Vous êtes sur la page 1sur 7

Original Study

Effect of body position on the arterial partial


pressures of oxygen and carbon dioxide in
spontaneously breathing, conscious dogs in an
intensive care unit
Matthew W. McMillan, BVM&S, MRCVS; Katie E. Whitaker, BVetMed, MRCVS; Dez Hughes, BVSc,
DACVECC, MRCVS; David C. Brodbelt, MA, VetMB, PhD, DVA, DECVA, MRCVS and Amanda K.
Boag, MA, VetMB, DACVIM, DACVECC, MRCVS
Abstract
Objective To evaluate the effect of body position on the arterial partial pressures of oxygen and carbon
dioxide (PaO
2
, PaCO
2
), and the efciency of pulmonary oxygen uptake as estimated by alveolar-arterial
oxygen difference (A-a difference).
Design Prospective, randomized, crossover study.
Setting University teaching hospital, intensive care unit.
Animals Twenty-one spontaneously breathing, conscious, canine patients with arterial catheters placed as
part of their management strategy.
Interventions Patients were placed randomly into lateral or sternal recumbency. PaO
2
and PaCO
2
were
measured after 15 minutes in this position. Patients were then repositioned into the opposite position and after
15 minutes the parameters were remeasured.
Measurements and Main Results Results presented as median (interquartile range). PaO
2
was signicantly
higher (P50.001) when patients were positioned in sternal, 91.2 mmHg (86.096.1 mmHg), compared with
lateral recumbency, 86.4 mmHg (73.990.9 mmHg). The median change was 5.4 mmHg (1.117.9 mmHg). All
7 dogs with a PaO
2
o80 mmHg in lateral recumbency had improved arterial oxygenation in sternal
recumbency, median increase 17.4 mmHg with a range of 3.829.7 mmHg. PaCO
2
levels when patients were
in sternal recumbency, 30.5 mmHg (27.332.7 mmHg) were not signicantly different from those in lateral
recumbency, 32.2 mmHg (28.336.0 mmHg) (P50.07). The median change was 1.9 mmHg ( 3.6
0.77 mmHg). A-a differences were signicantly lower (P50.005) when patients were positioned in sternal
recumbency, 21.7 mmHg (17.327.7 mmHg), compared with lateral recumbency, 24.6 mmHg (20.4
36.3 mmHg). The median change was 3.1 mmHg ( 14.60.9 mmHg).
Conclusions PaO
2
was signicantly higher when animals were positioned in sternal recumbency compared
with lateral recumbency, predominantly due to improved pulmonary oxygen uptake (decreased A-a
difference) rather than increased alveolar ventilation (decreased PaCO
2
). Patients with hypoxemia (dened as
PaO
2
o80 mmHg) in lateral recumbency may benet from being placed in sternal recumbency. Sternal
recumbency is recommended to improve oxygenation in hypoxemic patients.
(J Vet Emerg Crit Care 2009; 19(6): 564570) doi: 10.1111/j.1476-4431.2009.00480.x
Keywords: blood-gas analysis, hypoxemia, prone position, pulmonary function, recumbency
Introduction
The effect of body position on oxygenation and pul-
monary oxygen uptake has been studied in human
medicine with the current evidence suggesting that
oxygenation and pulmonary oxygen uptake improve
when a patient is moved from dorsal recumbency
(supine position) to sternal recumbency (prone posi-
tion).
13
Experimental studies in sheep, pigs, and dogs
The authors declare no conicts of interest and the study received no ex-
ternal funding. Preliminary ndings were presented in abstract form at the
International Veterinary Emergency and Critical Care Symposium 2007,
New Orleans, LA.
Address correspondence and reprint requests to
Dr. Matthew W. McMillan, Department of Clinical Veterinary Science, Queens
Veterinary School, University of Cambridge, Madingley Rd, Cambridge, UK.
Email: mwm32@cam.ac.uk
From the Department of Clinical Studies, Royal Veterinary College, Hert-
fordshire, UK.
Journal of Veterinary Emergency and Critical Care 19(6) 2009, pp 564570
doi:10.1111/j.1476-4431.2009.00480.x
& Veterinary Emergency and Critical Care Society 2009 564
have shown a signicantly higher PaO
2
when patients
are in sternal recumbency as compared with dorsal re-
cumbency.
414
A suggested mechanism for this change
is a reduction in ventilation:perfusion mismatch and
therefore an improvement in pulmonary oxygen up-
take.
414
The relevance of these studies to spontane-
ously breathing, conscious canine clinical patients is
uncertain. All the animal studies were performed on
animals under general anesthesia; therefore, the effects
of anesthetic agents on pulmonary function must be
considered. Most of the studies evaluated dorsal (su-
pine) recumbency versus sternal (prone) recumbency.
Dorsal recumbency is not a typical body position for a
conscious dog and fewer studies have assessed lateral
recumbencies. Previous studies (both experimental and
human clinical) have focused on mechanically venti-
lated patients.
114
Positive-pressure ventilation will
have effects on pulmonary oxygen uptake independent
of those caused by body position.
15
Therefore, it cannot
be inferred that similar improvements will be seen in
spontaneously breathing dogs. If sternal recumbency
improves oxygenation in spontaneously breathing, con-
scious dogs it could be considered as an adjunctive
measure in some hypoxemic patients where mechanical
ventilation is not possible.
The aim of this study was to investigate if partial
pressures of oxygen (PaO
2
) and carbon dioxide
(PaCO
2
), and the alveolar-arterial difference (A-a differ-
ence) were signicantly different between sternal and
lateral recumbency in a heterogeneous group of canine
patients.
Methods and Materials
Dogs hospitalized in an intensive care unit that had
arterial catheters placed as part of their clinical man-
agement were eligible for inclusion. Dogs were eligible
for inclusion regardless of whether or not they had
pulmonary disease. No arterial catheters were placed
solely for the purpose of this study. The study protocol
was approved by the local university elective project
committee that evaluates the ethical use and welfare of
animals in research.
Patients were not eligible for inclusion if they were
receiving oxygen supplementation. No dogs had oxy-
gen supplementation discontinued in order to be eligi-
ble for this study. Patients were excluded if they
displayed major discomfort when moved. Patients
were monitored for signs of clinical deterioration and
were excluded if this occurred. If patients had been
anesthetized, they were not eligible for participation in
the study until a minimum of 3 hours after the endo-
tracheal tube was removed. Animals were assessed and
considered to be fully recovered from anesthesia before
samples were taken.
Animals were randomly allocated to be placed ini-
tially in 1 of 2 recumbencies (lateral or sternal) using a
random number generator.
a
The animals were then
placed into the remaining recumbency after completion
of the rst study measurements. The lateral position
was subsequently randomly assigned as either right or
left lateral recumbency.
The signalment, body weight, primary complaint,
and presence of any respiratory disease or problems
were recorded for each patient. Rectal temperatures
were taken using a digital thermometer
b
and recorded
for each patient. During the sampling period the patient
was placed into the rst position and maintained in that
position for 15 minutes after which an arterial blood
sample was taken. The patient was immediately repo-
sitioned into the second position and maintained in this
position for a further 15 minutes before the second
sample was taken. Samples were taken anaerobically
from the arterial catheter using a 1-mL heparinized,
gas-permeable blood-gas syringe
c
in which the plunger
was drawn to 0.4 mL and allowed to ll by arterial
pressure. Awaste sample was taken in all cases using a
5-mL syringe prelled with 1 mL of heparinized saline
to avoid dilutional effects and the arterial catheter was
ushed following sampling using 25 mL of heparin-
ized saline. Samples were handled anaerobically and
analysis was performed within 90 seconds using a
point-of-care blood-gas analyzer.
d
Blood-gas analyzer
self-calibration occurred hourly and daily quality con-
trol was performed according to the manufacturers
recommendations. Corrections for body temperature
were not made. Using the measured PaO
2
and PaCO
2
values and barometric pressure (measured by the
blood-gas analyzer) the A-a difference was calculated
for each animal in each position. The A-a difference was
calculated from a modied alveolar gas equation as
follows:
A-a difference FiO
2
P
b
47 f g PaCO
2
=RQ
PaO
2
where FiO
2
50.21, RQ5respiratory quotient 50.9, and
P
b
is the barometric pressure, with 47 mmHg 5satura-
tion pressure of water at 371C.
Data were tested for normality using the Shapiro-
Wilk test. Wilcoxons signed rank tests were used for
comparisons between nonparametric or mixed data
sets. Paired t-tests were used to compare parametric
data sets. All statistical analyses were performed using
a commercially available computer package.
e
Nonpara-
metric data are reported as median (interquartile range)
and parametric data as mean (standard deviation). The
& Veterinary Emergency and Critical Care Society 2009, doi: 10.1111/j.1476-4431.2009.00480.x 565
Effect of body position on canine blood gases
null hypothesis for statistical analysis was that body
position does not affect PaO
2
, PaCO
2
, or A-a difference.
Dogs were further classied depending on their PaO
2
levels in lateral recumbency. Dogs with a PaO
2
above
80 mmHg were considered normal and not evaluated
further, and dogs with a PaO
2
below 80 mmHg were
considered hypoxemic. The differences in PaO
2
, PaCO
2
,
and A-a difference in the hypoxemic group were eval-
uated separately. Comparisons were also made be-
tween the rst and second samples taken to assess for
any bias regarding sample timing. A-a differences were
recalculated using an RQ of 0.7 to allow for errors in RQ
estimation. Signicance was set at a 55% (Po0.05).
Results
Twenty-four client-owned dogs were evaluated for in-
clusion in the study. One dog was excluded due to poor
compliance with positioning. Two dogs were excluded
as they appeared to be uncomfortable when moved into
the sternal position. Both of these dogs had undergone
median sternotomies for an exploratory thoracotomy.
The remaining 21 dogs were included. There was 1
PaCO
2
value not recorded due to an error with the
blood-gas analyzer. Consequently 1 fewer A-a differ-
ence was calculated. A high proportion of the data sets
were identied as nonparametric therefore Wilcoxons
signed ranked tests were used for all comparisons.
The median body weight was 28.4 kg (16.335.9 kg).
The median body temperature of the dogs was 37.61C
(37.438.11C). Six of the dogs had respiratory disease; 2
dogs had bilateral chest drains for the management of
pyothorax, 1 dog had resolving cardiogenic pulmonary
edema, 1 dog had a lung lobectomy with a unilateral
chest drain, 1 dog had aspiration pneumonia, and 1 dog
had penetrating thoracic trauma with a unilateral chest
drain. The remaining 15 dogs had no known respira-
tory disease. Of these, 11 dogs were recovering from
abdominal surgery (4 small intestinal perforation re-
pairs with septic peritonitis, 3 portosystemic shunt li-
gations, 2 biliary tract rupture repairs, 1 gastropexy due
to gastric dilatation and volvulus, and 1 splenectomy
due to hemangiosarcoma) and 3 dogs were recovering
from spinal surgery (2 hemilaminectomies for thoraco-
lumbar disc extrusion and 1 ventral slot procedure due
to cervical disc extrusion). The nal dog had brady-
cardia and seizures.
Ten of the dogs were placed in sternal recumbency rst
and 11 were placed in lateral (13 right and 8 left) rst. In
10 of the dogs the lateral position was the position they
had been in before the trial. The median PaO
2
in lateral
recumbency was 86.4mmHg (73.990.9mmHg) as com-
pared with 91.2mmHg (86.096.1mmHg) in sternal re-
cumbency (P50.001). Eighty-one percent (17/21) of the
dogs had a higher PaO
2
in sternal recumbency. The me-
dian change in PaO
2
between positions was 5.4mmHg
(1.117.9mmHg) (Figure 1).
PaCO
2
was measured in 20 dogs. The median PaCO
2
in lateral recumbency was 32.2 mmHg (28.336.0mmHg)
as compared with 30.5mmHg (27.332.7mmHg) in ster-
nal recumbency (P50.07). The median change in PaCO
2
between positions was 1.9 mmHg ( 3.60.77mmHg)
(Figure 2).
A-a difference was calculated in 20 dogs. Using an RQ
of 0.9, the median A-a difference in lateral recumbency
was 24.6 mmHg (20.436.3 mmHg) as compared with
21.7 mmHg (17.327.7 mmHg) in sternal recumbency
(P50.008). The median change between positions was
3.1 mmHg ( 14.60.9 mmHg). Eighty-ve percent
(17/20) of dogs had a lower A-a difference in sternal
recumbency, and all 17 dogs showed improvements in
PaO
2
levels. No dog in lateral recumbency had a normal
A-a difference and only 2 dogs had normal A-a differ-
ences in sternal recumbency (Figure 3).
Using an RQ of 0.7, the median A-a difference in lat-
eral recumbency was 14.7 mmHg (9.926.7 mmHg) as
compared with 13.1 mmHg (7.318.2 mmHg) in sternal
recumbency (P50.007). The median change between
positions was 2.4 (12.9 to 0.5 mmHg). Again 85%
(17/20) of dogs had a lower A-a difference in sternal
recumbency. Forty-ve percent (9/20) of dogs had ab-
normal A-a differences in lateral recumbency compared
with 20% (4/20) in sternal recumbency (Figure 4).
Seven of the dogs in the study were considered hypox-
emic (PaO
2
o80 mmHg) while in lateral recumbency
having a median (range) PaO
2
of 68.0 mmHg (62.4
77.0 mmHg). Five of these dogs had a primary pulmo-
nary disease process; of the remaining 2 dogs, 1 was
recovering from gastric dilatation and volvulus and the
other had septic peritonitis. All of these dogs had a
higher PaO
2
in sternal recumbency median (range) of
85.2 mmHg (71.8101.3 mmHg). The median change
Figure1: Comparison of PaO
2
levels in lateral versus sternal
recumbency. Individual animals results are connected by a line.
& Veterinary Emergency and Critical Care Society 2009, doi: 10.1111/j.1476-4431.2009.00480.x 566
M.W. McMillan et al.
(range) was 17.4 mmHg (3.829.7 mmHg). The median
(range) PaCO
2
for these 7 dogs was 29.6 mmHg (26.5
43.5 mmHg) in lateral as compared with 28.6 mmHg
(22.232.5 mmHg) in sternal recumbency. The median
(range) A-a difference (RQ50.9) for these 7 dogs was
46.3 mmHg (34.152.1 mmHg) in lateral as compared
with 27.7 mmHg (21.445.0 mmHg) in sternal recum-
bency.
Four dogs had a lower PaO
2
in sternal recumbency as
compared with lateral recumbency. All of these 4 dogs
had PaO
2
levels in the reference interval (80
110 mmHg) in both body positions. The decreases in
PaO
2
were 1.1, 1.6, 4.0, and 4.5 mmHg, respectively.
The PaO
2
, PaCO
2
, and A-a difference (RQ50.9) were
also compared between the rst and second body po-
sition. The median PaO
2
in the rst sample was
89.3 mmHg (85.193.0 mmHg) as compared with
87.9 mmHg (82.792.6 mmHg) for the second sample
(P50.75). The median PaCO
2
in the rst sample was
30.9 mmHg (26.835.1 mmHg) as compared with
32.2 mmHg (28.035.2 mmHg) for the second sample
(P50.11). The median A-a difference (RQ50.9) in the
rst sample was 22.2 mmHg (19.330.5 mmHg) as
compared with 24.5 mmHg (20.632.2 mmHg) for the
second sample (P50.76).
In 2 patients, 1 of the measured PaCO
2
values was
notably lower than the other. These differences were
signicantly larger than the general trend and therefore
considered potential errors. Statistical analysis was sub-
sequently repeated with these cases excluded; in the re-
maining dogs, changes in PaO
2
and A-a difference
(RQ50.9) between positions remained signicant, and
the changes in PaCO
2
between positions remained not
signicant. PaO
2
in lateral recumbency was 88.7mmHg
(83.491.2mmHg) as compared with 90.15mmHg (87.0
94.1mmHg) in sternal recumbency (Po0.01). The me-
dian change was 5.0 mmHg (1.114.5mmHg). PaCO
2
in
lateral recumbency was 31.9mmHg (29.234.8 mmHg)
as compared with 31.4mmHg (28.232.7mmHg)
in sternal recumbency (P50.19). The median change
was 1.35 mmHg ( 2.40.8 mmHg). A-a difference
(RQ50.9) in lateral recumbency was 24.1mmHg
(20.431.5mmHg) compared with 21.7mmHg (17.3
24.9mmHg) in sternal recumbency (P50.02). The me-
dian change was 2.8 ( 13.10.9mmHg).
Discussion
The aim of this study was to evaluate differences in
PaO
2
, PaCO
2
, and A-a difference between lateral and
sternal recumbency in a heterogeneous group of spon-
taneously breathing, conscious dogs. Human and ex-
perimental animal studies have demonstrated that
placement in the prone position is associated with a
signicantly higher PaO
2
as compared with the supine
position in mechanically ventilated patients with acute
lung injury and acute respiratory distress syndrome.
13
This study shows that a statistically signicant increase
Figure2: Comparison of PaCO
2
levels in lateral versus sternal
recumbency. Individual animals results are connected by a line.
Figure3: Comparison of calculated A-a differences (RQ50.9)
in lateral versus sternal recumbency. Individual animals results
are connected by a line.
Figure4: Comparison of calculated A-a differences (RQ50.7)
in lateral versus sternal recumbency. Individual animals results
are connected by a line.
& Veterinary Emergency and Critical Care Society 2009, doi: 10.1111/j.1476-4431.2009.00480.x 567
Effect of body position on canine blood gases
in PaO
2
occurred in our population of spontaneously
breathing, conscious dogs when maintained in sternal
recumbency as compared with lateral recumbency. Only
a small number of animals had PaO
2
levels that de-
creased when in sternal recumbency. This would suggest
that sternal positioning is not only associated with a sig-
nicantly higher PaO
2
across the patient population but
is also likely to cause only minimal drops in PaO
2
in
individual animals where improvements are not seen.
Also, sternal recumbency may have a greater effect in
increasing PaO
2
levels in animals who are hypoxemic
(PaO
2
o80 mmHg) when in lateral recumbency. Previous
trials in animals have been used to investigate the effects
of prone position on oxygenation. All the studies have
used anesthetized animals that were being mechanically
ventilated. All but one of the studies compared prone
and supine positions and none of the studies have been
carried out in clinical patients. Nonetheless, each of these
experimental animal studies has shown either a higher
PaO
2
or reduced histopathologic change in the lungs
when the animal is in sternal recumbency.
414
Improvements in PaO
2
may occur due to increased
alveolar ventilation or more efcient pulmonary oxy-
gen uptake due to improved ventilation:perfusion
matching. Alveolar ventilation can be estimated by
measuring PaCO
2
. A calculation derived from the al-
veolar gas equation was used to estimate the A-a
difference. This can be used as a measure of pulmonary
oxygen uptake efciency as it adjusts for the effects of
alveolar ventilation on oxygenation. The modied al-
veolar gas equation used assumed a FiO
2
of 0.21 (21%).
This assumption was justied on the basis that animals
were not being supplemented with oxygen. The initial
RQ used in A-a difference calculations was 0.9. The RQ
is determined by dividing CO
2
production in the body
by its O
2
consumption and varies with stress, metabolic
state, disease state, nutritional state, hormonal rhythms,
noise, breed, and exercise state.
1618
All dogs were fed
typical canine diets and had been fasted for 12 h or
more but they were in different disease and metabolic
states. It was not possible to measure RQ in this group
of clinical patients so RQ was estimated. Healthy dogs
have been shown to have an RQ of 0.9 but critically ill
dogs may have an RQ of closer to 0.7.
1618
The dogs in
this study likely had RQs falling between these 2 val-
ues. Assuming a RQ higher than it actually was would
make the A-a difference calculated falsely high and as-
suming a lower RQ would make the A-a difference
falsely low. Even so the calculation of A-a difference
remains useful in determining whether the hypoxemia
was a result of alveolar hypoventilation on an individ-
ual patient basis. Also, regardless of whether 0.7 or 0.9
was used as the RQ, changes in A-a difference re-
mained statistically signicant. Another point to note is
that 1 animal had a negative A-a difference in sternal
recumbency when using an RQ of 0.7. A-a difference
must be a positive value so this is a clear error. As this
was not a dog with a potentially erroneously low PaCO
2
it is likely that the negative value is a consequence of the
RQ being inappropriately low for in this individual.
A-a difference was signicantly lower in sternal re-
cumbency, whereas there was no signicant change in
PaCO
2
. It appears that improved pulmonary oxygen
uptake was the major mechanism responsible for the
higher PaO
2
in sternal recumbency. This is likely due to
better matching of ventilation and perfusion through-
out the lung in the sternal position.
Two of the dogs had abnormally low measured
PaCO
2
levels in sternal recumbency. These results were
probably due to alveolar hyperventilation in these dogs
as a consequence of increased respiratory rate or tidal
volume. Stress, pain, dysphoria secondary to opiate
administration, or increased stimulation in sternal
recumbency are potential causes for this. An alterna-
tive explanation would be sampling error, particularly
if there had been small air bubbles in the syringe in
contact with the sample.
19
Although each syringe was
checked for air and run within 90 seconds of sampling
this source of error cannot be totally excluded. These
results were included in the original analysis and sub-
sequent statistical analysis with these results excluded
conrmed minimal effects on the statistical signicance
of PaO
2
or A-a difference changes. One option that
would have avoided these potential discrepancies
would have been to run samples in duplicate.
There were a number of pretrial factors that could
have affected the measured PaO
2
and PaCO
2
. Many of
the patients underwent general anesthesia before in-
clusion and this may have been a confounding factor.
Most anesthetic drugs have dose-dependent respiratory
depressant effects, but atelectasis, respiratory failure
secondary to mechanical ventilation, pneumonia, la-
ryngeal oedema and spasm, bronchospasm, and re-
duced hypoxic pulmonary vasoconstriction are all
known to be pulmonary effects of general anesthe-
sia.
2026
Patients were not eligible for inclusion until 3
hours after endotracheal tube removal. A previous
study has shown minimal residual effects on arterial
blood-gas analysis in dogs undergoing routine neuter-
ing with a standardized anesthetic protocol after this
time.
27
Longer anesthetic times could have allowed
more inhaled anesthetic (and potentially other drugs) to
build up in the tissues, resulting in longer-lasting re-
spiratory effects. Although all the dogs appeared to be
fully recovered from anesthesia, subjectively appearing
to be free from any of the cited complications, it is not
certain if any of the dogs respiratory systems were still
affected by anesthetic agents.
& Veterinary Emergency and Critical Care Society 2009, doi: 10.1111/j.1476-4431.2009.00480.x 568
M.W. McMillan et al.
A heterogeneous group of dogs was used in order to
maximize the number of patients eligible for inclusion
therefore patients with a range of diseases including
some with identied pulmonary disease were included.
The distribution of lung pathology and its potential
effect on the blood-gas results in different body positions
was not assessed; a previous human study suggests this
may be an important factor determining PaO
2
levels in
lateral recumbency.
28
In the study, oxygenation tended to
be higher with the diseased lung uppermost.
28
Also, this trial did not evaluate any bias due to the
body position the patients had been in before starting
the trial. It is suggested that prolonged recumbency will
lead to deterioration of oxygenation and pulmonary
oxygen uptake in spontaneously breathing, conscious
animals due to dependent lobe atelectasis.
29
This would
mean the position of the dog before sampling and the
distribution of any respiratory disease would both be
factors in determining PaO
2
levels when in lateral body
positions. Respiratory disease, recumbency atelectasis,
or the affects of anesthesia could account for abnormal
PaO
2
in the sampled dogs. Differentiating between re-
spiratory complications caused by these 3 factors was
not possible in this study. Any bias caused by a failure
to standardize the anesthetic protocols and the waning
effects of anesthetic drugs would be reduced by the 3-
hour waiting period and by the randomization of the
initial positioning. Although it cannot be used to justify
a 3-hour waiting period, the lack of statistical difference
in PaO
2
, PaCO
2
, and A-a difference between rst and
second samples supports the conclusion that the sig-
nicant changes identied were due to the change in
body position rather than simply a time effect. The
effect of the animals position in the period before sam-
pling was reduced by the random designation of body
position order but could not be eliminated.
It should be noted that most of the animals included in
this study had calculated A-a differences outside the
published reference intervals (5.5 7.7 or 6.3 7.7
mmHg) when using an RQ of 0.9.
30,31
Although most
of the A-a differences improved when the patients were
positioned in sternal recumbency it only normalized in 2
cases. This would indicate some impairment in pulmo-
nary oxygen uptake across the sampled population,
despite only 6 dogs having respiratory disease identied.
As previously discussed the use of an RQ value of 0.9
will have caused overestimation of the A-a difference in
animals whose true RQ values were lower than this.
Hence calculations were repeated with an RQ of 0.7.
Alongside the dogs with respiratory disease, 3 dogs with
no diagnosed respiratory disease continued to have ab-
normal A-a differences while in lateral recumbency with
the RQ of 0.7. The A-a differences in these dogs were
normal in sternal recumbency. All 3 dogs had undergone
general anesthesia. Again prolonged effects of anesthetic
agents, recumbency atelectasis, or undiagnosed respira-
tory disease are the possible causes for the abnormal A-a
differences seen in these dogs.
There may be further limitations to the conclusions
that can be drawn from these results. The blood-gas
measurements were not temperature corrected; varia-
tions in body temperature may affect the solubility of
gases in vivo. Temperature correction of blood gases is
a topic of much debate but the available data suggests
that there is no clinical advantage to using values other
than those at 371C in almost all circumstances.
32
Fur-
thermore, this study compared the A-a differences of
the sampled animals. When the PaO
2
is calculated via
the alveolar gas equation a 371C temperature is as-
sumed and therefore the 371C PaO
2
value should be
used for comparison.
32
Measurements were not duplicated and this could
have ruled out sample error as the cause of the abnor-
mally low PaCO
2
levels in 2 patients. Animals with
more severe disease that required oxygen supplement-
ation were also not included. Their inclusion would
have complicated data analysis as an accurate FiO
2
cannot be determined easily. Therefore, any additional
effect on PaO
2
in addition to oxygen supplementation
could not be assessed. However, it is likely that, in an-
imals receiving oxygen supplementation, pulmonary
function would still benet through improved ef-
ciency of oxygen uptake.
The only published randomized prospective study in
humans suggested that the prone position was safe and
effective in improving PaO
2
but did not show a reduc-
tion in mortality.
3
The present study reports the short-
term effects of a recumbent dogs body position but did
not evaluate any longer term effects on oxygenation or
effects on morbidity and mortality. One of the few vet-
erinary reviews on the subject of managing recumbent
veterinary patients discusses that large and giant breed
dogs, especially when obese, could develop respiratory
dysfunction that can lead to hypoventilation and hy-
poxemia.
33
Turning of patients is mentioned to help
combat this but sternal positioning is recommended
only as a way to reduce aspiration.
33
A longer term
study to evaluate the effects of regular turning and
other interventions on oxygenation and atelectasis in
recumbent patients is warranted.
In conclusion, positioning canine patients in sternal
recumbency appears to be a useful intervention that can
improve PaO
2
in the short-term, principally due to im-
provements in pulmonary oxygen uptake. This study
suggests that there may be a benet to maintaining hy-
poxemic dogs (PaO
2
o80 mmHg) in sternal recum-
bency and to recovering dogs where there is a concern
over pulmonary function in sternal recumbency after
& Veterinary Emergency and Critical Care Society 2009, doi: 10.1111/j.1476-4431.2009.00480.x 569
Effect of body position on canine blood gases
general anesthesia. As not all patients will have im-
provements in their PaO
2
levels, patients should still be
monitored for clinical deterioration following any
change in body position. Positioning patients in ster-
nal recumbency is a cheap, easy, and freely available
adjunctive measure that can be used alongside oxygen
supplementation in the management of animals with
mild hypoxemia to improve PaO
2
levels. This may be
useful as an adjunctive measure in some hypoxemic
patients where mechanical ventilation is not possible,
giving time for disease-specic therapies to work and
for the patients pulmonary function to improve. Fur-
ther studies are warranted to evaluate patients with
more severe hypoxemia, to investigate the changes in
pulmonary function over a longer time frame and to
establish the effects of differing distributions of pul-
monary lesions.
Footnotes
a
Microsoft Excel, MS Ofce 2003, Seattle, WA.
b
Digital thermometer, Model VT-801SLEW, Jorgen Kruuse A/S,
Denmark.
c
BD Preset Critical Care Sampling Kit, BD Vacutainer Systems, BD UK/
Ireland, Oxford, UK.
d
STAT Prole Plus Nine, Critical Care Xpress (CCX), NOVA Biomedical,
Waltham, MA.
e
Graphpad Prism, Graphpad Software Inc, La Jolla, CA.
References
1. Pelosi P, Brazzi L, Gattinoni L. Prone position in acute respiratory
distress syndrome. Eur Respir J 2002; 20:10171028.
2. Wells D, Gillies D, Fitzgerald D. Positioning for acute respiratory
distress in hospitalised infants and children. Cochrane Database of
Systematic Reviews, 2005, Issue 2. Art. No.: CD003645. DOI:
10.1002/14651858.CD003645.pub2.
3. Gattinoni L, Tognoni G, Pesenti A, et al. Effect of prone positioning
on the survival of patients with acute respiratory failure. N Engl J
Med 2001; 345(8):568573.
4. Richter T, Bellani G, Harris RS, et al. Effect of prone position on
regional shunt, aeration, and perfusion in experimental acute lung
injury. Am J Respir Crit Care Med 2005; 172:480487.
5. Wang J, Abu-Zidan FM, Walther SM. Effects of prone and supine
posture on cardiopulmonary function after experimental chlorine
gas lung injury. Acta Anaesthesiol Scand 2002; 49(9):10941102.
6. Albert RK, Leasa D, Sanderson M, et al. The prone position im-
proves arterial oxygenation and reduces shunt in oleic-acid-in-
duced acute lung injury. Am Rev Respir Dis 1987; 135(3):628633.
7. Beck KC, Vettermann J, Rehder K. Gas exchange in dogs in the
prone and supine positions. J Applied Physiol 1992; 72(6):2292
2297.
8. Lamm WJ, Graham MM, Albert RK. Mechanism by which the
prone position improves oxygenation in acute lung injury. Am J
Respir Crit Care Med 1994; 150(1):184193.
9. Du HL, Yamada Y, Orii R, et al. Benecial effects of the prone
position on the incidence of barotraumas in oleic acid-induced
lung injury under continuous positive pressure ventilation. Acta
Anaesthesiol Scand 1997; 41(6):701707.
10. Broccard AF, Shapiro RS, Schmitz LL, et al. Inuence of prone
position on the extent and distribution of lung injury in a high
tidal volume oleic acid model of acute respiratory distress syn-
drome. Crit Care Med 1997; 25(1):1627.
11. Lim C-M, Koh Y, Chin JY, et al. Respiratory and haemodynamic
effects of the prone position at two different levels of PEEP in a
canine acute lung injury model. Eur Respir J 1999; 13:163168.
12. Cakar N, Van de Kloot T, Youngblood M, et al. Oxygenation re-
sponse to a recruitment maneuver during supine and prone po-
sitions in an oleic acid-induced lung injury model. Am J Respir
Crit Care Med 2000; 161:19491951.
13. Broccard AF, Shapiro RS, Schmitz LL, et al. Prone positioning at-
tenuates and redistributes ventilator-induced lung injury in dogs.
Crit Care Med 2000; 28(2):295303.
14. Hwang JH, Kwon YS, Kang EH, et al. Prone positioning improves
oxygenation without adverse hemodynamic effects during partial
liquid ventilation in a canine model of acute lung injury. Korean J
Intern Med 2004; 19(4):237242.
15. Hubmayr RD, Rodarte JR, Walters BJ, et al. Regional ventilation
during spontaneous breathing and mechanical ventilation in dogs.
J Appl Physiol 1987; 63:24672465.
16. Walters LM, Ogilvie GK, Salman MD, et al. Repeatability of energy
expenditure measurements in clinically normal dogs by use of
indirect calorimetry. Am J Vet Res 1993; 54:18811885.
17. Ogilvie GK, Salman MD, Kesel ML, et al. Effect of anesthesia and
surgery on energy expenditure determined by indirect calorimetry
in dogs with malignant and nonmalignant conditions. Am J Vet
Res 1996; 57:13211326.
18. Walton RS, Wingeld WE, Ogilvie GK, et al. Energy expenditure in
104 postoperative and traumatically injured dogs with indirect
calorimetry. J Vet Emerg Crit Care 1996; 6:7179.
19. Biswas CK, Ramos JM, Agroyannis B, et al. Blood gas analysis:
effect of air bubbles in syringe and delay in estimation. BMJ 1982;
284:923927.
20. Brainard BM, Alwood AJ, Kushner LI, et al. Postoperative pul-
monary complications in dogs undergoing laparotomy: anesthetic
and perioperative factors. J Vet Emerg Crit Care 2006; 16(3):184
191.
21. Brainard BM, Alwood AJ, Kushner LI, et al. Postoperative pul-
monary complications in dogs undergoing laparotomy: frequency
characterisation and disease related risks. J Vet Emerg Crit Care
2006; 16(3):176184.
22. Duggan M, Kavanagh BP. Pulmonary atelectasis: a pathogenic
perioperative entity. Anesthesiology 2005; 102(4):838854.
23. Staferi F, Franchini D, Carella GL, et al. Computed tomographic
analysis of the effects of two inspired oxygen concentrations on
pulmonary aeration in anesthetized and mechanically ventilated
dogs. Am J Vet Res 2007; 68(9):925931.
24. Domino KB, Borowec L, Alexander CM, et al. Inuence of isou-
rane on hypoxic pulmonary vasoconstriction in dogs. Anesthesi-
ology 1986; 64(4):423429.
25. Hedenstierna G. Airway closure, atelectasis and gas exchange
during anesthesia. Minerva Anestesiologica 2002; 68(5):332336.
26. Clercx C, van den Brom WE, de Vries HW. Effect of posture and
anesthesia on the distribution of pulmonary perfusion and lung
conguration in beagle dogs. Res Vet Sci 1989; 47:359366.
27. Campbell VL, Drobatz KJ, Perowski SZ. Postoperative hypoxemia
and hypercarbia in healthy dogs undergoing routine ovariohys-
terectomy or castration and receiving butorphanol or hydromor-
phone for analgesia. J Am Vet Med Assoc 2003; 222(3):330336.
28. Kim MJ, Hwang HJ, Song HH. A randomized trial on the effects of
body positions on pulmonary gas exchange with acute respiratory
failure patients. Int J Nurse Stud 2002; 39:549555.
29. Farrow CS. Postural atelectasis in a dog. Modern Vet Practice 1984;
65(6):474.
30. Haskins S, Pascoe PJ, Ilkiw JE, et al. The effect of moderate hypo-
volemia on cardiopulmonary function in dogs. J Vet Emerg Crit
Care 2005; 15(2):100109.
31. Haskins S, Pascoe PJ, Ilkiw JE, et al. Reference cardiopulmonary
values in normal dogs. Comp Med 2005; 55(2):156161.
32. Shapiro BA. Temperature correction of blood gas values. Resp Clin
N Am 1995; 1(1):6976.
33. Nicholl SA, Remedios AM. Recumbency in small animals: patho-
physiology and management. Compend Contin Edu Pract Vet
1995; 17(11):13671374.
& Veterinary Emergency and Critical Care Society 2009, doi: 10.1111/j.1476-4431.2009.00480.x 570
M.W. McMillan et al.