Feature

Nurse-Led Implementation of a
Safe and Effective Intravenous
Insulin Protocol in a Medical
Intensive Care Unit
Rabia Khalaila, RN, PhD
Eugene Libersky, RN, BSN
Dina Catz, RN, BSN
Elina Pomerantsev, RN, BSN
Abed Bayya, MD
David M. Linton, MD
Sigal Sviri, MD

H
BACKGROUND Recent evidence has linked tight glucose control to worsened clini- yperglycemia and
cal outcomes among adults in intensive care units. insulin resistance are
OBJECTIVE To evaluate the effectiveness and safety of a nurse-led intravenous common in critically
insulin protocol designed to achieve conservative blood glucose control in patients ill patients, even those
in a medical intensive care unit.
without a history of
METHODS A nurse-led intravenous insulin protocol was developed, targeting blood
diabetes mellitus.1,2 Patients in
glucose levels at 110 to 149 mg/dL. Hypoglycemia was defined as a blood glucose level
less than 70 mg/dL. Patients admitted to the medical intensive care unit who required
intensive care units (ICUs) experi-
an insulin infusion were enrolled in the study. Blood glucose levels in those patients ence significant physiological stress,
were compared with levels in 153 historical control patients admitted to the unit in the which leads to disturbances in the
12 months before the protocol was implemented who required an insulin infusion. endocrine axis such as changes in
RESULTS Ninety-six patients were enrolled and treated with the protocol. The the levels of epinephrine, cortisol,
protocol and control groups had similar characteristics at baseline. More measure- growth hormone, and glucagon.
ments in the protocol group than in the control group (46.3% vs 36.1%, P < .001) Such changes may subsequently
were within the target glucose range (110-149 mg/dL). Hyperglycemia (blood glu- cause hyperglycemia.3,4 Administra-
cose ≥ 200 mg/dL) occurred less often in the protocol group than in the control tion of certain medications such as
group (14.8% vs 20.1%, P = .003). Hypoglycemic events (blood glucose <70 mg/dL)
norepinephrine,5 corticosteroids,6
also occurred less often in the protocol group (0.07% vs 0.83%, P < .001).
and/or high-caloric enteral or par-
CONCLUSIONS Implementation of a nurse-led, conservative intravenous insulin
enteral nutrition3 may also aggra-
protocol in the medical intensive care unit is effective and safe and markedly
reduces the rate of hypoglycemia. (Critical Care Nurse. 2011;31[6]:27-35) vate hyperglycemia in these patients.
A growing body of evidence has
©2011 American Association of Critical-Care Nurses doi: http://dx.doi.org/10.4037/ccn2011934 linked hyperglycemia and insulin

www.ccnonline.org CriticalCareNurse Vol 31, No. 6, DECEMBER 2011 27

resistance to worsened clinical out- did not increase mortality,20 the blood glucose control were enrolled
comes in various groups of patients.1,7-9 NICE-SUGAR multicenter study23 in the prospective group. Patients
In particular, impaired glycemic con- recently showed that tight glucose admitted with diabetic ketoacidosis,
trol has been associated with increased control increased mortality and hyperglycemic hyperosmotic nonke-
hospital mortality in patients admit- increased the risk of severe hypo- totic coma, or who were being fed
ted to ICUs.10 The effects of blood glu- glycemia among adults in the ICU. orally were excluded. Patients
cose control with insulin on patients’ These conflicting results led to the enrolled in the retrospective control
outcomes have been investigated in recommendation to avoid lower group were those who had been
a number of studies,7,11 and results glucose targets in critically ill adults. admitted to the medical ICU during
have been markedly positive. In a We undertook this study to the 12 months preceding implemen-
prospective randomized controlled develop and evaluate a nurse-led, tation of the protocol who required
trial11 of 1548 surgical ICU patients, self-adjusting, standardized intra- intravenous insulin therapy. Patients
hospital mortality was reduced by venous insulin protocol designed to with ketoacidosis and hyperglycemic
34% after an intravenous insulin pro- achieve safe and effective control of hyperosmotic nonketotic coma were
tocol was introduced. Following this blood glucose levels in patients in excluded from the control group.
landmark study, various intensive the medical ICU by using a more Patients with a history of diabetes
insulin protocols aimed at achieving conservative glucose target and a mellitus were included in both groups.
inpatient glycemic control in the higher hypoglycemic threshold. Oral Hypoglycemic Agents. As a
ICU,1,7,12-18 including the medical routine, oral hypoglycemic agents
ICU,19,20 have been described. Methods are withheld during insulin adminis-
The optimal target range for blood Patients and Setting tration in the medical ICU. There-
glucose level in critically ill patients This prospective study with a fore, none of the patients in the
remains unclear. Most researchers retrospective control group was per- protocol group or the control group
prefer to use tight ranges of blood formed in a 9-bed medical ICU at received oral hypoglycemic agents
glucose control (80-110 mg/dL; mul- the Hadassah Hospital in Jerusalem, during the study.
tiply by 0.0555 to convert to mil- Israel. The medical ICU staff includes Standard Insulin Therapy (Control
limoles per liter) in order to prevent intensive care specialists, rotating Group). Before development of the
hyperglycemia; however, these medical residents in training, and protocol, elevated blood glucose
ranges increase the risk for hypo- ICU-trained and non–ICU-trained measurements were often untreated
glycemia.2,11,12,14,20-22 Others prefer to nurses. Approval for access to until they reached 180 mg/dL or
use less stringent ranges for blood patients’ data and a waiver of higher. When patients consistently
glucose (eg, 100-140 mg/dL) in order informed consent were obtained had blood glucose levels of 180
to minimize the risk for hypoglycemia from the institutional review board. mg/dL or greater, an insulin infusion
but still prevent hyperglycemia.1,7,16,19 Patients. All adult patients admit- was ordered. A typical order was to
Although the first randomized con- ted to the medical ICU during the adjust insulin to maintain blood glu-
trolled trial showed that tighter glu- 12 months after the initiation of the cose levels between 90 and 140
cose control (80-110 mg/dL) reduced nurse-led intravenous insulin proto- mg/dL. Orders of this type did not
morbidity and mortality,11 or at least col who required insulin infusion for direct the nurses when and how
much to increase or decrease the
insulin infusion; this resulted in glu-
Authors
Rabia Khalaila is the head nurse; Eugene Libersky, Dina Catz, and Elina Pomerantsev
cose control that varied with nurses’
are staff nurses; Abed Bayya and Sigal Sviri are consultant intensivists; and David M. experience and comfort level.
Linton is a professor and the director in the medical intensive care unit at Hadassah–
Hebrew University Medical Center, Jerusalem, Israel.
Development and Implementation
Corresponding author: Rabia Khalaila, RN, PhD, Head Nurse, Medical ICU, Hadassah–Hebrew University Medical of the Intravenous Insulin Protocol
Center, P. O. Box 91120, Jerusalem, Israel (e-mail: rabeik@hotmail.com).
The development of an insulin
To purchase electronic or print reprints, contact The InnoVision Group, 101 Columbia, Aliso Viejo, CA 92656.
Phone, (800) 899-1712 or (949) 362-2050 (ext 532); fax, (949) 362-2049; e-mail, reprints@aacn.org. infusion protocol was initiated by key

28 CriticalCareNurse Vol 31, No. 6, DECEMBER 2011 www.ccnonline.org

contributors from a team of nurses
and physicians from various disci-
plines. The protocol was written in a
logarithm format table that allowed
shifting between several algorithms
horizontally and vertically, depend-
ing on changes in capillary blood
glucose values from the previous
blood glucose measurement (see
Appendix—available online only,
at www.ccnonline.org). The proto-
col consisted of 5 algorithms. All
patients started in algorithm 0.
Nurses were instructed to start the
insulin protocol when 2 consecutive
blood glucose measurements were
150 mg/dL or greater, or when 1
blood glucose measurement was
200 mg/dL or greater.
The target blood glucose concen-
tration was set at an acceptable range
of 110 to 149 mg/dL,19 and hypo-
glycemia was defined as a blood
glucose level less than 70 mg/dL.
This target range for blood glucose
was selected to allow adequate glu-
cose control while preventing hypo-
glycemic episodes.
If the blood glucose level
decreased to less than 110 mg/dL,
the infusion was stopped, and blood
glucose levels were rechecked hourly
until they reached 150 mg/dL or
greater. Once the blood glucose
level was 150 mg/dL or greater, the
infusion was restarted at 1 algo-
rithm less than the last algorithm
used. Whenever the blood glucose
concentration decreased to less than
70 mg/dL, the infusion was turned
off and 25 mL of 25% dextrose injec-
tion was administered intravenously.
Capillary blood glucose levels were
then rechecked per protocol, and the
insulin infusion was restarted when
blood glucose level was 150 mg/dL
or greater, with algorithm 0 (see
www.ccnonline.org
Appendix—online only). No special
adjustments were made for the use
of corticosteroids, parenteral nutri-
tion, or substitution solutions with
glucose-containing infusions. All
patients who received an insulin
infusion also were receiving 1 or
more of the following: glucose infu-
sions, total parenteral nutrition
(TPN), or enteral feeding.
For measurement of blood glu-
cose levels, nurses used bedside glu-
cometers or undiluted heparinized
arterial blood (if systolic blood pres-
sure was <80 mm Hg). Capillary
blood glucose levels were measured
hourly in the first 8 hours of use of
the protocol, and then decreased to
every 2 hours until the glucose level
had been within the target range
(110-149 mg/dL) for 8 hours. Sub-
sequently, the frequency of capil-
lary blood glucose measurements
was changed to every 4 hours.
Insulin was administered by contin-
uous intravenous infusion through
a peripheral or a central venous
catheter with a 50-mL syringe pump.
The standard concentration of the
infusion was 1 IU/mL (50 IU of reg-
ular insulin in 50 mL of 0.9% saline).
In an effort to provide the nurses
with as much support as possible
before the protocol was imple-
mented, a plan was created to facili-
tate acceptance and correct use of
the protocol. Initially, the protocol
and updated literature about the
importance of glycemic control in
critical care patients were presented
in our bimonthly staff meeting. The
meeting involved all members of the
medical ICU (30 nurses and 4 physi-
cians), and lasted approximately 11⁄2
hours, including time for questions
and feedback. Subsequently, all
nursing staff in the medical ICU
CriticalCareNurse Vol 31, No. 6, DECEMBER 2011 29
received informal education for 2
weeks from the key project develop-
ers and contributors. The educa-
tional sessions lasted approximately
30 minutes and included explana-
tions about the proper use of the
protocol and its bedside implemen-
tation. In addition, during the study
period, a bedside chart containing
the protocol guidelines was avail-
able. This chart also was used to fill
in the date, hour, blood glucose
measurements, protocol number,
and the corresponding rate of the
insulin infusion.
Evaluation of the Protocol
The protocol was evaluated
with respect to effectiveness and
safety. Effectiveness was evaluated
by calculating the percentage and
mean percentage of blood glucose
values within the target range (110-
149 mg/dL) during implementation
of the protocol. The different blood
glucose levels were defined as hypo-
glycemic (<70 mg/dL), low (≥70
but <110), target (≥110 but <150),
acceptable hyperglycemic (≥150 but
<200), and severely hyperglycemic
(≥200). The higher the mean per-
centage of blood glucose measure-
ments within the target range, the
better the effectiveness. Safety was
assessed by determining the mean
percentage of blood glucose meas-
urements less than 70 mg/dL and
the percentage of patients with at
least 1 hypoglycemic episode.
Data Collection
Clinical data were obtained
from patients’ charts and records.
Data were collected prospectively
in the protocol group and retro-
spectively in the control group.
Patients’ baseline characteristics
included age, sex, history of diabetes,
and principal reason for ICU admis- Table 1 Comparison of baseline characteristics, diagnoses at admission,
clinical interventions, and illness severity in the protocol and control groupsa
sion. In addition, we calculated
Variables Protocol group (n = 96) Control group (n = 153)
patients’ score on the Acute Physiol-
Age, mean (SD), y 66 (19.9) 68 (16.2)
ogy and Chronic Health Evaluation
Male sex 46 (48) 83 (54)
II (APACHE II) and the predicted
History of diabetes 51 (53) 75 (49)
mortality for every patient at admis-
sion. For each patient, data on blood Diagnosis category
Respiratory problem 48 (50) 83 (54)
glucose levels, insulin doses, dura- Infection 32 (33) 57 (37)
tion of insulin therapy, blood glucose Sepsis/septic shock 14 (15) 27 (18)
Cardiovascular problem 15 (16) 24 (16)
level at admission and when insulin Gastrointestinal or liver problem 15 (16) 15 (10)
was started, time to reach target Hematologic/oncological disorder 6 (6) 5 (3)
glucose levels, relevant clinical inter- Renal problem 19 (20) 34 (22)
Neurological problemb 10 (10) 2 (1)
ventions (corticosteroids, vasopres- Metabolic disorder 6 (6) 6 (4)
sors, antibiotics, nutrition), and Other problem 14 (15) 32 (21)
calories administered were collected Clinical interventions
Enteral nutrition 79 (82) 120 (78)
from the active hospital chart and Corticosteroid therapy 65 (68) 105 (69)
the nursing records in the medical Vasopressor therapy 51 (53) 72 (47)
ICU. All blood glucose measure- Antibiotic therapy 95 99) 147 (96)
ments obtained during the period Nonprotein calories administered
on days 1-3, kcal/d
of intravenous insulin administra- Mean (SD) 700.7 (458.1) 670.9 (488.5)
tion were documented. Median 577.8 541.3
Finally, data on the total duration Illness severity
of mechanical ventilation, length of APACHE II score, mean (SD) 26.6 (7.9) 28.6 (9.3)
stay in the medical ICU, dialysis Abbreviation: APACHE II, Acute Physiology and Chronic Health Evaluation II.
after admission, and ICU mortality a Values in second and third columns are number (% of group) unless otherwise indicated.
b Difference between protocol group and control group is significant (P = .002) for this variable.
also were collected. All patients were
followed up until discharge from
the ICU or death. blood glucose control. All tests of of cerebrovascular accident that
significance were 2 tailed. P values required admission to the ICU dur-
Statistical Analysis less than .05 were considered statis- ing the protocol period.
Descriptive statistics were gener- tically significant. For the 96 patients in the proto-
ated by the SPSS Statistical Software col group, the median duration of
Package Version 12 (IBM SPSS Sta- Results insulin therapy was 7 days, with 84
tistics, Armonk, New York). Clinical Two hundred forty-nine patients (88%) continuing for more than 3
data are expressed as median, mean were enrolled in the study: 96 in the days. Overall for the protocol group,
(SD), or as a percentage. Subgroups prospective protocol group and 153 the median insulin dose used was
of patients were compared by using in the retrospective control group. 25.93 units per patient per 24 hours
c2 analysis or a Student independ- Baseline characteristics, diagnoses and the median blood glucose level
ent t test. The mean percentage of at admission, clinical interventions, at admission was 173 mg/dL. The
blood glucose levels in the protocol calories administered, and illness median blood glucose level at proto-
group and the control group were severity are presented in Table 1. col initiation was 212 mg/dL, and
compared by using a Student inde- The groups were well matched, the median time required to achieve
pendent t test. Hierarchical linear except for a larger group of neuro- target blood glucose levels (110-149
regression analysis was used to detect logical patients in the protocol mg/dL) was 8 hours (Table 2).
factors that could have influenced group (10% vs 1%, P=.002), attrib- During the 12-month study
hypoglycemia and the tightness of uted to a new protocol for treatment period, the 96 patients had a total

30 CriticalCareNurse Vol 31, No. 6, DECEMBER 2011 www.ccnonline.org


Table 2 Blood glucose measurements and insulin therapy in the 2 groups
Variable
Duration of insulin therapy, d
Mean (SD)
Median
Blood glucose level,a mg/dL
At admission
Mean (SD)
Median
At start of protocol use
Mean (SD)
Median
Time to goal, h
Mean (SD)
Median
Blood glucose samples and insulin administration
Total No. of measurements
Total No. of days of insulin therapy
No. of samples per patient, mean (SD)b
Hypoglycemic values (blood glucose <70
mg/dL), mean (SD), mg/dL
No. (%) of patients with at least 1 blood
glucose level <70 mg/dLc
Insulin dose per patient per 24 h, units
Mean (SD)
Median
a Blood glucose values can be converted to millimoles per liter by multiplying by 0.0555.
b Difference between protocol group and control group is significant (P = .03) for this variable.
c Difference between protocol group and control group is significant (P < .001) for this variable.
of 925 protocol days. Of the 9893
measurements obtained during
those 925 protocol implementation
days, 649 (6.56%) were in the low
range (≥70 but <110 mg/dL), 4546
(45.95%) in the target range (≥110
but <150 mg/dL), 3232 (32.66%) in
the acceptable hyperglycemic range
(≥150 but <200 mg/dL), and 1449
(14.64%) in the severely hyper-
glycemic range (≥200 mg/dL).
Protocol Safety
Hypoglycemia (blood glucose
<70 mg/dL) in the protocol group
was rare. Of the 9893 blood glucose
measurements, only 11 measure-
ments (0.11%) in 6 patients (6%)
were less than 70 mg/dL. The mean
blood glucose level in these meas-
urements was 57.0 (SD, 11.5) mg/dL
were within the target glucose range
(110-149 mg/dL; see Figure 1).
Protocol group Control group Hypoglycemic events (blood glu-
(n = 96) (n = 153)
cose <70 mg/dL) occurred less often
9.6 (6.5) 10.3 (9.3)
in the protocol group than in the
7 7 control group (7/10 000 measure-
ments vs 83/10 000 measurements,
respectively, P<.001; see Figure 2),
185.9 (77.1) 176.9 (82.3)
173 157 and fewer patients experienced 1
or more episodes of hypoglycemia
222.3 (52.4) 228.4 (68.3)
(6% vs 30%, respectively, P<.001).
212 208
Two separated hierarchical mul-
10.3 (11.5) 8.9 (7.5) tiple linear regressions were carried
8 7 out to determine the extent to which
the use of the protocol was predic-
9893 12 256
tive of 2 outcomes: (1) hypoglycemic
925 1577
103.10 (74.7) 80.11 (82.2) events, and (2) the tightness of
blood glucose control, after the fol-
57.0 (11.5) 58.5 (10.8)
lowing variables were controlled for:
6 (6) 46 (30) number of glucose measurements,
the duration of insulin therapy, his-
29.41 (18.98) 32.63 (21.1)
25.93 28.33 tory of diabetes, and neurological
diagnosis. Therefore, the predictors
of both outcomes were entered into
the model in 2 steps: (1) control
variables, and (2) the use of the
(Table 2). All episodes of hypo- protocol. The first regression showed
glycemia resulted in the administra- that only the use of the protocol
tion of a rescue bolus of dextrose (b=-.28) was independently associ-
per the protocol. All episodes of ated with a lower probability for
hypoglycemia lasted a mean of 30 hypoglycemic events, with an
(SD, 20) minutes. None of these adjusted R2 of 0.16 (P<.001), regard-
hypoglycemic events resulted in less of the effects of the control vari-
adverse consequences. ables. The second regression revealed
2 significant predictors of the tight-
Protocol Effectiveness ness of blood glucose control with
To assess the effectiveness and an adjusted R2 of 0.18 (P<.001): use
safety of the protocol compared with of the protocol (b=.35) and history
our previous practice standards, we of diabetes (b=-.26, data not shown).
used 153 patients consecutively (b indicates the relative weight of
admitted to our medical ICU during the standardized predictor variable
the 12 months before the protocol in question. A positive value indi-
was implemented. Figures 1 and 2 cates a positive relationship with the
show blood glucose levels in the 2 outcome, whereas a negative value
groups. More samples in the proto- indicates a negative association.
col group than in the control group Adjusted R2 measures the proportion
(46.3% vs 36.1%, respectively, P<.001) of the variation in the dependent

www.ccnonline.org CriticalCareNurse Vol 31, No. 6, DECEMBER 2011 31

 Discussion
P < .001
50 Our data have demonstrated
45 that a nurse-led intravenous insulin
protocol with a more conservative
40
blood glucose target is both effective
P = .11 35
and safe in medical ICU patients.

% of patients
30 When compared with historical
25 control patients, patients in the pro-
P = .003
P < .001 20
tocol group had more blood glucose
P < .001
measurements within the target
15
range, fewer blood glucose measure-
10 ments in the hyperglycemic range,
5 and significantly fewer hypoglycemic
0 episodes. Reduced hypoglycemia
70-109 110-149 150-199 ≥200 was significantly associated with
Blood glucose level, mg/dL use of the protocol. Lengths of stay
Control Protocol in the ICU and hospital were shorter
in the protocol group.
Figure 1 Blood glucose levels in the control and protocol groups. To convert to Studies on tight glycemic con-
millimoles per liter, multiply by 0.0555. trol in ICU patients have shown con-
flicting results, with both improved
outcomes and increased morbidity
90
and mortality reported.11,22-24 Studies
Frequency per 10 000 measurements

80 in which tight glycemic control was

P < .001
Control
Group
Figure 2 Frequency of blood glucose measurements less than 70 mg/dL in the
control and protocol groups. To convert to millimoles per liter, multiply by 0.0555.
variable accounted for by the explan-
atory variables.)
Severe hyperglycemia (blood glu-
cose ≥200 mg/dL) occurred less often
in the protocol group than in the con-
trol group (14.8% vs 20.1%, respec-
tively, P=.003), but no significant
reduction was observed in the accept-
70 targeted (80-110 mg/dL) were asso-
ciated with increased hypoglycemic
60
rates even with very low hypo-
50
glycemia thresholds.21,22,25 Alm-Kruse
40 et al21 reported that glucose targets
30 of 80 to 110 mg/dL resulted in an
20
increase in hypoglycemic episodes
(blood glucose <40 mg/dL) from
10
2.4% to 8.9%. The recent NICE-
0 SUGAR study23 demonstrated
Protocol
increased mortality in patients
treated with intensive glucose con-
trol compared with patients in whom
a conventional target of 180 mg/dL
or less was used (27.5% vs 24.9%,
able hyperglycemia range (Figure 1). respectively, P=.02), together with
Finally, no significant reduction was an increased risk of hypoglycemia
observed in mortality, duration of (6.8% vs 0.5%, P<.001). That study
mechanical ventilation, and dialysis raised pertinent questions regard-
after the implementation of the pro- ing the safety of the use of intensive
tocol. A significant reduction was glucose targets; however, it did not
observed in the length of stay in the address the safety of the use of
ICU and in the hospital (Table 3). higher glucose targets.

32 CriticalCareNurse Vol 31, No. 6, DECEMBER 2011 www.ccnonline.org


Table 3 Patients’ outcomes
Outcome
Total days of ventilation, mean (SD)
Hemodialysis after admission, No. (%) of patients
Days in intensive care unit, mean
Days in hospital, mean (SD)b
Deaths in intensive care unit, No. (%)
a Difference
b Difference
between protocol group and control group is significant (P = .04) for this variable.
between protocol group and control group is significant (P = .03) for this variable.
Use of an insulin protocol has
improved glycemic control in differ-
ent ICU environments17,18,26,27 and
reduced the frequency of hypo-
glycemic events.12,19,28 The safety and
effectiveness of such insulin proto-
cols may be related to 2 parameters:
the hypoglycemia threshold and
the target range of glucose control.29
Although most intensive insulin
protocols defined hypoglycemia as
a blood glucose level of less than
40 mg/dL, significant hypoglycemic
events still occurred.2,11,20,23 Our hypo-
glycemia threshold was defined as a
glucose level of less than 70 mg/dL.
Moreso, several measures were
undertaken to prevent hypoglycemic
episodes. Insulin rates were reduced
not only according to absolute glu-
cose levels but also according to the
rate of decline.9,17 The nursing staff
therefore reduced insulin adminis-
tration and administered intravenous
glucose earlier, thus avoiding a more
extreme decrease in glucose levels
and the associated symptoms. In
addition, special care was taken to
include guidelines for withholding
insulin during periods when feeding
was stopped (eg, gastric residual
checks, procedures). The fact that
the number of glucose measurements
per patient was higher in the proto-
col group also may have contributed
www.ccnonline.org
Protocol group Control group
(n = 96) (n = 153)
9.1 (7.9) 11.3 (12.4)
17 (18) 33 (22)
(SD)a 11.6 (7.8) 14.1 (12.7)
27.0 (15.4) 31.9 (23.8)
23 (24) 37 (24)
to the decreased rate of hypo-
glycemia. However, the results of
the logistic regression did not sup-
port this idea.
A conservative glucose target
range reduced the number of hypo-
glycemic events while still maintain-
ing clinically desirable glucose
control.19,27 Bode et al29 described
different glucose targets for differ-
ent groups of hospitalized patients
receiving intravenous insulin ther-
apy, with critically ill surgical
patients requiring the most inten-
sive glucose control. Goldberg et al19
used a target glucose range of 100
to 139 mg/dL and found that after
the initial decrease in glucose levels,
52% of patients were within the tar-
get range and 66% were within the
“clinically desirable” range of 80 to
139 mg/dL with only a minimal
occurrence (0.3%) of hypoglycemic
events. We opted to use a more con-
servative but clinically acceptable
glucose target of 110 to 149 mg/dL
in order to minimize the rate of
hypoglycemic events but still achieve
clinically acceptable glucose control.
The fact that the time to achieve
glucose control was similar in both
the protocol group and the control
group supports the idea that our
protocol was not more aggressive
than standard treatment. Moreover,
CriticalCareNurse Vol 31, No. 6, DECEMBER 2011 33
the increased amount of blood glu-
cose measurements in the protocol
group did not significantly influence
the tightness of glucose control, as
demonstrated by logistic regression;
however, use of the protocol did
influence the tightness of glucose
control. History of diabetes mellitus
was known in about half of the
patients in both groups (Table 1);
however, guidelines for insulin
administration did not differ in our
diabetic patients. Our analysis has
shown that the protocol was safe in
both patients with newly diagnosed
hyperglycemia and in diabetic
patients; however, we found a nega-
tive association between tightness
of glucose control and the presence
of diabetes, as glucose control in
these patients was probably harder
to achieve. In a recent study, mor-
tality and inflammatory response
did not differ in critically ill dia-
betic patients with severe sepsis
and nondiabetic patients.30 Until
further data are available, we think
that glucose targets should be stan-
dard to all patients.
The use of insulin protocols in
the ICU may enhance nurses’ auton-
omy and empowerment27 or it may
increase nursing workload and
patients’ discomfort because of the
increase in the number of glucose
measurements.31 Although using
the protocol may have been more
time-consuming for the nurses, as
reflected in the higher mean number
of glucose measurements per patient
in the protocol group (Table 2),
most of our ICU nurses reported
that the protocol was a useful and
instructive tool for glucose control
in their patients. In addition, we
have achieved high levels of satisfac-
tion among nurses and physicians,
standardization of treatment among
nurses, better cooperation among
interdisciplinary teams, greater con-
fidence with insulin administration,
and increased empowerment of the
nursing staff.
Recent developments have intro-
duced computerized systems, such
as the Glucommander32 and the
Clarion33 systems, in which insulin
doses and measurement intervals
are calculated on the basis of glu-
cose measurements and multipliers.
Although they are effective in
reaching blood glucose targets with
acceptable rates of hypoglycemia,
such computerized systems are
potentially expensive and have not
been shown to be superior to other
noncomputerized protocols.
Limitations
Our study has several limitations.
We used retrospective historical
control patients, admitted to the
medical ICU during the 12 months
preceding implementation of the
protocol, who required insulin ther-
apy. The number of patients in the
control group (n=153) was higher
than the number in the protocol
group (n=96). This difference may
have stemmed from a more liberal
and nonstandardized approach to
the administration of intravenous
insulin in the control group, as well
as the inclusion of patients taking
oral nutrition, who were excluded
from the protocol group (because
of safety considerations). Power
To learn more about insulin protocols, read
“Adherence to and Efficacy and Safety of an
Insulin Protocol in the Critically Ill” by
Oeyen et al in the American Journal of
Critical Care, 2007;16:599-608. Available at
www.ajcconline.org.
34 CriticalCareNurse Vol 31, No. 6, DECEMBER 2011
calculation for a difference of 0.76
in the mean percentage of the hypo-
glycemic level (blood glucose <70
mg/dL) between the protocol group
and the control group (Table 2),
with a standard deviation of 0.3 in
the protocol group and 1.7 in the
control group, and a sample size of
96 patients for the protocol group
and 153 patients for the control
group, provides a power of 99%.
Moreover, both groups were
clinically similar in terms of the his-
tory of diabetes, disease profiles,
severity of illness, and interventions
used (Table 1). The greater percent-
age of neurological patients in the
protocol group than in the control
group (10% vs 1%, P=.002) was
attributed to the implementation of
a new protocol for treatment of
cerebrovascular accidents that
required admission to the ICU dur-
ing the protocol period. The pres-
ence of a neurological diagnosis,
however, did not influence the safety
and effectiveness of the protocol. We
therefore think that the historical
control group is suitable for com-
parison with our protocol group.
Our population of patients con-
sisted of complicated medical patients
with a relatively high APACHE II
score, with about half requiring
inotropic support and more than
60% treated with corticosteroids. It
is therefore difficult to compare these
patients with patients in other trials
in which insulin protocols were
studied. Our study was not powered
for end points such as patients’ out-
come and length of stay. We there-
fore cannot attribute a shortened
stay in the ICU or hospital in the
protocol group to the use of the pro-
tocol alone, because other factors,
such as changes in overall patient
selection, discharge policies, and
optimization of hospital facilities
and resources may have been
involved. We did not find differences
in mortality between the 2 groups.
However, randomized controlled tri-
als should be performed in order to
detect significant changes in mortal-
ity when such protocols are used.
Conclusion
Results of this study indicate
that using a nurse-led intravenous
insulin protocol with a conservative
blood glucose target is effective in
controlling blood glucose levels in
critically ill medical patients and is
associated with a significantly lower
hypoglycemia rate than standard
practices. Our protocol can assist
nurses and physicians in safely tar-
geting blood glucose levels in an
ICU and can improve standardiza-
tion and optimization of continuous
intravenous insulin administration
by the nursing staff. Randomized
controlled trials should be done to
determine whether such a protocol
will result in improved outcomes for
medical ICU patients. CCN
Now that you’ve read the article, create or contribute
to an online discussion about this topic using eLetters.
Just visit www.ccnonline.org and click “Submit a
response” in either the full-text or PDF view of the
article.
Acknowledgments
We acknowledge and thank Prof Itamar Raz, Dia-
betes Unit, Hadassah-Hebrew University Medical
Center, for reviewing and editing the manuscript.
We thank all the nurses in the medical intensive
care unit for their assistance in performing the
study and collecting the data.
Financial Disclosures
None reported.
References
1. Byrum D. Why is it so important to treat
hyperglycemia in critically ill patients? Crit
Care Nurse. 2004;24(2):86-90.
2. Van den Berghe G, Wouters PJ, Bouillon R,
et al. Outcome benefit of intensive insulin
www.ccnonline.org
 therapy in the critically ill: insulin dose versus glycemic control. Crit 31. Osburne RC, Cook CB, Stockton L, et al. Improving hyperglycemia
Care Med. 2003;31(2):359-366. management in the intensive care unit: preliminary report of a nurse-
3. McCowen KC, Malhotra A, Bistrian BR. Stress-induced hyperglycemia. driven quality improvement project using a redesigned insulin infusion
Crit Care Clin. 2001;17(1):107-124. algorithm. Diabetes Educ. 2006;32(3):394-403.
4. Van den Berghe G. Tight blood glucose control with insulin in “real-life” 32. Davidson PC, Steed RD, Bode BW. Glucommander: a computer-
intensive care. Mayo Clin Proc. 2004;79(8):977-978. directed intravenous insulin system shown to be safe, simple, and effec-
5. Lewis KS, Kane-Gill SL, Bobek MB, Dasta JF. Intensive insulin therapy for tive in 120,618 h of operation. Diabetes Care. 2005;28(10):2418-2423.
critically ill patients. Ann Pharmacother. 2004;38(7-8):1243-1251. 33. Juneja R, Roudebush CP, Nasraway SA, et al. Computerized intensive
6. Bradley C. Intensive insulin therapy in critically ill patients. Intensive Crit insulin dosing can mitigate hypoglycemia and achieve tight glycemic
Care Nurs. 2002;18(2):128-129. control when glucose measurement is performed frequently and on
7. Krinsley JS. Effect of an intensive glucose management protocol on the mor- time. Crit Care. 2009;13(5):R163.
tality of critically ill adult patients. Mayo Clin Proc. 2004;79(8):992-1000.
8. Umpierrez GE, Isaacs SD, Bazargan N, You X, Thaler LM, Kitabchi AE.
Hyperglycemia: an independent marker of in-hospital mortality in patients
with undiagnosed diabetes. J Clin Endocrinol Metab. 2002;87(3):978-982.
9. Trence DL, Kelly JL, Hirsch IB. The rationale and management of hyper-
glycemia for in-patients with cardiovascular disease: time for change. J Clin
Endocrinol Metab. 2003;88(6):2430-2437.
10. Krinsley JS. Association between hyperglycemia and increased hospital
mortality in a heterogeneous population of critically ill patients. Mayo Clin
Proc. 2003;78(12):1471-1478.  T    
11. van den Berghe G, Wouters P, Weekers F, et al. Intensive insulin therapy in
the critically ill patients. N Engl J Med. 2001;345(19):1359-1367.     
12. Kanji S, Singh A, Tierney M, Meggison H, McIntyre L, Hebert PC. Stan-
dardization of intravenous insulin therapy improves the efficiency and
safety of blood glucose control in critically ill adults. Intensive Care Med.
2004;30(5):804-810. A
13. Laver S, Preston S, Turner D, McKinstry C, Padkin A. Implementing inten-
sive insulin therapy: development and audit of the Bath insulin protocol. 5
Anaesth Intensive Care. 2004;32(3):311-316.
14. Solano T, Totaro R. Insulin therapy in critically ill patients. Curr Opin Clin
Nutr Metab Care. 2004;7(2):199-205.
15. Whitehorn LJ. A review of the use of insulin protocols to maintain nor-
moglycaemia in high dependency patients. J Clin Nurs. 2007;16(1):16-27.
16. Zimmerman CR, Mlynarek ME, Jordan JA, Rajda CA, Horst HM. An
insulin infusion protocol in critically ill cardiothoracic surgery patients.
Ann Pharmacother. 2004;38(7-8):1123-1129.
17. Braithwaite SS, Edkins R, Macgregor KL, et al. Performance of a dose-
defining insulin infusion protocol among trauma service intensive care
unit admissions. Diabetes Technol Ther. 2006;8(4):476-488.
18. Markovitz LJ, Wiechmann RJ, Harris N, et al. Description and evaluation
of a glycemic management protocol for patients with diabetes undergo-
ing heart surgery. Endocr Pract. 2002;8(1):10-18.
19. Goldberg PA, Siegel MD, Sherwin RS, et al. Implementation of a safe and
effective insulin infusion protocol in a medical intensive care unit. Diabetes
Care. 2004;27(2):461-467.
20. Van den Berghe G, Wilmer A, Hermans G, et al. Intensive insulin therapy
in the medical ICU. N Engl J Med. 2006;354(5):449-461.
21. Alm-Kruse K, Bull EM, Laake JH. Nurse-led implementation of an insulin-
infusion protocol in a general intensive care unit: improved glycaemic control
with increased costs and risk of hypoglycaemia signals need for algorithm
revision. BMC Nurs. 2008;7:1.
22. Oeyen SG, Hoste EA, Roosens CD, Decruyenaere JM, Blot SI. Adherence
to and efficacy and safety of an insulin protocol in the critically ill: a
prospective observational study. Am J Crit Care. 2007;16(6):599-608.
23. Finfer S, Chittock DR, Su SY, et al. Intensive versus conventional glucose
control in critically ill patients. N Engl J Med. 2009;360(13):1283-1297.
24. Reed CC, Stewart RM, Sherman M, et al. Intensive insulin protocol
improves glucose control and is associated with a reduction in intensive
care unit mortality. J Am Coll Surg. 2007;204(5):1048-1055.
25. Wiener RS, Wiener DC, Larson RJ. Benefits and risks of tight glucose con-
trol in critically ill adults: a meta-analysis. JAMA. 2008;300(8):933-944.
26. Meijering S, Corstjens AM, Tulleken JE, Meertens JH, Zijlstra JG, Ligtenberg
JJ. Towards a feasible algorithm for tight glycaemic control in critically ill
patients: a systematic review of the literature. Crit Care.2006;10(1):R19.
27. Chant C, Wilson G, Friedrich JO. Validation of an insulin infusion nomo-
gram for intensive glucose control in critically ill patients. Pharmacother-
apy. 2005;25(3):352-359.
28. Ku SY, Sayre CA, Hirsch IB, Kelly JL. New insulin infusion protocol
improves blood glucose control in hospitalized patients without increas-
ing hypoglycemia. Jt Comm J Qual Patient Saf. 2005;31(3):141-147.
29. Bode BW, Braithwaite SS, Steed RD, Davidson PC. Intravenous insulin
infusion therapy: indications, methods, and transition to subcutaneous
insulin therapy. Endocr Pract. 2004;10(suppl 2):71-80.
30. Stegenga ME, Vincent JL, Vail GM, et al. Diabetes does not alter mortality
or hemostatic and inflammatory responses in patients with severe sepsis.
Crit Care Med. 2010;38(2):539-545.

www.ccnonline.org CriticalCareNurse Vol 31, No. 6, DECEMBER 2011 35