Original Article

INPATIENT MANAGEMENT OF HYPERGLYCEMIA:

THE NORTHWESTERN EXPERIENCE
Anthony J. DeSantis, MD,1 Lowell R. Schmeltz, MD,1
Kathleen Schmidt, MSN, APRN-BC,1 Eileen O’Shea-Mahler, MSN, APRN-BC,1
Connie Rhee, MD,2 Angela Wells, PA-C, MMS,1 Stephen Brandt, MD,1
Sara Peterson, BA,1 and Mark E. Molitch, MD1

ABSTRACT Conclusion: Validated protocols dedicated to the

achievement of strict glycemic goals were implemented
Objective: To describe a novel method of safe and by a GMS and resulted in substantial improvements in
effective intensive management of inpatient hyper- glycemic control on the surgical inpatient services, with a
glycemia with use of cost-effective protocols directed by a reduced frequency of hypoglycemia. The protocols and
glucose management service (GMS). the GMS have been well received by the inpatient nursing
Methods: An intravenous insulin protocol was and surgical staff members, and all of this has been done
designed to achieve a glycemic target of 80 to 110 mg/dL. in a cost-effective manner. (Endocr Pract. 2006;12:491-
When stable inpatients were transferred from the intra- 505)
venous protocol to a subcutaneous insulin protocol, which
consisted of basal long-acting and prandial and supple-
mental rapid-acting insulins, the blood glucose target was Abbreviations:
80 to 150 mg/dL. Glucose levels were reviewed by the BMI = body mass index; CVICU = cardiovascular
GMS at least daily for protocol adjustments, when neces- intensive care unit; DM = diabetes mellitus; GMS =
sary. glucose management service; ICU = intensive care unit;
Results: The intravenous insulin protocol was used in SICU = surgical intensive care unit; TPN = total par-
276 patients, and 4,058 capillary blood glucose levels enteral nutrition
were recorded. Glycemic target levels (80 to 110 mg/dL)
were achieved, on average, 10.6 ± 5.2 hours after initiation
of insulin drip therapy. The mean capillary blood glucose
level during the study interval was 135.3 ± 49.9 mg/dL. INTRODUCTION
Hypoglycemia (≤60 mg/dL) was recorded in 1.5% of glu-
cose values, and hyperglycemia (≥400 mg/dL) was Guidelines for the diagnosis and management of dia-
recorded in only 0.06%. The subcutaneous insulin proto- betes mellitus (DM), supported by evidence from random-
col was used in 922 patients, and 18,067 capillary glucose ized clinical trials, are well established in the outpatient
levels were documented. The mean blood glucose level setting. In contrast, corresponding guidelines for the man-
was 145.6 ± 55.8 mg/dL during the study period. The agement of DM in the inpatient arena are haphazard at
blood glucose target of 80 to 150 mg/dL was achieved in best, and practice patterns differ considerably. This dis-
58.6%, whereas 74.3% of glycemic values were in the crepancy is due, in part, to a lack of evidence to support
clinically acceptable range (80 to 180 mg/dL). the control of hyperglycemia in some inpatient popula-
Hypoglycemia (≤60 mg/dL) occurred in 1.3% of capillary tions and the belief by most health-care providers that
blood glucose values, and hyperglycemia (≥400 mg/dL) hyperglycemia is a normal physiologic response to illness
occurred in 0.4% of values. or hospital-induced stress. Recent evidence belies this
belief. Hyperglycemia is common in hospitalized patients,
with a prevalence of approximately 25%, and is an inde-
Submitted for publication November 9, 2005
pendent risk factor for poor clinical outcome in multiple
Accepted for publication February 22, 2006 patient populations (1,2). In addition to stress-induced
From the 1Division of Endocrinology, Metabolism and Molecular hyperglycemia, contributing factors to inpatient hyper-
Medicine, Northwestern University Feinberg School of Medicine, Chicago,
Illinois, and 2Department of General Internal Medicine, Oregon Health and
glycemia include pharmacologic agents (3), enteral and
Sciences University, Portland, Oregon. parenteral nutrition (4,5), and glucocorticoid therapy (6).
Address correspondence and reprint requests to Dr. Mark E. Molitch, Recent clinical trials have shown clear benefits rela-
Division of Endocrinology, Metabolism and Molecular Medicine,
Northwestern University Feinberg School of Medicine, 676 North St. Clair,
tive to morbidity and mortality from intensive manage-
303 E. Chicago Ave., Terry 15-731, Chicago, IL 60611. ment of inpatient hyperglycemia, even in patients without
© 2006 AACE. a prior history of DM (7-10). The medical community is

ENDOCRINE PRACTICE Vol 12 No. 5 September/October 2006 491

492 Inpatient Hyperglycemia, Endocr Pract. 2006;12(No. 5)

obligated to respond to this growing literature supporting hourly drip rate, is also administered at the initiation of the
intensive treatment of inpatient hyperglycemia. The devel- infusion (see Table 1 in Appendix 1). The insulin drip is
opment of easy-to-follow, effective, and safe management subsequently titrated on the basis of the current capillary
strategies must become a priority for inpatient providers. blood glucose value and the rate of change from the pre-
In this report, we describe our experience at Northwestern ceding capillary blood glucose value. Nurses refer to the
Memorial Hospital, a 776-bed tertiary care center in appropriate table for insulin drip adjustment, depending on
Chicago, Illinois, with the development and promulgation whether the current glucose value had increased (see Table
of protocols directed by an inpatient glucose management 2 in Appendix 1) or decreased (see Table 3 in Appendix 1)
service (GMS) to provide safe inpatient glycemic control. from the previous value. The degree of change from the
previous blood glucose value (with a change in capillary
METHODS blood glucose of 60 mg/dL as a cutoff value) determines
which column within each table is referenced for insulin
Development of Protocols to Treat Hyperglycemia drip titration orders, consisting of rate changes and addi-
Our goal was to design a system of care for accurate tional boluses as indicated (see Tables 2 and 3 in
identification and treatment of hyperglycemia among hos- Appendix 1).
pitalized patients. A retrospective review estimated that For example, if the previous blood glucose value was
21% of patients requiring admission to our institution in 100 mg/dL and the current blood glucose value is 170
September 2003 had documented glucose levels >200 mg/dL, the nurse would reference Table 2 in Appendix 1
mg/dL. For our protocols, glycemic targets were set because the glucose level increased. Because the degree of
according to preexisting guidelines (11,12). In the critical- change in the glucose level was 70 mg/dL, the orders in
ly ill population, a target glycemic range of 80 to 110 the far right column of Table 2 (“Greater than 60 mg/dL”)
mg/dL was established, on the basis of evidence from would be followed, and the insulin drip rate should be
other studies supporting a reduction of morbidity and mor- increased by 0.5 U/h with a 2-U bolus administered. The
tality with achievement of similar glycemic levels (7). In next blood glucose assessment would be in 1 hour, and
the non-critically ill population, a glycemic target of 80 to further insulin drip titration would be performed.
150 mg/dL was established. Blood glucose levels ranging The intravenous insulin protocol order set stipulates
from 80 to 150 mg/dL and from 80 to 180 mg/dL are con- the frequency of capillary blood glucose monitoring.
sidered clinically acceptable in the critically ill and the Initially, capillary blood glucose levels are monitored
non-critically ill populations, respectively, as defined by a every hour, and then the frequency of monitoring is
recent consensus conference (12). decreased to every 2 to 4 hours, depending on the stabili-
Orally administered hypoglycemic agents are rela- ty of blood glucose values and the clinical situation. The
tively contraindicated in the hospital setting (13). The protocol incorporates variables for intravenous fluid
extended duration of action and the potential for harmful administration, including glucose and potassium repletion
drug interactions, coupled with the rapidly changing clin- to minimize hypoglycemia and hypokalemia. Guidelines
ical scenario of hospitalized patients, make these agents for the treatment of hypoglycemia are explicit (Appendix
less desirable for achieving strict glycemic goals. Insulin 1). Orders to notify the managing service of various clini-
therapy can achieve strict glycemic control efficiently and cal scenarios—specifically, persistent hypoglycemia or
predictably with the least amount of drug interactions. A hyperglycemia, rapid changes in glucose values (increase
literature search of extant inpatient hyperglycemia man- or decrease >100 mg/dL), hypokalemia, or impending
agement protocols was performed (14-16). We adapted dietary changes—are included.
elements of these protocols to fit our patient and provider
populations optimally, specifically emphasizing safety, Subcutaneous Insulin Protocol
efficacy, and ease of application. Separate protocols were The subcutaneous insulin protocol was designed for
designed for the delivery and management of intravenous effective and safe achievement of target blood glucose val-
insulin therapy and subcutaneous insulin therapy. ues of 80 to 150 mg/dL, with insulin delivery modeling
physiologic insulin secretion (Appendix 2). Similar to pre-
Intravenous Insulin Protocol viously published protocols (17), insulin needs were
The intravenous protocol was developed to achieve a subdivided into basal, prandial, and supplemental require-
glycemic target (80 to 110 mg/dL) while minimizing the ments. Basal insulin controls hepatic glucose output and
risk of hypoglycemia (Appendix 1). The efficacy of the gluconeogenesis in the fasting state, prandial insulin com-
protocol for the management of diabetic ketoacidosis has pensates for blood glucose elevations from caloric intake,
not been validated and cannot be recommended. and supplemental insulin is used as a corrective factor for
Intravenous insulin therapy is administered as a regular preprandial hyperglycemia. The GMS routinely uses the
insulin solution (100 U of regular insulin in 100 mL of iso- long-acting insulin analogue, glargine (Lantus), for basal
tonic saline). The initial insulin drip rate is determined by insulin requirements because of its once-daily dosing and
the initial capillary blood glucose value (Appendix 1). A relatively flat insulin profile. Guided by the hospital
loading bolus dose of regular insulin, equivalent to the formulary, the GMS generally uses the rapid-acting
 Inpatient Hyperglycemia, Endocr Pract. 2006;12(No. 5) 493

insulin analogue, aspart (NovoLog), for prandial and sup- safety were collected. After 1 month of protocol use on the
plemental insulin requirements. CVICU and SICU services, modifications were made to
Capillary blood glucose monitoring is ordered before improve efficacy, adherence, and safety. Once safety of
every meal and at bedtime in those patients tolerating an the protocols was demonstrated, use of the protocols was
oral diet, every 6 hours if the patient is allowed nothing expanded to additional ICUs and the general medical and
orally or is receiving continuous enteral or parenteral surgical hospital wards. In-service education was provid-
nutrition, and with greater frequency as clinically indicat- ed to the nursing staff of each unit before implementation
ed. To minimize the need to call a physician for a patient of the insulin protocols.
with abnormal glucose values, the nursing staff adminis- Capillary blood glucose monitoring allows immediate
ters supplemental insulin doses according to the ordered point-of-care assessments with which to base clinical deci-
supplemental insulin scale given concurrently with sched- sions. The SICU and CVICU nursing staffs were instruct-
uled prandial insulin (if eating) or alone (if the patient is ed to perform capillary blood glucose monitoring for all
allowed nothing by mouth or is receiving enteral or par- patients. The initiation and frequency of glucose monitor-
enteral therapy) (see Order 4 in Appendix 2). Daily review ing (preoperatively, intraoperatively, or postoperatively)
of bedside capillary blood glucose values is performed by were based on the clinical discretion of the managing sur-
the GMS so that titration of basal, prandial, or supplemen- gical and anesthesia staffs. Most commonly, capillary
tal insulin doses can achieve the aforementioned glycemic blood glucose monitoring was performed every 4 hours
goals. The subcutaneous insulin protocol provides guide- and begun in the postoperative setting. Management of
lines regarding hypoglycemia management and service hyperglycemia was considered necessary if capillary
notification for persistent capillary blood glucose values blood glucose values exceeded 110 mg/dL on 2 separate
outside the target range and if dietary changes are planned. occasions or exceeded 200 mg/dL once. Once hyper-
glycemia was established, the GMS was consulted, and
Development of the Glucose Management Service insulin administration (intravenous or subcutaneous) was
The GMS was formed with a team-oriented approach determined on the basis of the clinical scenario. In most
to cover all aspects of hyperglycemia management. The patients with hyperglycemia, intravenous insulin therapy
team consists of the following health-care providers: an was begun in the immediate postoperative period (24 to 72
advanced practice nurse (or nurses) or a physician’s assis- hours), at a time when the patients had limited dietary
tant, responsible for initial patient consultation and daily intake, were intubated, and/or were receiving intravenous-
management; an endocrinology attending, responsible for ly administered pressor agents. Transition to a subcuta-
initial patient consultation, evaluation, and management; neous insulin regimen occurred once the critical illness
an endocrinology fellow, responsible for night and week- had resolved and dietary intake was adequate. The insulin
end coverage of the service; and an administrator, respon- protocols proved to be safe and effective (see the
sible for collection and storage of pertinent data. The team “Results” section). It became apparent after 1 month of
functions in coordination with ancillary health-care protocol use that the SICU and CVICU nursing staffs were
providers critical to comprehensive management of DM, sufficiently qualified to use the protocols without prereq-
including inpatient dietitians and diabetes educators. uisite consultation from the GMS. Members of the GMS
Approval for use of the insulin protocols as directed were available for consultation, and protocol outcomes
by the GMS was granted by the Medication Safety, were continually monitored.
Pharmacy, and Therapeutics Subcommittee and the Eventually, approval for the use of the insulin proto-
Quality Assurance Subcommittee at Northwestern cols was granted hospitalwide by the various hospital care
Memorial Hospital. The initial strategy was to limit use of committees. Consultation with the GMS was no longer
the insulin protocols in a select inpatient group, to monitor mandatory for insulin protocol implementation. The most
safety and efficacy of the protocols over a specified peri- common scenario for postsurgical hyperglycemia identifi-
od and to make necessary adjustments, and then to expand cation, insulin protocol initiation, and GMS consultation
the use of the service in a stepwise fashion. Initially, the at our institution is schematically represented in Figure 1.
insulin protocols were restricted to the GMS and available
as a patient care tool only after consultation with the GMS Hyperglycemia Management Strategies
was obtained. Postoperative capillary blood glucose monitoring has
The pilot groups for the GMS and insulin protocols now become routine in the ICUs and as clinically indicat-
were the surgical intensive care unit (SICU) and cardio- ed on the general surgical nursing units. In the SICU and
vascular intensive care unit (CVICU) because intensive CVICU, a blood glucose level is determined within an
glycemic control already had been shown to provide ben- hour after transfer from the recovery room. SICU and
efit in such intensive care unit (ICU) populations (7,8). CVICU protocol indicates initiation of an insulin drip if
Educational sessions describing protocol rationale and use blood glucose values exceed 200 mg/dL once or 110
were given to the SICU and CVICU provider staffs by a mg/dL on 2 occasions independent of the GMS. In the
member of the GMS. Data regarding efficacy to achieve non-ICU areas, the GMS is consulted for assistance with
glycemic targets, adherence to protocol guidelines, and glycemic control in patients with hyperglycemia at the dis-
494 Inpatient Hyperglycemia, Endocr Pract. 2006;12(No. 5)

Fig. 1. Algorithm for identification and treatment of inpatients with hyperglycemia. BG = blood glu-
cose; CBG = capillary blood glucose; D/C = discontinue; GMS = glucose management service; IV =
intravenous; NPO = nothing allowed by mouth; qAM = every morning; SQ = subcutaneous.

cretion of the managing service. The initial hyperglycemia A subcutaneous insulin treatment strategy is optimal
treatment strategy is dependent on the clinical scenario for inpatients with hyperglycemia who are receiving a sta-
and multiple variables. ble caloric source or in whom orally administered hypo-
Intravenously administered insulin is most appropri- glycemic agents are contraindicated. Many clinical
ate, but not exclusively, for critically ill patients with variables are taken into account by the GMS to determine
blood glucose values exceeding 110 mg/dL, those with a the basal, prandial, and supplemental insulin requirements
history of DM unable to receive oral caloric support, and to be met by a subcutaneous insulin strategy. Patients
patients with severe hyperglycemia (glucocorticoids, sep- receiving total parenteral nutrition (TPN) will commonly
sis, pressor agents). In the postoperative critical care set- have regular insulin added to their TPN after GMS con-
ting, the patient is often intubated, and dietary intake is sultation with the TPN nurse. Patients with hyperglycemia
restricted. Nursing personnel initiate the intravenous receiving continuous enteral nutrition are usually treated
insulin protocols and titrate infusion rates accordingly for with a long-acting basal insulin. Supplemental insulin is
those patients identified as having hyperglycemia. The administered as needed on the basis of capillary blood glu-
insulin infusions are maintained while the patient is cose monitoring every 6 hours.
receiving nothing orally. The GMS is consulted if the tar-
get glycemic range is not achieved within 8 hours after Conversion From Intravenous Insulin Therapy
insulin drip initiation or for assistance with conversion to When the patient’s therapy is being converted from an
subcutaneous insulin therapy after the critical illness intravenous insulin drip, the drip rate is used as a guide to
resolves and the decision to advance diet is made by the determine total daily insulin requirements. The insulin
managing service. The GMS continues insulin manage- drip rate for the preceding 6 hours is averaged to obtain a
ment throughout the rest of the hospitalization. stable hourly rate. The average rate is multiplied by 24
 Inpatient Hyperglycemia, Endocr Pract. 2006;12(No. 5) 495

hours to calculate the total daily insulin requirement. The insulin levels. Data analysis revealed inadequate glycemic
basal insulin dose ordered is 80% of the total daily insulin control 6 hours after the glargine injection despite the
requirement and usually administered as a once-daily sub- overlap period with the insulin drip. In addition, it became
cutaneous injection of glargine insulin. Most commonly, apparent that practical necessities outweighed this physio-
the prandial insulin dose for each meal is 10% of the logic reasoning, in that transfer of the patient from intra-
glargine dose, usually given as insulin aspart per hospital venous to subcutaneous insulin therapy often coincided
formulary. The prandial insulin dose is adjusted accord- with transfer of the patient out of the ICU, with the result
ingly as the patient’s appetite improves postoperatively. that the insulin infusion was simply stopped without an
The proportion of insulin given for prandial dosing is sub- overlap period. Therefore, we changed our protocol to
stantially less than that recommended by others because include a conversion dose of aspart insulin that was 10%
these patients are generally consuming only a clear liquid of the glargine dose given simultaneously to discontinuing
diet initially with a reduced caloric content. The dose is the intravenous infusion and the first administration of
maintained the next day because the overall insulin glargine to maintain adequate glycemic control.
requirement is generally decreasing substantially as the
stress of the surgical procedure or acute illness abates. The Initiation of Subcutaneous Insulin Protocol When No
initial doses of basal and prandial insulin are given at sep- Intravenous Insulin Therapy Has Been Given
arate injection sites concomitant with the discontinuation In the non-critical care surgical population, the GMS
of the intravenous insulin drip and ingestion of the first is consulted for management of hyperglycemia in patients
postoperative meal (see Example 1 in Table 1). During the with known DM or those in whom postoperative hyper-
first few months of using the protocols, it was mandated glycemia develops. Often these patients are receiving a
that the insulin infusion be continued for at least 2 hours stable caloric source and have not received insulin intra-
after the injection of glargine to maintain adequate serum venously during the current hospitalization. A subcuta-

Table 1
Calculation of Subcutaneous Insulin Need

Example 1. Conversion From Intravenous Insulin Therapy

Step 1. Intravenous insulin drip rate averaged 1.8 U/h with final glucose level 98 mg/dL

Step 2. Calculate average insulin infusion rate for last 6 h = 2.1 U/h and multiply × 24 to get total daily insulin
requirement (2.1 × 24 = 50 U/24 h)

Step 3. Multiply this 24-h dose (50 U) × 80% to obtain glargine dose = 40 U, which is given and the infusion is
stopped

Step 4. Multiply the glargine dose by 10% to give as a rapid-acting insulin (e.g., aspart, lispro, or glulisine) at
the time the glargine is given and the infusion is stopped

Step 5. Give 10% of the glargine dose as prandial doses before each meal

Example 2. Estimating Insulin Doses When No Intravenous Insulin Therapy Has Been Given
Step 1. Calculate estimated total daily dose of insulin as follows:
Type 2 diabetes (known): 0.5 to 0.7 U/kg
Type 1 diabetes (known): 0.3 to 0.5 U/kg
Unknown: 0.3 to 0.5 U/kg

Step 2. Divide total daily dose of insulin into 50% basal as glargine and 50% prandial as aspart, lispro, or
glulisine

Step 3. Divide prandial insulin into 3 equal doses to be given with meals
496 Inpatient Hyperglycemia, Endocr Pract. 2006;12(No. 5)

neous insulin strategy is developed by the GMS. The total September 2004, for which no data are available).
daily insulin requirement is influenced by multiple clinical Demographic data for these patients are shown in Table 2.
variables (for example, prior history of DM, type of DM, Most patients were men (63%), and cardiovascular
body mass index [BMI], outpatient hypoglycemia regi- surgery was the managing service most frequently
men, surgical stress, concomitant medications, caloric requesting consultation (47%). The majority of patients
intake) that affect insulin sensitivity and secretion. Body were in the CVICU or SICU while receiving intravenous
weight can be used as a guideline to calculate initial insulin therapy. The mean age was 59.6 years, and the
insulin dosing requirements (0.5 U/kg for patients with mean BMI was 29.1 kg/m2. The majority of patients had
type 2 DM and 0.3 U/kg for those with type 1 DM or with- no prior history of diabetes (64%), whereas 32% had a his-
out a prior history of DM). Half of the calculated daily tory of type 2 DM and only 4% a history of type 1 DM
insulin requirement is given as a daily subcutaneous injec- (Table 2). It bears repeating that the intravenous insulin
tion of glargine insulin to meet basal insulin coverage. The protocol was not intended for use in the treatment of dia-
remainder is divided into 3 equal doses administered as betic ketoacidosis, and any use of the protocol outside the
aspart insulin with meals to fulfill prandial insulin needs postoperative setting should be done with caution.
(see Example 2 in Table 1). Glycemic target levels (80 to 110 mg/dL) were
A supplemental insulin scale is determined on the achieved, on average, 10.6 ± 5.2 hours after initiation of
basis of multiple clinical factors (for example, type and insulin drip therapy. The mean capillary blood glucose
severity of hyperglycemia, BMI, insulin requirements, concentration for the study interval was 135.3 ± 49.9
medical stress, concomitant medications) and is delivered mg/dL. In these 276 patients, 4,058 capillary blood glu-
in addition to the standing prandial insulin dose in the cose levels were recorded. Hypoglycemia (≤60 mg/dL)
form of aspart insulin to correct preprandial hyper- was recorded in 1.5% of capillary blood glucose values,
glycemia. Capillary blood glucose is monitored at meals and hyperglycemia (≥400 mg/dL) was recorded in only
and at bedtime. Daily adjustment of basal or bolus insulin 0.06%.
doses on the basis of individual glycemic results is a criti- Glycemic control in the CVICU and SICU from
cal function of the GMS. September 1 through September 30, 2003, before the
development of the GMS and implementation of the intra-
ANALYSES venous insulin protocol, was used as a historic control for
comparative purposes. The mean blood glucose level in
The major efficacy end points for the intravenous patients in the CVICU and SICU during that period was
insulin protocol were the time from initiation of drip 169.0 ± 69.0 mg/dL, with 0.6% and 0.4% of capillary
insulin therapy to target glycemia (80 to 110 mg/dL) and blood glucose values (N = 526) ≤60 mg/dL and ≥400
the mean blood glucose level during the infusion. The mg/dL, respectively.
incidences of hypoglycemia (≤60 mg/dL) and hyper- With respect to protocol adherence, insulin drip rates
glycemia (≥400 mg/dL) were tallied. Adherence to the were appropriately initiated in accordance with the proto-
protocol was also monitored. col 84% of the time. Insulin drip rates were titrated in
The efficacy of the subcutaneous insulin protocol was
determined by percentages of capillary blood glucose
measurements in the target glycemic range (80 to 150 Table 2
mg/dL) and the clinically acceptable range (80 to 180 Demographic Data
mg/dL). The incidences of hypoglycemia (≤60 mg/dL) for 276 Patients Managed
and hyperglycemia (≥400 mg/dL) were recorded as well. With the Intravenous Insulin Protocol
Approval to publish data was granted by the Institutional
Factor Data
Review Board of the Northwestern University, Feinberg
School of Medicine. Mean age (yr) 59.6 ± 6.8
Adherence to the protocol was defined as a function
Male patients 63%
of accuracy of insulin dosing and accuracy of service noti-
Admitting service, no. (%)
fication, as outlined by the protocol order sets. Adherence
of the nursing staff to the intravenous insulin protocol was Cardiovascular surgery 129 (46.7)
monitored retrospectively. Transplantation surgery 51 (18.5)
General surgery 12 (4.3)
RESULTS Surgical oncology 17 (6.2)
Other 67 (24.3)
Intravenous Insulin Protocol History of diabetes, no. (%)
Two hundred seventy-six patients were managed with Known type 1 diabetes 10 (3.6)
the intravenous insulin protocol after consultation with the Known type 2 diabetes 88 (31.9)
GMS between June 2004 and June 2005 (excluding No previous history of diabetes 178 (64.5)
 Inpatient Hyperglycemia, Endocr Pract. 2006;12(No. 5) 497

Table 3. The mean age at the time of GMS consultation

Table 3 was 60.5 years, and the mean BMI was 29.2 kg/m2. The
Demographic Data majority of patients with hyperglycemia had no prior his-
for 922 Patients Managed tory of diabetes (51.2%); 42.3% and 6.5% had a history of
With the Subcutaneous Insulin Protocol type 2 DM and type 1 DM, respectively.
In these 922 patients, 18,067 capillary blood glucose
Factor Data levels were obtained. The mean blood glucose concentra-
tion was 145.6 ± 55.8 mg/dL during the study period. The
Mean age (yr) 60.5 ± 13.5 majority of capillary blood glucose measurements (58.6%)
Male patients 61% were in the target range of 80 to 150 mg/dL, and 74.3%
Admitting service, no. (%) were in the clinically acceptable range (80 to 180 mg/dL).
Cardiovascular surgery 388 (42.1) Hypoglycemia (≤60 mg/dL) was documented in 1.3% of
Transplantation surgery 161 (17.5) capillary blood glucose values, with an incidence of 0.25
General surgery 62 (6.7) episode per patient. Hyperglycemia (≥400 mg/dL) was
Surgical oncology 101 (11.0) less frequent; it was documented in 0.4% of capillary
Other 210 (22.8) blood glucose measurements and occurred with a frequen-
History of diabetes, no. (%) cy of 0.09 episode per patient (Fig. 2).
Known type 1 diabetes 60 (6.5) A historical comparison of glycemic control on the
Known type 2 diabetes 390 (42.3) same surgical services from September 1 through
No previous history of diabetes 472 (51.2) September 30, 2003, a period before the development of
the GMS and implementation of the subcutaneous insulin
protocol, was made. During that period, 2,379 capillary
blood glucose levels were determined. The mean blood
accordance with the protocol 58% of the time, with the glucose level was 163.5 ± 68.3 mg/dL, with the glycemic
most common error (22%) being underdosing of insulin. target range (80 to 150 mg/dL) being achieved in 48.4% of
capillary blood glucose values and the clinically accept-
Subcutaneous Insulin Protocol able range (80 to 180 mg/dL) being achieved in 67%.
A total of 922 patients received subcutaneous insulin Hypoglycemia (≤60 mg/dL) occurred in 1.4%, and hyper-
management in consultation with the GMS from June glycemia (≥400 mg/dL) occurred in 0.88% of these values.
2004 through June 2005 (excluding September 2004, for
which no data are available). Similar to those receiving the Improvement With Protocol Use Over Time
intravenous insulin protocol, the majority were male A basic tenet of the GMS is to conduct periodic
patients (61%), and the most frequent service requesting review of outcomes and to adjust protocols and manage-
GMS consultation was cardiovascular surgery (42%). ment strategies as needed for improvement. A comparison
Demographic data for these patients are summarized in of outcomes for both the intravenous insulin protocol and

Fig. 2. Glycemic ranges in hospitalized patients treated with the subcutaneous insulin protocol
between June 2004 and June 2005 (excluding September 2004). The results consist of 18,067
capillary glucose measurements performed in 922 patients. The percentages of glucose values
in the ranges of <80 mg/dL, 80 to 150 mg/dL, 150 to 180 mg/dL, and >180 mg/dL are shown.
Overall, 74% of values were in the clinically acceptable range of 80 to 180 mg/dL. It should be
noted that only 1.3% of values were ≤60 mg/dL and only 0.4% of values were ≥400 mg/dL.
498 Inpatient Hyperglycemia, Endocr Pract. 2006;12(No. 5)

Table 4
Comparison of Glucose Values in the Initial Quarter*
and the Latest Quarter† for the
Intravenous and Subcutaneous Insulin Protocols

Initial Latest
Protocol quarter quarter

Intravenous
Total number of patients 34 114
Total number of glucose measurements 674 1,543
Time to glucose goal (h) 18.2 8.3
Mean glucose level (mg/dL) 148.4 133.2
Glucose levels (mg/dL)
80-110 (%) 24.1 27.6
80-150 (%) 57.3 65.8
≤60 (%) 3.1 1.0
≥400 (%) 0.2 0.3
Subcutaneous
Total number of patients 120 331
Total number of glucose measurements 2,337 6,524
Mean glucose level (mg/dL) 153.6 139.7
Glucose levels (mg/dL)
80-110 (%) 47.0 66.2
80-150 (%) 65.5 79.8
≤60 (%) 1.6 1.1
≥400 (%) 0.5 0.3

*June 1, 2004 to August 31, 2004.

†April 1, 2005 to June 30, 2005.

the subcutaneous insulin protocol for the first quarter Hyperglycemia has been shown to be an independent
(June 1, 2004 to August 31, 2004) and the most recent risk factor for a poor clinical outcome in multiple inpatient
quarter (April 1, 2005 to June 30, 2005) analyzed is out- settings. In patients who have undergone a cardiac surgi-
lined in Table 4. An improvement in glycemic end points cal procedure, DM is an independent predictor of pro-
is seen, despite an increase in the number of consultations. longed ICU stay, sternal wound infection, postoperative
delirium, perioperative stroke, renal dysfunction, and need
DISCUSSION for postoperative reintubation (20). Patients with a mean
blood glucose level >150 mg/dL during the 3 days after a
The number of hospitalizations for DM or DM-relat- cardiovascular surgical procedure have double the infec-
ed complications in the United States was estimated at 4.6 tion rate and up to 13 times the mortality of their normo-
million in 2001, with an estimated cost of $40 billion (18). glycemic counterparts (8). Hyperglycemia is associated
In the hospital setting, however, DM is often unrecog- with a 9-fold and a 2-fold increase in mortality during hos-
nized, and treatment is seldom standardized. A retrospec- pitalization among those with new-onset hyperglycemia
tive review of patients admitted to a Philadelphia hospital and those with a known history of diabetes, respectively
revealed that 31% met the American Diabetes Association (1). The results of these studies emphasize the importance
criteria for DM, 41% of whom were unrecognized as hav- of recognition of hyperglycemia in all inpatients, even
ing DM at the time of dismissal from the hospital (19). A those without a prior history of DM. As noted previously,
retrospective analysis of blood glucose measurements (N less than half of the patients with hyperglycemia treated by
= 3,092) on the medicine services at Northwestern us had a prior diagnosis of DM.
Memorial Hospital from September 1 through September Treatment of hyperglycemia with insulin has been
30, 2003, revealed a mean blood glucose level of 182.6 ± shown to reduce mortality, sepsis, ICU stay, need for dial-
87.8 mg/dL, with only 53.1% of the capillary blood glu- ysis, need for transfusion, and duration of ventilation in
cose levels measured being in the clinically acceptable the postoperative setting (7). Furthermore, postoperative
range of 80 to 180 mg/dL. glycemic control has been shown to decrease postopera-
 Inpatient Hyperglycemia, Endocr Pract. 2006;12(No. 5) 499

tive wound infections (8). On the basis of these studies, the data in patients managed with use of these protocols are
American Diabetes Association and the American now in progress.
Association of Clinical Endocrinologists have put forward
recommendations for inpatient management of hyper- DISCLOSURE
glycemia (11,12).
In this report, we describe a practical approach to Mark E. Molitch, MD, has received research support
identify and treat inpatient hyperglycemia. We have from Sanofi-Aventis Pharmaceuticals, Genentech, and
shown a substantial improvement in mean blood glucose Amgen. In addition, he is a consultant for Abbott
level and percentage of blood glucose values reported in Laboratories and Sanofi-Aventis Pharmaceuticals.
the target glucose range (80 to 150 mg/dL) and in what has Anthony J. DeSantis, MD, is a consultant for Sanofi-
been termed the clinically acceptable range (80 to 180 Aventis Pharmaceuticals.
mg/dL) (12) at our institution with the use of easy-to-fol-
low insulin protocols guided by a formal management ser- ACKNOWLEDGMENT
vice. This change has been accomplished safely with no
increase in hypoglycemia in comparison with historical We acknowledge the support of Gerald Baumann,
methods of glycemic management. MD, Herman Blomeier, MD, Allison Hahr, MD, Rekha
The limited number of patients treated thus far has not Ramamurthy, MD, and Joy Springer, RN, CDE.
allowed us to evaluate the effect of this service on mor-
bidity and mortality outcomes. On the basis of the results REFERENCES
of other studies, however, we would expect similar clini-
cal benefits. Our results are limited to the inpatient surgi- 1. Umpierrez GE, Isaacs SD, Barzagan N, You X, Thaler
LM, Kitabchi AE. Hyperglycemia: an independent mark-
cal services, and extrapolation to other inpatient
er of in-hospital mortality in patients with undiagnosed dia-
populations may not be appropriate. Nevertheless, one betes. J Clin Endocrinol Metab. 2002;87:978-982.
could argue that the complicated clinical environment 2. Capes SE, Hunt D, Malmberg K, Pathak P, Gerstein
characteristic of the current inpatient experience necessi- HC. Stress hyperglycemia and prognosis of stroke in non-
tates a focused specialized approach to hyperglycemia in diabetic and diabetic patients: a systematic overview.
all such patients. Hyperglycemia has been associated with Stroke. 2001;32:2426-2432.
3. Luna B, Feinglos MN. Drug-induced hyperglycemia.
a poor clinical outcome in a variety of other patient popu- JAMA. 2001;286:1945-1948.
lations, including those admitted to the hospital with a 4. Rosmarin DK, Wardlaw GM, Mirtallo J. Hypergly-
diagnosis of congestive heart failure (21), stroke (2), and cemia associated with high, continuous infusion rates of
acute myocardial infarction (22), and improved glycemic total parenteral nutrition dextrose. Nutr Clin Pract. 1996;
control has been shown to benefit the last-mentioned 11:151-156.
5. Cheung NW, Napier B, Zaccaria C, Fletcher JP. Hyper-
group of patients. glycemia is associated with adverse outcomes in patients
The use of advanced practice nurses as the corner- receiving total parenteral nutrition. Diabetes Care.
stone of inpatient hyperglycemia management, in con- 2005;28:2367-2371.
junction with supervision by a board-certified 6. Hollingdal M, Juhl CB, Dall R, et al. Glucocorticoid
endocrinologist, has proved effective and financially induced insulin resistance impairs basal but not glucose
entrained high-frequency insulin pulsatility in humans.
viable. Revenue generated by the GMS consultation activ- Diabetologia. 2002;45:49-55.
ity at the current census has been able to provide salary 7. Van den Berghe G, Wouters P, Weekers F, et al.
support for 2 full-time advanced practice nurses, an Intensive insulin therapy in the critically ill patients. N
administrator, and 25% of a supervising physician’s Engl J Med. 2001;345:1359-1367.
salary. 8. Furnary AP, Wu Y, Bookin SO. Effect of hyperglycemia
and continuous intravenous insulin infusions on outcomes
of cardiac surgical procedures: the Portland Diabetic
CONCLUSION Project. Endocr Pract. 2004;10(Suppl 2):21-33.
9. Wahab NN, Cowden EA, Pearce NJ, Gardner MJ,
In summary, we have developed strategies to identify Merry H, Cox JL (ICONS Investigators). Is blood glu-
inpatient hyperglycemia and validated protocols dedicated cose an independent predictor of mortality in acute
to the achievement of strict glycemic goals. With use of myocardial infarction in the thrombolytic era? J Am Coll
Cardiol. 2002;40:1748-1754.
these interventions, we have made substantial improve- 10. Guvener M, Pasaoglu I, Demircin M, Oc M.
ments in glycemic control on our surgical inpatient ser- Perioperative hyperglycemia is a strong correlate of post-
vices. These efforts have been undertaken without operative infection in type II diabetic patients after coro-
jeopardizing patient safety and have, in fact, reduced the nary artery bypass grafting. Endocr J. 2002;49:531-537.
frequency of hypoglycemia. The protocols and GMS have 11. American Diabetes Association. Standards of medical
care in diabetes [erratum in Diabetes Care. 2005;28:990].
been well received by the inpatient nursing and surgical Diabetes Care. 2005;28(Suppl 1):S4-S36.
staff members, and all of this has been done in a cost- 12. Moghissi ES, Hirsch IB. Hospital management of dia-
effective manner. Analyses of long-term health outcome betes. Endocrinol Metab Clin North Am. 2005;34:99-116.
500 Inpatient Hyperglycemia, Endocr Pract. 2006;12(No. 5)

13. Clement S, Braithwaite SS, Magee MF, et al (American 18. Hogan P, Dall T, Nikolov P (American Diabetes
Diabetes Association Diabetes in Hospitals Writing Association). Economic costs of diabetes in the US in
Committee). Management of diabetes and hyperglycemia 2002. Diabetes Care. 2003;26:917-932.
in hospitals [errata in Diabetes Care. 2004;27:856 and 19. Levetan CS, Passaro M, Jablonski K, Kass M, Ratner
Diabetes Care. 2004;27:1255]. Diabetes Care. 2004;27: RE. Unrecognized diabetes among hospitalized patients.
553-591. Diabetes Care. 1998;21:246-249.
14. Markovitz LJ, Wiechmann RJ, Harris N, et al. 20. Bucerius J, Gummert JF, Walther T, et al. Impact of
Description and evaluation of a glycemic management pro- diabetes mellitus on cardiac surgery outcome [erratum in
tocol for patients with diabetes undergoing heart surgery. Thorac Cardiovasc Surg. 2003;51:113]. Thorac
Endocr Pract. 2002;8:10-18. Cardiovasc Surg. 2003;51:11-16.
15. Goldberg PA, Siegel MD, Sherwin RS, et al. 21. Bhatia V, Wilding GE, Dhindsa G, et al. Association of
Implementation of a safe and effective insulin infusion pro- poor glycemic control with prolonged hospital stay in
tocol in a medical intensive care unit. Diabetes Care. 2004; patients with diabetes admitted with exacerbation of con-
27:461-467. gestive heart failure. Endocr Pract. 2004;10:467-471.
16. The Portland protocol. Available at: http://www.provi- 22. Malmberg K, Norhammar A, Wedel H, Ryden L.
dence.org/Oregon/Programs_and_Service/Heart/ Glucometabolic state at admission: important risk marker
portlandprotocol/default.htm. Accessed for verification of mortality in conventionally treated patients with dia-
August 27, 2006. betes mellitus and acute myocardial infarction; long-term
17. Trence DL, Kelly JL, Hirsch IB. The rationale and man- results from the Diabetes and Insulin-Glucose Infusion in
agement of hyperglycemia for in-patients with cardiovas- Acute Myocardial Infarction (DIGAMI) study.
cular disease: time for a change. J Clin Endocrinol Metab. Circulation. 1999;99:2626-2632.
2003;88:2430-2437.
 Inpatient Hyperglycemia, Endocr Pract. 2006;12(No. 5) 501

APPENDIX 1
502 Inpatient Hyperglycemia, Endocr Pract. 2006;12(No. 5)
Inpatient Hyperglycemia, Endocr Pract. 2006;12(No. 5) 503
504 Inpatient Hyperglycemia, Endocr Pract. 2006;12(No. 5)

APPENDIX 2
Inpatient Hyperglycemia, Endocr Pract. 2006;12(No. 5) 505