Insulin Infusion During Critical Care: O True Apothecary! Thy Drugs Are Quick

Silvio E Inzucchi, MD

A review of the literature reveals abundant evidence regarding the relationship between blood glucose (BG) levels and clinical outcomes in intensive care units. Observational trials have demonstrated the link between hyperglycemia and poor clinical outcomes and spurred the initiation of randomized clinical trials to investigate the effects of interventional glycemic control on health outcomes. Data from interventional studies support the treatment and prevention of hyperglycemia to improve clinical outcomes among intensive care patients. The controlled clinical trials have also guided development of recommendations for glycemic targets and procedures to attain better glycemic control in intensive care settings. In addition to reviewing key studies and providing an overview of published protocols for intravenous (IV) insulin infusion, Dr Inzucchi discussed barriers to implementing protocols for glycemic control and strategies for overcoming these barriers.


A number of observational studies have examined the relationship between hyperglycemia and mortality among critically ill patients in different intensive care unit (ICU) settings. Krinsley analyzed retrospective data of 1826 consecutive Stamford Hospital (Stamford, CT) ICU patients whose plasma glucose (PG) values were obtained.[1] The Stamford Hospital ICU is a mixed unit, admitting general medical, surgical, and coronary patients. For each patient, a mean PG level was calculated from all PG values recorded during the patient's ICU stay. Compared with patients having a mean PG of 80 to 99 mg/dL, the mortality rate was 2-fold higher for patients with mean PG values of 100 to 199 mg/dL, 3-fold higher for patients with a mean BG of 200 to 299 mg/dL, and 4-fold higher for patients with mean PG >300 mg/dL.[1] Furnary et al. observed that mortality increased with increasing average postoperative glucose among diabetic coronary artery bypass graft (CABG) patients.[2] The mortality rate among patients with glucose levels >250 mg/dL was 14.5% vs 0.9% among patients with glucose levels <150 mg/dL.[2] Kosiborod et al. reviewed data from the Cooperative Cardiovascular Project (CCP) and reported similar findings for elderly patients, with and without diabetes, who were hospitalized with acute myocardial infarction (MI).[3] The study cohort was comprised of 141,680 Medicare beneficiaries discharged from nongovernmental hospitals in the US. Thirty-day and 1-year mortality increased with increasing admission BG level, and the relationship was more pronounced for patients without recognized diabetes.[3] Higher BG levels were associated with a significantly greater mortality risk increase in patients without a diagnosis of diabetes than in patients known to have diabetes (P<.001).[3] It is important to note that a large number of hyperglycemic patients did not have a previous diagnosis of diabetes, and these patients were less likely to receive insulin, even when hyperglycemia was severe.[3]


The observational studies reveal a relationship between hyperglycemia and mortality but do not clarify whether hyperglycemia itself increases death rates or is simply an indicator of more severe illness. If hyperglycemia directly contributes to increased rates of morbidity and mortality, then correcting hyperglycemia should decrease both. Dr Inzucchi reviewed several interventional studies designed to test this hypothesis.


The Portland Diabetes Project was a prospective interventional study of 2467 diabetes patients admitted to Portland St Vincent Medical Center (Portland, OR) for open-heart surgery between January 1987 and November 1997.[4] The study group consisted of patients (N = 1499) admitted between September 1991 and November 1997, after implementation of the Portland Continuous Insulin Infusion (CII) protocol. Patients admitted from January 1987 to September 1991 were included in the historical control group (n = 968). The control group received insulin every 4 hours, according to the sliding-scale method, in order to maintain BG below 200 mg/dL. The study group received continuous insulin infusion to maintain BG levels between 150 and 200 mg/dL.[4] The study group had better glycemic control and decreased risk of deep sternal wound infections relative to the control group (see Figure 1).[4]


In a later phase of the project, Furnary et al. tested the hypothesis that perioperative continuous glucose infusion would reduce mortality in diabetes patients undergoing CABG.[2] The study population consisted of diabetes patients who had isolated CABG, with no concomitant procedures, between January 1987 and December 2001. The study included patients with newly diagnosed diabetes in addition to patients with a known history of diabetes. Patients received a new diagnosis of diabetes if they experienced persistently elevated postoperative BG levels and had a discharge requirement for pharmacologic glycemic control. As in the earlier study, the historical control group was comprised of patients who had surgery from January 1987 to September 1991, prior to implementation of the Portland CII protocol. In addition to observing a direct relationship between elevated BG and mortality, the study demonstrated that glycemic control was better (177 ± 30 mg/dL vs 214 ± 41 mg/dL, P<.0001) and mortality lower (2.5% vs 5.3%, P<.0001) with continuous IV insulin infusion than with subcutaneous insulin.[2]


Krinsley also published a study comparing ICU outcomes before and after implementation of an intensive glucose management protocol but in a mixed medical-surgical ICU setting.[5] The protocol goal was the maintenance of BG levels to <140 mg/dL. Blood glucose was monitored by fingerstick; continuous intravenous insulin infusion was used if BG exceeded 200 mg/dL on 2 successive occasions, and subcutaneous insulin was used for lower BG levels.[5] The study included 800 consecutive patients who were admitted immediately before implementation of the glycemic control protocol and the first 800 patients admitted after implementation of the protocol.[5] Overall mortality was reduced 29.3% following protocol implementation. When the data were analyzed according to patient subgroups, significantly lower mortality rates were seen in patients with septic shock or a general surgical or neurologic diagnosis.[5]


The Diabetes Insulin-Glucose Infusion in Acute Myocardial Infarction (DIGAMI) study was a prospective, randomized study that investigated the long-term effect of intensive metabolic treatment on mortality in diabetes patients following acute MI.[6] The study included 620 patients with or without a previous diagnosis of diabetes and a BG concentration >198 mg/dL (11 mmol/L). Patients were assigned to 1 of 4 groups based on previous insulin use and risk level (high or low): no insulin-low risk, no insulin-high risk, insulin-low risk, and insulin-high risk. The classification of high risk was given to patients with at least 2 of the following characteristics: >70 years old, previous MI, history of CHF, current treatment with digitalis. Patients were subsequently randomized to intensive or standard care groups. Intensive care consisted of standard care plus insulin-glucose infusion for ≥24 hours followed by multidose insulin treatment for ≥3 months.[6] At randomization, standard and intensive care group A1C values were 8 ± 2% and 8.1 ± 1.9% (P = .4), respectively. Blood glucose levels were 282.6 ± 75.6 mg/dL for the standard care group and 277.2 ± 73.8 mg/dL for the intensive care group (P = .2).[6] Blood glucose levels were significantly lower in the intensive care group at 24 hours after randomization (172.8 ± 59.4 vs 210.6 ± 73.8 mg/dL, P<.0001) and at hospital discharge (147.6 ± 55.8 vs 162 ± 54 mg/dL, P<.01). After 1 year, fasting BG did not differ between the 2 groups.[6] The mean duration of follow-up was 3.4 years.[6] The long-term follow-up data revealed a 28% decrease in mortality risk after acute MI for patients receiving intensive care compared with those receiving standard care (P = .011; see Figure 2). For patients in the no insulin-low risk category, the difference was even more pronounced, with a relative risk reduction of 51% for the intensive care group relative to the standard care group (P = .004).[6]


Because mortality rates were lower than expected, the DIGAMI study did not have statistical power to clearly identify a specific reason for reduced mortality in the intensive care group. The DIGAMI study was also unable to address the question of whether the decreased long-term mortality rate was due to acute insulin-glucose infusion, post-hospitalization insulin-based metabolic control, or both. Therefore, the DIGAMI 2 study was designed to compare 3 treatment regimens: intensive inpatient/outpatient glycemic control, intensive inpatient glycemic control alone, and conventional care.[7] The DIGAMI 2 study was a prospective, randomized open trial with blinded evaluation. Patients in the DIGAMI 2 study had acute MI and established type 2 diabetes or an admission BG >198 mg/dL. Intensive inpatient/outpatient glycemic control consisted of glucose-insulin infusion for ≥24 hours to achieve BG levels of 126 to 180 mg/dL. Patients were then switched to long-term insulin-based glycemic control with a BG goal of 90 to 126 mg/dL fasting and <180 mg/dL nonfasting. Intensive inpatient treatment included the same glucose-insulin infusion as the inpatient/outpatient group, but follow-up care was left to the discretion of the responsible physician. For conventional care, treatment for glycemic control was left up to the responsible physician. The median study duration was 2.1 years.[7] Neither mortality nor long-term glycemic control differed among the treatment groups, so the study did not identify a significant benefit for any of the treatment regimens.[7] The small difference between study groups was in part attributed to the fact that the target BG level of 90 to 126 mg/dL was never reached for the intensive inpatient/outpatient group.[7] In addition, glycemic control was better than anticipated for all patients.[7] Other study limitations included lower recruitment and mortality than expected.[7] Despite study limitations, hyperglycemia still emerged as an important predictor of mortality.[7]


The Leuven study is a landmark study that focused on the effects of hyperglycemia, rather than diabetes, on mortality and morbidity in surgical ICU patients.[8] Participants (N = 1548) were randomly assigned to intensive therapy or conventional care. For intensive therapy, insulin infusion was initiated when BG exceeded 110 mg/dL and was adjusted to maintain levels of 80 to 110 mg/dL. Conventional care also used insulin infusion, but the initiation level was 215 mg/dL with maintenance goals of 180 to 200 mg/dL.[8] Patients in the tight glucose control arm had lower mortality and morbidity, as illustrated in Figure 3.[8] Leuven 2 was similar to the Leuven study but was conducted in a medical ICU setting.[9] Intensive insulin therapy decreased morbidity, particularly in terms of time required for weaning from mechanical ventilation, time to discharge from the ICU, and time to discharge from the hospital.[9] Intensive therapy did not decrease overall mortality but did decrease mortality among patients in the ICU for ≥3 days (38.1% in the conventional treatment group vs 31.3% in the intensive group).[9] Although these results are interesting, their utility in determining treatment is not clear as it is not possible to identify at the time of admission which patients will remain in the ICU for ≥3 days. The results of these studies suggest that, although tight glycemic control improves outcomes in ICU settings as a general rule, protocols may need to be tailored to meet the needs of the different types of patients in different ICU settings.


There are still unanswered questions regarding control of inpatient hyperglycemia. In position statements on inpatient diabetes and metabolic control, the ACE, AACE, and ADA have included lists of areas needing further research.[10,11] These lists indicate needs for clinical trials to better define optimal glucose levels and to refine insulin protocols. Some ongoing trials are attempting to address these needs. For example, the Normoglycemia in Intensive Care Evaluation and Survival Using Glucose Algorithm Regulation (NICE-SUGAR) study is a multicenter trial looking at different goals for glucose levels, and the Glucontrol Study is a multicenter trial comparing the effects of two insulin-based glycemic control regimens in the ICU.


Recommended BG targets for ICU settings are provided in Table 1. Intravenous insulin administration is the recommended method to achieve these targets, but tight glucose control with IV insulin is associated with risks. These include sympathetic nervous system stimulation, increased stroke volume, potassium/phosphate shifts, potential mitogenic effects, and, most notably, hypoglycemia. Hypoglycemia rates for published IV insulin protocols are presented in Table 2. The data in Table 2 suggest that the rate of hypoglycemia may depend on the infusion protocol.


Table 1. Blood Glucose Targets for ICU Settings

ADA Technical Review ACE Position Statement ADA Practice Guidelines
80-110* mg/dL 80-110 mg/dL As close to 110 mg/dL as possible, and generally
<180† mg/dL

*4.4-6.1 mmol/L. †10.0 mmol/L.

Table 2. Hypoglycemia Rates in Published IV Insulin Protocols

Protocol Hypoglycemia Def. % Patients RR*
Leuven[4] <40 mg/dL 5.1 7
Leuven 2[9] <40 mg/dL 19 6
Yale MICU[12] <40, <50, <60 mg/dL 4.3, 8.7, 17 -
Yale CTICU[13] <40, <50, <60 mg/dL 0, 0.7, 3.6 -
DIGAMI 2[7] <54 mg/dL 9.6-12.7 10
ECLA[14] “symptomatic” 0.4 3
Glucommander[15] <40, <50, <60 mg/dL 2.6, 7.8, 16.5 -

*Relative risk.


Other protocols have been published in addition to those listed in Table 2, including the Portland Protocol[16], a protocol by Markovitz et al.[17], and the Texas Diabetes Council protocol.[18] Implementing a protocol is important for overcoming some barriers to achieving tight glycemic control in the ICU, such as fear of hypoglycemia, lack of appreciation for the benefits of strict glycemic control, and sporadic documentation. The ideal IV insulin protocol has the following characteristics:

  • Standardized (thresholds, targets)
  • Easily ordered
  • Easily implemented
  • Effective
  • Safe (minimal risk of hypoglycemia)
  • Takes into account current BG, rate of change in BG, current insulin infusion rate

The Yale Insulin Infusion Protocol is presented in Figure 4 and the results of implementation are depicted in Figure 5.[12]


Protocols should also provide direction for continuing insulin therapy when a patient is transferred from the ICU to general medical and surgical wards. If IV infusion is not permitted on the wards, the protocol may need to include information regarding the transition to a subcutaneous, multiple-dose injection regimen. Subcutaneous insulin should be administered prior to discontinuing IV infusion in order to maintain effective blood levels of insulin. Short- or rapid-acting insulin should be administered 1 to 2 hours before IV discontinuation, and intermediate- or long-acting insulin should be administered 2 to 3 hours prior to discontinuing IV insulin. Insulin with the appropriate duration of action, when given as one injection or repeated injections, should be used to maintain the basal effect in the period between discontinuation of the IV insulin infusion and the time that the preferred basal insulin is usually administered.[19]


Successful implementation of tight glycemic control in intensive care settings requires a standardized and comprehensive protocol that accommodates patients’ changing requirements, aggressive educational efforts, and multidisciplinary involvement. Physicians can facilitate adoption of insulin infusion protocols by becoming physician champions, educating nursing allies, recruiting and educating clinical allies in each unit, dispelling myths of hypoglycemia, and encouraging forethought and troubleshooting.



  1. Krinsley JS. Association between hyperglycemia and increased hospital mortality in a heterogeneous population of critically ill patients. May Clin Proc. 2003;78:1471-1478.
  2. Furnary AP, Gao G, Grunkemeier GL, et al. Continuous insulin infusion reduces mortality in patients with diabetes undergoing coronary artery bypass grafting. J Thorac Cardiovasc Surg. 2003;125:1007-1021.
  3. Kosiborod M, Rathore SS, Inzucchi SL. Admission glucose and mortality in elderly patients hospitalized with acute myocardial infarction: implications for patients with and without recognized diabetes. Circulation. 2005;111:3078-3086.
  4. Furnary AP, Zerr KJ, Grunkemeier GL, Starr A. Continuous intravenous insulin infusion reduces the incidence of deep sternal wound infection in diabetic patients after cardiac surgical procedures. Ann Thorac Surg. 1999;67:352-360.
  5. Krinsley JS. Effect of an intensive glucose management protocol on the mortality of critically ill adult patients. Mayo Clin Proc. 2004;79:992-1000.
  6. Malmberg K for the DIGAMI Study Group. Prospective randomized study of intensive insulin treatment on long term survival after acute myocardial infarction in patients with diabetes mellitus. BMJ. 1997;314:1512-1515.
  7. Malmberg K, Rydén L, Wedel H, et al. for the DIGAMI 2 Investigators. Intense metabolic control by means of insulin in patients with diabetes mellitus and acute myocardial infarction (DIGAMI 2): effects on mortality and morbidity. Eur Heart J. 2005;26:651-661.
  8. Van den Berghe G, Wouters P, Weekers F, et al. Intensive insulin therapy in critically ill patients. N Engl J Med. 2001;345:1359-1367.
  9. Van den Berghe G, Wilmer A, Hermans G, et al. Intensive insulin therapy in the medical ICU. N Engl J Med. 2006;354:449-461.
  10. American College of Endocrinology. Position statement on inpatient diabetes and metabolic control. Clin Endocrinol. 2004;10:77-82.
  11. American Association of Clinical Endocrinologists, American Diabetes Association. Position statement based on consensus development conference recommendations from, “Inpatient diabetes and glycemic control: a call to action conference.” Available at:
    InpatientDMGlycemicControlPositionStatement02.01.06.REV.pdf. Accessed July 13, 2006.
  12. 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;27:461-467.
  13. Goldberg PA, Sakharova OV, Barrett PW, et al. Improving glycemic control in the cardiothoracic intensive care unit: clinical experience in two hospital settings. J Cardiothorac Vasc Anes. 2004;6:690-697.
  14. Mehta SR, Yusuf S, Diaz R, et al. Effect of glucose-insulin-potassium infusion on mortality in patients with acute ST-segment elevation myocardial infarction: the CREATE-ECLA randomized controlled trial. JAMA. 2005;293:437-446.
  15. Davidson PC, Steed RD, Bode BW. Glucommander: a computer-directed intravenous insulin system shown to be safe, simple, and effective in 120,618 h of operation. Diabetes Care. 2005;28:2418-2423.
  16. Furnary AP, Wu Y, Bookin SO. Effect of hyperglycemia and continuous intravenous insulin infusions on outcomes of cardiac surgical procedures: the Portland Diabetic Project. Endocr Pract. 2004;10:21-33.
  17. Markovitz LJ, Wiechmann RJ, Harris N, et al. Description and evaluation of a glycemic management protocol for patients with diabetes undergoing heart surgery. Endocr Pract. 2002;8:10-18.
  18. Texas Department of State Health Services. IV insulin infusion protocol for critically ill adult patients in the ICU setting. Available at: Accessed July 16, 2006.
  19. Clement S, Braithwaite S, Magee M, et al. Management of diabetes and hyperglycemia in hospitals. Diabetes Care. 2004;27:553-591.



Figure 1

    Figure 2

Figure 3

Figure 4


Figure 5

Figure 6