.2 (10.7 to 11.7) 1.4 (1.2 to 1.6) 11.2 (10.7 to 11.7) 1.5 (1.3 to 1.7) 0 (⫺0.7 to 0.7) ⫺0.1 (⫺0.4 to 0.1)
Secondary outcomes
Pain intensity 5.3 (4.9 to 5.6) ⫺0.7 (⫺0.9 to ⫺0.6) 5.5 (5.2 to 5.9) ⫺0.9 (⫺1.0 to ⫺0.7) ⫺0.3 (⫺0.8 to 0.2) 0.1 (⫺0.1 to 0.3)
Pain at rest 2.3 (1.9 to 2.8) ⫺0.2 (⫺0.4 to ⫺0.1) 3.2 (2.8 to 3.6) ⫺0.6 (⫺0.7 to ⫺0.4) ⫺0.9 (⫺1.4 to ⫺0.3)* 0.3 (0.1 to 0.5)*
Pain on movement 4.5 (4.0 to 4.9) ⫺0.5 (⫺0.7 to ⫺0.3) 4.9 (4.5 to 5.4) ⫺0.7 (⫺0.9 to ⫺0.5) ⫺0.5 (⫺1.1 to 0.2) 0.2 (0 to 0.5)
Interference 3.0 (2.5 to 3.5) ⫺0.4 (⫺0.6 to ⫺0.2) 3.9 (3.4 to 4.4) ⫺0.7 (⫺0.9 to ⫺0.5) ⫺0.9 (⫺1.6 to ⫺0.2)* 0.3 (0.1 to 0.6)*
Global treatment 3.0 (2.8 to 3.2) ⫺0.1 (⫺0.1 to 0) 2.4 (2.2 to 2.6) 0.1 (0.1 to 0.2) 0.6 (0.3 to 0.9)† ⫺0.2 (⫺0.3 to ⫺0.1)†
effectiveness
Adverse effects b
13.9 (12.0 to 15.7) ⫺2.9 (⫺3.6 to ⫺2.2) 16.3 (14.5 to 18.1) ⫺4.0 (⫺4.6 to ⫺3.3) ⫺2.4 (⫺5.0 to 0.1) 1.0 (0.0 to 2.0)‡
QOR, Quality of Recovery Score.
a
Higher QOR scores represent better recovery after anesthesia and surgery; higher global measure of effectiveness scores represent better effectiveness of pain
intervention.
b
Higher adverse effect scores represent worse experience with opioid-related side effects.
* P ⱕ 0.01; † P ⱕ 0.001; ‡ P ⫽ 0.04.

and interference with daily activities than did the CWPS

Table 3. Incidence (%; 95% Confidence Interval) group over the first 3 days after surgery (Table 2).
of Moderate to Severe Pain at Rest and on The incidence of moderate to severe pain (NRS ⬎3 on a 0
Movement After Major Surgery by Groups to 10 scale) at rest and on movement after major surgery in the
Conventional 2 groups is shown in Table 3. The incidences of “poor”
Acute ward P
pain service pain service value treatment effectiveness on the first day after surgery in the
Day 1 APS and CWPS groups were 0.5% (95% CI, 0.1 to 2.9) and
At rest 25.8 (20.1 to 32.4) 34.7 (28.4 to 41.5) 0.06 4.8% (95% CI, 2.5 to 8.8), respectively. The proportion of
On movement 56.1 (49.0 to 63.0) 57.0 (50.1 to 63.7) 0.86 patients with 1 or more days of highly effective pain manage-
Day 2
At rest 12.4 (8.3 to 18.0) 20.8 (15.7 to 27.0) 0.03
ment (i.e., treatment effectiveness rated as very good and
On movement 41.9 (34.9 to 49.2) 45.1 (38.3 to 52.1) 0.53 excellent) was higher in the APS group than in the control
Day 3 group (86% vs. 75%; absolute risk difference 11%; 95% CI, 3%
At rest 19.7 (14.6 to 26.0) 14.9 (10.6 to 20.5) 0.22 to 20%; ␹2 test P ⬍ 0.01; Fig. 2). This is equivalent to a number
On movement 37.2 (30.5 to 44.5) 33.8 (27.6 to 40.7) 0.50
needed to treat of 9 (95% CI, 5 to 33). There were no significant

1046 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

 severity of opioid-related side effects on the first day after
surgery tended to be less in the APS group than in the CWPS
group (Table 2; P ⫽ 0.06). Overall, both groups experienced 1
day of no opioid-related side effects (Table 4).

Figure 2. The number of days experiencing a highly effective
treatment (“very good” and “excellent” ratings) in the acute pain
service and conventional ward pain service groups. There was a
significant difference between the 2 groups (P ⬍ 0.01).
differences in the other pain-free outcomes (Table 4) between
the 2 groups. The mean duration of hospital stay (days) was
similar between the APS and CWPS groups (12 [11] vs. 10 [12],
respectively; 2-sample t test P ⫽ 0.13).
Adverse Events
One patient in the APS group had respiratory depression
due to PCA with morphine (incidence 0.5%, 95% CI, 0.1% to
2.8%) and required IV 0.8 mg naloxone treatment. During
the study, 1 patient in the APS group died after coronary
artery bypass surgery in the intensive care unit because of
uncontrolled bleeding from the surgical site.
The risk of opioid-related side effects at any time during
the follow-up was similar between groups (41% in the APS
group and 40% in the CWPS group; absolute risk difference
1%, 95% CI, ⫺9% to 12%; ␹2 test P ⫽ 0.76). However, the
Costs
As was expected, the costs of analgesia, medications to treat
opioid-related side effects, and APS staff costs were signifi-
cantly higher in the APS group than in the CWPS group
(Table 5). The mean difference in the total cost of pain
treatment was US$46 (95% CI, $44 to $48) per patient (P ⬍
0.001). Because there was no significant extra day gained
for being pain free at rest, pain free during movement, no
opioid-related side effects, and no interference with daily
activity measures in the APS group over the CWPS group,
incremental cost-effectiveness ratios were not estimated.
The incremental cost-effectiveness ratio for costs per 1 day
with highly effective treatment gained was US$151 (95% CI,
$87 to $546) per patient. Decision makers who are willing to
pay less than US$87 per patient per 1 day with highly
effective treatment can be 95% confident that the APS
represents bad value; between US$87 and US$546 per
patient per 1 day with highly effective treatment, the
decision maker cannot be 95% confident that the 2 inter-
ventions differ in value; for those willing to pay more than
US$546 per patient per 1 day with highly effective treat-
ment, they can be 95% confident that the APS represents
good value in comparison with CWPS (Fig. 3).
DISCUSSION
We conducted a cost-effectiveness analysis alongside a
randomized controlled trial of APS versus CWPS. Previous
studies included in systematic reviews7,27,28 examining the
effect of APS on postoperative outcomes were likely to be
biased because the studies were observational in design
(before–after studies, matched comparisons). In this trial of
422 patients, there were no significant differences in the

Table 4. Number of Outcome-Free Days (Median, IQR) During the First 3 Days After Surgery
Acute pain service Conventional ward pain service
group group
Outcome (n ⴝ 199) (n ⴝ 203) P value
Pain freea 1 (0 to 2) 1 (0 to 2) 0.62
Pain free at resta 3 (2 to 3) 3 (2 to 3) 0.63
Pain free on movementa 2 (0 to 3) 2 (1 to 3) 0.69
Pain-free interference on daily activitiesa 0 (0 to 1) 0 (0 to 1) 0.80
Free of opioid-related side effectsb 1 (0 to 1) 1 (0 to 1) 0.81
IQR, interquartile range.
a
An outcome-free day was defined as numeric rating ⱕ3 on a 0 to 10 scale.
b
Defined as adverse effect score ⫽ 0.

Table 5. Mean Pain Treatment Use and Costs (in US$) per Patient
Acute pain service Conventional ward pain service
group group Mean difference
Cost category a
(n ⴝ 199) (n ⴝ 203) (95% CI) P value
Analgesia 18.74 1.21 17.53 (17.00 to 18.05) ⬍0.001
Medications to treat side effects 2.19 0.94 1.25 (0.07 to 2.42) 0.04
Acute pain service staff costs 27.34 0.50 26.84 (25.63 to 28.05) ⬍0.001
Total cost of pain treatment 48.27 2.65 45.62 (43.52 to 47.71) ⬍0.001
a
One patient in the acute pain service group did not receive the intervention because of lack of staff to set up the IV patient-controlled analgesia in the recovery
room. Three patients in the conventional ward pain service group received acute pain service intervention after initial pain treatment was inadequate. Refer to
METHODS section in text for costing methodology.

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 1047

Cost Effectiveness of Acute Pain Service
Figure 3. Cost-effectiveness acceptability curve showing the probability
that acute pain service (APS) is cost effective for a range of decision
makers’ maximum willingness to pay (U.S. dollars) for 1 day with highly
effective treatment gained. The observed mean incremental cost-
effectiveness ratio of US$151 per patient (x axis) corresponds to a 50%
probability of the acute pain service being cost effective (y axis).
QOR score on the first day after surgery or in the rate of
improvement in the QOR score between the 2 groups. It is
possible that the 9-item QOR instrument lacked responsive-
ness and discrimination when compared with the 40-item
QOR instrument.29 Nevertheless, we found lower pain
scores at rest, less interference with daily activities because
of pain, and better global measure of treatment effective-
ness for the APS group on the first day after surgery. These
results suggest that there are some initial benefits associ-
ated with APS over CWPS in the early postoperative
period, but that they disappear on the second and third day
after surgery. Although we did not measure patients’ locus
of control in pain, we believe that PCA afforded greater
perceived control over pain relief30 and that this benefit
extended past the first day after surgery. Not surprisingly,
APS was associated with additional costs, mainly from staff
costs, which were 57% of the total cost of pain treatment.
Although up to one third of patients in this trial expe-
rienced moderate to severe pain at rest on the first day after
surgery, the mean difference in pain at rest scores (⫺0.9,
95% CI, ⫺1.4 to ⫺0.3) between groups was statistically and
clinically significant (28% reduction) if one considers a 20%
reduction to represent a minimum clinically important
difference.31 Half of our patients experienced moderate to
severe pain during movement on the first day after surgery.
This incidence appears to be high in comparison with the
results of a meta-analysis of 33 studies by Dolin et al. (95%
CI, 25% to 40%).2 We found differences between the 2
groups for pain at rest, but not pain during movement on
the first day after surgery. These conflicting findings may
suggest that many patients are unable to distinguish be-
tween pain at rest and pain during movement, because this
may be influenced by coughing, need for physiotherapy,
and dressing changes.2 Therefore, our results for pain
during movement require cautious interpretation.
Pain interference with daily activities—such as walking
ability, mood, sleep, relations with others, and ability to
concentrate—are increasingly being used in conjunction
with pain intensity outcomes. This outcome represents the
physical and mental functions affected by pain during the
recovery process from anesthesia and surgery. We showed
1048 www.anesthesia-analgesia.org
that APS was associated with less pain interference on daily
activities on the first day of surgery, but this effect over
subsequent days was less than that in the CWPS group and
made no difference to the quality of recovery.
During the trial, 1 patient died from surgical complica-
tions. The risk of respiratory depression associated with
PCA in the APS group (0.5%) was similar to that reported
in Werner et al.’s systematic review.27 We found no evidence
to support APS in preventing or reducing the incidence of
opioid-related side effects. However, in comparison with the
CWPS group, there was some weak evidence to suggest
that APS was associated with milder opioid-related side
effects.
We used a global measure of treatment effectiveness
outcome for our cost-effectiveness analysis. The global
measure of treatment effectiveness allows patients to bal-
ance the unpleasantness or inconvenience of the pain
service intervention, the personal meaningfulness of im-
provements in pain and function, and the unpleasantness
and meaning of any opioid-related side effects.8 We
showed that for every 9 patients treated with the APS, 1
would experience 1 or more days of highly effective pain
treatment. A previous study12 showed that the global
measure of treatment effectiveness is valid and can provide
estimates of analgesic efficacy equivalent to pain relief.
When interpreting whether APS is cost effective, it is
important to ask how much decision makers are willing to
pay for APS to be highly effective rather than spend funds
on improving analgesic techniques per se for being pain
free at rest or during movement. Our trial suggests that we
can be certain that APS is cost effective when the WTP for
an extra day gained from a highly effective treatment was
more than US$546 per patient. This estimate is higher than
the value of what ⬎90% of Canadian patients were willing
to pay, which was no more than US$245* for APS to
provide PCA treatment.32 The differences in results may be
due to differences in median household income, patient
population, and the WTP methodology. It is likely that our
anesthesiologist-led nurse-based APS is cost effective be-
cause funding of APS reflects public preferences for acute
pain relief and perceived aversion to major complications
related to inadequate pain relief.33,34
This study has several limitations that need to be
considered. We expanded the coverage of APS care to a
specific group of major surgical procedures for which the
benefits of PCA were uncertain. The IV PCA mode ac-
counted for 92% of the pain techniques given by the APS in
2009 at our hospital. It could be argued that this study
compared IV PCA to IM analgesia, rather than a compari-
son between APS and CWPS care. However, we believe
that our study is the latter because we chose to focus on the
health service infrastructure providing the postoperative
pain management rather than the pain technique per se.
In the design of economic analysis alongside clinical
trials, it is recommended that the study should be more
naturalistic to increase the external validity of the economic
results for which the objective is to determine the value for
*Conversion of CAN$200 in 1997 to US$245 at 2008 value from
http://eppi.ioe.ac.uk/costconversion/default.aspx. Accessed April 26,
2010.
ANESTHESIA & ANALGESIA
money (i.e., lowest cost per unit benefit).20,35 That is,
pragmatic trials evaluate effectiveness in comparison with
standard care in “real-world” patient populations and
practice settings, and although internal validity may be
lower than a standard randomized controlled trial, they are
more suitable for collection of health economic data.20,35
Although PCA is a common device used by APS, many
patients (63%) who are not cared for by an anesthesiologst-
based service also receive PCA.6 Given the same availabil-
ity of oral/IM analgesia and PCA technology, a dedicated
APS will prescribe and use these pain techniques differ-
ently than will non-APS physicians.36 For example, if sur-
geons prescribed a regular parenteral opioid to patients on the
first day after surgery, we would expect pain relief outcomes
to be similar to those of the APS group. However, because
there were significant differences in pain outcomes between
the groups on the first day after surgery, it is likely that
patients were prescribed PRN analgesia in the CWPS group.
Although surgeons want to maintain an active role in post-
operative pain management,37 extending the role of APS is
likely to be more cost effective than the CWPS by surgeons.
The results of this trial are not generalizable to different
organizational structures of APS. Our APS is an
anesthesiologist-led, nurse-based model that is the typical
model found in Hong Kong and Australian hospitals. In
many centers, the anesthesiologist in the APS may only
attend ward rounds once or twice a week, with most care
given by nurses who specialize in pain medicine. This
would likely reduce the staffing costs associated with an
anesthesiologist-led, nurse-based APS program. An eco-
nomic analysis of a nursed-based APS33 concluded that it
was cost effective. However, we believe that there is still
uncertainty about its cost effectiveness given that a cost-
effectiveness acceptability curve was not drawn. We esti-
mate that, if we changed our APS model to a nurse-based
APS, we would reduce the overall pain cost by 25%.
However, we are uncertain about what corresponding
changes in clinical benefits would arise.
The nature of the pain service intervention meant that
patients, health care professionals and the research assistants
could not be blinded. Therefore, we cannot exclude the
possibility of measurement bias and the Hawthorne effect of
the APS visits in this trial. However, we believe that selection
bias and confounding would be minimal in this randomized
controlled trial. Because the pain score measured every 4
hours was not part of routine postoperative care on the ward
at the time of the study, we collected the daily pain scores
from patients; however, this may be prone to recall bias. We
used multiple primary outcomes measures to adequately
capture the complexity of the pain experience and how it may
be modified by pain management interventions consistent
with best practices in pain research.8 Furthermore, we mini-
mized the risk of making a false positive conclusion about the
clinical benefits of APS using methods recommended by the
IMMPACT (Initiative on Methods, Measurement, and Pain
Assessment in Clinical Trials) guidelines.25
Improving postoperative pain management by the use
of APS remains a public health challenge.38 Many APS face
structural, political, cultural, educational, emotional, and
physical/technological challenges despite considerable ef-
forts by APS members.38 Many surgeons are satisfied with
October 2010 • Volume 111 • Number 4
the service provided by the APS but many want to main-
tain an active role in postoperative pain management.37 We
have shown that the anesthesiologist-led, nurse-based APS
is likely to be more cost effective than is the CWPS.
RECUSE NOTE
Tony Gin is section editor of Anesthetic Clinical Pharmacology
for the Journal. This manuscript was handled by Spencer S.
Liu, section editor of Pain Medicine, and Dr. Gin was not
involved in any way with the editorial process or decision.
REFERENCES
1. Weiser TG, Regenbogen SE, Thompson KD, Haynes AB,
Lipsitz SR, Berry WR, Gawande AA. An estimation of the
global volume of surgery: a modelling strategy based on
available data. Lancet 2008;372:139 – 44
2. Dolin SJ, Cashman JN, Bland JM. Effectiveness of acute post-
operative pain management: I. Evidence from published data.
Br J Anaesth 2002;89:409 –23
3. Cheung CW, Ying CL, Lee LH, Tsang SF, Tsui SL, Irwin MG.
An audit of postoperative intravenous patient-controlled anal-
gesia with morphine: evolution over the last decade. Eur J Pain
2009;13:464 –71
4. Hung CT, Lau LL, Chan CK, Chow B, Chui PT, Ho B, Kung
MC, Lui J, Hui T, Ho E, Chan SF, Chen PP. Acute pain services
in Hong Kong: facilities, volume, and quality. Hong Kong Med J
2002;8:196 –201
5. McDonnell A, Nicholl J, Read SM. Acute Pain Teams in
England: current provision and their role in postoperative pain
management. J Clin Nurs 2003;12:387–93
6. Miaskowski C, Crews J, Ready LB, Paul SM, Ginsberg B.
Anesthesia-based pain services improve the quality of postop-
erative pain management. Pain 1999;80:23–9
7. Lee A, Chan S, Chen PP, Gin T, Lau AS. Economic evaluations
of acute pain service programs: a systematic review. Clin J Pain
2007;23:726 –33
8. Australian and New Zealand College of Anaesthetists and
Faculty of Pain Med. Acute Pain Management: Scientific
Evidence. 3nd ed. Melbourne: Australian and New Zealand
College of Anaesthetists, 2010
9. Schug SA, Kehlet H, Bonnet F, Camu F, Fischer B, Joshi G,
Neugebauer EAM, Rawal N, Simanski C. Procedure specific
pain management after surgery—“PROSPECT.” Acute Pain
2007;9:55–7
10. Hong Kong Government. Hong Kong Government Gazette
2003;13/2003(Special Suppl 4, Annex IV): D4668 –704
11. Mendoza TR, Chen C, Brugger A, Hubbard R, Snabes M,
Palmer SN, Zhang Q, Cleeland CS. The utility and validity of
the modified brief pain inventory in a multiple-dose postop-
erative analgesic trial. Clin J Pain 2004;20:357– 62
12. Collins SL, Edwards J, Moore RA, Smith LA, McQuay HJ.
Seeking a simple measure of analgesia for mega-trials: is a
single global assessment good enough? Pain 2001;91:189 –94
13. Myles PS, Hunt JO, Nightingale CE, Fletcher H, Beh T, Tanil D,
Nagy A, Rubinstein A, Ponsford JL. Development and psycho-
metric testing of a quality of recovery score after general
anesthesia and surgery in adults. Anesth Analg 1999;88:83–90
14. Apfelbaum JL, Gan TJ, Zhao S, Hanna DB, Chen C. Reliability
and validity of the perioperative opioid-related symptom
distress scale. Anesth Analg 2004;99:699 –709
15. Chang AM, Ip WY, Cheung TH. Patient-controlled analgesia
versus conventional intramuscular injection: a cost effective-
ness analysis. J Adv Nurs 2004;46:531– 41
16. Hoch JS, Dewa CS. A clinician’s guide to correct cost-
effectiveness analysis: think incremental not average. Can
J Psychiatry 2008;53:267–74
17. Bartha E, Carlsson P, Kalman S. Evaluation of costs and
effects of epidural analgesia and patient-controlled intrave-
nous analgesia after major abdominal surgery. Br J Anaesth
2006;96:111–7
18. Singer JD, Willet JB. Applied Longitudinal Data Analysis. New
York: Oxford University Press, 2003
www.anesthesia-analgesia.org 1049
Cost Effectiveness of Acute Pain Service
19. Fitzmaurice GM, Ravichandran C. A primer in longitudinal
data analysis. Circulation 2008;118:2005–10
20. Marshall DA, Hux M. Design and analysis issues for economic
analysis alongside clinical trials. Med Care 2009;47:S14 –20
21. Glick HA, Doshi JA, Sonnad SS, Polsky D. Economic evalua-
tion in clinical trials. New York: Oxford University Press, 2007
22. Hoch JS, Rockx MA, Krahn AD. Using the net benefit regres-
sion framework to construct cost-effectiveness acceptability
curves: an example using data from a trial of external loop
recorders versus Holter monitoring for ambulatory monitoring
of “community acquired” syncope. BMC Health Serv Res
2006;6:68
23. Hoch JS, Briggs AH, Willan AR. Something old, something
new, something borrowed, something blue: a framework for
the marriage of health econometrics and cost-effectiveness
analysis. Health Econ 2002;11:415–30
24. Hoch JS. Improving efficiency and value in palliative care with
net benefit regression: an introduction to a simple method for
cost-effectiveness analysis with person-level data. J Pain Symp-
tom Manage 2009;38:54 – 61
25. Turk DC, Dworkin RH, McDermott MP, Bellamy N, Burke LB,
Chandler JM, Cleeland CS, Cowan P, Dimitrova R, Farrar JT,
Hertz S, Heyse JF, Iyengar S, Jadad AR, Jay GW, Jermano JA,
Katz NP, Manning DC, Martin S, Max MB, McGrath P,
McQuay HJ, Quessy S, Rappaport BA, Revicki DA, Rothman
M, Stauffer JW, Svensson O, White RE, Witter J. Analyzing
multiple endpoints in clinical trials of pain treatments:
IMMPACT recommendations. Pain 2008;139:485–93
26. Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE
II: a severity of disease classification system. Crit Care Med
1985;13:818 –29
27. Werner MU, Soholm L, Rotboll-Nielsen P, Kehlet H. Does an
acute pain service improve postoperative outcome? Anesth
Analg 2002;95:1361–72
1050 www.anesthesia-analgesia.org
28. McDonnell A, Nicholl J, Read SM. Acute pain teams and the
management of postoperative pain: a systematic review and
meta-analysis. J Adv Nurs 2003;41:261–73
29. Myles PS, Weitkamp B, Jones K, Melick J, Hensen S. Validity
and reliability of a postoperative quality of recovery score: the
QoR-40. Br J Anaesth 2000;84:11–5
30. Chumbley GM, Hall GM, Salmon P. Patient-controlled analge-
sia: an assessment by 200 patients. Anaesthesia 1998;53:216 –21
31. Cepeda MS, Africano JM, Polo R, Alcala R, Carr DB. What
decline in pain intensity is meaningful to patients with acute
pain? Pain 2003;105:151–7
32. Badner NH, Komar WE, Craen RA. Patient attitudes regarding
pca and associated costs. Can J Anaesth 1997;44:255– 8
33. Stadler M, Schlander M, Braeckman M, Nguyen T, Boogaerts
JG. A cost-utility and cost-effectiveness analysis of an acute
pain service. J Clin Anesth 2004;16:159 – 67
34. Macario A, Fleisher LA. Is there value in obtaining a patient’s
willingness to pay for a particular anesthetic intervention?
Anesthesiology 2006;104:906 –9
35. Ramsey S, Willke R, Briggs A, Brown R, Buxton M, Chawla A,
Cook J, Glick H, Liljas B, Petitti D, Reed S. Good research
practices for cost-effectiveness analysis alongside clinical trials:
the ISPOR RCT-CEA Task Force report. Value Health
2005;8:521–33
36. Stacey BR, Rudy TE, Nelhaus D. Management of patient-
controlled analgesia: a comparison of primary surgeons and a
dedicated pain service. Anesth Analg 1997;85:130 – 4
37. Chan SK, Chui PT, Lee A, Lai PB, Li TY, Gin T. Surgeons’
attitudes and perception of an acute pain service. Hong Kong
Med J 2008;14:342–7
38. Powell AE, Davies HT, Bannister J, Macrae WA. Challenge of
improving postoperative pain management: case studies of
three acute pain services in the UK National Health Service.
Br J Anaesth 2009;102:824 –31
ANESTHESIA & ANALGESIA
 Pain Mechanisms
Section Editor: Tony L. Yaksh/Quinn H. Hogan

The Effect of Ketamine Anesthesia on the Immune

Function of Mice with Postoperative Septicemia
Tetsuya Takahashi, MD, PhD,* Manabu Kinoshita, MD, PhD,† Satoshi Shono, MD,† Yoshiko Habu, PhD,†
Takahiro Ogura, MD,* Shuhji Seki, MD, PhD,† and Tomiei Kazama, MD, PhD*

BACKGROUND: It is unknown how ketamine anesthesia immunologically affects the outcome of

patients with postoperative septicemia. We investigated the effects of ketamine anesthesia on
mice with an Escherichia coli or lipopolysaccharide (LPS) challenge after laparotomy, focusing on
phagocytosis by liver macrophages (Kupffer cells) and cytokine production.
METHODS: C57BL/6 mice received ketamine or sevoflurane anesthesia during laparotomy, which
was followed by an E. coli or LPS challenge; thereafter, mouse survival rates and cytokine secretions
were examined. The effects of a ␤-adrenoceptor antagonist, nadolol, on ketamine anesthesia were
also assessed to clarify the mechanisms of ketamine-induced immunosuppressive effects.
RESULTS: Ketamine anesthesia increased the mouse survival rate after LPS challenge after
laparotomy compared with sevoflurane anesthesia, whereas such an effect of ketamine was not
observed after E. coli challenge. Ketamine suppressed tumor necrosis factor (TNF) and interferon
(IFN)-␥ secretion after LPS and E. coli challenge. When bacterial growth was inhibited using an
antibiotic, ketamine anesthesia effectively improved mouse survival after E. coli challenge
compared with sevoflurane anesthesia. Neutralization of TNF also improved survival and
decreased IFN-␥ secretion after bacterial challenge in antibiotic-treated mice with sevoflurane
anesthesia, suggesting that ketamine’s suppression of TNF may improve survival. Ketamine also
suppressed in vivo phagocytosis of microspheres by Kupffer cells in LPS-challenged mice.
Concomitant use of nadolol with an anesthetic dose of ketamine did not restore TNF suppression
in LPS-challenged mice, suggesting a mechanism independent of the ␤-adrenergic pathway.
However, it restored TNF secretion under low-dose ketamine (10% anesthetic dose). In contrast,
nadolol restored the decrease in phagocytosis by Kupffer cells, which was induced by the
anesthetic dose of ketamine via the ␤-adrenergic pathway, suggesting distinct mechanisms.
CONCLUSION: Ketamine suppresses TNF production and phagocytosis by Kupffer cells/macrophages.
Therefore, unless bacterial growth is well controlled (by an antibiotic), postoperative infection might
not improve despite reduction of the inflammatory response. (Anesth Analg 2010;111:1051–8)

A nesthesia affects the immune functions of patients.

The decision to use one or a combination of anes-
thetics might affect both patient outcome and prog-
nosis.1 A combination of anesthetics also might have
affected the tumor metastasis in experimental animals.2
Ketamine is an anesthetic commonly used for patients and
laboratory animals with septicemia or trauma. Ketamine
reduces proinflammatory cytokine production, and thereby
improves the survival rate in lipopolysaccharide (LPS)-
induced septicemia models.3–5 Whereas ketamine reportedly
also improves the survival rate of animals after E. coli infec-
tion,4 other reports have demonstrated that ketamine does not
affect the outcome of sepsis using not only C3H/HeN mice
From the *Department of Anesthesiology, National Defense Medical Col-
lege, 3-2 Namiki Tokorozawa Saitama Japan 359-8513, †Department of
Immunology and Microbiology, National Defense Medical College, 3-2
Namiki Tokorozawa Saitama Japan 359-8513.
Accepted for publication June 2, 2010.
This work was supported in part by a grant-in-aid for Special Research Program
(Host stress responses to internal and external factors) from the National
Defense Medical College (to S.S.).
Disclosure: The authors report no conflict of interest.
Address correspondence to Shuhji Seki, MD, Department of Immunology
and Microbiology, National Defense Medical College, 3-2 Namiki, To-
korozawa, 359-8613 Japan. Address e-mail to btraums@ndmc.ac.jp
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181ed12fc
but also C3H/HeJ mice initiated by cecal ligation and punc-
ture.6 Moreover, it has been reported that ketamine
increases mouse mortality after cecal ligation and punc-
ture.7 In addition, studies of human polymorphonuclear
leukocytes have demonstrated that ketamine attenuates
not only proinflammatory cytokine secretion but also
phagocytosis and bactericidal activity in vitro.8 –11 These
equivocal results may be attributed in part to the differ-
ent septic models used. Nonetheless, a detailed immu-
nological analysis of ketamine anesthesia is required to
fully understand the ramifications of its clinical use.
Sepsis is defined as a systemic inflammatory response
syndrome (SIRS) induced by infection.12 If SIRS occurs in
the absence of massive bacterial infection, suppressing
inflammatory responses and cytokine production may
increase host survival.13,14 In contrast, when the inflamma-
tory response is insufficient and production of proinflam-
matory cytokines is inadequate, the host cannot eliminate
the invading bacteria.15–17 Thus, inflammatory responses
induced in the presence and absence of bacterial infection
must be considered separately to fully understand the
processes that are critical to survival.
Ketamine stimulates the ␤-adrenergic pathway in
sympathetic nerves and induces the production of cat-
echolamines.18,19 ␤-Adrenoceptor agonists such as adrena-
line and isoproterenol (catecholamine) suppress LPS-induced

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 1051

Ketamine Anesthesia and Postoperative Infection
tumor necrosis factor (TNF) production20 –22 and natural killer
(NK) cell activity.23–25 Therefore, ketamine-induced sympa-
thetic nerve stimulation may impair the host immune
system. Nevertheless, there is only limited understanding
of how ketamine anesthesia affects cytokine production
and phagocytosis by macrophages and liver Kupffer cells
and how the ketamine-stimulated ␤-adrenergic pathway
affects host immune functions. We hypothesized that
ketamine anesthesia during surgery may change the
proinflammatory cytokine response against LPS and
bacteria challenges and may also affect phagocytosis of
bacteria by Kupffer cells. Based on this hypothesis, we
found the immune-suppressive effect of ketamine on
Kupffer cells and its possible mechanism.
METHODS
Animals and Regents
All experiments were approved by the National Defense
Medical College Institution Animal Care and Use Commit-
tee. Male C57BL/6 mice (10 weeks old, 25 g) were obtained
from SLC, Inc. (Shizuoka, Japan). Escherichia coli strain B
(ATCC11303, Sigma-Aldrich, St. Louis, MO) and LPS (E.
coli 0111: B4, Sigma-Aldrich) were used for experiments.
Ketamine (preservative-free; Daiichi Sankyo Propharma,
1052 www.anesthesia-analgesia.org
Figure 1. Experimental design of ketamine anesthe-
sia in mice.
Kanagawa, Japan) and sevoflurane (Maruishi, Osaka, Ja-
pan) were used as anesthetics. ␣-Galactosylceramide
(␣-GalCer) was kindly provided by Pharmaceutical Re-
search Laboratory, Kirin Brewery, Takasaki, Japan.
Ketamine Anesthesia and Surgical Intervention
Mice were injected intraperitoneally (IP) with an anesthetic
dose of ketamine (100 mg/kg/0.5 mL) or phosphate buffer
sodium (PBS) (0.5 mL) 10 minutes before laparotomy.
During laparotomy, mice were anesthetized with approxi-
mately 0.2% to 0.5% sevoflurane (ketamine group, n ⫽ 25)
or approximately 2% to 3% sevoflurane (ketamine plus
sevoflurane group, n ⫽ 25) for 5 minutes after ketamine
injection or approximately 2% to 3% sevoflurane for 5
minutes (sevoflurane group, n ⫽ 25) in room air via a
vaporizer (Fig. 1A). During laparotomy, we must some-
times change the concentration of sevoflurane to regulate
the movement and respiratory conditions of mice during
surgical maneuvers, as described in our previous study.2
The 3-cm midline incision was made for the laparotomy
and was closed in layers using 4 to 0 silk sutures. An
anesthetic dose of ketamine alone cannot satisfactorily
sedate mouse movement during laparotomy. Also, ket-
amine is not clinically used without other anesthetics. We
ANESTHESIA & ANALGESIA
studied the effects of anesthetic methods using ketamine
but not the direct pharmacological effect of ketamine per se.
Therefore, a small dose of sevoflurane inhalant (approxi-
mately 0.2%– 0.5%) was administered to ketamine-treated
mice during the laparotomy to obtain the appropriate
sedative state.
E. coli or LPS Challenge
Immediately after wound closure, mice were challenged IP
with a lethal (1 ⫻ 109 colony-forming units (CFU)/mouse)
or sublethal (1 ⫻ 108 CFU/mouse) dose of E. coli (n ⫽ 25 in
each group) (Fig. 1B). Antibiotic-treated mice were also
challenged IP with 2 ⫻ 108 CFU/mouse of E. coli immedi-
ately after laparotomy (Fig. 1C). Similarly, mice were
challenged IP with LPS (200 ␮g/mouse) after laparotomy
(Fig. 1A, n ⫽ 25 in each group).
Antibiotic Treatment (Double and
Single Injections)
Tail vein injections of cefazolin sodium (50 mg/kg) were
administered to mice in both the ketamine (with approxi-
mately 0.2%– 0.5% sevoflurane) and sevoflurane (ap-
proximately 2%–3%) groups, one immediately before the
laparotomy and the second 6 hours after the E. coli
challenge (n ⫽ 20 in each group). A single injection with
cefazolin sodium (50 mg/kg) was also given to ketamine-
anesthetized mice just before the laparotomy but not
after bacterial challenge (Fig. 1C, n ⫽ 25).
Neutralization of TNF with Antibiotic Treatment
Anti-TNF neutralizing antibody (0.5 mg/mouse, MP6-XT3;
BD Pharmingen) or rat IgG was injected via IV 1 hour
before E. coli challenge (2 ⫻ 108 CFU/mouse) after lapa-
rotomy. The mice received double or single injections with
cefazolin sodium, as described above (Fig. 1D, n ⫽ 20 in
each group).
Pretreatment with Nadolol
Nadolol is a nonselective ␤-adrenoceptor antagonist. Mice
were injected with nadolol (10 mg/kg/0.2 mL) or 0.2 mL
PBS via the tail vein 5 minutes before ketamine/PBS
treatment. The mice were then laparotomized, followed by
LPS challenge (Fig. 1E, n ⫽ 5 in each group). To precisely
evaluate the relationship between ketamine and nadolol,
some mice did not receive laparotomy, thereby avoiding
sevoflurane treatment. Ten minutes after ketamine/PBS
treatment, these mice were similarly challenged IP with
LPS (n ⫽ 5 in each group). To examine the affect of nadolol
on mice treated with a low dose of ketamine, mice were
also injected IP with 10 mg/kg ketamine (10% of the
anesthetic induction dose) 5 minutes after the nadolol/PBS
treatment (n ⫽ 5 in each group).
Cytokine Measurements and Bacterial Counts
Blood samples of individual mice were obtained from the
retro-orbital sinus immediately before anesthesia adminis-
tration and at 1, 3, 6, 12, 24, and 48 hours after E. coli or LPS
challenge. Sera and culture supernatants of mononuclear
cells (MNCs) were detected using Enzyme-Linked Immuno-
Sorbent Assay (ELISA) kits (TNF and interferon [IFN]-␥,
BD Pharmingen, San Diego, CA; IL-12, Endogen, Woburn,
MA). Livers were aseptically removed and homogenized in
October 2010 • Volume 111 • Number 4
PBS. Homogenates were serially diluted 10-fold with PBS,
plated on brain heart infusion agar, and then incubated at
37°C for 24 hours for colony counting.
Isolation of Peripheral Blood Leukocytes, Liver,
and Spleen MNCs, and In Vitro LPS Stimulation
Blood samples were obtained from mice by cardiac punc-
ture, and leukocytes were separated by hemolysis. Liver
MNCs were prepared as previously described.13,26,27 In
brief, liver specimens were minced, placed in 0.05% colla-
genase solution (Wako, Osaka, Japan), shaken for 20 min-
utes in a 37°C water bath, and passed through steel mesh.
After mixing in 33% Percoll, the samples were centrifuged
at 500g for 20 minutes at room temperature. Liver MNCs
were obtained after the red blood cells were lysed. Spleno-
cytes were also obtained after the spleen was passed
through a mesh and the red blood cells were lysed. The
cells (5 ⫻ 105 per well in 200 ␮L Roswell Park Memorial
Institute [RPMI] 1640 supplemented with 10% fetal bovine
serum in 96 well flat– bottomed plates) were incubated with
ketamine (1 ⫻ 10⫺6 M) at 37°C, in 5% CO2 for 1 hour.
Subsequently, the cells were incubated with LPS (200
ng/mL) for 2 hours. Three individual experiments were
performed, with three or four samples per each group.
In Vivo Phagocytosis of Microspheres by
Kupffer Cells
Mice were injected with 20 ␮L of Fluoresbrite YG (FITC)
Carboxylate Microspheres (75-nm diameter; Polysciences
Europe, Eppelheim, Germany) (hereafter called FITC-
microspheres) via tail veins, immediately after LPS/PBS
challenge (n ⫽ 5 in each group). One hour later, liver MNCs
were isolated. After incubation for 10 minutes with an
Fc-blocker (2.4 G2; Pharmingen) to prevent nonspecific
binding, the MNCs were labeled with PE-conjugated anti-
CD11b mAb and PC5-conjugated anti-Gr-1 mAb. Kupffer
cells stain positively with CD11b but are negative for Gr-1
staining.15 CD11b⫹Gr-1⫺ Kupffer cells were analyzed for in
vivo phagocytosis by flow cytometry (EPICS XL, Coulter,
Miami, FL).
Statistical Analysis
The data are presented as means ⫾ SE. Statistical analysis
was performed using Graphpad-Prism software, version
4.00 (Graphpad Software, Inc., San Diego, CA). Survival
frequencies were compared using log rank tests. Changes
in serum cytokine concentrations were compared using
two-way analysis of variance (ANOVA). Other cytokine
levels between groups or among several groups were
compared using the Student t test or one-way ANOVA
followed by the Newman-Keuls test. The survival of
antibiotic-treated mice was evaluated using Fisher exact
test. The level of significance was set at P ⬍ 0.05. Sample
sizes were chosen based on data in other papers in this
field.3,5,28 –30
RESULTS
Ketamine Anesthesia Improves Survival After
Laparotomy with LPS Challenge and Suppresses
Elevation of Serum Proinflammatory Cytokines
Both ketamine-treated groups (anesthetized either with
0.2%– 0.5% sevoflurane or 2%–3% sevoflurane) showed
www.anesthesia-analgesia.org 1053
Ketamine Anesthesia and Postoperative Infection
Figure 3. The effects of ketamine anesthesia on mouse survival
after E. coli challenge after laparotomy. Mice received ketamine
anesthesia (⫹0.2%– 0.5% sevoflurane) or sevoflurane (2%–3%) an-
esthesia. Thereafter, they were challenged IP with sublethal (1 ⫻
108 colony-forming units (CFU)/mouse) or lethal (1 ⫻ 109
CFU/mouse) doses of E. coli after laparotomy; n ⫽ 25 in each group.
significantly higher survival rates and greater suppres-
sion of serum peaks of TNF, interleukin (IL)-12, and
IFN-␥ compared with the sevoflurane (approximately
2%–3%) group (Fig. 2). However, the two ketamine-
treated groups showed similar survival rates and serum
cytokine levels, suggesting that ketamine increases
mouse survival and suppresses proinflammatory cyto-
kine secretion. The differences observed between the
mice treated with ketamine/low-dose sevoflurane (ket-
amine group) versus high-dose sevoflurane alone
(sevoflurane group) might not have been due to the use
of a lower dose of sevoflurane but due to the use of
ketamine. Therefore, the affect of ketamine anesthesia on
the ketamine (with low-dose sevoflurane) and sevoflu-
rane groups was compared in further experiments.
Ketamine Anesthesia Does Not Increase
Survival After Laparotomy with E. coli
Challenge Despite Suppression of Serum
TNF and IFN-␥
Mice in the ketamine group did not survive at higher rates
after laparotomy with E. coli (1 ⫻ 108 CFU/mouse) chal-
lenge compared with those in the sevoflurane group (Fig.
3), despite significant suppression of TNF and IFN-␥ (Table
1). Ketamine anesthesia also failed to increase survival after
laparotomy with lethal E. coli challenge (1 ⫻ 109
1054 www.anesthesia-analgesia.org
Figure 2. The effect of ketamine anesthesia on
survival (A), serum tumor necrosis factor (TNF) (B),
interleukin (IL)-12 (C), and interferon (IFN)-␥ (D)
levels in mice after lipopolysaccharide (LPS) chal-
lenge after laparotomy. Mice were anesthetized
with ketamine ⫹0.2% to 0.5% sevoflurane (ket-
amine group), ketamine ⫹2% to 3% sevoflurane
(sevoflurane ⫹ ketamine group) or 2% to 3%
sevoflurane (sevoflurane group). Data are the
means ⫾ SE, n ⫽ 25 in each group. *P ⬍ 0.01,
†P ⬍ 0.05 vs other groups.
Table 1. Serum TNF and IFN-␥ Levels in Mice
After E. coli Challenge Following Laparotomy
Sevoflurane Ketamine
Serum TNF at 1 h
1 ⫻ 108 CFU 119 ⫾ 14 66 ⫾ 2*
1 ⫻ 109 CFU 404 ⫾ 53 146 ⫾ 19*
Serum IFN-␥ at 6 h
1 ⫻ 108 CFU 396 ⫾ 63 145 ⫾ 32*
1 ⫻ 109 CFU 1103 ⫾ 181 530 ⫾ 51*
TNF ⫽ tumor necrosis factor; IFN ⫽ interferon.
Sera were obtained from the mice 1 hour and 6 hours after E. coli challenge
to measure serum TNF and IFN-␥, respectively. Data are means ⫾ SE (pg/mL)
from 25 mice in each group. * P ⬍ 0.01 vs sevoflurane.
CFU/mouse) (Fig. 3), despite suppression of TNF and
IFN-␥ (Table 1).
Ketamine Anesthesia Plus Antibiotic Therapy
Increases Survival After Laparotomy with
E. coli Challenge
Unlike LPS challenge, bacterial growth and proliferation
could affect the survival of E. coli– challenged mice after
ketamine treatment. Because perioperative septic patients
are usually treated with antibiotics, E. coli– challenged mice
were given two injections of cefazolin. In addition, TNF
was depleted in mice receiving sevoflurane anesthesia and
cefazolin injections to determine if TNF levels decreased by
ketamine might explain the increased survival of cefazolin-
treated mice in postoperative infection. Concomitant use of
cefazolin with ketamine anesthesia significantly increased
survival after E. coli challenge with suppression of TNF and
IFN-␥ compared with sevoflurane anesthesia alone (Fig. 4,
Table 2). The TNF-depleted mice also showed a marked
increase in survival after postlaparotomy E. coli challenge
after sevoflurane anesthesia (Fig. 4) with significant sup-
pression of IFN-␥ (Table 2).
Suppression of TNF Decreases Survival in
E. coli–Challenged Mice Under Insufficient
Regulation of Bacterial Growth
TNF-neutralizing antibodies (Ab) was administered to
single cefazolin-treated mice before laparotomy (but not
after E. coli challenge) followed by sevoflurane anesthesia.
ANESTHESIA & ANALGESIA
 Table 4. The Effect of Antibiotic Treatment on
Survival After E. coli Challenge Following
Laparotomy in Ketamine-Anesthetized Mice
Survival
Without injection Single injection Double injections
36 h 10/25* 20/25 21/25
72 h (3d) 5/25* 17/25 20/25
168 h (7d) 4/25* 12/25§ 19/25
Ketamine-anesthetized mice were treated with a single injection of cefazolin,
double injections, or without injection. * P ⬍ 0.01 vs other groups, § P ⬍ 0.05
vs double injections.
Figure 4. The effect of ketamine anesthesia and neutralizing tumor
necrosis factor (TNF) ⫹ antibiotic therapy on mouse survival after
sublethal E. coli challenge after laparotomy. Mice received ketamine Table 5. The Effect of In Vitro Ketamine
anesthesia, sevoflurane anesthesia, or sevoflurane anesthesia after Treatment on LPS-Stimulated TNF Production
TNF neutralization. They were then treated twice with cefazolin and
challenged IP with 2 ⫻ 108 E. coli colony-forming units (CFU)/mouse Ketamine
after laparotomy; n ⫽ 20 in each group, *P ⬍ 0.01. Without ketamine With ketamine (10ⴚ6 M)
Circulating leukocytes 240 ⫾ 20 111 ⫾ 20*
Liver MNCs 548 ⫾ 31 278 ⫾ 20*
Table 2. Serum TNF and IFN-␥ Levels in Mice Spleen MNCs 200 ⫾ 12 82 ⫾ 5*
After Antibiotic-Treated E. coli Challenge LPS ⫽ lipopolysaccharide; MNCs ⫽ mononuclear cells.
Following Laparotomy Data are means ⫾ SE (pg/mL) from three individual experiments with two
Sevoflurane ⴙ samples per each group. * P ⬍ 0.05 vs without ketamine.
Sevoflurane Anti-TNF Ab Ketamine
Serum TNF 172.0 ⫾ 23.2* Undetected 58.6 ⫾ 2.2*
Serum IFN-␥ 649.3 ⫾ 86.7* 124.5 ⫾ 13.1 146.7 ⫾ 18.1 Table 6. The Effect of Ketamine Treatment
TNF ⫽ tumor necrosis factor; IFN ⫽ interferon; Ab ⫽ antibodies. on Phagocytosis by Kupffer Cells in
Cefazolin (double injections)-treated mice were injected IV with anti-TNF LPS-Challenged Mice
neutralizing Ab or rat IgG, as a control, 1 hour before E. coli challenge. Sera Without laparotomy
were obtained from the mice at 1 hour and 6 hour to measure serum TNF and
Treatment PBS Ketamine PBS Ketamine
IFN-␥, respectively. Data are means ⫾ SE (pg/mL) from 20 mice in each
group. * P ⬍ 0.01 vs other groups. Challenge PBS PBS LPS LPS
Phagocytosis 13.6 ⫾ 1 7.5 ⫾ 0.8* 20.7 ⫾ 1.1* 14.02 ⫾ 1.1
(%)
Laparotomy
Table 3. The Effect of Neutralizing Anti-TNF Anesthesia Sevoflurane Ketamine Sevoflurane Ketamine
Antibody on Survival After E. coli Challenge of Challenge PBS PBS LPS LPS
Mice Treated by a Single Injection of Antibiotic Phagocytosis 14.9 ⫾ 0.5 9.4 ⫾ 0.8* 21.4 ⫾ 0.7* 14.6 ⫾ 0.6
Survival (%)

Sevoflurane Sevoflurane ⴙ Anti-TNF-Ab PBS ⫽ phosphate buffer sodium; LPS ⫽ lipopolysaccharide.

36 h 15/20 17/20
72 h (3 d) 13/20 4/20*
168 h (7 d) 12/20 3/20*
TNF ⫽ tumor necrosis factor; Ab ⫽ antibodies.
Mice were injected IV with anti-TNF neutralizing antibody or rat IgG 1 hour
before E. coli challenge following laparotomy. Cefazolin was injected once
immediately before laparotomy. * P ⬍ 0.05 vs sevoflurane.
Most survived the initial 36-hour period but eventually
died 7 days after infection (Table 3). Bacteria were then
counted in the livers of TNF-depleted mice treated with
cefazolin (one or two injections) or without cefazolin 24
hours after E. coli challenge. A single injection of cefazolin
did not effectively suppress bacterial growth (counts) in the
liver compared with double injections (single; 1.1 ⫾ 0.4
versus double; 0.1 ⫾ 0.1 ⫻ 106 CFU, P ⬍ 0.05, n ⫽ 6 in each
group), although growth was significantly inhibited in
antibiotic-treated mice (without antibiotic injection; 5.1 ⫾
0.3 ⫻ 106 CFU, P ⬍ 0.01, n ⫽ 6). Because ketamine
suppressed TNF secretion, survival after E. coli challenge
was examined in the mice receiving ketamine anesthesia
that were treated with a single injection of cefazolin.
During the initial 36 hours, these mice exhibited a rate of
Proportion of microsphere phagocytosis by Kupffer cells are shown as
means ⫾ SE from 5 mice in each group. * P ⬍ 0.05 vs other groups.
survival comparable to the mice with two cefazolin injec-
tions but showed significantly lower survival at 7 days,
although both antibiotic treatments increased mouse sur-
vival after bacterial challenge (Table 4). These findings
suggest that unless bacterial growth/proliferation is well
controlled (by repeated antibiotic injections), suppression
of TNF (by neutralizing Ab or ketamine) may adversely
affect surgical patients with bacterial infections.
Ketamine Suppresses LPS-Induced TNF
Production by Cultured Lymphocytes In Vitro
Coculturing with ketamine significantly suppressed TNF
production from blood leukocytes, liver, or spleen MNCs
stimulated by LPS in vitro, suggesting that ketamine di-
rectly suppresses LPS-induced TNF production by these
cells (Table 5).
Ketamine Attenuates In Vivo Phagocytic Activity
of Kupffer Cells in LPS-Challenged Mice
Ketamine treatment significantly decreased phagocytosis
by Kupffer cells in mice challenged with PBS (Table 6).

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 1055

Ketamine Anesthesia and Postoperative Infection
Figure 5. The effect of nadolol on serum tumor necrosis factor (TNF)
levels after lipopolysaccharide (LPS) challenge in mice given the
anesthetic or low dose of ketamine. After pretreatment with nadolol
or phosphate buffer sodium (PBS), mice were treated with ketamine
(100 mg kg⫺1 ⫽ anesthetic induction dose; 10 mg kg⫺1 ⫽ 10%
anesthetic induction dose) or PBS followed by LPS challenge. Sera
were obtained from the mice 1 hour after LPS challenge. Data are
the means ⫾ SE from five mice in each group. §P ⬍ 0.05, *P ⬍ 0.01
vs other groups.
Table 7. The Effect of Nadolol on the Serum TNF
Levels After LPS Challenge Following Laparotomy
Pretreatment PBS PBS Nadolol Nadolol
Anesthesia Sevoflurane Ketamine Sevoflurane Ketamine
Serum TNF 989 ⫾ 74 248 ⫾ 47* 1020 ⫾ 109 239 ⫾ 34*
levels at 1 h
TNF ⫽ tumor necrosis factor; LPS ⫽ lipopolysaccharide;
Data are means ⫾ SE (pg/mL) from 5 mice in each group. * P ⬍ 0.01 vs
sevoflurane.
Although LPS challenge increased phagocytosis by Kupffer
cells, ketamine treatment also significantly decreased their
phagocytosis (Table 6). Although laparotomy alone did not
significantly affect the phagocytic activity by Kupffer cells,
ketamine anesthesia significantly decreased the phagocyto-
sis after laparotomy (Table 6).
The Effect of Nadolol Pretreatment on
Ketamine-Suppressed Proinflammatory Cytokine
Response or Phagocytosis in LPS-Challenged Mice
Although the anesthetic dose of ketamine (100 mg/kg)
markedly suppressed serum TNF levels 1 hour after LPS
challenge in mice (without laparotomy), nadolol pretreat-
ment did not abolish the ketamine-induced suppression of
TNF (Fig. 5). A low dose of ketamine (10 mg/kg, 10% of the
anesthetic dose) also suppressed serum TNF levels after
LPS challenge; however, nadolol pretreatment significantly
restored TNF levels in mice (Fig. 5). Further examination of
LPS-induced TNF secretion after laparotomy revealed that
laparotomy obviously suppressed LPS-induced TNF eleva-
tion in sevoflurane-anesthetized mice (Table 7, Fig. 5).
Many investigators have also demonstrated that surgical
1056 www.anesthesia-analgesia.org
Table 8. The Effect of Nadolol on Phagocytosis
by Kupffer Cells in With or Without Ketamine-
Treated Mice
Without laparotomy
Pretreatment PBS PBS Nadolol Nadolol
Anesthesia PBS Ketamine PBS Ketamine
Phagocytosis (%) 14.5 ⫾ 0.7 8.4 ⫾ 0.5* 15.8 ⫾ 1.3 15.8 ⫾ 1.5
Laparotomy
Pretreatment PBS PBS Nadolol Nadolol
Anesthesia Sevoflurane Ketamine Sevoflurane Ketamine
Phagocytosis (%) 15.8 ⫾ 0.9 9.6 ⫾ 0.8* 16.7 ⫾ 1.0 15.0 ⫾ 0.9
PBS ⫽ phosphate buffer sodium.
Proportion of microsphere phagocytosis by Kupffer cells are shown as
means ⫾ SE from 5 mice in each group. * P ⬍ 0.05 vs other groups.
stress such as surgery or traumatic injury induces hypore-
sponsiveness of TNF secretion to LPS, in particular, the
early phase after surgery,31–34 but TNF suppression by
ketamine anesthesia (100 mg/kg) did not significantly
differ in mice with (Table 7) or without laparotomy (Fig. 5).
Phagocytosis by Kupffer cells was examined in LPS-
challenged mice (without laparotomy) that received an
anesthetic dose of ketamine. Unlike the TNF response,
nadolol almost completely restored the ketamine-
suppressed phagocytic activity of Kupffer cells (Table 8).
This restoration by nadolol was also observed in laparoto-
mized mice (Table 8).
DISCUSSION
Ketamine anesthesia improved survival in LPS-challenged
mice but not in mice challenged with viable E. coli. Never-
theless, the survival rate after E. coli challenge was
improved by the concomitant use of an antibiotic and
ketamine anesthesia, suggesting that inhibition of bacterial
growth critically influences the ketamine-induced benefi-
cial effect on survival.
Since the septic condition induced by LPS is merely a
result of SIRS without infection, suppression of proinflam-
matory cytokines by ketamine improves mouse survival. In
contrast, septicemia after E. coli challenge is a result of
infection accompanied by SIRS. TNF and IFN-␥ are impor-
tant for the elimination of invading bacteria, although they
also induce shock and organ injury.13,14,16 The suppression
of proinflammatory cytokines by ketamine may thereby
attenuate bacterial elimination. Ketamine also decreased
the phagocytic activity of Kupffer cells, indicating further
impairment of bacterial clearance. These findings may
explain why ketamine (without antibiotic) could not im-
prove the survival of E. coli–infected mice. A subanesthetic
dose of ketamine reportedly suppressed LPS-stimulated
TNF production by peripheral blood MNCs in patients,11
suggesting that ketamine, even at a low dose, suppresses
LPS-induced TNF production in humans and mice.
IL-12 and ␣-GalCer, a synthetic NKT cell ligand, were used
to determine whether ketamine directly affects IFN-␥ produc-
tion from NK and/or NKT cells. Exogenous IL-12 directly
induces IFN-␥ production by mouse NKT cells.35–37 ␣-GalCer
also directly stimulates IFN-␥ production from NKT cells and
subsequently from NK cells.2,28,38 – 41 Ketamine anesthesia did
not suppress serum IFN-␥ levels after IL-12 or ␣-GalCer
challenge in mice (data not shown), suggesting that the IFN-␥
ANESTHESIA & ANALGESIA
producing capacities of NK/NKT cells are not directly af-
fected by ketamine anesthesia but are suppressed primarily
through reduced IL-12 production from macrophages/
Kupffer cells.
Sevoflurane anesthesia for 15 to 30 minutes reportedly
suppresses LPS-induced TNF secretion in mice.42,43 In this
study, sevoflurane was only administered for 5 minutes.
The TNF elevation observed in the ketamine plus high-dose
sevoflurane group was similar to that seen after LPS
challenge to the ketamine plus low-dose sevoflurane group,
suggesting that ketamine-induced suppression of TNF is
independent of sevoflurane dose.
Nadolol did not inhibit suppression of TNF secretion by
an anesthetic dose of ketamine (100 mg/kg), although it
significantly restored TNF secretion under low-dose ket-
amine (10 mg/kg). Low-dose ketamine also reportedly
increased serum adenosine levels (0.4 ␮M) and thereby
inhibited secretion of proinflammatory cytokines after LPS
challenge,5 as evidenced by use of an adenosine receptor
antagonist. However, this study did not examine how the
anesthetic dose of ketamine affects serum adenosine levels
or proinflammatory cytokine levels. Propranolol (another
␤-adrenoceptor blocker) inhibited the secretion of adeno-
sine from the kidneys after electric stimulation of the
periarterial sympathetic nerve.44 Therefore, adenosine and,
presumably, adrenaline appear to be responsible for the
reduction of TNF production by activated-macrophages
after a low dose of ketamine. However, the direct effect
(non-␤-adrenergic pathway) of an anesthetic dose of ket-
amine may overcome or overwhelm the inhibitory effect
via the ␤-adrenergic pathway by nadolol (as seen with
low-dose ketamine) and further suppress TNF production.
The addition of ketamine to cultures consistently suppressed
LPS-induced TNF production from isolated MNCs in vitro,
suggesting a direct suppression of TNF by ketamine.
The results of this study also demonstrated for the first
time that ketamine anesthesia significantly inhibited in vivo
phagocytic activity by Kupffer cells. Previous in vitro studies
had shown that ketamine did not inhibit phagocytosis by
either polymorphonuclear leukocytes or macrophages at clini-
cal concentrations, although it could suppress at higher con-
centrations.9,10,45 Interestingly, inhibiting the ␤-adrenergic
pathway by nadolol restored ketamine-suppressed phagocy-
tosis by Kupffer cells. The anesthetic dose of ketamine might
directly suppress TNF secretion in LPS-challenged mice inde-
pendent of the ␤-adrenergic pathway, while also indirec-
tly suppressing phagocytosis by Kupffer cells via the
␤-adrenergic pathway. Inhibiting the ␤-adrenergic pathway
by nadolol may be effective in ketamine anesthesia for septic
patients because it might abrogate the enhanced catechol-
amine effect. However, when nadolol was administered to E.
coli– challenged mice under ketamine anesthesia without us-
ing an antibiotic, survival was not improved (unpublished
observations). Nadolol may induce heart failure and severe
hypotension in mice with septicemia under ketamine anes-
thesia and thereby mask restoration of the phagocytic func-
tion of macrophages.
Ketamine anesthesia may be preferred for septic patients
because of its suppressive effects of proinflammatory cyto-
kines. However, ketamine also might decrease bacterial
October 2010 • Volume 111 • Number 4
phagocytosis by macrophages/Kupffer cells, which pre-
sumably enhance bacterial growth/proliferation in periop-
erative septic patients. Therefore, antibiotic treatment could
be important for septic patients who received ketamine
anesthesia. It should also be noted that ketamine reportedly
injures Kupffer cells and endothelial cells and may affect
liver functions.46
REFERENCES
1. Exadaktylos AK, Buggy DJ, Moriarty DC, Mascha E, Sessler DI.
Can anesthetic technique for primary breast cancer surgery affect
recurrence or metastasis? Anesthesiology 2006;105:660 – 4
2. Wada H, Seki S, Takahashi T, Kawarabayashi N, Higuchi H,
Habu Y, Sugahara S, Kazama T. Combined spinal and general
anesthesia attenuates liver metastasis by preserving TH1/TH2
cytokine balance. Anesthesiology 2007;106:499 –506
3. Taniguchi T, Takemoto Y, Kanakura H, Kidani Y, Yamamoto
K. The dose-related effects of ketamine on mortality and
cytokine responses to endotoxin-induced shock in rats. Anesth
Analg 2003;97:1769 –72
4. Shaked G, Czeiger D, Dukhno O, Levy I, Artru AA, Shapira Y,
Douvdevani A. Ketamine improves survival and suppresses
IL-6 and TNFalpha production in a model of Gram-negative
bacterial sepsis in rats. Resuscitation 2004;62:237– 42
5. Mazar J, Rogachev B, Shaked G, Ziv NY, Czeiger D, Chaimo-
vitz C, Zlotnik M, Mukmenev I, Byk G, Douvdevani A.
Involvement of adenosine in the antinflammatory action of
ketamine. Anesthesiology 2005;102:1174 – 81
6. Imai T, Takahashi K, Masuo F, Goto F. Anaesthesia affects
outcome of sepsis in mice. Can J Anaesth 1998;45:360 – 6
7. Hansbrough JF, Zapata-Sirvent RL, Bartle EJ, Anderson JK, Elliott
L, Mansour MA, Carter, WH. Alterations in splenic lymphocyte
subpopulations and increased mortality from sepsis following
anesthesia in mice. Anesthesiology 1985;63:267–73
8. Krumholz W, Endrass J, Hempelmann G. Inhibition of phago-
cytosis and killing of bacteria by anaesthetic agents in vitro.
Br J Anaesth 1995;75:66 –70
9. Heller A, Heller S, Blecken S, Urbaschek R, Koch T. Effects of
intravenous anesthetics on bacterial elimination in human
blood in vitro. Acta Anaesthesiol Scand 1998;42:518 –26
10. Nishina K, Akamatsu H, Mikawa K, Shiga M, Maekawa N,
Obara H, Niwa Y. The inhibitory effects of thiopental, mida-
zolam, and ketamine on human neutrophil functions. Anesth
Analg 1998;86:159 – 65
11. Beilin B, Rusabrov Y, Shapira Y, Roytblat L, Greemberg L,
Yardeni IZ, Bessler H. Low-dose ketamine affects immune
responses in humans during the early postoperative period.
Br J Anaesth 2007;99:522–7
12. Am College of Chest Physicians/Society of Critical Care Med
Consensus Conference: definitions for sepsis and organ failure
and guidelines for the use of innovative therapies in sepsis.
Crit Care Med 1992;20:864 –74
13. Ogasawara K, Takeda K, Hashimoto W, Satoh M, Okuyama R,
Yanai N, Obinata M, Kumagai K, Takada H, Hiraide H, Seki S.
Involvement of NK1⫹ T cells and their IFN-gamma produc-
tion in the generalized Shwartzman reaction. J Immunol
1998;160:3522–7
14. Sato K, Kinoshita M, Motegi A, Habu Y, Takayama E,
Nonoyama S, Hiraide H, Seki S. Critical role of the liver CD8⫹
CD122⫹ T cells in the generalized Shwartzman reaction of
mice. Eur J Immunol 2005;35:593– 602
15. Kinoshita M, Uchida T, Nakashima H, Ono S, Seki S, Hiraide
H. Opposite effects of enhanced tumor necrosis factor-alpha
production from Kupffer cells by gadolinium chloride on liver
injury/mortality in endotoxemia of normal and partially hepa-
tectomized mice. Shock 2005;23:65–72
16. Kinoshita M, Seki S, Ono S, Shinomiya N, Hiraide H. Paradoxical
effect of IL-18 therapy on the severe and mild Escherichia coli
infections in burn-injured mice. Ann Surg 2004;240:313–20
17. Watanabe H, Numata K, Ito T, Takagi K, Matsukawa A. Innate
immune response in Th1- and Th2-dominant mouse strains.
Shock 2004;22:460 – 6
www.anesthesia-analgesia.org 1057
Ketamine Anesthesia and Postoperative Infection
18. Zsigmond EK, Kelsch RC, Kothary SP. Rise in plasma free-
norepinephrine during anesthetic induction with ketamine.
Behav Neuropsychiatry 1974;6:81– 4
19. Clanachan AS, McGrath JC. Effects of ketamine on the periph-
eral autonomic nervous system of the rat. Br J Pharmacol
1976;58:247–52
20. Severn A, Rapson NT, Hunter CA, Liew FY. Regulation of
tumor necrosis factor production by adrenaline and beta-
adrenergic agonists. J Immunol 1992;148:3441–5
21. Spengler RN, Chensue SW, Giacherio DA, Blenk N, Kunkel SL.
Endogenous norepinephrine regulates tumor necrosis factor-
alpha production from macrophages in vitro. J Immunol
1994;152:3024 –31
22. Verhoeckx KC, Doornbos RP, van der Greef J, Witkamp RF,
Rodenburg RJ. Inhibitory effects of the beta-adrenergic recep-
tor agonist zilpaterol on the LPS-induced production of TNF-
alpha in vitro and in vivo. J Vet Pharmacol Ther 2005;28:531–7
23. Schedlowski M, Hosch W, Oberbeck R, Benschop RJ, Jacobs R,
Raab HR, Schmidt RE. Catecholamines modulate human NK
cell circulation and function via spleen-independent beta
2-adrenergic mechanisms. J Immunol 1996;156:93–9
24. Shakhar G, Ben-Eliyahu S. In vivo beta-adrenergic stimulation
suppresses natural killer activity and compromises resistance
to tumor metastasis in rats. J Immunol 1998;160:3251– 8
25. Kappel M, Poulsen TD, Galbo H, Pedersen BK. Effects of
elevated plasma noradrenaline concentration on the immune
system in humans. Eur J Appl Physiol Occup Physiol
1998;79:93– 8
26. Dobashi H, Seki S, Habu Y, Ohkawa T, Takeshita S, Hiraide H,
Sekine I. Activation of mouse liver natural killer cells and
NK1.1(⫹) T cells by bacterial superantigen-primed Kupffer
cells. Hepatology 1999;30:430 – 6
27. Habu Y, Seki S, Takayama E, Ohkawa T, Koike Y, Ami K, Majima
T, Hiraide H. The mechanism of a defective IFN-gamma response
to bacterial toxins in an atopic dermatitis model, NC/Nga mice,
and the therapeutic effect of IFN-gamma, IL-12, or IL-18 on
dermatitis. J Immunol 2001;166:5439 – 47
28. Inui T, Nakagawa R, Ohkura S, Habu Y, Koike Y, Motoki K,
Kuranaga N, Fukasawa M, Shinomiya N, Seki S. Age-
associated augmentation of the synthetic ligand- mediated
function of mouse NK1.1 ag(⫹) T cells: their cytokine produc-
tion and hepatotoxicity in vivo and in vitro. J Immunol
2002;169:6127–32
29. Gurfinkel R, Czeiger D, Douvdevani A, Shapira Y, Artru AA,
Sufaro Y, Mazar J, Shaked G. Ketamine improves survival in
burn injury followed by sepsis in rats. Anesth Analg
2006;103:396 – 402
30. Shibakawa YS, Sasaki Y, Goshima Y, Echigo N, Kamiya Y,
Kurahashi K, Yamada Y, Andoh T. Effects of ketamine and
propofol on inflammatory responses of primary glial cell
cultures stimulated with lipopolysaccharide. Br J Anaesth
2005;95:803–10
31. Majetschak M, Flach R, Kreuzfelder E, Jennissen V, Heukamp
T, Neudeck F, Schmit-Neuerburg KP, Obertacke U, Schade FU.
The extent of traumatic damage determines a graded depres-
sion of the endotoxin responsiveness of peripheral blood
mononuclear cells from patients with blunt injuries. Crit Care
Med 1999;27:313– 8
32. Ogata M, Okamoto K, Kohriyama K, Kawasaki T, Itoh H,
Shigematsu A. Role of interleukin-10 on hyporesponsiveness
of endotoxin during surgery. Crit Care Med 2000;28:3166 –70
1058 www.anesthesia-analgesia.org
33. Kawasaki T, Ogata M, Kawasaki C, Tomihisa, T, Okamoto K,
Shigematsu A. Surgical stress induces endotoxin hyporespon-
siveness and an early decrease of monocyte mCD14 and
HLA-DR expression during surgery. Anesth Analg 2001;
92:1322– 6
34. Lahat N, Rahat MA, Brod V, Cohen S, Weber G, Kinarty A,
Bitterman H. Abdominal surgery reduces the ability of rat
spleen cells to synthesize and secrete active tumour necrosis
factor-alpha (TNF-alpha) by a multilevel regulation. Clin Exp
Immunol 1999;115:19 –25
35. Seki S, Hashimoto W, Ogasawara K, Satoh M, Watanabe H,
Habu Y, Hiraide H, Takeda K. Antimetastatic effect of NK1⫹ T
cells on experimental haematogenous tumour metastases in
the liver and lungs of mice. Immunology 1997;92:561– 6
36. Hashimoto W, Takeda K, Anzai R, Ogasawara K, Sakihara H,
Sugiura K, Seki S, Kumagai K. Cytotoxic NK1.1 Ag⫹ alpha
beta T cells with intermediate TCR induced in the liver of mice
by IL-12. J Immunol 1995;154:4333– 40
37. Habu Y, Uchida T, Inui T, Nakashima H, Fukasawa M, Seki S.
Enhancement of the synthetic ligand-mediated function of
liver NK1.1Ag⫹ T cells in mice by interleukin-12 pretreatment.
Immunology 2004;113:35– 43
38. Nakagawa R, Nagafune I, Tazunoki Y, Ehara H, Tomura H,
Iijima R, Motoki K, Kamishohara M, Seki S. Mechanisms of the
antimetastatic effect in the liver and of the hepatocyte injury
induced by alpha-galactosylceramide in mice. J Immunol
2001;166:6578 – 84
39. Carnaud C, Lee D, Donnars O, Park SH, Beavis A, Koezuka Y,
Bendelac A. Cutting edge: Cross-talk between cells of the
innate immune system: NKT cells rapidly activate NK cells.
J Immunol 1999;163:4647–50
40. Nakagawa R, Inui T, Nagafune I, Tazunoki Y, Motoki K,
Yamauchi A, Hirashima M, Habu Y, Nakashima H, Seki S.
Essential role of bystander cytotoxic CD122⫹CD8⫹ T cells for
the antitumor immunity induced in the liver of mice by
alpha-galactosylceramide. J Immunol 2004;172:6550 –7
41. Kawabata T, Kinoshita M, Inatsu A, Habu Y, Nakashima H,
Shinomiya N, Seki S. Functional alterations of liver innate
immunity of mice with aging in response to CpG-
oligodeoxynucleotide. Hepatology 2008;48:1586 –97
42. Kidani Y, Taniguchi T, Kanakura H, Takemoto Y, Tsuda K,
Yamamoto K. Sevoflurane pretreatment inhibits endotoxin-
induced shock in rats. Anesth Analg 2005;101:1152– 6
43. Hofstetter C, Boost KA, Flondor M, Basagan-Mogol E, Betz C,
Homann M, Muhl H, Pfeilschifter J, Zwissler B. Anti-
inflammatory effects of sevoflurane and mild hypothermia in
endotoxemic rats. Acta Anaesthesiol Scand 2007;51:893–9
44. Mi Z, Jackson EK. Effects of alpha- and beta-adrenoceptor
blockade on purine secretion induced by sympathetic nerve
stimulation in the rat kidney. J Pharmacol Exp Ther
1999;288:295–301
45. Davidson JA, Boom SJ, Pearsall FJ, Zhang P, Ramsay G.
Comparison of the effects of four i.v. anaesthetic agents on
polymorphonuclear leucocyte function. Br J Anaesth
1995;74:315– 8
46. Thompson JS, Brown SA, Khurdayan V, Zeynalzadedan A,
Sullivan PG, Scheff SW. Early effects of tribromoethanol,
ketamine/xylazine, pentobarbitol, and isoflurane anesthesia
on hepatic and lymphoid tissue in ICR mice. Comp Med
2002;52:63–7
ANESTHESIA & ANALGESIA
 Regional Anesthesia
Section Editor: Terese T. Horlocker

Estimation and Pharmacodynamic Consequences of

the Minimum Effective Anesthetic Volumes for
Median and Ulnar Nerve Blocks: A Randomized,
Double-Blind, Controlled Comparison Between
Ultrasound and Nerve Stimulation Guidance
Matthieu Ponrouch, MD,* Nicolas Bouic, MD,* Sophie Bringuier, PharmD, PhD,†
Philippe Biboulet, MD,* Olivier Choquet, MD,* Michèle Kassim, MD,* Nathalie Bernard, MD, MSc,*
and Xavier Capdevila, MD, PhD‡

BACKGROUND: Nerve stimulation and ultrasound guidance are the most popular techniques for
peripheral nerve blocks. However, the minimum effective anesthetic volume (MEAV) in selected
nerves for both techniques and the consequences of decreasing the local anesthetic volume on the
pharmacodynamic characteristics of nerve block remain unstudied. We designed a randomized,
double-blind controlled comparison between neurostimulation and ultrasound guidance to estimate
the MEAV of 1.5% mepivacaine and pharmacodynamics in median and ulnar nerve blocks.
METHODS: Patients scheduled for carpal tunnel release were randomized to ultrasound guidance
(UG) or neurostimulation (NS) groups. A step-up/step-down study model (Dixon method) was used to
determine the MEAV with nonprobability sequential dosing based on the outcome of the previous
patient. The starting dose of 1.5% mepivacaine was 13 and 11 mL for median and ulnar nerves at
the humeral canal. Block success/failure resulted in a decrease/increase of 2 mL. A blinded
physician assessed sensory blockade at 2-minute intervals for 20 minutes. Block onset time and
duration were noted.
RESULTS: The MEAV50 (SD) of the median nerve was lower in the UG group 2 (0.1) mL (95%
confidence interval [CI] ⫽ [1, 96] to [2, 04]) than in the NS group 4 (3.8) mL (95% CI ⫽ [2, 4] to [5,
6]) (P ⫽ 0.017). There was no difference for the ulnar nerve between UG group 2 (0.1) mL (95% CI ⫽
[1, 96] to [2, 04]) and NS group 2.4 (0.6) mL (95% CI ⫽ [2, 1] to [2, 7]). The duration of sensory
blockade was significantly correlated to local anesthetic volume, but onset time was not modified.
CONCLUSION: Ultrasound guidance selectively provided a 50% reduction in the MEAV of mepiva-
caine 1.5% for median nerve sensory blockade in comparison with neurostimulation. Decreasing the
local anesthetic volume can decrease sensory block duration but not onset time. (Anesth Analg
2010;111:1059 –64)

N erve stimulation is an indirect technique of nerve

identification but is still one of the most popular
techniques for peripheral nerve blocks. The suc-
cess rate is 91% to 98%, depending on the trials. Ultrasound
guidance may be of benefit for peripheral nerve blocks.1–3
Ultrasound guidance permits a dynamic vision of nerves,
From the *Department of Anesthesiology and Critical Care, Montpellier
University Hospital, Montpellier, France; †Department of Anesthesiology
and Critical Care Medicine, Lapeyronie University Hospital, and Epidemi-
ology and Clinical Research Department, Arnaud de Villeneuve University
Hospital Montpellier, France; and ‡Department of Anesthesiology and
Critical Care, Montpellier I University and Montpellier University Hospital;
Institut National de la Sante et de la Recherche Médicale, Montpellier,
France.
Accepted for publication June 1, 2010.
The study has been presented in part at the American Society of Anesthe-
siologists meeting, New Orleans, Louisiana, October 17–21, 2009.
Address correspondence to Prof. Xavier Capdevila, Department of Anesthe-
siology, Lapeyronie University Hospital, Route de Ganges, 34295 Montpel-
lier Cedex 5, France. Address e-mail to x-capdevila@chu-montpellier.fr.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181eb6372
vessels, muscles, and needle movements and allows the
volume distribution of local anesthetic to be controlled.4
The local anesthetic volume injected near a nerve is a factor
determining the rate of successful nerve block.5,6 The
volume and concentration affect the absorption of local anes-
thetic.7–9 The possibility of decreasing local anesthetic vol-
umes for peripheral nerve blocks is a relevant question.
Several studies have reported that multiple neurostimulations
for locating nerves permitted reduction of the local anesthetic
volume.10 –16 The study of Casati et al.6 and the systematic
review of Koscielniak-Nielsen2 found that decreasing the local
anesthetic volume seems possible using ultrasound guidance.
O’Donnell and Iohom17 recently reported in a descriptive
study that 1 mL of 2% lidocaine on each nerve component
during an ultrasound-guided axillary block was sufficient to
promote complete anesthetic block. However, we have only
sparse data supporting the consequences of such low volumes
on the pharmacodynamic parameters of sensory blockade.18
Furthermore, some authors have claimed that it is man-
datory to develop studies comparing neurostimulation
with ultrasound.2,6,19 –21

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 1059

Minimum Effective Anesthetic Volumes for Median and Ulnar Nerve Blocks
Therefore, we conducted a prospective, randomized,
double-blind controlled study to determine the minimum
effective anesthetic volume (MEAV) necessary to achieve
median and ulnar nerve blocks, by using neurostimulation
or ultrasound guidance. We tested the hypothesis that
ultrasound guidance reduces MEAV by 20%. The second-
ary end point provides information on the influence of the
local anesthetic volume on the pharmacodynamic charac-
teristics of the nerve block.
METHODS
After obtaining ethics committee approval (Comité de
protection des personnes Sud Méditérannée 3) and written
informed consent, ASA physical status I–III patients ages 18
to 90 years scheduled to undergo ambulatory endoscopic or
open pit carpal tunnel release surgery were recruited for
this randomized controlled study. Patients who did not
cooperate and those who had psychological disorders or
linguistic difficulties that might interfere with sensory
blockade were excluded. Medical exclusion criteria were
coagulopathies, known allergy to the trial drugs, infection
at the puncture site, a body mass index ⬎40 or ⬍19 kg/m2,
diabetes mellitus or known neuropathies, patients who
received opiates for chronic pain, and cardiac conduction
problems (third-degree atrioventricular block).
Patients were included in the ultrasound guidance (UG)
group or neurostimulation (NS) group using a random list
at the preanesthetic consultation. On the day of surgery,
patients were premedicated with 1 mg/kg hydroxyzine,
and 500 mL of saline at the rate of 4 to 6 mL/kg/h was
infused with an IV 20-G catheter on the contralateral
forearm. Patients were monitored using a noninvasive
arterial blood pressure measurement, a continuous electro-
cardiogram, and pulse oximetry. A high-concentration oxy-
gen mask at 6 l/minute was put on the patient’s face. The
patients were sedated with 0.05 ␮g/kg IV sufentanil. The
operating arm was positioned at 80° abduction and external
rotation. The blocks of the median and ulnar nerves were
done at the junction between the upper and middle thirds
of the arm (i.e., humeral canal) with a 15 mg/mL mepiva-
caine solution. All blocks were always placed by 1 of the
same 2 investigators (Michèle Kassim or Matthieu
Ponrouch), who had substantial expertise in regional anes-
thesia techniques. Patients were blinded to the technique at
the beginning of the procedure by the passage of the
ultrasound probe in the NS group and a single stimulation
at 1 mA of the median nerve in the UG group. In the 2
groups the patient could not see the screen of the ultra-
sound device. The anesthesiologist in charge of the block
procedure turned off the ultrasound machine in the NS
group. The ultrasound probe was placed at the beginning
of the nerve stimulation procedure. A conventional aseptic
procedure was used for peripheral nerve blocks. The anes-
thesiologist wore a mask, cap, and gloves. The puncture
site was prepared with an alcohol povidone–iodine solu-
tion, and surrounding areas were disinfected. The probe
was covered with a film-type sterile Tegaderm®.
In the NS group, nerve location was made with a 50-mm
22-G needle (Uniplex nanoLine Facet威, Pajunk©, Germany)
and a nerve stimulator (MultiStim Sensor威, Pajunk©, Ger-
many) initially set at pulse duration 0.1 ms, intensity 1.5
1060 www.anesthesia-analgesia.org
mA, and stimulation frequency 2 Hz. Nerve blocks were
achieved as has been previously demonstrated5 at the
humeral canal by a palmar flexion of the first 3 fingers and
a carpal pronation for the median nerve, and a flexion of
the fifth and fourth fingers, and a thumb adduction and
flexion of carpi ulnaris for the ulnar nerve. The procedure
started at the median nerve. The puncture site was located
at the humeral canal immediately above the brachial artery.
The needle was inserted tangentially to the skin until
minimum stimulus intensity between 0.4 and 0.6 mA was
achieved. The starting dose of 1.5% mepivacaine was 13
and 11 mL for median and ulnar nerves. The predefined
local anesthetic volume was injected after an aspiration test
repeated between each bolus of 2 mL, until the final
volume. The needle was withdrawn without leaving the
skin, and the intensity of stimulation was increased again to
1.5 mA. The needle was redirected medially and posteriorly
and the same procedure was performed to locate and block
the ulnar nerve separately.
In the UG group, localization of both nerves was per-
formed with a 50-mm 22-G needle (Uniplex nanoLine
Facet威, Pajunk©, Germany) and an ultrasound machine
(Logic E威, GE Healthcare©) with a linear probe set at a
frequency of 12 MHz. After analysis of different anatomical
elements, the probe was positioned perpendicularly to the
skin to obtain a cross-section of the humeral canal. The
nerves were visualized in their short axis. The needle was
inserted at the lateral end of the probe to keep it in the
plane of the sonogram. The needle bevel and shaft were
viewed throughout the approach to the selected nerve. The
predefined local anesthetic volume was injected after an
aspiration test repeated between each bolus of 2 mL, until
the final volume. The injection was slow and at low
pressure. The absence of intraneural injection was con-
trolled by ultrasound. After a unique needle puncture,
needle repositioning was allowed to optimize the distribu-
tion of local anesthetic around each nerve. A circumferen-
tial spread was required without exceeding the defined
volume. For both groups the volume of 1 mL was consid-
ered the lowest possible volume.
At the end of the last nerve injection, the sensory
blockade was tested every 2 minutes for 20 minutes by an
observer blinded to the technique and the volume injected.
The sensory block was assessed by the patient’s ability to
distinguish hot and cold and to discriminate a light touch in
the center of the skin area innervated by each nerve: the
thenar eminence and the anterior surface of the distal end
of the index finger for the median nerve, and the hypothe-
nar eminence and the anterior surface of the distal end of
the fifth finger for the ulnar nerve. A comparison with the
contralateral area was used to evaluate sensory blockade. A
value of 0 was noted if the sensation was the same on both
sides; 1 in case of decreased sensation in the anesthetized
hand and 2 in case of no sensation (complete sensory
block). When the sum of the variables was equal to 4 within
20 minutes, the block was defined as complete. A sum ⬍4
was considered a failure. Adverse events (i.e., paresthesia,
pain during injection, intravascular injection, and cardio-
vascular and neurologic events) were noted during the
procedure and until the end of the sensory block. The
ANESTHESIA & ANALGESIA

Figure 1. Flowchart summarizing the design of both parts of the
study. A, Comparison of the MEAV measured by the Dixon method for
both selected nerves in each group. B, Pharmacodynamics of
successful nerve blocks (onset time and duration of blocks) related
to local anesthetic volume. MEAV ⫽ minimum effective anesthetic
volume for 50% of the patients; BMI ⫽ body mass index; UG ⫽
ultrasound guidance; NS ⫽ nerve stimulation.
sensory blockade was assessed on arrival in the postanes-
thetic care unit, then every 15 minutes until the end of the
block (score returned to 0).
The definition of sensory onset time was the time from
performance of the block to a value of 4 in sensory blockade
in the selected areas, and the definition of duration of
sensory blockade was the time from performance of the
block to a value of 0 for the sensory block. Motor blockade
was not noted because it was not necessary for surgery, and
we wanted to avoid interfering with the blinded evaluation
of the sensory blockade. If the block was ineffective, the
surgeon performed a wrist infiltration with 8 mL of 15
mg/mL mepivacaine. In case of failure, the patient was
anesthetized with sufentanil 0.2 ␮g/kg and propofol in a
target-controlled infusion, and a laryngeal mask airway
was inserted. Postoperative analgesia was provided by 1 g
paracetamol every 6 hours for 48 hours. Patients were
discharged home in the evening of the day of surgery.
The primary hypothesis was to determine the MEAV
necessary to achieve median and ulnar nerve blocks using
neurostimulation or ultrasound guidance. We tested the
hypothesis that ultrasound guidance reduces MEAV by
20%. We compared the MEAV measured by the Dixon
method for both selected nerves in each group. Secondary
end points of the study examined the pharmacodynamic
characteristics of all successful nerve blocks (onset time and
duration of blocks) related to the local anesthetic volume
injected. A flowchart summarizing the design of both parts
of the study is presented in Figure 1.
Statistics
The primary end point chosen for the 2 techniques was a
20% decrease in the MEAV in the UG group. The descrip-
tive study of Carles et al. reported that the mean volume
required to block the ulnar nerve successfully using a nerve
October 2010 • Volume 111 • Number 4
Table 1. Patient Characteristics of the
Neurostimulation (NS) and Ultrasound Guidance
(UG) Groups (n ⴝ 42)
NS group UG group
(n ⴝ 21) (n ⴝ 21)
Age (years) 56 (17) 55 (17)
Weight (kg) 70 (22.1) 67.5 (20.2)
Height (cm) 164 (36) 167 (36)
BMI (kg/m2) 26 (6.9) 25.1 (6.9)
Sex (male/female) 8/13 6/15
ASA physical status I/II/III 11/10/0 9/11/1
Data are presented as mean ⫾ SD. P value ⬎0.05. ASA ⫽ American Society
of Anesthesiology; BMI ⫽ body mass index. There were no significant
differences between groups.
stimulator was 11 mL and 13 mL for the ulnar and median
nerves.5 We used these values as the initial volumes. A
difference of 2 mL in this volume in the UG group was
considered clinically significant. The Dixon method was
used.22 The MEAV50 corresponds to the MEAV that suc-
cessfully anesthetized 50% of the patients. Accepting an ␣
risk of 5% and a power (1–␤) of 80%, 12 subjects in each
group were necessary for a significant difference. It has
been demonstrated that beyond 11 patients, the signifi-
cance of the results increases with the number of patients
included.22 Considering these factors, the number of patients
enrolled was arbitrarily set at 42. Patients were divided into 2
groups, 21 in the NS group and 21 in the UG group. The
distribution was made according to a computer-randomized
list in clusters of 6 patients. Statistical analysis was per-
formed using SAS威 version 9 software (SAS Institute, Cary,
North Carolina). Comparison between qualitative variables
was performed by ␹2 test or Fisher’s exact test, if necessary.
Comparisons of quantitative variables were performed
using the Student’s t test or Mann–Whitney Wilcoxon test,
if necessary. The relationship between quantitative vari-
ables was evaluated using Spearman’s correlation coeffi-
cients. Correlation of the volume of 1.5% mepivacaine in
relation to the sensory onset time and duration of sensory
blockade was done by linear regression. P ⬍ 0.05 was
considered significant.
RESULTS
Forty-two patients were included in the study, 21 in the NS
group and 21 in the UG group. There were no significant
differences in the patients’ characteristics in both groups
(Table 1). The MEAV50 (SD) of the median nerve was lower
in the UG group 2 (0.1) mL (95% confidence interval [CI] ⫽
[1, 96] to [2, 04]) than in NS group 4 (3.8) mL (95% CI ⫽
[2, 4] to [5, 6]) (P ⫽ 0.017). There was no difference for the
ulnar nerve between UG group 2 (0.1) mL (95% CI ⫽ [1, 96]
to [2, 04]) and NS group 2.4 (0.6) mL (95% CI ⫽ [2, 1] to
[2, 7]) (Fig. 2). With both nerve localization techniques,
there was a positive significant correlation between dura-
tion of sensory blockade and the volume of mepivacaine
1.5% injected for both ulnar and median nerves. For both
nerves, the duration of sensory block decreased when less
local anesthetic was injected. In some patients a volume of
1 mL allowed a sensory blockade for ⬍60 minutes. Con-
versely, onset times of sensory blockade of the median and
ulnar nerves were not significantly correlated to the local
www.anesthesia-analgesia.org 1061
Minimum Effective Anesthetic Volumes for Median and Ulnar Nerve Blocks
anesthetic volume injected (Fig. 3). For 1 patient in the UG
group, an unplanned subcutaneous infiltration of local
anesthetic in the musculocutaneous nerve skin area was
necessary for surgical incision. Fifteen blocks showed a
negative response 20 minutes after block placement. After
recording the presence of a negative response to the
up-and-down sequence, these patients received a wrist
infiltration with 8 mL of 15 mg/mL mepivacaine. No
1062 www.anesthesia-analgesia.org
Figure 2. Minimum effective anesthetic volume
(MEAV)50 measured by the Dixon up-and-down
method for the median (A) and ulnar (B) nerves. A,
P value ⫽ 0.017. B, P value ⫽ 0.441. The volume
X0 was chosen at 13 mL for the median nerve and
11 mL for the ulnar nerve. The interval between
volumes in 2 different patients was set at 2 mL.
The next volume X1 was determined by the per-
formance of the previous volume, X0. If X0 was
effective, X1 ⫽ X0 – 2 mL; if X0 was ineffective,
X1 ⫽ X0 ⫹ 2 mL. UG ⫽ ultrasound guidance;
NS ⫽ nerve stimulation.
Figure 3. A, B, Significant correlations
between local anesthetic volume and du-
ration of sensory blockade for median
and ulnar nerves (*P ⫽ 0.03). C, D, No
correlation between local anesthetic vol-
ume and onset time for median and ulnar
nerves (P ⫽ 0.71).
general anesthesia was required. No adverse events were
noted in either group.
DISCUSSION
We report that ultrasound guidance can reach an MEAV50
value lower than neurostimulation for the median nerve
but not for the ulnar nerve. In addition, the reduction in
local anesthetic volume caused a decrease in the duration of
ANESTHESIA & ANALGESIA
sensory blockade but did not modify the onset time of the
block for both nerves. Ultrasound guidance can reduce the
local anesthetic volume for a selected nerve block in
comparison with neurostimulation. Our results confirm
data reported in those previous studies.1–3,6,21,23 It is inter-
esting to note that if we apply the same methodology for
the calculation of MEAV for both guidance methods,
neurostimulation allows the same volume reduction as
does ultrasound guidance for some isolated nerves. For the
ulnar nerve, the results show that the MEAV50 is compa-
rable for both techniques of nerve location. For the femoral
nerve, Casati et al.6 reported that the MEAV50 was signifi-
cantly lower in an UG group (15 mL) than in a NS group (26
mL). The same authors reported in another study that the
MEAV50 using neurostimulation was significantly de-
creased in a multiple-neurostimulations and multiple-
injections group (14 mL) than in a single-neurostimulation
and single-injection group (23 mL) for a femoral nerve
block.24 In other words, a precise identification of nerves
reduces MEAV50. Ultrasound guidance permits direct vi-
sualization of the spread of local anesthetic around and in
contact with the nerve.20
The location of the median nerve and its anatomical
relationship into the humeral canal may explain the differ-
ence in the MEAV50 in the UG group in comparison with
the NS group. The median nerve is close to the brachial
artery.1,25,26 The circumferential distribution of a small
local anesthetic volume around the median nerve is avail-
able with ultrasound guidance through needle redirections
and visualization of the local anesthetic spread. The inti-
mate relationship of the median nerve and the brachial
artery, by using neurostimulation, suggests that the circum-
ferential spread around the nerve seems unpredictable and
difficult to obtain with a low volume of local anesthetic.
Conversely, the ulnar nerve is located farther from the
artery, allowing that needle movements are not always
necessary to obtain a circumferential distribution of local
anesthetic around the ulnar nerve.25,26 This may explain
why there was no difference in the MEAV50 for the ulnar
nerve in both guidance groups. Our randomized compara-
tive study reports that, depending on the nerves and
surrounding anatomical structure, ultrasound guidance
allowed a smaller volume than did neurostimulation. The
MEAV obtained was lower by 50% in the UG group. These
results confirm the results in the Danelli et al. study.27
These authors compared neurostimulation and ultrasound
guidance for sciatic nerve blocks using a subgluteal ap-
proach. They reported a 37% decrease of MEAV in the UG
group.
The decrease in volumes of local anesthetic leads to
changes in some of the nerve block pharmacodynamic
characteristics. When the volume decreases, the onset time
of the sensory blockade does not vary. Conversely, the
duration of nerve sensory blockade is closely correlated to
the local anesthetic volume. Serradell et al., using 20 to 36
mL of local anesthetic, did not report a difference in the
duration of the axillary block, corresponding to the right
side of the correlation curve in Figure 3.13 In a recent study
of 20 volunteers scheduled for sciatic nerve block with
ultrasound guidance, Latzke et al.18 reported that a local
anesthetic volume of 0.10 mL/mm cross-sectional nerve
October 2010 • Volume 111 • Number 4
area had no effect on sensory onset time, whereas the
duration of sensory block was shorter. Marhofer et al.23
reported that ultrasound can reduce the local anesthetic
volume and onset time of the femoral sensory blockade.
However, the authors did not report significant correlations
between the local anesthetic volume and onset time of the
block. Onset time and quality of the blocks also depend on
the nerve and approach.28,29
The spread of local anesthetic in contact with the nerve
is crucial for obtaining the MEAV. It is conceptualized as
the minimum circumferential area of nerve exposed to
MEAV associated with the surface of the nerve. Eichen-
berger et al.21 studied the concept of MEAV (in milliliters
per millimeter) on the basis of the diameter of the nerve.
They reported a MEAV of 0.11 mL/mm for mepivacaine
1% for ulnar nerve block at the proximal forearm. This is
interesting but does not predict MEAV values in other parts
of the same nerve. The variability of the structure of a single
nerve from the brachial plexus to its distal part explains the
different expected MEAV values. Moayeri et al.30 reported
from a cadaveric study that the nonneural tissue composi-
tion of the nerves within the epineurium increased from
their proximal to distal parts. The neuronal tissue/
nonneural tissue ratio increases from 1:1 to 1:2. These
studies demonstrated that for the same dose of local
anesthetic (volume and concentration), the onset time de-
pends on the quality of nerve localization using ultrasound
guidance or neurostimulation, the nerve studied, and the
local anesthetic spread close to the nerve.25,26
Some limitations of our study deserve comment. The
clinical relevance of a MEAV50 value might be questioned,
because in the method used, about 50% of the studied
patients had an incomplete nerve block. Nonetheless, the
ED50 concept has been used to determine volume– effect
relationships in the field of peripheral nerve blocks.6,22,24
On the other hand, we reported a correlation between
volume of local anesthetic solution and duration of the
blocks. However, the study was not designed for a com-
parison in block duration for fixed high or low volumes of
local anesthetics. Further comparative randomized studies
are necessary to determine whether the volume of local
anesthetic solution really influences the duration of nerve
blocks.
This double-blind, randomized, clinical trial shows
that the use of ultrasound guidance was effective in
reducing the local anesthetic volume for the median
nerve. Neurostimulation can provide similar results for
the ulnar nerve. The ability to visualize anatomical struc-
tures, to control the position of the needletip, and to control
the circumferential spread of local anesthetic may partly
explain these results. The decrease in volume helps to
highlight a strong correlation between local anesthetic
volume and duration of the nerve block; the onset time
does not vary with the local anesthetic volume. The precise
control of these variables and knowledge of variations of
pharmacodynamic characteristics based on the local anes-
thetic volume injected, concentration, and approach of the
block should assist practitioners to adapt the procedure to
different clinical situations.
www.anesthesia-analgesia.org 1063
Minimum Effective Anesthetic Volumes for Median and Ulnar Nerve Blocks
REFERENCES
1. Abrahams MS, Aziz MF, Fu RF, Horn JL. Ultrasound guidance
compared with electrical neurostimulation for peripheral
nerve block: a systematic review and meta-analysis of random-
ized controlled trials. Br J Anaesth 2009;102:408 –17
2. Koscielniak-Nielsen ZJ. Ultrasound-guided peripheral nerve
blocks: what are the benefits? Acta Anaesthesiol Scand 2008;
52:727–37
3. Marhofer P, Chan VW. Ultrasound-guided regional anesthesia:
current concepts and future trends. Anesth Analg 2007;104:
1265–9
4. Retzl G, Kapral S, Greher M, Mauritz W. Ultrasonographic
findings of the axillary part of the brachial plexus. Anesth
Analg 2001;92:1271–5
5. Carles M, Pulcini A, Macchi P, Duflos P, Raucoules-Aime M,
Grimaud D. An evaluation of the brachial plexus block at the
humeral canal using a neurostimulator (1417 patients): the
efficacy, safety, and predictive criteria of failure. Anesth Analg
2001;92:194 – 8
6. Casati A, Baciarello M, Di Cianni S, Danelli G, De Marco G,
Leone S, Rossi M, Fanelli G. Effects of ultrasound guidance on
the minimum effective anaesthetic volume required to block
the femoral nerve. Br J Anaesth 2007;98:823–7
7. Brown DL, Ransom DM, Hall JA, Leicht CH, Schroeder DR,
Offord KP. Regional anesthesia and local anesthetic-induced
systemic toxicity: seizure frequency and accompanying cardio-
vascular changes. Anesth Analg 1995;81:321– 8
8. Becker DE, Reed KL. Essentials of local anesthetic pharmacol-
ogy. Anesth Prog 2006;53:98 –108
9. Mather LE, Copeland SE, Ladd LA. Acute toxicity of local
anesthetics: underlying pharmacokinetic and pharmacody-
namic concepts. Reg Anesth Pain Med 2005;30:553– 66
10. Casati A, Danelli G, Baciarello M, Corradi M, Leone S, Di
Cianni S, Fanelli G. A prospective, randomized comparison
between ultrasound and nerve stimulation guidance for mul-
tiple injection axillary brachial plexus block. Anesthesiology
2007;106:992– 6
11. Sia S, Bartoli M, Lepri A, Marchini O, Ponsecchi P. Multiple-
injection axillary brachial plexus block: a comparison of two
methods of nerve localization–nerve stimulation versus pares-
thesia. Anesth Analg 2000;91:647–51
12. Sia S, Lepri A, Ponzecchi P. Axillary brachial plexus block
using peripheral nerve stimulator: a comparison between
double- and triple-injection techniques. Reg Anesth Pain Med
2001;26:499 –503
13. Serradell A, Herrero R, Villanueva JA, Santos JA, Moncho JM,
Masdeu J. Comparison of three different volumes of mepiva-
caine in axillary plexus block using multiple nerve stimulation.
Br J Anaesth 2003;91:519 –24
14. Inberg P, Annila I, Annila P. Double-injection method using
peripheral nerve stimulator is superior to single injection in
axillary plexus block. Reg Anesth Pain Med 1999;24:509 –13
15. Deleuze A, Gentili ME, Marret E, Lamonerie L, Bonnet F. A
comparison of a single-stimulation lateral infraclavicular
plexus block with a triple-stimulation axillary block. Reg
Anesth Pain Med 2003;28:89 –94
16. Benhamou D. Axillary plexus block using multiple nerve
stimulation: a European view. Reg Anesth Pain Med
2001;26:495– 8
1064 www.anesthesia-analgesia.org
17. O’Donnell BD, Iohom G. An estimation of the minimum
effective anesthetic volume of 2% lidocaine in ultrasound-
guided axillary brachial plexus block. Anesthesiology 2009;
111:25–9
18. Latzke D, Marhofer P, Zeitlinger M, Machata A, Neumann F,
Lackner E, Kettner SC. Minimal local anaesthetic volumes for
sciatic nerve block: evaluation of ED 99 in volunteers. Br J
Anaesth 2009;104:239 – 44
19. Macaire P, Singelyn F, Narchi P, Paqueron X. Ultrasound- or
nerve stimulation-guided wrist blocks for carpal tunnel re-
lease: a randomized prospective comparative study. Reg
Anesth Pain Med 2008;33:363– 8
20. Hadzic A, Dewaele S, Gandhi K, Santos A. Volume and dose of
local anesthetic necessary to block the axillary brachial plexus
using ultrasound guidance. Anesthesiology 2009;111:8 –9
21. Eichenberger U, Stockli S, Marhofer P, Huber G, Willimann P,
Kettner SC, Pleiner J, Curatolo M, Kapral S. Minimal local
anesthetic volume for peripheral nerve block: a new
ultrasound-guided, nerve dimension-based method. Reg
Anesth Pain Med 2009;34:242– 6
22. Dixon WJ. Staircase bioassay: the up-and-down method. Neu-
rosci Biobehav Rev 1991;15:47–50
23. Marhofer P, Schrogendorfer K, Koinig H, Kapral S, Weinstabl
C, Mayer N. Ultrasonographic guidance improves sensory
block and onset time of three-in-one blocks. Anesth Analg
1997;85:854 –7
24. Casati A, Fanelli G, Beccaria P, Magistris L, Albertin A, Torri G.
The effects of single or multiple injections on the volume of
0.5% ropivacaine required for femoral nerve blockade. Anesth
Analg 2001;93:183– 6
25. Sites BD, Neal JM, Chan V. Ultrasound in regional anesthesia:
where should the “focus” be set? Reg Anesth Pain Med
2009;34:531–3
26. van Geffen GJ, Moayeri N, Bruhn J, Scheffer GJ, Chan VW,
Groen GJ. Correlation between ultrasound imaging, cross-
sectional anatomy, and histology of the brachial plexus: a
review. Reg Anesth Pain Med 2009;34:490 –7
27. Danelli G, Ghisi D, Fanelli A, Ortu A, Moschini E, Berti M,
Ziegler S, Fanelli G. The effects of ultrasound guidance and
neurostimulation on the minimum effective anesthetic volume
of mepivacaine 1.5% required to block the sciatic nerve using
the subgluteal approach. Anesth Analg 2009;109:1674 – 8
28. Taboada Muniz M, Rodriguez J, Bermudez M, Valino C,
Blanco N, Amor M, Aguirre P, Masid A, Cortes J, Alvarez J,
Atanassoff PG. Low volume and high concentration of local
anesthetic is more efficacious than high volume and low
concentration in Labat’s sciatic nerve block: a prospective,
randomized comparison. Anesth Analg 2008;107:2085– 8
29. Casati A, Fanelli G, Borghi B, Torri G. Ropivacaine or 2%
mepivacaine for lower limb peripheral nerve blocks. Study
group on orthopedic anesthesia of the Italian Society of Anes-
thesia, Analgesia, and Intensive Care. Anesthesiology 1999;
90:1047–52
30. Moayeri N, Bigeleisen PE, Groen GJ. Quantitative architecture
of the brachial plexus and surrounding compartments, and
their possible significance for plexus blocks. Anesthesiology
2008;108:299 –304
ANESTHESIA & ANALGESIA
 BRIEF REPORT

Pulse-Oximetric Measurement of Prilocaine-Induced

Methemoglobinemia in Regional Anesthesia
Peter Soeding, MD,* Matthias Deppe,* and Hartmut Gehring, MD, PhD*

BACKGROUND: The Masimo Radical 7® is a new pulse CO oximeter designed to measure

methemoglobin. The device has not been evaluated in a clinical setting.
METHODS: In this prospective observational study we compared the arterial methemoglobin
levels and the corresponding pulse CO-oximetric values of the Radical 7® in regional anesthesia
with prilocaine.
RESULTS: We analyzed 360 data pairs with methemoglobin values up to 6.6%. The mean bias
and limits (⫾1.96 SD) of the device were 0.27% (⫾1.33%).
CONCLUSION: We found a high degree of agreement in measurement of methemoglobin
between the 2 methods. (Anesth Analg 2010;111:1065–8)

P rilocaine, in comparison with all other local anesthet-

ics, has the lowest direct systemic toxicity, but may lead
to an increased formation of methemoglobin
(MetHb).1,2 Whereas in healthy individuals higher concentra-
tions of MetHb are usually well tolerated, it may endanger
oxygen supply in patients with diminished cardiopulmonary
reserves or anemia.3– 6 Because of their 2-wavelength technol-
ogy, conventional pulse oximeters are not able to identify the
extent of dyshemoglobinemias.7–9 The optical analysis of a
new pulse CO oximeter (Masimo, Radical 7®)) is based on
absorbance measurements at several wavelengths. Conse-
quently, dyshemoglobinemias might also be recorded by a
pulse oximeter with sufficient precision. This has been dem-
onstrated in 1 preclinical trial10 and 2 case reports.11,12
This prospective study is the first to evaluate the efficacy
of this pulse CO oximeter in a clinical setting. We hypoth-
esized that the results for MetHb are the same for the pulse
oximetric method and arterial blood sample measurement
using a CO oximeter.
MetHb by pulse oximetry (SpMet%).10 These values could
be read using a laptop and stored after conversion into an
Excel spreadsheet. The continuous monitoring of patients
with the Radical 7® was started before the onset of regional
anesthesia.
At the times 0, 15, 30, 60, 120, 180, 240, 300, and 360
minutes after the first injection of prilocaine, 2 mL of
arterial blood was taken and immediately analyzed in a
blood gas analyzer (ABL-625, Radiometer America, Copen-
hagen, Denmark) with an integrated CO oximeter as a
reference device. The data for MetHb content (cMetHb) and
for Sao2 were compared with the corresponding values
obtained with the Radical 7®.
For statistical analysis using the Software Package for
Social Sciences (SPSS) 15.0 for Windows, we averaged the 9
repeated measurements for each of the 40 patients accord-
ing to Bland and Altman.13
A probability of P ⬍ 0.05 was considered statistically
significant.

METHODS RESULTS
After approval by the ethics committee and written One patient after interscalene block and 1 patient after
informed consent, we investigated 40 patients, physical combined femoral–sciatic nerve block required general
status ASA I–III, having orthopedic surgery: 20 patients
received an interscalene plexus block with 30 mL prilo-
caine 1% (i.e., 300 mg) and 20 patients a combined
femoral–sciatic nerve blockade with 2 ⫻ 30 mL prilocaine
1% (i.e., 600 mg in all). All blocks were performed using
a nerve stimulator (Stimuplex HNS 11, Braun, Germany).
Injection was only performed if a contraction of indicator
muscles could be demonstrated at 0.3 to 0.5 mA (stimu-
lus duration: 0.1 ms).
The pulse CO oximeter Radical 7® is a device manufac-
tured by Masimo Corp. (Irvine, California), which in addi-
tion to oxygen saturation (Spo2 in %) can also measure

From the *Department of Anesthesiology, University Clinic of Schleswig—

Holstein, Campus Luebeck, Luebeck, Germany.
Accepted for publication June 1, 2010.
Parts of the work were presented at the 2008 American Society of Anesthe-
siologists annual meeting in Baltimore, Maryland, and at the 2008 European
Society of Anesthesiologists annual meeting in Copenhagen, Denmark.
Address correspondence to Peter Soeding, MD, Department of Anesthesiology,
University Clinic of Schleswig—Holstein, Campus Luebeck, Ratzeburger Allee Figure 1. Methemoglobin (MetHb) levels after regional anesthesia
160, D-23,538 Luebeck, Germany. Address e-mail to peter.soeding@web.de. with prilocaine: plots show mean MetHb levels after 300 mg
Copyright © 2010 International Anesthesia Research Society prilocaine for interscalene block (ISB) and after 600 mg prilocaine for
DOI: 10.1213/ANE.0b013e3181eb6239 combined femoral–sciatic nerve block (FSNB).

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 1065

 BRIEF REPORT

Table 1. Methemoglobin Levels and Success Rate After Regional Anesthesia with Prilocaine for
Interscalene or Femoral–Sciatic Nerve Blocks
Methemoglobin
Peak mean ⴞ SD
Nerve blockade No. of patients Prilocaine Success rate (range) Time to peak
Interscalene 20 300 mg (30 mL) 95% (19/20) 2.3 ⫾ 0.8% (1.1–4.9) 120 minutes
Femoral–sciatic 20 600 mg (2 ⫻ 30 mL) 95% (19/20) 4.1 ⫾ 1.5% (2.0–6.6) 300 minutes

Table 2. Interclass Statistical Comparison of SpMet (%) Versus cMetHb (%) for Each Subject and for
Pooled Data
Regression analysis Bland–Altman
n Slope y intercept SEE Bias Limits (ⴞ1.96 SD) r
Pooled data 360 1.19 ⫺0.14 0.61 0.27 1.33 0.95
Subject
1 9 1.09 ⫺0.35 0.21 ⫺0.12 0.47 0.99
2 9 1.12 ⫺0.32 0.21 ⫺0.01 0.48 0.98
3 9 0.84 0.36 0.41 0.13 0.79 0.84
4 9 1.30 ⫺0.86 0.27 ⫺0.42 0.59 0.94
5 9 1.05 ⫺0.004 0.54 0.21 1.05 0.98
6 9 0.79 0.61 0.68 0.37 1.28 0.4
7 9 1.34 ⫺0.68 0.15 ⫺0.08 0.64 0.99
8 9 1.21 ⫺0.04 0.18 0.2 0.39 0.96
9 9 1.24 ⫺0.17 0.10 0.04 0.23 0.96
10 9 1.20 0.53 0.82 1.16 1.76 0.96
11 9 1.27 ⫺1.02 0.96 0.04 2.18 0.95
12 9 1.26 ⫺0.64 0.52 ⫺0.08 1.09 0.93
13 9 1.17 0.24 1.04 0.74 2.05 0.92
14 9 1.19 ⫺0.40 0.24 0.13 0.68 0.99
15 9 1.11 ⫺0.21 0.07 ⫺0.01 0.23 0.99
16 9 1.16 ⫺0.31 0.53 0.2 1.15 0.98
17 9 1.18 ⫺0.34 0.13 ⫺0.12 0.3 0.98
18 9 1.21 ⫺0.44 0.47 0.11 1.07 0.97
19 9 0.61 1.27 0.72 0.61 1.43 0.51
20 9 1.38 ⫺0.39 0.75 0.89 2.09 0.97
21 9 1.15 0.78 0.94 1.34 1.93 0.96
22 9 1.43 ⫺0.08 0.34 0.77 0.9 0.96
23 9 1.09 ⫺0.34 0.16 ⫺0.18 0.36 0.99
24 9 0.88 0.25 0.38 0.06 0.73 0.82
25 9 1.26 ⫺0.49 0.37 0.1 0.92 0.97
26 9 1.18 ⫺0.12 0.19 0.36 0.66 0.99
27 9 1.81 ⫺0.73 0.37 0.4 1.15 0.95
28 9 0.89 0.46 0.94 0.31 1.76 0.44
29 9 0.59 0.23 0.19 ⫺0.12 0.4 0.57
30 9 0.84 0.70 0.44 0.4 0.87 0.85
31 9 0.65 0.87 0.94 0.38 1.79 0.28
32 9 1.19 ⫺0.37 0.20 ⫺0.04 0.46 0.97
33 9 1.12 0.18 0.31 0.38 0.61 0.93
34 9 1.20 0.03 0.41 0.52 0.98 0.98
35 9 1.21 ⫺0.45 0.41 0.16 1.03 0.98
36 9 1.03 ⫺0.05 0.44 0 0.82 0.90
37 9 1.22 ⫺0.28 0.24 0.14 0.57 0.98
38 9 1.46 0.13 0.77 1.26 1.77 0.92
39 9 0.62 0.74 0.42 0.02 0.95 0.75
40 9 1.26 ⫺0.40 0.38 0.49 1.37 0.99
SpMet ⫽ Radical 7® measurement of methemoglobin; cMetHb% ⫽ CO oximeter measurement of methemoglobin; n ⫽ number of data points for each subject;
SEE ⫽ SE of the estimate; r ⫽ correlation coefficient.

anesthesia for block failure (success rate 95%). Figure 1
shows MetHb levels over time for both types of blocks.
Peak levels were reached for interscalene blocks after 120
minutes and for combined femoral–sciatic nerve blocks
after 5 hours (Table 1).
The statistical agreement of the MetHb measurement
between the laboratory method as a reference method
and the pulse oximetric measurement using the Radical
7® for all 40 patients is shown in Table 2 and Figure 2.
According to Bland–Altman analysis (Fig. 2), the bias
was 0.27%, and the 95% confidence limits (⫾1.96 sd)
1.33%. With the increasing rise in MetHb, there is a clear
gap between the values reported for functional oxygen
saturation by the Radical 7® and the values reported by
the CO oximeter in the blood gas analyzer device (Fig. 3).
The Radical 7® displays Spo2 readings that run in parallel

1066 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

Pulse-Oximetric Measurement of Methemoglobin
Figure 2. Bland and Altman analysis for repeated measurements:
bias plot of the difference of pulse-oximeter estimate of methe-
moglobin (MetHb) (SpMet [%]) and cMetHb% versus the average
of SpMet and cMetHb. Averaged data of the 9 repeated measure-
ments. Lines show values of bias (mean of the differences)
and ⫾1.96 SD. cMetHb% ⫽ CO-oximeter measurement of
methemoglobin.
Figure 3. Regression analysis: scatter plot of the functional satura-
tion measured by Radical 7® (SPO2) and the functional saturation
measured by the CO oximeter (SAO2) versus cMetHb%. Averaged data
of the 9 repeated measurements of each of the 40 patients.
cMetHb% ⫽ CO oximeter measurement of methemoglobin. SEE ⫽ SE
of the estimate
with and slightly higher than the values for fractional
saturation (Fig. 4).
DISCUSSION
In this prospective study we evaluated a new pulse Co
oximeter (Radical 7®, Masimo, Inc.) after regional anesthe-
sia with high doses of prilocaine. We compared the pulse
oximetric method for noninvasive monitoring of MetHb
with direct arterial measurements by using a reference
procedure. A high degree of agreement between the 2
methods could be shown for MetHb values up to 6.6%
(mean correlation coefficient ⫽ 0.95, bias ⫽ 0.27, 95%
limits ⫽ ⫾1.33).
October 2010 • Volume 111 • Number 4
Figure 4. Regression analysis: scatter plot of the functional satura-
tion measured by Radical 7® (SPO2) and the fractional saturation
measured by the CO oximeter (O2Hb) versus cMetHb%. Averaged
data of the 9 repeated measurements of each of the 40 patients.
cMetHb% ⫽ CO oximeter measurement of methemoglobin. SEE ⫽ SE
of the estimate
In comparison with lidocaine and mepivacaine, prilo-
caine has the advantage of far less cardiac or central
nervous system toxicity.14,15 Nevertheless, prilocaine is not
available in much of the world. The presence of acquired
methemoglobinemia is often assumed.3,16,17 It is triggered
not only by prilocaine but also by many other drugs,3,18 –21
in particular by benzocaine.17 The symptoms are nonspe-
cific and often unrecognized.3,9 Methemoglobinemia is
associated with the reduction in the fractional oxygen
saturation. Concentrations ⬍15% are usually well tolerated
by healthy individuals, but in patients with anemia or
cardiopulmonary diseases, clinical symptoms can occur
when the concentration exceeds 8%.3 Despite administra-
tion of the recommended threshold dose of prilocaine, it
was only possible to assess MetHb values up to 6.6%.
Further investigations are necessary to analyze the accu-
racy of the pulse oximeter at higher levels.
Dyshemoglobinemias cannot be identified by conven-
tional pulse oximeters, because their measurements, based
on 2 wavelengths of light absorption, only allow the
recording of oxyhemoglobin and desoxyhemoglobin.7,22
The pulse CO oximeter Radical 7® measures the light
absorption of 8 different wavelengths. In a preclinical study
in healthy volunteers, Barker et al. evaluated a predecessor
of the pulse oximeter that we used.10 The results (bias 0%,
sd ⫾ 0.45%) differ only slightly from the data presented in
our clinical study.
Although the Radical 7®, according to the manufactu-
rer’s information, only displays the functional oxygen
saturation,23 the data collected for the Spo2 display during
the study period tend to follow the fractional rather than
the functional oxygen saturation (Figs. 3 and 4). This
difference might be a result of the analysis algorithm of the
device.
In conclusion, we found a high degree of agreement in
measurement of MetHb with a CO oximeter and a nonin-
vasive and readily available pulse-oximetric procedure.
This may facilitate early diagnosis and treatment, when
necessary, of dyshemoglobinemia.
www.anesthesia-analgesia.org 1067
 BRIEF REPORT
DISCLOSURE
Financial support for the work: Peter Soeding and Hartmut
Gehring are employees of the University Clinic of Schleswig—
Holstein, Campus Luebeck, Luebeck, Germany. Device was
provided by Masimo, Inc., Germany.
REFERENCES
1. Scott DB, Owen JA, Richmond J. Methaemoglobinaemia due to
prilocaine. Lancet 1964;3:728 –9
2. Wright RO, Lewander WJ, Woolf AD. Methemoglobinemia:
etiology, pharmacology, and clinical management. Ann Emerg
Med 1999;34: 646 –56
3. Ash-Bernal R, Wise R, Wright SM. Acquired
methemoglobinemia—A retrospective series of 138 cases at 2
teaching hospitals. Med 2004;83:265–73
4. Bellamy MC, Hopkins PM, Halsall PJ, Ellis FR. A study into
incidence of methemoglobinaemia after “three-in-one” block
with prilocaine. Anaesthesia 1992;47:1084 –5
5. Knobeloch L, Goldring J, LeMay W, Anderson H. Three cases
of methemoglobinemia associated with dental anesthesia. Wis
Dent Assoc J 1994;70:34 –5
6. Kreeftenberg HG, Braams R, Nauta P. Methemoglobinemia
after low-dose prilocaine in an adult patient receiving barbitu-
rate comedication. Anesth Analg 2007;104:459 – 60
7. Reynolds KJ, Palayiwa E, Moyle JT, Sykes MK, Hahn CE. The
effect of dyshemoglobins on pulse oximetry: part I. theoretical
approach and part II. experimental results using an in vitro test
system. J Clin Monit 1993;9:81–90
8. Barker SJ, Tremper KK, Hyatt J. Effects of methemoglobinemia
on pulse oximetry and mixed venous oximetry. Anesthesiol-
ogy 1989;70:112–7
9. Yang JJ, Lin N, Lv R, Sun J, Zhao F, Zhang J, Xu JG.
Methemoglobinemia misdiagnosed as ruptured ectopic preg-
nancy. Acta Anesth Scand 2005;49:586 – 8
1068 www.anesthesia-analgesia.org
10. Barker SJ, Curry J, Redford D, Morgan S. Measurement of
carboxyhemoglobin and methemoglobin by pulse oximetry.
Anesthesiology 2006;105:892–7
11. Annabi EH, Barker SJ. Severe methemoglobinemia detected by
pulse oximetry. Anesth Analg 2009;108:898 –9
12. Macknet M, Kimball-Jones P, Applegate R. Benzocaine in-
duced methemoglobinemia after TEE. Resp Care 2007;52:2007
Open Forum Abstracts [e-abstracts]
13. Bland JM, Altman DG. Agreement between methods of mea-
surement with multiple observations per individual. J Biop-
harm Stat 2007;17:571– 82
14. Zink W, Graf BM. Toxikologie der Lokalanästhetika. Anaes-
thesist 2003;52:1102–23
15. Scott DB, Jerson. PJR, Braid DP, Örtengren B, Frisch P. Factors
affecting plasma levels of lignocaine and prilocaine. Br J
Anaesth 1972;44:1040 –9
16. Weinberg GL. Banning benzocaine: of bananas, bureaucrats,
and blue men. Anesth Analg 2009;108:699 –701
17. Guay J. Methemoglobinemia related to local anesthetics: a
summary of 242 episodes. Anesth Analg 2009;108:837– 45
18. Anderson ST, Hajduczek J, Barker SJ. Benzocaine-induced met-
hemoglobinemia in an adult. Anesth Analg 1988;67:1099 –101
19. Rehman HU. Methemoglobinemia. West J Med 2001;175:193– 6
20. Fung HT, Lai CH, Wong OF, Lam KK, Kam CW. Two cases of
methemoglobinemia following zoplicone ingestion. Clin Toxi-
col 2008;46:167–70
21. White CD, Weiss LD. Varying presentations of methemoglo-
binemia: two cases. J Emerg Med Suppl 1991;1:45–9
22. Rieder HU, Frei FJ, Zbinden AM, Thomson DA. Pulse oximetry
in methaemoglobinaemia. Failure to detect low oxygen satu-
ration. Anaesthesia 1989;44:326 –7
23. Radical-7 color display signal extraction pulse co-oximeter
with Rainbow Technology. Operator‘s manual. Irvine, CA:
Masimo Corporation, 2007
ANESTHESIA & ANALGESIA
 CASE REPORT

Symptomatic Axillary Hematoma After Ultrasound-

Guided Infraclavicular Block in a Patient with
Undiagnosed Upper Extremity Mycotic Aneurysms
Dave Gleeton, MD, Simon Levesque, MD, FRCPC, Claude A. Trépanier, MD, FRCPC,
Jean-Luc Gariépy, MD, FRCPC, Jean Brassard, MD, FRCPC, and Nicolas Dion, MD, FRCPC

We present a case of axillary hematoma complicating an ultrasound-guided infraclavicular

block in a patient with undiagnosed mycotic aneurysms of the peripheral arteries. Mycotic
aneurysm is a rare medical condition with well-identified risk factors. When performing
regional anesthesia in patients with these risk factors, clinicians should have a high degree of
suspicion about the possible existence of vascular anomalies. A preprocedure Doppler study of the
block area and real-time guidance of the needle using ultrasound may be useful. (Anesth Analg
2010;111:1069 –71)

U ltrasound guidance can detect variations in normal

anatomy, thus allowing the anesthesiologist to
tailor the technique to the patient-specific condi-
tion.1–3 Despite these advantages, complications still occur
during ultrasound-guided peripheral nerve blocks.4 We
present a case in which the operator using ultrasound failed
to detect mycotic aneurysms of the upper extremity arter-
ies, leading to a large hematoma of the axillary region after
an ultrasound-guided infraclavicular block.
CASE DESCRIPTION
A 44-year-old women, ASA physical status III, was admitted
in our hospital with a diagnosis of infected olecranon bursa of
the right elbow. Her recent medical history included a mitral
valve replacement for mitral endocarditis (Streptococcus viri-
dans) 4 months previously. Her medical history revealed IV
drug abuse, recurrent cellulitis of the right upper limb, high
blood pressure, Type 2 diabetes, gastroesophageal reflux, and
mild asthma. Her preoperative medication included warfarin,
rosuvastatin, metformin, diltiazem, acetylsalicylic acid,
budesonide/formoterol, salbutamol, and pantoprazole. Be-
cause surgery to drain the bursa was planned, warfarin was
discontinued 24 hours before surgery and the patient was
given 5 mg IV of vitamin K after which the international
normalized ratio was 1.3. While she awaited surgery, prophy-
lactic IV heparin was started and stopped 7 hours before
surgery. No other coagulation test was done immediately
before surgery.
At her arrival in the operating room, standard monitor-
ing and an IV line were started. Because the patient did not
show any sign of systemic infection (normal white blood
cell count and temperature), an anesthetic technique con-
sisting of an ultrasound-guided infraclavicular block was
From the Département d’Anesthésie-Réanimation, Centre de Recherche du
CHA, Unité de Recherche en Traumatologie-Urgence-Soins Intensifs, Hôpi-
tal de l’Enfant-Jésus, Université Laval, Québec, Canada.
Accepted for publication June 1, 2010.
Supported by intramural department sources.
Disclosure: The authors report no conflicts of interest.
Address correspondence and reprint requests to Dave Gleeton, MD, Depart-
ment of Anesthesiology, Hôpital de l’Enfant-Jésus, Université Laval, 1401,
18ème rue, Quebec, QC, G1J 1Z4, Canada. Address e-mail to g.dave@live.ca.
Copyright © 2010 International Anesthesia Research Society
DOI: 10.1213/ANE.0b013e3181ee80b3
offered and accepted by the patient. After sedation with
midazolam 2 mg, standard skin asepsis was accomplished
with 2% wt/vol chlorhexidine gluconate and 70% vol/vol
isopropyl alcohol and a sterile sheath was used to cover the
ultrasound probe. A 5- to 12-MHz linear probe was posi-
tioned in a parasagittal plane, medial to the coracoid
process and adjusted to give a transverse view of the
axillary artery using an ultrasound device (Zone Ultra;
Zonare Medical Systems, Mountain View, CA). Using an
in-plane technique, an 8.89-cm 20-gauge Tuohy needle (B.
Braun, Bethlehem, PA) was advanced to the posterior side
of the axillary artery until a fascial click was perceived and
30 mL of mepivacaine 1.5% was injected slowly after
multiple negative aspirations. After documentation of suc-
cessful sensory block of the upper arm, surgery was
performed using a tourniquet (pressure of 250 mm Hg).
The surgeon opened and drained the olecranon bursa, and
the procedure lasted 10 minutes. The patient was then
directed to the postanesthesia care unit for a 30-minute
observation period and was subsequently discharged to the
ward. With agreement of the surgeon, IV heparin was
resumed 2 hours after the end of the surgery and warfarin
was restarted on the first postoperative day. The patient
was followed daily and discharged from the hospital 6 days
later, after a therapeutic level of anticoagulation (interna-
tional normalized ratio 2.6) had been obtained. At that time,
the patient complained of pain in the right shoulder but a
physical examination did not reveal any abnormality.
Two weeks later, the patient consulted at emergency
room for severe pain in her right shoulder. A physical
examination demonstrated a significant swelling of the
anterior part of the shoulder, the axilla, and the upper part
of the right arm. There was no redness or discoloration of
the skin. A neurological examination of the upper arm did
not demonstrate any motor or sensory deficit but the range
of movement of the right shoulder was limited by severe
pain. The radial pulse was present at the right wrist. A
superficial sonogram of the upper part of the right arm
showed a solid hyperechogenic mass of 7.8 ⫻ 3.5 ⫻ 4.7 cm
containing a small amount of liquid. A Doppler study did
not reveal any flow in this mass. A computed tomographic
angiography showed a 5.4-cm hematoma located at the
axillo-humeral junction of the axillary artery within which

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 1069

 CASE REPORT
Figure 1. A, Angiography of the shoulder area before the emboliza-
tion; black arrow ⫽ 13-mm mycotic aneurysm; white arrow ⫽ 5-mm
mycotic aneurysm. B, Angiography performed after the embolization;
black arrow ⫽ previous site of the 13-mm mycotic aneurysm showing
complete exclusion of the aneurysm.
was a 20-mm pool of contrast dye suggesting an aneurysm
just underneath the wound left by a recent cutaneous
puncture below the right clavicle (consistent with the
puncture site of the infraclavicular block). A second aneu-
rysm was suspected in the deltoid area. Digital angiogra-
phy confirmed the presence of 2 aneurysms, 1 near the
humeral neck and the other in the prescapular region (5
and 13 mm). These were supplied by a right lateral branch
of the dorsoscapular artery and the right thoracoacromial
artery, respectively. Supraselective catheterization of the
feeding artery and embolization with liquid adhesive glue
(Indermil; Tyco, Norwalk, CT) and lipiodol was performed.
Control angiogram confirmed the occlusion of both aneu-
rysms (Fig. 1). The patient was discharged from the hospi-
1070 www.anesthesia-analgesia.org
tal the following day without any complication. Three
follow-up visits revealed a complete regression of the
swelling and the pain. One month later, the patient had no
sequel.
DISCUSSION
Mycotic aneurysms are defined as an infectious break in the
wall of an artery with formation of a blind saccular
outpouching that is contiguous with the arterial lumen,
usually in an area of bifurcation or narrowing. The aorta,
peripheral arteries, cerebral arteries, and visceral arteries
are involved in descending order of frequency.5 In the
pre–antibiotic era, ⬎85% of mycotic aneurysms were asso-
ciated with bacterial endocarditis. Currently, the majority
of mycotic aneurysms occur in IV drug users or after invasive
medical procedures. Depressed host immunity secondary to
systemic disease (diabetes, cirrhosis, collagen vascular dis-
ease) and corticosteroid therapy are also contributing fac-
tors.6 Only 15 mycotic aneurysms of the subclavian artery
have been reported since 1923.7 The diagnosis is often
difficult because of the insidious nature of the disease. Pain,
erythema, palpable mass, or ischemia distal to the affected
area is sometimes present.6 Computed tomographic an-
giography is the imaging modality of choice for evaluation
of mycotic aneurysms but Doppler sonography has both
good sensitivity and specificity for detection of mycotic
aneurysms located in peripheral arteries.8 The usual treat-
ment is surgical excision but endovascular techniques have
also been reported.9
One of the main advantages of ultrasound-guided nerve
blocks is the possibility of a real-time visualization of the
needle, nerve, and surrounding structures, notably the
vessels. It has been demonstrated that ultrasound guidance
diminishes the rate of vascular puncture during infracla-
vicular nerve block compared with a nerve stimulation
technique.10 Moreover, it sometimes allows the detection of
abnormal anatomy, thus offering the possibility of making
adjustments to the anesthetic technique planned.1,3 How-
ever, complications have not always been prevented by the
use of ultrasound guidance to perform nerve blockade.4,11
The location of the mycotic aneurysms just below the scar
left by the needle on the skin most likely suggests that in
our case 1 of the 2 preexisting mycotic aneurysms was
punctured by the needle during the technique. This acci-
dental puncture during the course of an apparently uncom-
plicated technique can occur in 2 different situations. First,
it is possible that the needle was not adequately visualized
by the anesthesiologist (one of the most frequent errors of
clinicians performing ultrasound-guided nerve block12)
and could have punctured the mycotic aneurysms just
outside the ultrasound visualization plane. It is also pos-
sible that the mycotic aneurysms were located within the
ultrasound visualization plane but were not recognized
and diagnosed, leading to a witnessed but unrecognized
puncture during the technique.
The use of color Doppler study during sonographic
examination results in a characteristic yin-yang sign in the
presence of mycotic aneurysms (Fig. 2), thus greatly en-
hancing the sensibility of ultrasound to detect mycotic
aneurysms.5 The blood stasis inside the mycotic aneurysms
produces an unusual gray ultrasound image of arterial
ANESTHESIA & ANALGESIA
Undiagnosed Upper Extremity Mycotic Aneurysms
Figure 2. Sonograms show a 5.5-cm complex lesion. On the color
Doppler image, the hypoechoic center has turbulent flow (“ying-yang”
sign), a finding indicative of a patent aneurysm lumen. The thick,
heterogeneous, hypoechoic rind is attributable to hematoma and
inflammatory tissue. (Reprinted with permission from Lee et al.5 [Lee
W-K, Mossop PJ, Little AF, et al. Infected (mycotic) aneurysms:
spectrum of imaging appearances and management. Radiographics
2008;28:1853– 68].)
lumen that might complicate its recognition when Doppler
is not used. In our case, a preprocedure Doppler study was
not performed before the technique. Preprocedure diagno-
sis of the mycotic aneurysms in the needle path could have
resulted in modification of the anesthetic technique (choice
of another block site, general anesthesia, use of smaller
needle, etc.) that could have prevented the occurrence of
the complication. However, it should be kept in mind that
ultrasound is a relatively recent technology in the anesthe-
siology field. Because of its rare use for diagnostic pur-
poses, certain rare medical conditions such as the one
reported herein may still go unrecognized.
In conclusion, this case report illustrates that the use of
ultrasound cannot completely eliminate the occurrence of
October 2010 • Volume 111 • Number 4
complications. Despite the advantages of this technology, a
high degree of suspicion should be maintained to minimize
the risk of complications.13 Finally, IV drug users or
patients with a history of endocarditis are at a high risk of
mycotic aneurysms and may benefit from a formal prepro-
cedure examination before a peripheral nerve block.14
AUTHOR CONTRIBUTIONS
All authors helped with manuscript preparation.
REFERENCES
1. Duggan E, Brull R, Lai J, Abbas S. Ultrasound-guided brachial
plexus block in a patient with multiple glomangiomatosis. Reg
Anesth Pain Med 2008;33:70 –3
2. Kessler J, Gray AT. Sonography of scalene muscle anomalies
for brachial plexus block. Reg Anesth Pain Med 2007;32:172–3
3. Sites BD, Spence BC, Gallagher JD, Beach ML. On the edge of
the ultrasound screen: regional anesthesiologists diagnosing
nonneural pathology. Reg Anesth Pain Med 2006;31:555– 62
4. Zetlaoui PJ, Labbe JP, Benhamou D. Ultrasound guidance for
axillary plexus block does not prevent intravascular injection.
Anesthesiology 2008;108:761
5. Lee W, Mossop PF, Little AF, Fitt GJ, Vrazas JI, Hoang JK,
Hennessy OF. Infected (mycotic) aneurysms: spectrum of
imaging appearances and management. Radiographics
2008;28:1853– 68
6. Kearney RA, Eisen HJ, Wolf JE. Nonvalvular infections of the
cardiovascular system. Ann Intern Med 1994;121:219 –30
7. Tsao JW, Marder SR, Goldstone J, Bloom AI. Presentation,
diagnosis, and management of arterial mycotic pseudoaneu-
rysms in injection drug users. Ann Vasc Surg 2002;16:652– 62
8. Coughlin BF, Paushter DM. Peripheral pseudoaneurysms:
evaluation with duplex US. Radiology 1988;168:339 – 42
9. Leon LR, Psalms SB, Labropoulos N, Mills JL. Infected upper
extremity aneurysms: a review. Eur J Vasc Endovasc Surg
2008;35:320 –31
10. Maalouf D, Gordon M, Paroli L, Tong-Ngork S. Ultrasound-
guidance vs. nerve stimulation for the infraclavicular blockade
of the brachial plexus: a comparison of the vascular puncture
rate. Reg Anesth Pain Med 2006;30:A46
11. Loubert C, Williams SR, Hélie F, Arcand G. Complication
during ultrasound-guided regional block: accidental intra-
vascular injection of local anesthetic. Anesthesiology
2008;108:759 – 60
12. Sites BD, Spence BC, Gallagher JD, Wiley CW, Bertrand ML,
Blike GT. Characterizing novice behaviour associated with
learning ultrasound-guided peripheral regional anesthesia.
Reg Anesth Pain Med 2007;32:107–15
13. Benitez PR, Newell MA. Vascular trauma in drug abuse:
patterns of injury. Ann Vasc Surg 1986;1:175– 81
14. Manickam BP, Perlas A, Chan VW, Brull R. The role of a
preprocedure systematic sonographic survey in ultrasound-
guided regional anesthesia. Reg Anesth Pain Med
2008;33:566 –70
www.anesthesia-analgesia.org 1071
 COCHRANE CORNER
Infraclavicular Brachial Plexus
Block for Regional Anaesthesia of
the Lower Arm
Ki Jinn Chin, Mandeep Singh, Veerabadran
Velayutham, and Victor Chee
BACKGROUND: Several approaches exist to produce local
anaesthetic blockade of the brachial plexus. It is not clear
which is the technique of choice for providing surgical anaes-
thesia of the lower arm although infraclavicular blockade (ICB)
has several purported advantages. We therefore performed a
systematic review of ICB compared to the other brachial plexus
blocks (BPBs).
OBJECTIVES: To evaluate the efficacy and safety of ICB
compared to other BPBs in providing regional anaesthesia of
the lower arm.
SEARCH STRATEGY: We searched CENTRAL (The Cochrane
Library 2008, Issue 3), MEDLINE (1950 to September 22nd
2008) and EMBASE (1980 to September 22nd 2008). We also
searched conference proceedings (from 2004 to 2008) and
the www.clinicaltrials.gov registry. No language restriction was
applied.
SELECTION CRITERIA: We included any randomized con-
trolled trials (RCTs) that compared ICB with other BPBs as the
sole anaesthetic techniques for surgery on the lower arm.
DATA COLLECTION AND ANALYSIS: The primary outcome
was adequate surgical anaesthesia within 30 minutes of block
completion. Secondary outcomes included sensory block of
individual nerves, tourniquet pain, onset time of sensory
blockade, block performance time, block-associated pain and
complications related to the block.
MAIN RESULTS: We identified 15 studies with 1020 partici-
pants, of whom 510 received ICB and 510 received other
BPBs. The control group intervention was the axillary block in
10 studies, mid-humeral block in two studies, supraclavicular
block in two studies and parascalene block in one study. Three
studies employed ultrasound-guided ICB. The risk of failed
surgical anaesthesia and of complications were low and similar
for ICB and all other BPBs. Tourniquet pain was less likely with
ICB (risk ratio (RR) 0.47, 95% CI 0.24 to 0.92, P ⫽ 0.03).
When compared to a single-injection axillary block, ICB was
better at providing complete sensory block of the musculocu-
taneous nerve (RR for failure 0.46, 95% CI 0.27 to 0.60, P ⬍
0.0001) and the axillary nerve (RR of failure 0.37, 95% CI 0.24
to 0.58, P ⬍ 0.0001). ICB was faster to perform than
multiple-injection axillary (mean difference (MD) ⫺2.7 min,
95% CI ⫺4.2 to ⫺1.1, P ⫽ 0.0006) or midhumeral blocks (MD
⫺4.8 min, 95% CI ⫺6.0 to ⫺3.6, P ⬍ 0.00001) but this was
offset by a longer sensory block onset time (MD 3.9 min, 95%
CI 3.2 to 4.5, P ⬍ 0.00001).
AUTHORS’ CONCLUSIONS: ICB is a safe and simple tech-
nique for providing surgical anaesthesia of the lower arm, with
an efficacy comparable to other BPBs. The advantages of ICB
include a lower likelihood of tourniquet pain during surgery, and
more reliable blockade of the musculocutaneous and axillary
nerves when compared to a single-injection axillary block. The
efficacy of ICB is likely to be improved if adequate time is
allowed for block onset (at least 30 minutes) and if a volume of
at least 40 ml is injected. Since publication of many of the
trials included in this review, it has become clear that a distal
posterior cord motor response is the appropriate endpoint for
electrostimulation-guided ICB; we recommend it be used in all
1072 www.anesthesia-analgesia.org
future comparative studies. There is also a need for additional
RCTs comparing ultrasound-guided ICB with other BPBs.
Chin KJ, Singh M, Velayutham V, Chee V. Infraclavicular
brachial plexus block for regional anaesthesia of the lower arm
published, in the Cochrane Database Syst Rev 2010, Issue 2. Art.
No.: CD005487.
DOI: 10.1002/14651858.CD005487.pub2
Heated Humidification Versus Heat
and Moisture Exchangers for
Ventilated Adults and Children
Margaret Kelly, Donna Gillies, David A. Todd, and
Catherine Lockwood
BACKGROUND: Humidification by artificial means must be
provided when the upper airway is bypassed during mechanical
ventilation. Heated humidification (HH) and heat and moisture
exchangers (HMEs) are the most commonly used types of
artificial humidification in this situation.
OBJECTIVES: To determine whether HHs or HMES are more
effective in preventing mortality and other complications in
people who are mechanically ventilated.
SEARCH STRATEGY: We searched the Cochrane Central
Register of Controlled Trials (The Cochrane Library 2010, Issue
4) and MEDLINE, EMBASE and CINAHL (January, 2010) to
identify relevant randomized controlled trials.
SELECTION CRITERIA: We included randomized controlled tri-
als comparing HMEs to HHs in mechanically ventilated adults and
children. We included randomized crossover studies.
DATA COLLECTION AND ANALYSIS: We assessed the quality
of each study and extracted the relevant data. Where appro-
priate, results from relevant studies were meta-analyzed for
individual outcomes.
MAIN RESULTS: We included 33 trials with 2833 participants;
25 studies were parallel group design (n ⫽ 2710) and 8
crossover design (n ⫽ 123). Only 3 included studies reported
data for infants or children. There was no overall effect on
artificial airway occlusion, mortality, pneumonia, or respiratory
complications; however, the PaCO2 and minute ventilation
were increased when HMEs were compared to HHs and body
temperature was lower. The cost of HMEs was lower in all
studies that reported this outcome. There was some evidence
that hydrophobic HMEs may reduce the risk of pneumonia and
that blockages of artificial airways may be increased with the
use of HMEs in certain subgroups of patients.
AUTHORS’ CONCLUSIONS: There is little evidence of an
overall difference between HMEs and HHs. However, hydropho-
bic HMEs may reduce the risk of pneumonia and the use of an
HMEs may increase artificial airway occlusion in certain sub-
groups of patients. Therefore, HMEs may not be suitable for
patients with limited respiratory reserve or prone to airway
blockage. Further research is needed relating to hydrophobic
versus hygroscopic HMEs and the use of HMEs in the pediatric
and neonatal populations. As the design of HMEs evolves,
evaluation of new generation HMEs will also need to be
undertaken.
Kelly M, Gillies D, Todd DA, Lockwood C. Heated humidifi-
cation versus heat and moisture exchangers for ventilated
adults and children. Cochrane Database Syst Rev 2010, Issue
4. Art. No.: CD004711.
DOI: 10.1002/14651858.CD004711.pub2.
October 2010 • Volume 111 • Number 4
 COCHRANE CORNER
Infraclavicular Brachial Plexus
Block for Regional Anaesthesia of
the Lower Arm
Ki Jinn Chin, Mandeep Singh, Veerabadran
Velayutham, and Victor Chee
BACKGROUND: Several approaches exist to produce local
anaesthetic blockade of the brachial plexus. It is not clear
which is the technique of choice for providing surgical anaes-
thesia of the lower arm although infraclavicular blockade (ICB)
has several purported advantages. We therefore performed a
systematic review of ICB compared to the other brachial plexus
blocks (BPBs).
OBJECTIVES: To evaluate the efficacy and safety of ICB
compared to other BPBs in providing regional anaesthesia of
the lower arm.
SEARCH STRATEGY: We searched CENTRAL (The Cochrane
Library 2008, Issue 3), MEDLINE (1950 to September 22nd
2008) and EMBASE (1980 to September 22nd 2008). We also
searched conference proceedings (from 2004 to 2008) and
the www.clinicaltrials.gov registry. No language restriction was
applied.
SELECTION CRITERIA: We included any randomized con-
trolled trials (RCTs) that compared ICB with other BPBs as the
sole anaesthetic techniques for surgery on the lower arm.
DATA COLLECTION AND ANALYSIS: The primary outcome
was adequate surgical anaesthesia within 30 minutes of block
completion. Secondary outcomes included sensory block of
individual nerves, tourniquet pain, onset time of sensory
blockade, block performance time, block-associated pain and
complications related to the block.
MAIN RESULTS: We identified 15 studies with 1020 partici-
pants, of whom 510 received ICB and 510 received other
BPBs. The control group intervention was the axillary block in
10 studies, mid-humeral block in two studies, supraclavicular
block in two studies and parascalene block in one study. Three
studies employed ultrasound-guided ICB. The risk of failed
surgical anaesthesia and of complications were low and similar
for ICB and all other BPBs. Tourniquet pain was less likely with
ICB (risk ratio (RR) 0.47, 95% CI 0.24 to 0.92, P ⫽ 0.03).
When compared to a single-injection axillary block, ICB was
better at providing complete sensory block of the musculocu-
taneous nerve (RR for failure 0.46, 95% CI 0.27 to 0.60, P ⬍
0.0001) and the axillary nerve (RR of failure 0.37, 95% CI 0.24
to 0.58, P ⬍ 0.0001). ICB was faster to perform than
multiple-injection axillary (mean difference (MD) ⫺2.7 min,
95% CI ⫺4.2 to ⫺1.1, P ⫽ 0.0006) or midhumeral blocks (MD
⫺4.8 min, 95% CI ⫺6.0 to ⫺3.6, P ⬍ 0.00001) but this was
offset by a longer sensory block onset time (MD 3.9 min, 95%
CI 3.2 to 4.5, P ⬍ 0.00001).
AUTHORS’ CONCLUSIONS: ICB is a safe and simple tech-
nique for providing surgical anaesthesia of the lower arm, with
an efficacy comparable to other BPBs. The advantages of ICB
include a lower likelihood of tourniquet pain during surgery, and
more reliable blockade of the musculocutaneous and axillary
nerves when compared to a single-injection axillary block. The
efficacy of ICB is likely to be improved if adequate time is
allowed for block onset (at least 30 minutes) and if a volume of
at least 40 ml is injected. Since publication of many of the
trials included in this review, it has become clear that a distal
posterior cord motor response is the appropriate endpoint for
electrostimulation-guided ICB; we recommend it be used in all
1072 www.anesthesia-analgesia.org
future comparative studies. There is also a need for additional
RCTs comparing ultrasound-guided ICB with other BPBs.
Chin KJ, Singh M, Velayutham V, Chee V. Infraclavicular
brachial plexus block for regional anaesthesia of the lower arm
published, in the Cochrane Database Syst Rev 2010, Issue 2. Art.
No.: CD005487.
DOI: 10.1002/14651858.CD005487.pub2
Heated Humidification Versus Heat
and Moisture Exchangers for
Ventilated Adults and Children
Margaret Kelly, Donna Gillies, David A. Todd, and
Catherine Lockwood
BACKGROUND: Humidification by artificial means must be
provided when the upper airway is bypassed during mechanical
ventilation. Heated humidification (HH) and heat and moisture
exchangers (HMEs) are the most commonly used types of
artificial humidification in this situation.
OBJECTIVES: To determine whether HHs or HMES are more
effective in preventing mortality and other complications in
people who are mechanically ventilated.
SEARCH STRATEGY: We searched the Cochrane Central
Register of Controlled Trials (The Cochrane Library 2010, Issue
4) and MEDLINE, EMBASE and CINAHL (January, 2010) to
identify relevant randomized controlled trials.
SELECTION CRITERIA: We included randomized controlled tri-
als comparing HMEs to HHs in mechanically ventilated adults and
children. We included randomized crossover studies.
DATA COLLECTION AND ANALYSIS: We assessed the quality
of each study and extracted the relevant data. Where appro-
priate, results from relevant studies were meta-analyzed for
individual outcomes.
MAIN RESULTS: We included 33 trials with 2833 participants;
25 studies were parallel group design (n ⫽ 2710) and 8
crossover design (n ⫽ 123). Only 3 included studies reported
data for infants or children. There was no overall effect on
artificial airway occlusion, mortality, pneumonia, or respiratory
complications; however, the PaCO2 and minute ventilation
were increased when HMEs were compared to HHs and body
temperature was lower. The cost of HMEs was lower in all
studies that reported this outcome. There was some evidence
that hydrophobic HMEs may reduce the risk of pneumonia and
that blockages of artificial airways may be increased with the
use of HMEs in certain subgroups of patients.
AUTHORS’ CONCLUSIONS: There is little evidence of an
overall difference between HMEs and HHs. However, hydropho-
bic HMEs may reduce the risk of pneumonia and the use of an
HMEs may increase artificial airway occlusion in certain sub-
groups of patients. Therefore, HMEs may not be suitable for
patients with limited respiratory reserve or prone to airway
blockage. Further research is needed relating to hydrophobic
versus hygroscopic HMEs and the use of HMEs in the pediatric
and neonatal populations. As the design of HMEs evolves,
evaluation of new generation HMEs will also need to be
undertaken.
Kelly M, Gillies D, Todd DA, Lockwood C. Heated humidifi-
cation versus heat and moisture exchangers for ventilated
adults and children. Cochrane Database Syst Rev 2010, Issue
4. Art. No.: CD004711.
DOI: 10.1002/14651858.CD004711.pub2.
October 2010 • Volume 111 • Number 4
 LETTERS TO THE EDITOR
Airway Management: Standardization,
Simplicity, and Daily Practice Are the
Keys to Success
To the Editor
T he editorial by Hung and Murphy,1 although inter-
esting, contains potentially misleading statements.
Some have opined that careful evaluation of the
airway as part of a preplanned strategy may lead to
improved outcome.2 However, it is potentially dangerous
to suggest that the choice of the technique and hence the
choice of the equipment is or should be influenced primar-
ily by different circumstances. Advocating such an ap-
proach would result in multiple strategies using many
different and, at times, unfamiliar airway devices.
Only a few studies have focused on effective airway
management and, in these, outcome was investigated un-
der the prevailing clinical conditions.3–5 A common char-
acteristic of each of these studies was a limitation of
techniques and devices and that deviation from the pre-
defined algorithm was recorded infrequently.
A key factor regarding safety recorded by highly reliable
organizations such as aviation is a standardized process.6
However, the “recommendations” of Hung and Murphy
that airway management is primarily “context sensitive”
would guide us in the wrong direction. There are only a
few situations wherein we might deviate from our difficult
airway guidelines. For example, it is very unlikely that you
can perform an awake intubation in an uncooperative
patient. Importantly, such situations should be managed by
the most experienced physicians. However, if deviation
from, rather than adherence to, the guidelines is current
practice, the guidelines should be modified accordingly.
A second point is the missing commitment to fiberoptic
intubation (“unwritten truth”). If the authors mean that
there are no prospective randomized studies showing the
effectiveness of fiberoptic intubation, they are correct. Un-
fortunately, there are no prospective randomized studies
demonstrating the effectiveness (not efficacy) of most tech-
niques used in daily practice. However, it is generally
agreed among airway management practitioners and rec-
ommended by many anesthesia societies that fiberoptic
intubation should be used for management of the antici-
pated difficult airway.7 So, questioning this technique is
probably potentially misleading for the average anesthesi-
ologist in practice.
Regarding airway management, the message should be:
Standardization, simplicity, and daily practice are the keys
to success.
Thomas Heidegger, MD
Department of Anesthesia
Spitalregion Rheintal Werdenberg Sarganserland
Walenstadt, Switzerland
thomas.heidegger@srrws.ch
October 2010 • Volume 111 • Number 4
Section Editor: Lawrence Saidman
REFERENCES
1. Hung O, Murphy M. Context-sensitive airway management.
Anesth Analg 2010;110:982–3
2. American Society of Anesthesiologists Task Force on Manage-
ment of the Difficult Airway (2003). Practice guidelines for
management of the difficult airway: an updated report by the
American Society of Anesthesiologists Task Force on Manage-
ment of the Difficult Airway. Anesthesiology 2003;98:1269 –77
3. Hopkins A. Measuring the Quality of Medical Care. 1st ed.
Oxford: Royal College of Physicians of London, 1990
4. Combes X, Le Roux B, Suen P, Dumerat M, Motamed C, Sauvat
S, Duvaldestin P, Dhonneur G. Unanticipated difficult airway in
anesthetized patients: prospective validation of a management
algorithm. Anesthesiology 2004;100:1146 –50
5. Heidegger T, Gerig HJ, Ulrich B, Kreienbühl G. Validation of a
simple algorithm for tracheal intubation: daily practice is the
key to success in emergencies—an analysis of 13,248 intuba-
tions. Anesth Analg 2001;92:517–22
6. Reason J. Human error: models and management. BMJ
200;320:768 –70
7. Heidegger T, Gerig HJ, Henderson JJ. Strategies and algorithms
for management of the difficult airway. Best Pract Res Clin
Anaesthesiol 2005;19:661–74
DOI: 10.1213/ANE.0b013e3181ec312a
In Response
No one disagrees that the fundamental goal of airway man-
agement is oxygenation and ventilation, not devices and tools.
A systematic approach to airway management includes air-
way evaluation, selection of an appropriate course of action
likely to succeed (Plan A) and preparation for failure (i.e., Plan
B, Plan C, etc.). This approach is, and must be, consistent with
the “context.” In other words, one must accept that airways
present themselves in a variety of forms, circumstances, and
locations, to a panoply of health care providers with varying
skill sets. Although there is a recommended “strategy,” there
are varying tactics or techniques one may use. The tactic or
technique is virtually always dependent on the circumstances
and skill set of the airway practitioner.
Apart from the oxygenator of a bypass pump, clinicians
basically use only 4 methods of ventilation and oxygenation:
through a bag-mask, an extraglottic device (e.g., a laryngeal
mask airway), a tracheal tube, or a surgical airway. As stated
in our editorial, “context-sensitive airway management im-
plies that managing a difficult or failed airway should be
driven by the principles of ‘gas exchange’ and not be ‘device-
dependent.’”1 In addition, we also stated that “clinicians must
be trained to understand the basic principles of airway
management using basic techniques and learn how to apply
these techniques properly in an appropriate environment.”
These statements were structured to inform clinicians that
they must learn basic airway techniques and apply them
properly. Nowhere in the editorial did we advocate that “…
an approach would result in multiple strategies using many
different and at times, unfamiliar airway devices.”
Dr. Heidegger correctly states that awake tracheal intu-
bation is considered to be the most prudent approach in a
patient with an anticipated difficult airway.2 But, in addi-
tion to the flexible bronchoscope, awake intubation can be
performed safely and effectively by many techniques, in-
cluding the rigid fiberoptic laryngoscopes, videolaryngo-
scopes, and Macintosh laryngoscope with the Eschmann
www.anesthesia-analgesia.org 1073
 LETTERS TO THE EDITOR
Airway Management: Standardization,
Simplicity, and Daily Practice Are the
Keys to Success
To the Editor
T he editorial by Hung and Murphy,1 although inter-
esting, contains potentially misleading statements.
Some have opined that careful evaluation of the
airway as part of a preplanned strategy may lead to
improved outcome.2 However, it is potentially dangerous
to suggest that the choice of the technique and hence the
choice of the equipment is or should be influenced primar-
ily by different circumstances. Advocating such an ap-
proach would result in multiple strategies using many
different and, at times, unfamiliar airway devices.
Only a few studies have focused on effective airway
management and, in these, outcome was investigated un-
der the prevailing clinical conditions.3–5 A common char-
acteristic of each of these studies was a limitation of
techniques and devices and that deviation from the pre-
defined algorithm was recorded infrequently.
A key factor regarding safety recorded by highly reliable
organizations such as aviation is a standardized process.6
However, the “recommendations” of Hung and Murphy
that airway management is primarily “context sensitive”
would guide us in the wrong direction. There are only a
few situations wherein we might deviate from our difficult
airway guidelines. For example, it is very unlikely that you
can perform an awake intubation in an uncooperative
patient. Importantly, such situations should be managed by
the most experienced physicians. However, if deviation
from, rather than adherence to, the guidelines is current
practice, the guidelines should be modified accordingly.
A second point is the missing commitment to fiberoptic
intubation (“unwritten truth”). If the authors mean that
there are no prospective randomized studies showing the
effectiveness of fiberoptic intubation, they are correct. Un-
fortunately, there are no prospective randomized studies
demonstrating the effectiveness (not efficacy) of most tech-
niques used in daily practice. However, it is generally
agreed among airway management practitioners and rec-
ommended by many anesthesia societies that fiberoptic
intubation should be used for management of the antici-
pated difficult airway.7 So, questioning this technique is
probably potentially misleading for the average anesthesi-
ologist in practice.
Regarding airway management, the message should be:
Standardization, simplicity, and daily practice are the keys
to success.
Thomas Heidegger, MD
Department of Anesthesia
Spitalregion Rheintal Werdenberg Sarganserland
Walenstadt, Switzerland
thomas.heidegger@srrws.ch
October 2010 • Volume 111 • Number 4
Section Editor: Lawrence Saidman
REFERENCES
1. Hung O, Murphy M. Context-sensitive airway management.
Anesth Analg 2010;110:982–3
2. American Society of Anesthesiologists Task Force on Manage-
ment of the Difficult Airway (2003). Practice guidelines for
management of the difficult airway: an updated report by the
American Society of Anesthesiologists Task Force on Manage-
ment of the Difficult Airway. Anesthesiology 2003;98:1269 –77
3. Hopkins A. Measuring the Quality of Medical Care. 1st ed.
Oxford: Royal College of Physicians of London, 1990
4. Combes X, Le Roux B, Suen P, Dumerat M, Motamed C, Sauvat
S, Duvaldestin P, Dhonneur G. Unanticipated difficult airway in
anesthetized patients: prospective validation of a management
algorithm. Anesthesiology 2004;100:1146 –50
5. Heidegger T, Gerig HJ, Ulrich B, Kreienbühl G. Validation of a
simple algorithm for tracheal intubation: daily practice is the
key to success in emergencies—an analysis of 13,248 intuba-
tions. Anesth Analg 2001;92:517–22
6. Reason J. Human error: models and management. BMJ
200;320:768 –70
7. Heidegger T, Gerig HJ, Henderson JJ. Strategies and algorithms
for management of the difficult airway. Best Pract Res Clin
Anaesthesiol 2005;19:661–74
DOI: 10.1213/ANE.0b013e3181ec312a
In Response
No one disagrees that the fundamental goal of airway man-
agement is oxygenation and ventilation, not devices and tools.
A systematic approach to airway management includes air-
way evaluation, selection of an appropriate course of action
likely to succeed (Plan A) and preparation for failure (i.e., Plan
B, Plan C, etc.). This approach is, and must be, consistent with
the “context.” In other words, one must accept that airways
present themselves in a variety of forms, circumstances, and
locations, to a panoply of health care providers with varying
skill sets. Although there is a recommended “strategy,” there
are varying tactics or techniques one may use. The tactic or
technique is virtually always dependent on the circumstances
and skill set of the airway practitioner.
Apart from the oxygenator of a bypass pump, clinicians
basically use only 4 methods of ventilation and oxygenation:
through a bag-mask, an extraglottic device (e.g., a laryngeal
mask airway), a tracheal tube, or a surgical airway. As stated
in our editorial, “context-sensitive airway management im-
plies that managing a difficult or failed airway should be
driven by the principles of ‘gas exchange’ and not be ‘device-
dependent.’”1 In addition, we also stated that “clinicians must
be trained to understand the basic principles of airway
management using basic techniques and learn how to apply
these techniques properly in an appropriate environment.”
These statements were structured to inform clinicians that
they must learn basic airway techniques and apply them
properly. Nowhere in the editorial did we advocate that “…
an approach would result in multiple strategies using many
different and at times, unfamiliar airway devices.”
Dr. Heidegger correctly states that awake tracheal intu-
bation is considered to be the most prudent approach in a
patient with an anticipated difficult airway.2 But, in addi-
tion to the flexible bronchoscope, awake intubation can be
performed safely and effectively by many techniques, in-
cluding the rigid fiberoptic laryngoscopes, videolaryngo-
scopes, and Macintosh laryngoscope with the Eschmann
www.anesthesia-analgesia.org 1073
 LETTERS TO THE EDITOR
Tracheal Introducer. Dr. Heidegger also correctly states
that “. . . it is generally agreed among airway management
practitioners and recommended by many anesthesia soci-
eties that fiberoptic intubation should be used for manage-
ment of the anticipated difficult airway.” However, it
would be grossly incorrect and perhaps dangerous for Dr.
Heidegger to imply that “awake fiberoptic intubation
should be standardized in the management of the antici-
pated difficult airway.” In an uncooperative patient, in the
presence of blood, in emergency situations, or in an envi-
ronment with limited resources, fiberoptic intubation
would be difficult, if not impossible.
Having taught airway management to thousands of
practitioners, we recognize that there is tremendous vari-
ability within individual skill sets. Because of this variabil-
ity, the principles of airway management and the strategy
for managing a difficult airway have been well elucidated
by guidelines promulgated by various societies including
the American Society of Anesthesiologists.3,4 It is critical for
all practitioners to recognize that these guidelines are not
recipes to be rigidly followed. Rather, as stated clearly in
the ASA Practice Guidelines for Management of the Diffi-
cult Airway,3,4 “. . . these recommendations may be adopted,
modified, or rejected according to clinical needs and constraints”
(context-sensitive). Furthermore, “Practice guidelines are
not intended as standards or absolute requirements. The
use of practice guidelines cannot guarantee any specific
outcome. Practice guidelines are subject to revision as
warranted by the evolution of medical knowledge, technol-
ogy, and practice.”3,4
Orlando Hung, MD
Professor Anesthesiology, Surgery, and Pharmacology
Dalhousie University
Queen Elizabeth II Health Sciences
Halifax, Nova Scotia, Canada
hungorla@dal.ca
Michael Murphy, MD
Professor and Chair Anesthesiology
Professor Emergency Medicine
Dalhousie University
District Chief Anesthesiology
Capital District Health Authority
Queen Elizabeth II Health Sciences
Halifax, Nova Scotia, Canada
murphymf1@gmail.com
REFERENCES
1. Hung O, Murphy M. Context-sensitive airway management.
Anesth Analg 2010;110:982–3
2. Heidegger T. Airway management: standardization, simplicity,
and daily practice are the keys to success. Anesth Analg
2010;111:1073
3. Practice guidelines for management of the difficult airway: a
report by the American Society of Anesthesiologists Task Force
on Management of the Difficult Airway. Anesthesiology
1993;78:597– 602
4. American Society of Anesthesiologists Task Force on Manage-
ment of the Difficult Airway. Practice guidelines for manage-
ment of the difficult airway: an updated report by the American
Society of Anesthesiologists Task Force on Management of the
Difficult Airway. Anesthesiology 2003;98:1269 –77
DOI: 10.1213/ANE.0b013e3181ec3153
1074 www.anesthesia-analgesia.org
Achieving Full Risk Disclosure in
Pediatric Anesthesia Research
To the Editor
H ong et al.’s1 study using fluoroscopy on 73 ASA
physical status I children, ages 1 to 5 years, to
assess ropivacaine–radiopaque dye solution spread
in caudal injections for postorchiopexy pain raises a vexing
question: Were all known risks fully disclosed to their
subjects’ parents? The absence of details about fluoroscopy
equipment, radiation reduction measures, radiation doses,
and subject radiation risks suggests that they were not.
Strauss and Kaste2 stated (and Linet et al.3 concurred)
that the ALARA concept means radiation should be “As
Low As Reasonably Achievable.” Children, they noted,
“might be as much as 10 times more radiosensitive than
adults.” Cohen4 compared 2 fluoroscopic machines and
noted that “the radiologist can greatly increase patient
exposure by merely altering settings over which they have
immediate control” and cited exposure variations of 3486%
and 4479% on the basis of such factors. Petterson et al.5
found a 2.23 testicular cancer relative risk for children with
undescended testes operated on before age 13. Presumably,
their risk is augmented by radiation exposure.
Hong et al. experimented to answer a clinical question.
Their study conferred no additional benefit to subjects, only
a potential of harm. Science alone benefited. Beecher6
warned researchers away from such practices. In research
as in clinical practice, Primum Non Nocere still applies.
Vincent J. Kopp, MD
Michael G. Danekas, MD
Division of Pediatric Anesthesia
Department of Anesthesiology
School of Medicine
University of North Carolina at Chapel Hill
Chapel Hill, North Carolina
vkopp@aims.unc.edu
REFERENCES
1. Hong J-Y, Han SW, Kim WO, Cho JS, Kil HJ. A comparison of
high volume/low concentration and low volume/high concen-
tration ropivacaine in caudal analgesia for pediatric orchiopexy.
Anesth Analg 2009;109:1073– 8
2. Strauss KJ, Kaste SC. The ALARA concept in pediatric interven-
tional fluoroscopic imaging: striving to keep radiation doses as
low as possible during fluoroscopy of pediatric patients. A
white paper executive summary. AJR 2006;187:818 –9
3. Linet MS, Kim KP, Rajaraman P. Children’s exposure to diag-
nostic medical radiation and cancer risk: epidemiologic and
dosimetric considerations. Pediatr Radiol 2009;39(Suppl 1):S4 –26;
Epub 2008 Dec 16
4. Cohen M. Are we doing enough to minimize fluoroscopic
radiation exposure in children? Pediatr Radiol 2007;37:1020 – 4
5. Petterson A, Richardi L, Nordensklod, Kaijser M, Akre O. Age
at surgery for undescended testis and risk of testicular cancer.
N Engl J Med 2007;356:1835– 41
6. Beecher HK. Ethics and clinical research. N Engl J Med
1966;274:367–72
DOI: 10.1213/ANE.0b013e3181ed17ff
ANESTHESIA & ANALGESIA
 LETTERS TO THE EDITOR
Tracheal Introducer. Dr. Heidegger also correctly states
that “. . . it is generally agreed among airway management
practitioners and recommended by many anesthesia soci-
eties that fiberoptic intubation should be used for manage-
ment of the anticipated difficult airway.” However, it
would be grossly incorrect and perhaps dangerous for Dr.
Heidegger to imply that “awake fiberoptic intubation
should be standardized in the management of the antici-
pated difficult airway.” In an uncooperative patient, in the
presence of blood, in emergency situations, or in an envi-
ronment with limited resources, fiberoptic intubation
would be difficult, if not impossible.
Having taught airway management to thousands of
practitioners, we recognize that there is tremendous vari-
ability within individual skill sets. Because of this variabil-
ity, the principles of airway management and the strategy
for managing a difficult airway have been well elucidated
by guidelines promulgated by various societies including
the American Society of Anesthesiologists.3,4 It is critical for
all practitioners to recognize that these guidelines are not
recipes to be rigidly followed. Rather, as stated clearly in
the ASA Practice Guidelines for Management of the Diffi-
cult Airway,3,4 “. . . these recommendations may be adopted,
modified, or rejected according to clinical needs and constraints”
(context-sensitive). Furthermore, “Practice guidelines are
not intended as standards or absolute requirements. The
use of practice guidelines cannot guarantee any specific
outcome. Practice guidelines are subject to revision as
warranted by the evolution of medical knowledge, technol-
ogy, and practice.”3,4
Orlando Hung, MD
Professor Anesthesiology, Surgery, and Pharmacology
Dalhousie University
Queen Elizabeth II Health Sciences
Halifax, Nova Scotia, Canada
hungorla@dal.ca
Michael Murphy, MD
Professor and Chair Anesthesiology
Professor Emergency Medicine
Dalhousie University
District Chief Anesthesiology
Capital District Health Authority
Queen Elizabeth II Health Sciences
Halifax, Nova Scotia, Canada
murphymf1@gmail.com
REFERENCES
1. Hung O, Murphy M. Context-sensitive airway management.
Anesth Analg 2010;110:982–3
2. Heidegger T. Airway management: standardization, simplicity,
and daily practice are the keys to success. Anesth Analg
2010;111:1073
3. Practice guidelines for management of the difficult airway: a
report by the American Society of Anesthesiologists Task Force
on Management of the Difficult Airway. Anesthesiology
1993;78:597– 602
4. American Society of Anesthesiologists Task Force on Manage-
ment of the Difficult Airway. Practice guidelines for manage-
ment of the difficult airway: an updated report by the American
Society of Anesthesiologists Task Force on Management of the
Difficult Airway. Anesthesiology 2003;98:1269 –77
DOI: 10.1213/ANE.0b013e3181ec3153
1074 www.anesthesia-analgesia.org
Achieving Full Risk Disclosure in
Pediatric Anesthesia Research
To the Editor
H ong et al.’s1 study using fluoroscopy on 73 ASA
physical status I children, ages 1 to 5 years, to
assess ropivacaine–radiopaque dye solution spread
in caudal injections for postorchiopexy pain raises a vexing
question: Were all known risks fully disclosed to their
subjects’ parents? The absence of details about fluoroscopy
equipment, radiation reduction measures, radiation doses,
and subject radiation risks suggests that they were not.
Strauss and Kaste2 stated (and Linet et al.3 concurred)
that the ALARA concept means radiation should be “As
Low As Reasonably Achievable.” Children, they noted,
“might be as much as 10 times more radiosensitive than
adults.” Cohen4 compared 2 fluoroscopic machines and
noted that “the radiologist can greatly increase patient
exposure by merely altering settings over which they have
immediate control” and cited exposure variations of 3486%
and 4479% on the basis of such factors. Petterson et al.5
found a 2.23 testicular cancer relative risk for children with
undescended testes operated on before age 13. Presumably,
their risk is augmented by radiation exposure.
Hong et al. experimented to answer a clinical question.
Their study conferred no additional benefit to subjects, only
a potential of harm. Science alone benefited. Beecher6
warned researchers away from such practices. In research
as in clinical practice, Primum Non Nocere still applies.
Vincent J. Kopp, MD
Michael G. Danekas, MD
Division of Pediatric Anesthesia
Department of Anesthesiology
School of Medicine
University of North Carolina at Chapel Hill
Chapel Hill, North Carolina
vkopp@aims.unc.edu
REFERENCES
1. Hong J-Y, Han SW, Kim WO, Cho JS, Kil HJ. A comparison of
high volume/low concentration and low volume/high concen-
tration ropivacaine in caudal analgesia for pediatric orchiopexy.
Anesth Analg 2009;109:1073– 8
2. Strauss KJ, Kaste SC. The ALARA concept in pediatric interven-
tional fluoroscopic imaging: striving to keep radiation doses as
low as possible during fluoroscopy of pediatric patients. A
white paper executive summary. AJR 2006;187:818 –9
3. Linet MS, Kim KP, Rajaraman P. Children’s exposure to diag-
nostic medical radiation and cancer risk: epidemiologic and
dosimetric considerations. Pediatr Radiol 2009;39(Suppl 1):S4 –26;
Epub 2008 Dec 16
4. Cohen M. Are we doing enough to minimize fluoroscopic
radiation exposure in children? Pediatr Radiol 2007;37:1020 – 4
5. Petterson A, Richardi L, Nordensklod, Kaijser M, Akre O. Age
at surgery for undescended testis and risk of testicular cancer.
N Engl J Med 2007;356:1835– 41
6. Beecher HK. Ethics and clinical research. N Engl J Med
1966;274:367–72
DOI: 10.1213/ANE.0b013e3181ed17ff
ANESTHESIA & ANALGESIA
Letters to the Editor
In Response
We agree with Kopp et al.1 that the ALARA (as low as
reasonably achievable) concept must be considered in
pediatric fluoroscopic imaging. Of course, in our study2 we
obtained consent from the parents of the children after
explaining details of the study and radiation exposure. The
genital area was protected from radiation by a lead plate
during fluoroscopy using a new pulsed unit from Phillips
(Eindhoven, The Netherlands). The study by Cohen3 de-
scribed possible radiation exposure variations as much as
3486% and 4479% with the maximum magnification, great-
est pulse rate, intensifier high, grid in, and greatest inten-
sifier exposure rate. We believe that the exposure variations
were not as great because we did not alter the basal
magnification setting and used the lowest pulse rate, grid
out, and lowest intensifier exposure rate. The exposure
frequency to obtain 1 image (exposure rate) was 1 or 2, and
overall duration of fluoroscopy was ⬍30 seconds. How-
ever, because the average estimated entrance surface dose
(ESD) and dose area product increases with age, the ESD
also increases in children in comparison with adults4; we
investigated the ESD in 5 randomly selected children in our
previous study.5 The ESD was 0.05 to 0.26 mSv, which was
less than the results of a United Kingdom nationwide
survey in 2005.6 Although our procedures did not diverge
from the ALARA concept when we substituted the proce-
dure to the flow diagram for managing patient dose by
Strauss and Kaste,7 we believe that the radiation exposure
should be strictly limited in children.
Hae K. Kil, MD
Jeong Y. Hong, MD
Won O. Kim, MD
Department of Anesthesiology and Pain Medicine
Yonsei University College of Medicine
Seoul, South Korea
hkkil@yuhs.ac
REFERENCES
1. Kopp VJ, Danekas MG. Achieving full risk disclosure in pedi-
atric anesthesia research. Anesth Analg 2010;111:1074
2. Hong JY, Han SW, Kim WO, Cho JS, Kil HK. A comparison of
high volume/low concentration and low volume/high concen-
tration ropivacaine in caudal analgesia for pediatric orchiopexy.
Anesth Analg 2009;109:1073– 8
3. Cohen M. Are we doing enough to minimize fluoroscopic
radiation exposure in children? Pediatr Radiol 2007;37:
1020 – 4
4. Linet MS, Kim KP, Rajaraman R. Children’s exposure to diagnostic
medical radiation and cancer risk: epidemiologic and dosimetric
considerations. Pediatr Radiol 2009;39(Suppl 1):S4 –26
5. Shin SK, Hong JY, Kim WO, Koo BN, Kim JE, Kil HK. Ultra-
sound evaluation of the acral area and comparison of sacral
interspinous and hiatal approach for caudal block in children.
Anesthesiology 2009;111:1135– 40
6. Hart D, Hillier MC, Wall BF. National reference doses for
common radiographic, fluoroscopic and dental X-ray examina-
tions in the UK. Br J Radiol 2009;82:1–12
7. Strauss KJ, Kaste SC. The ALARA (as low as reasonably
achievable) concept in pediatric interventional and fluoroscopic
imaging: striving to keep radiation doses as low as possible
during fluoroscopy of pediatric patients. A white paper execu-
tive summary. Pediatr Radiol 2006;36(Suppl 2):110 –2
DOI: 10.1213/ANE.0b013e3181ed1811
October 2010 • Volume 111 • Number 4
Pulse Dye Densitometry and
Indocyanine Green Plasma
Disappearance: The Issue of
“Normal” Values
To the Editor
R eekers et al.1 investigated normal values of plasma
disappearance rate of indocyanine green (PDR-ICG)
measured by pulse dye densitometry in a population
of ASA physical status I and II patients, assumed free of
liver disease. They also evaluated the noninvasive mea-
surement of PDR-ICG using a nose and finger probe and
compared these results with those using invasive arterial
blood measurements.
Reekers et al. confirmed that, as has been shown before,
PDR-ICG is adequately measured noninvasively by pulse
dye densitometry.2,3 In addition, they found a mean PDR-
ICG value of 23.1%/min (SD ⫾ 7.9%/min), which is in line
with previous studies.4 – 6 However, they found that PDR-
ICG varied widely with a range from 9.7%/min to
43.2%/min. This is in contrast to earlier publications in
which PDR-ICG ranged from 18.7%/min to 30.1%/min.4 – 6
As explained by the authors, a PDR-ICG ⬍18%/min is
generally considered to indicate impaired hepatic function.5,7
Although the authors argue that only patients without overt
liver pathology were examined, there is a lack of objective
evidence of normal hepatic function in the entire population
studied. Biochemical liver function tests were available in
only 22% of the patients. The authors recognize this limitation
and argue that the patients studied are representative of a
population with normal liver function based on the absence of
any clinical evidence of hepatic dysfunction. This is a rational
argument, but the clinical evidence for normal liver function is
not a surrogate for laboratory testing because patients with
elevated liver enzymes do not necessarily present clinical
signs of hepatic dysfunction.8 Because PDR-ICG is very
sensitive for hepatic dysfunction, changes in PDR-ICG can
even precede changes in serum bilirubin, as the authors
stated.1 Given the fact that the authors ultimately want to
argue for a testing threshold to identify normal liver function,
the study would have been more compelling if all patients
had received a careful hepatic testing.
Unfortunately, Reekers et al. also did not present any data
describing hemodynamics of their patients. This is, however,
of utmost importance, because PDR-ICG is, as is ICG liver
extraction capacity, primarily dependent on splanchnic blood
flow and total plasma volume.2,9 It is also not clear why
Reekers et al. studied both awake and anesthetized patients.
There are clearly potential differences between the groups
with respect to hemodynamics, which may influence the
obtained PDR-ICG values. Reekers et al. argue that the results
in the anesthetized patients were not different from the awake
patients when analyzed separately so that the data were
pooled. However, it is well known that induction of anesthe-
sia with propofol results in a significant decrease in cardiac
output decreasing splanchnic blood flow and subsequently
impairing PDR-ICG, even in a relatively young group of ASA
physical status I and II patients.10
www.anesthesia-analgesia.org 1075
 LETTERS TO THE EDITOR
In conclusion, we believe the relatively wide and low
ranged PDR-ICG “normal” values found by Reekers et al.
might be the result of nonhomogeneity of the patient group
with nonconstant hemodynamics and thus should not lead to
a premature rejection of a well-established method.
Jaap J. Vos, BSc
Thomas W. L. Scheeren, MD, PhD
Götz J. K. Wietasch, MD, PhD
Department of Anesthesiology
University Medical Center Groningen
University of Groningen
Groningen, The Netherlands
j.k.g.wietasch@anest.umcg.nl
REFERENCES
1. Reekers M, Simon MJ, Boer F, Mooren RA, van Kleef JW,
Dahan A, Vuyk J. Pulse dye densitometry and indocyanine
green plasma disappearance in ASA physical status I-II pa-
tients. Anesth Analg 2010;110:466 –72
2. von Spiegel T, Scholz M, Wietasch G, Hering R, Allen SJ, Wood
P, Hoeft A. Perioperative monitoring of indocyanine green
clearance and plasma disappearance rate in patients undergo-
ing liver transplantation. Anaesthesist 2002;51:359 – 66
3. Sakka SG, Reinhart K, Meier-Hellmann A. Comparison of
invasive and noninvasive measurements of indocyanine green
plasma disappearance rate in critically ill patients with me-
chanical ventilation and stable hemodynamics. Intensive Care
Med 2000;26:1553– 6
4. Rowell LB, Blackmon JR, Bruce RA. Indocyanine green clear-
ance and estimated hepatic blood flow during mild to maximal
exercise in upright man. J Clin Invest 1964;43:1677–90
5. Hori T, Iida T, Yagi S, Taniguchi K, Yamamoto C, Mizuno S,
Yamagiwa K, Isaji S, Uemoto S. K(ICG) value, a reliable
real-time estimator of graft function, accurately predicts out-
comes in adult living-donor liver transplantation. Liver
Transpl 2006;12:605–13
6. de Liguori Carino N, O’Reilly DA, Dajani K, Ghaneh P, Poston
GJ, Wu AV. Perioperative use of the LiMON method of
indocyanine green elimination measurement for the prediction
and early detection of post-hepatectomy liver failure. Eur
J Surg Oncol 2009;35:957– 62
7. Kuntz H, Schregel W. Indocyanine green: evaluation of liver
function—application in intensive care medicine. In: Lewis F,
Pfeiffer U, eds. Practical Applications of Fiberoptics in Critical
Care Monitoring. 2nd ed. New York: Springer, 1990:57– 62
8. Hultcrantz R, Glaumann H, Lindberg G, Nilsson LH. Liver
investigation in 149 asymptomatic patients with moderately
elevated activities of serum aminotransferases. Scand J Gastro-
enterol 1986;21:109 –13
9. Caesar J, Shaldon S, Chiandussi L, Guevara L, Sherlock S. The
use of indocyanine green in the measurement of hepatic blood
flow and as a test of hepatic function. Clin Sci 1961;21:43–57
10. Lange H, Stephan H, Rieke H, Kellermann M, Sonntag H, Bircher
J. Hepatic and extrahepatic disposition of propofol in patients
undergoing coronary bypass surgery. Br J Anaesth 1990;64:563–70
DOI: 10.1213/ANE.0b013e3181ef35ba
In Response
We agree with Vos et al.1 that the absence of liver enzyme
measurements in some of our patients may be a drawback
in our study. However, biochemical hepatic function test-
ing may not offer the key information on hepatocellular
dysfunction2 as Vos et al. suggest. Hultcrantz et al.3 described
that chronically elevated liver enzymes without symptoms or
physical signs of liver disease correspond with various forms
of liver disease preferably in the presence of a positive history
on alcohol consumption, drug abuse, hepatitis, or obesity. All
1076 www.anesthesia-analgesia.org
Figure 1. Representation of cardiac output measurements based on
arterial blood indocyanine green (ICG) concentrations versus ICG–
plasma disappearance rate (PDR) values determined in patients who
underwent simultaneous arterial ICG sampling. The circles represent
the measurements in awake subjects; the diamonds represent the
measurements during propofol induction.
patients displaying these factors were excluded from our
study population, thus removing those patients that may
exhibit elevated hepatic enzymes in the absence of physical
signs or symptoms of liver disease.
Regarding the importance of hemodynamics and indocya-
nine green plasma disappearance rate (ICG-PDR), we origi-
nally intended to use the noninvasive method of ICG mea-
surement, pulse dye densitometry, to measure cardiac output
in our study population. To validate the transcutaneous
method versus intraarterial measurement of ICG, we per-
formed simultaneous measurements in a subpopulation. In
this subpopulation, we had to conclude that, for individual
measurement of cardiac output, the transcutaneous measure-
ment of ICG by pulse dye densitometry is not accurate
enough.4 For the purpose of this discussion, the measure-
ments of cardiac output versus the ICG-PDR value in the
patients in whom we performed arterial blood sampling is
shown in Figure 1. The open diamonds represent the patients
receiving a propofol induction. Inspection of Figure 1 leads to
the conclusion that the absence or presence of propofol did
not induce significant hemodynamic changes in this other-
wise healthy population nor did it affect ICG-PDR.
We thus maintain our conclusion4 that ICG-PDR values
in a population without clinical signs of liver failure range
well below 18% min⫺1, cited as the cutoff value for hepatic
failure and propagated as criterion for clinical intervention.
This cutoff value needs to be reconsidered as has been
suggested elsewhere.5 We agree with the reviewers that
further studies are needed and that the final word on this
subject has not yet been written.
Marije Reekers, MD
Fred Boer, MD, PhD
Jaap Vuyk, MD, PhD
Department of Anesthesiology
Leiden University Medical Centre
Leiden, The Netherlands
m.reekers@lumc.nl
ANESTHESIA & ANALGESIA
Letters to the Editor
REFERENCES
1. Vos JJ, Scheeren TWL, Wietasch GJK. Pulse dye densitometry
and indocyanine green plasma disappearance: the issue of
“normal” values. Anesth Analg 2010;111:1075– 6
2. Sakka SG. Assessing liver function. Curr Opin Crit Care
2007;13:207–14
3. Hultcrantz H, Glaumann H, Lindberg G, Nilsson LH. Liver
investigation in 149 asymptomatic patients with moderately
elevated activities of serum aminotransferases. Scand J Gastro-
enterol 1986;21:109 –13
4. Reekers M, Simon MJ, Boer F, Mooren FA, van Kleef JW, Dahan A,
Vuyk J. Cardiovascular monitoring by pulse dye densitometry or
arterial indocyanine green dilution. Anesth Analg 2009;109:441– 6
5. Merle U, Sieg O, Stremmel W, Encke J, Eisenbach C. Sensitivity
and specificity of plasma disappearance rate of indocyanine green
as a prognostic indicator in acute liver failure BMC Gastroenterol-
ogy 2009;9:91
DOI: 10.1213/ANE.0b013e3181ef35e7
Serotonin Syndrome in the Perioperative
Period: Role of Tramadol
To the Editor
A ltman and Jahangiri1 describe an important com-
plication associated with certain psychiatric as
well as other serotonergic medications used in
the perioperative period. However, in addition to the
various drugs mentioned by the authors, an important
drug also implicated in this setting is tramadol. It is a
centrally acting analgesic frequently used for treating
moderate to severe postoperative pain, especially in
third world countries. Tramadol, a weak agonist at the
␮-opioid receptor, also has a non-opioid mechanism of
action that includes release of serotonin and inhibition of
reuptake of norepinephrine and is therefore likely to
contribute to the development of serotonin syndrome.2
Satinder Gombar, MD
Nidhi Bhatia, MD
Department of Anaesthesia & Intensive Care
Government Medical College and Hospital
Chandigarh, India
dr_sgombar@rediffmail.com
REFERENCES
1. Altman CS, Jahangiri MF. Serotonin syndrome in the perioper-
ative period. Anesth Analg 2010;110:526 – 8
2. Takeshita J, Litzinger MH. Serotonin syndrome associated with
tramadol. Prim Care Companion J Clin Psychiatry 2009;11:273
DOI: 10.1213/ANE.0b013e3181eb02e8
Pressure-Rated Needleless Access
Connectors Slow IV Flow Rate
To the Editor
O ur hospital recently added pressure-rated needleless
access connectors (PNACs) to all of our IV fluid sets
to facilitate postoperative withdrawal of blood and
medication administration. Shortly thereafter, some of our
anesthesiologists noted decreased flow rate and asked just
how much fluid administration was diminished by these
October 2010 • Volume 111 • Number 4
Figure 1. Testing the effect of a pressure-rated needleless access
connector (PNAC) on IV flow rate. Two identical saline bags with IV
tubing of equal length were primed, suspended 7 feet above the
catheter tips, then allowed to simultaneously drain to gravity. The
only difference between the 2 setups was the presence of a PNAC
on the distal aspect of the IV tubing attached to 1 bag but not the
other. The time required for complete drainage of each saline bag
was measured by means of a stopwatch; flow rate was calculated
by dividing the volume of saline drained by time elapsed.
Table 1. The Effect of Pressure-Rated Needleless
Access Connectors on Gravity-Driven Flow Rate
Through Catheters of Various Internal Diameters
Catheter Control setup, PNAC setup, Difference, mL/min
parameter mL/min mL/min (% change)
22 gaugea 40.8 40.0 ⫺0.8 (⫺2%)
20 gaugea 54.5 56.8 ⫹2.3 (⫹4%)
16 gaugeb 214 121 ⫺93 (⫺44%)
14 gaugeb 223 138 ⫺85 (⫺38%)
PNAC ⫽ pressure-rated needleless access connector.
a
Introcan Safety® IV Catheter (B. Braun Medical Inc., Bethlehem, PA).
b
CATHLON® IV Catheter (Smiths Medical North America, Dublin, OH).
devices. We sought to answer this question by measuring IV
flow rates through 2 nearly identical IV setups. The only
difference between the 2 systems was the presence
(“PNAC setup”) or absence (“control setup”) of a PNACa
on the distal aspect of the IV tubingb (Fig. 1). Although
our results agree with that of the PNAC manufacturerc for
22-gauge catheter sets, we observed a substantially decreased
gravity-driven flow rate when the device was used with 14-
and 16-gauge catheters (Table 1); device performance is not
reported by the manufacturer for either of these large-bore
catheters. Admittedly, these observations merely illustrate a
well-known fact: flow rate is a function of various parame-
ters,1– 4 yet they also underscore the importance of critically
evaluating modifications to IV fluid sets before wholesale
adoption. Our aim is not to discourage the use of PNACs, but
a
b
MaxPlus® Clear (Maximus Medical, Ontario, CA).
LifeShield® Bifurcated Blood Set (Hospira, Morgan Hill, CA) and Clearlink
Extension Set (Baxter, Deerfield, IL).
c
Available at: http://www.maximusmedical.com/pdf/ML3084FlowRate
Performance.pdf. Accessed May 27, 2010.
www.anesthesia-analgesia.org 1077
Letters to the Editor
REFERENCES
1. Vos JJ, Scheeren TWL, Wietasch GJK. Pulse dye densitometry
and indocyanine green plasma disappearance: the issue of
“normal” values. Anesth Analg 2010;111:1075– 6
2. Sakka SG. Assessing liver function. Curr Opin Crit Care
2007;13:207–14
3. Hultcrantz H, Glaumann H, Lindberg G, Nilsson LH. Liver
investigation in 149 asymptomatic patients with moderately
elevated activities of serum aminotransferases. Scand J Gastro-
enterol 1986;21:109 –13
4. Reekers M, Simon MJ, Boer F, Mooren FA, van Kleef JW, Dahan A,
Vuyk J. Cardiovascular monitoring by pulse dye densitometry or
arterial indocyanine green dilution. Anesth Analg 2009;109:441– 6
5. Merle U, Sieg O, Stremmel W, Encke J, Eisenbach C. Sensitivity
and specificity of plasma disappearance rate of indocyanine green
as a prognostic indicator in acute liver failure BMC Gastroenterol-
ogy 2009;9:91
DOI: 10.1213/ANE.0b013e3181ef35e7
Serotonin Syndrome in the Perioperative
Period: Role of Tramadol
To the Editor
A ltman and Jahangiri1 describe an important com-
plication associated with certain psychiatric as
well as other serotonergic medications used in
the perioperative period. However, in addition to the
various drugs mentioned by the authors, an important
drug also implicated in this setting is tramadol. It is a
centrally acting analgesic frequently used for treating
moderate to severe postoperative pain, especially in
third world countries. Tramadol, a weak agonist at the
␮-opioid receptor, also has a non-opioid mechanism of
action that includes release of serotonin and inhibition of
reuptake of norepinephrine and is therefore likely to
contribute to the development of serotonin syndrome.2
Satinder Gombar, MD
Nidhi Bhatia, MD
Department of Anaesthesia & Intensive Care
Government Medical College and Hospital
Chandigarh, India
dr_sgombar@rediffmail.com
REFERENCES
1. Altman CS, Jahangiri MF. Serotonin syndrome in the perioper-
ative period. Anesth Analg 2010;110:526 – 8
2. Takeshita J, Litzinger MH. Serotonin syndrome associated with
tramadol. Prim Care Companion J Clin Psychiatry 2009;11:273
DOI: 10.1213/ANE.0b013e3181eb02e8
Pressure-Rated Needleless Access
Connectors Slow IV Flow Rate
To the Editor
O ur hospital recently added pressure-rated needleless
access connectors (PNACs) to all of our IV fluid sets
to facilitate postoperative withdrawal of blood and
medication administration. Shortly thereafter, some of our
anesthesiologists noted decreased flow rate and asked just
how much fluid administration was diminished by these
October 2010 • Volume 111 • Number 4
Figure 1. Testing the effect of a pressure-rated needleless access
connector (PNAC) on IV flow rate. Two identical saline bags with IV
tubing of equal length were primed, suspended 7 feet above the
catheter tips, then allowed to simultaneously drain to gravity. The
only difference between the 2 setups was the presence of a PNAC
on the distal aspect of the IV tubing attached to 1 bag but not the
other. The time required for complete drainage of each saline bag
was measured by means of a stopwatch; flow rate was calculated
by dividing the volume of saline drained by time elapsed.
Table 1. The Effect of Pressure-Rated Needleless
Access Connectors on Gravity-Driven Flow Rate
Through Catheters of Various Internal Diameters
Catheter Control setup, PNAC setup, Difference, mL/min
parameter mL/min mL/min (% change)
22 gaugea 40.8 40.0 ⫺0.8 (⫺2%)
20 gaugea 54.5 56.8 ⫹2.3 (⫹4%)
16 gaugeb 214 121 ⫺93 (⫺44%)
14 gaugeb 223 138 ⫺85 (⫺38%)
PNAC ⫽ pressure-rated needleless access connector.
a
Introcan Safety® IV Catheter (B. Braun Medical Inc., Bethlehem, PA).
b
CATHLON® IV Catheter (Smiths Medical North America, Dublin, OH).
devices. We sought to answer this question by measuring IV
flow rates through 2 nearly identical IV setups. The only
difference between the 2 systems was the presence
(“PNAC setup”) or absence (“control setup”) of a PNACa
on the distal aspect of the IV tubingb (Fig. 1). Although
our results agree with that of the PNAC manufacturerc for
22-gauge catheter sets, we observed a substantially decreased
gravity-driven flow rate when the device was used with 14-
and 16-gauge catheters (Table 1); device performance is not
reported by the manufacturer for either of these large-bore
catheters. Admittedly, these observations merely illustrate a
well-known fact: flow rate is a function of various parame-
ters,1– 4 yet they also underscore the importance of critically
evaluating modifications to IV fluid sets before wholesale
adoption. Our aim is not to discourage the use of PNACs, but
a
b
MaxPlus® Clear (Maximus Medical, Ontario, CA).
LifeShield® Bifurcated Blood Set (Hospira, Morgan Hill, CA) and Clearlink
Extension Set (Baxter, Deerfield, IL).
c
Available at: http://www.maximusmedical.com/pdf/ML3084FlowRate
Performance.pdf. Accessed May 27, 2010.
www.anesthesia-analgesia.org 1077
Letters to the Editor
REFERENCES
1. Vos JJ, Scheeren TWL, Wietasch GJK. Pulse dye densitometry
and indocyanine green plasma disappearance: the issue of
“normal” values. Anesth Analg 2010;111:1075– 6
2. Sakka SG. Assessing liver function. Curr Opin Crit Care
2007;13:207–14
3. Hultcrantz H, Glaumann H, Lindberg G, Nilsson LH. Liver
investigation in 149 asymptomatic patients with moderately
elevated activities of serum aminotransferases. Scand J Gastro-
enterol 1986;21:109 –13
4. Reekers M, Simon MJ, Boer F, Mooren FA, van Kleef JW, Dahan A,
Vuyk J. Cardiovascular monitoring by pulse dye densitometry or
arterial indocyanine green dilution. Anesth Analg 2009;109:441– 6
5. Merle U, Sieg O, Stremmel W, Encke J, Eisenbach C. Sensitivity
and specificity of plasma disappearance rate of indocyanine green
as a prognostic indicator in acute liver failure BMC Gastroenterol-
ogy 2009;9:91
DOI: 10.1213/ANE.0b013e3181ef35e7
Serotonin Syndrome in the Perioperative
Period: Role of Tramadol
To the Editor
A ltman and Jahangiri1 describe an important com-
plication associated with certain psychiatric as
well as other serotonergic medications used in
the perioperative period. However, in addition to the
various drugs mentioned by the authors, an important
drug also implicated in this setting is tramadol. It is a
centrally acting analgesic frequently used for treating
moderate to severe postoperative pain, especially in
third world countries. Tramadol, a weak agonist at the
␮-opioid receptor, also has a non-opioid mechanism of
action that includes release of serotonin and inhibition of
reuptake of norepinephrine and is therefore likely to
contribute to the development of serotonin syndrome.2
Satinder Gombar, MD
Nidhi Bhatia, MD
Department of Anaesthesia & Intensive Care
Government Medical College and Hospital
Chandigarh, India
dr_sgombar@rediffmail.com
REFERENCES
1. Altman CS, Jahangiri MF. Serotonin syndrome in the perioper-
ative period. Anesth Analg 2010;110:526 – 8
2. Takeshita J, Litzinger MH. Serotonin syndrome associated with
tramadol. Prim Care Companion J Clin Psychiatry 2009;11:273
DOI: 10.1213/ANE.0b013e3181eb02e8
Pressure-Rated Needleless Access
Connectors Slow IV Flow Rate
To the Editor
O ur hospital recently added pressure-rated needleless
access connectors (PNACs) to all of our IV fluid sets
to facilitate postoperative withdrawal of blood and
medication administration. Shortly thereafter, some of our
anesthesiologists noted decreased flow rate and asked just
how much fluid administration was diminished by these
October 2010 • Volume 111 • Number 4
Figure 1. Testing the effect of a pressure-rated needleless access
connector (PNAC) on IV flow rate. Two identical saline bags with IV
tubing of equal length were primed, suspended 7 feet above the
catheter tips, then allowed to simultaneously drain to gravity. The
only difference between the 2 setups was the presence of a PNAC
on the distal aspect of the IV tubing attached to 1 bag but not the
other. The time required for complete drainage of each saline bag
was measured by means of a stopwatch; flow rate was calculated
by dividing the volume of saline drained by time elapsed.
Table 1. The Effect of Pressure-Rated Needleless
Access Connectors on Gravity-Driven Flow Rate
Through Catheters of Various Internal Diameters
Catheter Control setup, PNAC setup, Difference, mL/min
parameter mL/min mL/min (% change)
22 gaugea 40.8 40.0 ⫺0.8 (⫺2%)
20 gaugea 54.5 56.8 ⫹2.3 (⫹4%)
16 gaugeb 214 121 ⫺93 (⫺44%)
14 gaugeb 223 138 ⫺85 (⫺38%)
PNAC ⫽ pressure-rated needleless access connector.
a
Introcan Safety® IV Catheter (B. Braun Medical Inc., Bethlehem, PA).
b
CATHLON® IV Catheter (Smiths Medical North America, Dublin, OH).
devices. We sought to answer this question by measuring IV
flow rates through 2 nearly identical IV setups. The only
difference between the 2 systems was the presence
(“PNAC setup”) or absence (“control setup”) of a PNACa
on the distal aspect of the IV tubingb (Fig. 1). Although
our results agree with that of the PNAC manufacturerc for
22-gauge catheter sets, we observed a substantially decreased
gravity-driven flow rate when the device was used with 14-
and 16-gauge catheters (Table 1); device performance is not
reported by the manufacturer for either of these large-bore
catheters. Admittedly, these observations merely illustrate a
well-known fact: flow rate is a function of various parame-
ters,1– 4 yet they also underscore the importance of critically
evaluating modifications to IV fluid sets before wholesale
adoption. Our aim is not to discourage the use of PNACs, but
a
b
MaxPlus® Clear (Maximus Medical, Ontario, CA).
LifeShield® Bifurcated Blood Set (Hospira, Morgan Hill, CA) and Clearlink
Extension Set (Baxter, Deerfield, IL).
c
Available at: http://www.maximusmedical.com/pdf/ML3084FlowRate
Performance.pdf. Accessed May 27, 2010.
www.anesthesia-analgesia.org 1077
 LETTERS TO THE EDITOR

rather to raise awareness of the flow impedance caused by

these (and other) IV devices.

ACKNOWLEDGMENTS
Lawrence J. Saidman, MD, is acknowledged for his guidance in
the preparation of this letter.
Jorge A. Caballero, MD
Stanford University School of Medicine
Stanford, California
Frain Rivera, MD
Joshua Edwards, MD
John G. Brock-Utne, MD, PhD
Department of Anesthesia
Stanford University School of Medicine
Stanford, California
brockutn@stanford.edu

REFERENCES
1. Brown N, Duttchen KM, Caveno JW. An evaluation of flow
rates of normal saline through peripheral and central venous
catheters. American Society of Anesthesiologists Annual Meet-
ing, Orlando. Anesthesiology 2008:A1484
2. Andersen HW, Benumof JL, Trousdale FR, Ozaki GT. Increasing
the functional gauge on the side port of large catheter sheath
introducers. Anesthesiology 1982;56:57–9
3. Benumof JL, Trousdale FR, Alfery DD, Ozaki GT. Larger Figure 1. Arterial anatomy of the hand.
catheter sheath introducers and their side port functional gauge.
Anesth Analg 1981;60:216 –7
4. Benumof JL, Wyte SR, Rogers SN. A large catheter sheath
introducer with an increased side-port functional gauge. Crit
Care Med 1983;11:660 –2
DOI: 10.1213/ANE.0b013e3181f0948c

Anatomic Snuffbox Radial

Artery Cannulation
To the Editor

F ailure of radial artery cannulation at the wrist,

although uncommon, is not infrequent and is sec-
ondary to vasospasm, hematoma formation, and
intimal dissection or thickening. We describe an alterna-
tive successful ultrasound-guided approach to dorsal
radial artery cannulation at the “anatomic snuffbox.” A
55-year-old woman, with a medical history significant
for diabetes mellitus, coronary artery disease (postcoro-
nary artery bypass surgery), and end-stage renal disease
receiving hemodialysis via a left arm arteriovenous
fistula, was scheduled for renal transplant. After induc-
tion of general anesthesia, several attempts to percuta-
neously insert a right radial artery catheter failed be-
cause of hematoma formation. Cannulation of the
brachial and axillary arteries of the right arm was
considered but might have precluded future surgical
fistula formation. We decided not to cannulate the ulnar
artery because of the radial artery hematoma. The radial
artery was, however, palpable as the dorsal radial artery
in the anatomic snuffbox. Using ultrasound guidance,
the diameter of the dorsal radial artery was measured at
2.2 mm and a 22-gauge (0.9-mm outer diameter) cannula
was inserted into the snuffbox radial artery.
In 1982, Pyles et al.1 published their clinical experi-
ence with cannulation of the dorsal radial artery. The
Figure 2. Semiprone position of the hand for cannulation.
cannulation site in the anatomic snuffbox is distal to the
division of the radial artery that provides collateral flow to the
hand through the superficial palmar (volar) arch (Fig. 1).2
Cannulation distal to that separation would be expected to
reduce the potential for digital ischemia. The dorsal radial
artery does provide arterial flow to the deep palmar arch and
cannulation of the same may be preferable to a proximal site,
which could interfere with both the deep and superficial
palmar arch flows. For successful cannulation, we recom-
mend semiprone position of hand (Fig. 2) and small-sized
catheters. To conclude, cannulation of the dorsal radial artery
is a viable alternative to the radial artery at the wrist and other
possible arterial access sites.
Krishnaprasad Deepika, MD
Dhamodaran Palaniappan, MD

1078 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

 LETTERS TO THE EDITOR

rather to raise awareness of the flow impedance caused by

these (and other) IV devices.

ACKNOWLEDGMENTS
Lawrence J. Saidman, MD, is acknowledged for his guidance in
the preparation of this letter.
Jorge A. Caballero, MD
Stanford University School of Medicine
Stanford, California
Frain Rivera, MD
Joshua Edwards, MD
John G. Brock-Utne, MD, PhD
Department of Anesthesia
Stanford University School of Medicine
Stanford, California
brockutn@stanford.edu

REFERENCES
1. Brown N, Duttchen KM, Caveno JW. An evaluation of flow
rates of normal saline through peripheral and central venous
catheters. American Society of Anesthesiologists Annual Meet-
ing, Orlando. Anesthesiology 2008:A1484
2. Andersen HW, Benumof JL, Trousdale FR, Ozaki GT. Increasing
the functional gauge on the side port of large catheter sheath
introducers. Anesthesiology 1982;56:57–9
3. Benumof JL, Trousdale FR, Alfery DD, Ozaki GT. Larger Figure 1. Arterial anatomy of the hand.
catheter sheath introducers and their side port functional gauge.
Anesth Analg 1981;60:216 –7
4. Benumof JL, Wyte SR, Rogers SN. A large catheter sheath
introducer with an increased side-port functional gauge. Crit
Care Med 1983;11:660 –2
DOI: 10.1213/ANE.0b013e3181f0948c

Anatomic Snuffbox Radial

Artery Cannulation
To the Editor

F ailure of radial artery cannulation at the wrist,

although uncommon, is not infrequent and is sec-
ondary to vasospasm, hematoma formation, and
intimal dissection or thickening. We describe an alterna-
tive successful ultrasound-guided approach to dorsal
radial artery cannulation at the “anatomic snuffbox.” A
55-year-old woman, with a medical history significant
for diabetes mellitus, coronary artery disease (postcoro-
nary artery bypass surgery), and end-stage renal disease
receiving hemodialysis via a left arm arteriovenous
fistula, was scheduled for renal transplant. After induc-
tion of general anesthesia, several attempts to percuta-
neously insert a right radial artery catheter failed be-
cause of hematoma formation. Cannulation of the
brachial and axillary arteries of the right arm was
considered but might have precluded future surgical
fistula formation. We decided not to cannulate the ulnar
artery because of the radial artery hematoma. The radial
artery was, however, palpable as the dorsal radial artery
in the anatomic snuffbox. Using ultrasound guidance,
the diameter of the dorsal radial artery was measured at
2.2 mm and a 22-gauge (0.9-mm outer diameter) cannula
was inserted into the snuffbox radial artery.
In 1982, Pyles et al.1 published their clinical experi-
ence with cannulation of the dorsal radial artery. The
Figure 2. Semiprone position of the hand for cannulation.
cannulation site in the anatomic snuffbox is distal to the
division of the radial artery that provides collateral flow to the
hand through the superficial palmar (volar) arch (Fig. 1).2
Cannulation distal to that separation would be expected to
reduce the potential for digital ischemia. The dorsal radial
artery does provide arterial flow to the deep palmar arch and
cannulation of the same may be preferable to a proximal site,
which could interfere with both the deep and superficial
palmar arch flows. For successful cannulation, we recom-
mend semiprone position of hand (Fig. 2) and small-sized
catheters. To conclude, cannulation of the dorsal radial artery
is a viable alternative to the radial artery at the wrist and other
possible arterial access sites.
Krishnaprasad Deepika, MD
Dhamodaran Palaniappan, MD

1078 www.anesthesia-analgesia.org ANESTHESIA & ANALGESIA

Letters to the Editor

Thomas Fuhrman, MD has been brought to our attention that some of the data that
Bruce Saltzman, MD we reported have been included in a series of 4 cases
Department of Anesthesiology, Perioperative Medicine and published earlier this year.2
Pain Management We were not aware of this publication at the time our
Jackson Memorial Hospital manuscript was accepted for publication and therefore we
University of Miami did not cite it in our case report. Therefore, our communi-
Miami, Florida cation was inaccurate inasmuch as it was not the second
KDeepika@med.miami.edu time it has been used in an emergency. We would empha-
REFERENCES
size that the use of the device as described by Strueber et
1. Pyles ST, Scher KS, Vega ET, Harrah JD, Rubis LJ. Cannulation al.2 is a novel application. Using the Novalung for right
of the dorsal radial artery: a new technique. Anesth Analg ventricular support mandates central placement as de-
1982;61:876 – 8 scribed, regardless of the size of the patient.
2. Gray H, Lewis WH. Gray’s Anatomy of the Human Body. 20th US
ed. Philadelphia: Lea & Febiger, originally published in 1918
DOI: 10.1213/ANE.0b013e3181ef343a Helen Holtby, MB, BS, FRCPC
Director of Cardiac Anesthesia
Hospital for Sick Children
Correction to the Case Report Toronto, Ontario, Canada
helen.holtby@sickkids.ca
“Emergency Interventional Lung
Assist for Pulmonary Hypertension” REFERENCES
Anesth Analg 2009;109:382–5 1. Taylor K, Holtby H. Emergency interventional lung assist for
pulmonary hypertension. Anesth Analg 2009;109:382–5
To the Editor 2. Strueber M, Hoeper MM, Fischer S, Cypel M, Warnecke G,
Gottlieb J, Pierre A, Welte T, Haverich A, Simon AR, Keshavjee

W e recently described our experience of the anes-

thetic management of a patient using an emer-
gency lung assist device.1 Although the above
report focused on the anesthetic considerations for that case, it
S. Bridge to thoracic organ transplantation in patients with
pulmonary arterial hypertension using a pumpless lung assist
device. Am J Transplant 2009;9:853–7
DOI: 10.1213/ANE.0b013e3181ec2bea

October 2010 • Volume 111 • Number 4 www.anesthesia-analgesia.org 1079

 BOOK, MULTIMEDIA, AND MEETING REVIEWS
Section Editor: Paul F. White
An Introductory Curriculum for
Ultrasound-Guided Regional Anesthesia
Brian A. Pollard Vincent W. S. Chan; illustrations by
Diana Kryski; photographs by Michele Dalgarno. Toronto:
B. A. Pollard, 2010. ISBN: 978-0-7727-8735-4. $95.00.
orty years after Dr. Alon Winnie described the paresthe-
F sia technique for interscalene blockade, evolving tech-
nology enables clinicians to see and differentiate anatomic
structures in real time. The growing interest in ultrasound-
guided regional anesthesia (UGRA) has been an exciting,
landmark development in the practice of regional anesthe-
sia and pain medicine. The authors’ of An Introductory
Curriculum for Ultrasound-Guided Regional Anesthesia, Drs.
Pollard and Chan, are well-known and accomplished lead-
ers in this emerging field.
With this new technology comes responsibility. Spirited
debate now surrounds the definition of core competencies,
training requirements, competency assessment, institu-
tional certification for UGRA practice, and strategies for
quality improvement in UGRA. In writing these introduc-
tory curriculums on UGRA, the authors have appropriately
side-stepped these issues and focused instead on a more
important founding principle for their text—patient safety.
As is stated by the authors, the goal in writing this book
was “to create an educational curriculum that will permit a
self-directed foundational path for clinicians, from novice
to expert, community and academic-based, to build on
existing regional anesthesia expertise by integrating essen-
tial ultrasound techniques into daily practice.” They accom-
plish this task through nine chapters in five well-organized
sections.
Section 1 (39 pages) is the largest and constitutes almost
half of the text. Starting with physics as applied to ultra-
sonography, this section then confronts the fundamentals
of ultrasound scanning and needle techniques. Sections 2, 3,
and 4 categorize nerve blocks into introductory, intermedi-
ate, and advanced designations. Chapters within these
sections cover the most commonly performed ultrasound-
guided nerve blocks. Finally, section 5 describes the use of
UGRA to assist with conventionally performed epidural
and subarachnoid blocks.
An ubiquitous debate continues among practitioners as
to what constitutes an “introductory” block versus an
“advanced” block, which can be seen as a potential weak-
ness in this presentation. For example, many clinicians may
not find a selective block of the radial nerve “easier” than a
conventional ultrasound-guided approach to the brachial
plexus.
Each chapter ends with a salient summary, as well as 4
to 6 key references for the self-directed learner. The funda-
mental learning points of the chapter are then listed as
useful “knowledge keys.” Esthetics were not overlooked in
the production of the text, which presents superior illustra-
tions and an excellent collection of images. Often, images
and illustrations are combined into impressive and thoughtful
1080 www.anesthesia-analgesia.org
schematics detailing phenomena such as reverberation arti-
fact. The high-quality images and illustrations are a major
strength of the book.
Each chapter reads easily, yet with only 81 pages and
lacking references, this is not meant to be a comprehensive
textbook on the topic. Rather, it nicely complements exist-
ing resources, and as such represents one of the finest
presentations on basic UGRA concepts available to practi-
tioners. The emphasis on competency with imaging is
appropriate and is reflected in the fact that the bulk of the
text material addresses this skill set. Specific clinical recom-
mendations, such as type of local anesthetic and volume
selection, are not emphasized.
As interest in the use of UGRA in clinical practice and
education continues to expand, efforts to optimize patient
safety and minimize risk should be at the forefront. Al-
though no evidence exists to support this claim, one could
reasonably hypothesize that poor UGRA technique may
increase the risk of associated complications. To that end,
this text is an essential resource for UGRA educators and
would be an excellent addition to all libraries on regional
anesthesia.
Jonathan C. Beathe, MD
Gregory A. Liguori, MD
Weill Medical College of Cornell University
Hospital for Special Surgery
New York, New York
liguorig@hss.edu
Visionaries and Dreamers: The
Story of Founding Fathers of
Anesthesiology in Israel
Gabrial M. Gurman. Beer-Sheva, Israel: Ben-Gurion Uni-
versity of the Negev Press, 2008. ISBN: 978-9-6534-
2963-5. 228 pages. $16.53.
isionaries and Dreamers is a series of accounts about
V pioneers of Israeli anesthesiology. The author, Gabriel
M. Gurman, MD, is a Professor Emeritus of Anesthesiology
and Critical Care at Ben-Gurion University of the Negev,
and both Hebrew and English translations were edited by
Lior Granot. Dr. Gurman provides the details of the lives of
innovative anesthesiologists through a series of interviews
and is able to capture compelling stories of adversity and
eventual triumphs. As the lives of these 12 anesthesiolo-
gists are described, the reader is pulled into what seems
like an almost “sacred circle of leaders.” As an anesthesi-
ologist, it was helpful to see how the book breaks down the
elements of their leadership, and sheds light on each of
these qualities separately: the importance of building a
vision, the value of being bold and unconstrained by
established rules, and having the foresight to build a solid,
trustworthy team. These are invaluable lessons that would
help the professional careers of all aspiring anesthesiologists.
October 2010 • Volume 111 • Number 4
 BOOK, MULTIMEDIA, AND MEETING REVIEWS
Section Editor: Paul F. White
An Introductory Curriculum for
Ultrasound-Guided Regional Anesthesia
Brian A. Pollard Vincent W. S. Chan; illustrations by
Diana Kryski; photographs by Michele Dalgarno. Toronto:
B. A. Pollard, 2010. ISBN: 978-0-7727-8735-4. $95.00.
orty years after Dr. Alon Winnie described the paresthe-
F sia technique for interscalene blockade, evolving tech-
nology enables clinicians to see and differentiate anatomic
structures in real time. The growing interest in ultrasound-
guided regional anesthesia (UGRA) has been an exciting,
landmark development in the practice of regional anesthe-
sia and pain medicine. The authors’ of An Introductory
Curriculum for Ultrasound-Guided Regional Anesthesia, Drs.
Pollard and Chan, are well-known and accomplished lead-
ers in this emerging field.
With this new technology comes responsibility. Spirited
debate now surrounds the definition of core competencies,
training requirements, competency assessment, institu-
tional certification for UGRA practice, and strategies for
quality improvement in UGRA. In writing these introduc-
tory curriculums on UGRA, the authors have appropriately
side-stepped these issues and focused instead on a more
important founding principle for their text—patient safety.
As is stated by the authors, the goal in writing this book
was “to create an educational curriculum that will permit a
self-directed foundational path for clinicians, from novice
to expert, community and academic-based, to build on
existing regional anesthesia expertise by integrating essen-
tial ultrasound techniques into daily practice.” They accom-
plish this task through nine chapters in five well-organized
sections.
Section 1 (39 pages) is the largest and constitutes almost
half of the text. Starting with physics as applied to ultra-
sonography, this section then confronts the fundamentals
of ultrasound scanning and needle techniques. Sections 2, 3,
and 4 categorize nerve blocks into introductory, intermedi-
ate, and advanced designations. Chapters within these
sections cover the most commonly performed ultrasound-
guided nerve blocks. Finally, section 5 describes the use of
UGRA to assist with conventionally performed epidural
and subarachnoid blocks.
An ubiquitous debate continues among practitioners as
to what constitutes an “introductory” block versus an
“advanced” block, which can be seen as a potential weak-
ness in this presentation. For example, many clinicians may
not find a selective block of the radial nerve “easier” than a
conventional ultrasound-guided approach to the brachial
plexus.
Each chapter ends with a salient summary, as well as 4
to 6 key references for the self-directed learner. The funda-
mental learning points of the chapter are then listed as
useful “knowledge keys.” Esthetics were not overlooked in
the production of the text, which presents superior illustra-
tions and an excellent collection of images. Often, images
and illustrations are combined into impressive and thoughtful
1080 www.anesthesia-analgesia.org
schematics detailing phenomena such as reverberation arti-
fact. The high-quality images and illustrations are a major
strength of the book.
Each chapter reads easily, yet with only 81 pages and
lacking references, this is not meant to be a comprehensive
textbook on the topic. Rather, it nicely complements exist-
ing resources, and as such represents one of the finest
presentations on basic UGRA concepts available to practi-
tioners. The emphasis on competency with imaging is
appropriate and is reflected in the fact that the bulk of the
text material addresses this skill set. Specific clinical recom-
mendations, such as type of local anesthetic and volume
selection, are not emphasized.
As interest in the use of UGRA in clinical practice and
education continues to expand, efforts to optimize patient
safety and minimize risk should be at the forefront. Al-
though no evidence exists to support this claim, one could
reasonably hypothesize that poor UGRA technique may
increase the risk of associated complications. To that end,
this text is an essential resource for UGRA educators and
would be an excellent addition to all libraries on regional
anesthesia.
Jonathan C. Beathe, MD
Gregory A. Liguori, MD
Weill Medical College of Cornell University
Hospital for Special Surgery
New York, New York
liguorig@hss.edu
Visionaries and Dreamers: The
Story of Founding Fathers of
Anesthesiology in Israel
Gabrial M. Gurman. Beer-Sheva, Israel: Ben-Gurion Uni-
versity of the Negev Press, 2008. ISBN: 978-9-6534-
2963-5. 228 pages. $16.53.
isionaries and Dreamers is a series of accounts about
V pioneers of Israeli anesthesiology. The author, Gabriel
M. Gurman, MD, is a Professor Emeritus of Anesthesiology
and Critical Care at Ben-Gurion University of the Negev,
and both Hebrew and English translations were edited by
Lior Granot. Dr. Gurman provides the details of the lives of
innovative anesthesiologists through a series of interviews
and is able to capture compelling stories of adversity and
eventual triumphs. As the lives of these 12 anesthesiolo-
gists are described, the reader is pulled into what seems
like an almost “sacred circle of leaders.” As an anesthesi-
ologist, it was helpful to see how the book breaks down the
elements of their leadership, and sheds light on each of
these qualities separately: the importance of building a
vision, the value of being bold and unconstrained by
established rules, and having the foresight to build a solid,
trustworthy team. These are invaluable lessons that would
help the professional careers of all aspiring anesthesiologists.
October 2010 • Volume 111 • Number 4
Books, Multimedia, and Meeting Reviews
Dr. Gurman began his career with a generation of
physicians who had already survived a horrific time. Many
of Dr. Gurman’s friends and colleagues narrowly escaped
the front lines of the extermination camps of World War II
through unimaginable acts of courage. These physicians
were in some cases the only survivors of their entire
families. Most of them migrated to the new state of Israel
shortly after its establishment in1948. In the beginning, the
scarcity of anesthesiologists in the new State of Israel meant
that one anesthesiologist might cover an entire hospital.
This is epitomized in the accounts of Dr. Thomas Gesztes,
who had his own incarcerated hernia repaired under local
anesthesia, and who afterwards anesthetized his patients
while taking call duties the same night as his own surgery.
This is an example of the ultimate “ambulatory” surgery
experience! Each one of these anesthesiologists was instru-
mental in building departments (and societies) of anesthe-
sia from “the ground up.”
When Dr. Gurman migrated to Israel in 1972, anesthesia
clinical practices were still in a rapid flux from what
seemed archaic practices. The transformation was nothing
short of miraculous, and clinical anesthesia care is now
considered among the most advanced specialties in the
field of medicine. In his book, Dr. Gurman interrupts his
storytelling periodically to offer vignettes and historical
commentaries on the various personalities and their per-
sonal and professional stories. Thus, the action of the book
does not proceed chronologically; instead, it moves back
and forth in time, depending on which colleague he is
memorializing. This strategy of tying present-day individuals
to events in the past is very helpful to those readers who may
not be very familiar with the geography and history of Israel.
October 2010 • Volume 111 • Number 4
There are many stories of success and significant ad-
vances in clinical care in the field of anesthesiology, most of
which were achieved through hard work, perseverance,
dedication, and an unwillingness to succumb to the “status
quo.” However, anesthesia was far from ideal in Israel,
especially from an academic and professional standpoint.
This is illustrated by Dr. Shamay Cotev, head of the first
intensive care unit (ICU) in Israel, who is quoted in the
book as having said, “In retrospect, I wouldn’t have gone
into anesthesia.” This is perhaps the only part of this bright
and optimistic book that touches on the harsh reality that
much more work is still needed—in anesthesia and in
medicine. Currently, ⬍2% of the graduates of medical
schools in Israel choose to pursue postgraduate training in
anesthesiology!
In this well-written book, Dr. Gurman takes the reader
into the middle of Israeli anesthesia history. The tremen-
dous strides of the past 50 years are apparent, but the
summit is yet to be achieved. Dr. Gurman’s book provides
us with direction regarding where we need to be in the
future. This book is dedicated to the 60th anniversary of the
State of Israel, and is recommended reading for young and
old anesthesiologists not only for its historical value, but
more importantly, for the optimistic glimpse it provides
into the future of anesthesiology, medicine, and humanity.
Sorin J. Brull, MD
Joseph A. Cartwright, MD
Mayo Clinic College of Medicine
Jacksonville, Florida
cartwright.joseph@mayo.edu
www.anesthesia-analgesia.org 1081