Special Feature

Is Insulin the ICU Magic Bullet’?

By Uday B. Nanavaty, MD

A wide variety of therapeutic interventions have failed to produce a significant change in the mortality of critically ill patients. Studies of these interventions include numerous trials of anti-inflammatory agents in sepsis, the trial involving growth hormone in critically ill patients, as well as a host of other investigations. Recently, from one single study, it was reported that tightly controlling blood glucose levels with insulin infusion in the surgical critical care unit resulted in significant reductions in mortality and morbidity. Numerous publications and further analysis of this one study have been published to convince intensivists that insulin infusion, adjusted based on a standard protocol, will dramatically change the outlook for patients across all ICUs. I will review the physiologic role of insulin and the limited evidence available to support this concept of tight regulation of blood glucose. My objective is to suggest that caution needs to be applied and that careful monitoring is needed if one applies this new and aggressive treatment approach—especially outside the surgical critical care setting.

Insulin: An Important Regulator of Metabolism

Insulin is a hormone normally secreted by beta cells of the islets of Langerhans of the pancreas. It is a small protein with a molecular weight of 5808 daltons. Insulin has profound effects on carbohydrate, protein, and fat metabolism. Under physiologic conditions insulin secretion is associated with energy abundance, especially with carbohydrate sources of energy. In turn, it facilitates the entry of glucose into most cells, and helps to store excess energy substances such as glycogen in liver and muscle. It causes fat storage in adipose tissue. If excess carbohydrate cannot be converted to glycogen, insulin helps to store it as adipose tissue. Overall, insulin favors use of carbohydrates as a source of energy and supports fat storage. It also helps to increase storage of proteins, almost an anabolic effect, through unclear mechanisms. By facilitating the use of glucose as the source of energy, and by facilitating glucose entry into most cells, insulin lowers blood glucose levels.

There are 4 important counter-regulatory hormones, as far as blood glucose levels are concerned. These include growth hormone and glucagon as well as cortisol and epinephrine. Epinephrine has potent glycogenolytic actions in the liver as well as more potent lipolytic effects, raising fatty acid levels. Overall, epinephrine tips the balance in favor of fat use as the source of energy during times of stress, even though it causes hyperglycemia. It is also suggested that tumor necrosis factor alpha and interleukin 1 counteract effects of Insulin.

Hyperglycemia in the Critically Ill: Stress Hyperglycemia

Hyperglycemia is considered an important predictor of increased mortality in a variety of critical illnesses. A relationship between elevated plasma glucose levels and poor outcome has been established in patients with acute myocardial infarction and stroke. In diabetic patients with acute myocardial infarction, admission hyperglycemia is considered a predictor of increased mortality, and an intensive insulin regimen to control hyperglycemia has been shown to reduce mortality at one year. This mortality benefit was most evident in diabetic patients who had not received insulin in past, and who had otherwise low cardiovascular risk.

The incidence of stress hyperglycemia is hard to define in large or general ICU populations. Studies in the 1980s suggested that stress hyperglycemia (using a threshold random blood glucose value of > 200 mg/dL) occurs in nearly half of trauma patients needing critical care and also in patients with septic shock.

In a recent report by Kitabchi et al, about 12% of all hospitalized patients had evidence of hyperglycemia, defined either as a fasting blood glucose of 126 mg/dL or 2 or more values of random glucose > 200 mg/dL. Of these patients, about a third (29%) required ICU admission. These groups of patients with stress hyperglycemia had longer hospital lengths of stay and were less often discharged to home when compared to diabetics with hyperglycemia or patients with normoglycemia. Regardless of whether they were admitted to ICU, these hyperglycemic patients without known diabetes experienced higher mortality rate. Among ICU patients, patients with new onset hyperglycemia experienced 3-fold higher mortality rates. These newly hyperglycemic patients also experienced a higher rate of death due to infection and neurological events.

Hyperglycemia is also found in large numbers of patients undergoing cardiac surgery. It is not very well known whether these patients have a higher mortality, as most patients will have hyperglycemia in the immediate postoperative period. The incidence of hyperglycemia in medical ICUs is not well established, and its relationship with mortality or poor outcome is uncertain.

In all the ICU situations, hyperglycemia is induced as a result of the stress response. Sicker patients are more likely to have stress hyperglycemia. In the past, even if hyperglycemia was observed, it was considered a marker of severity of illness and not necessarily as a cause of increased mortality. That concept has now changed.

Tight Control of Blood Glucose: Is it Desirable?

Insulin is not the first hormone to be tried in improving the mortality of patients admitted to the ICU. Previously, human growth hormone has been tried but not recommended as it was found to lead to higher mortality in treated patients. Initial enthusiasm in evaluating another major metabolic hormone, thyroid hormone, has waned with description of terms such as sick euthyroid syndrome.

Insulin infusion was tried in cardiopulmonary bypass patients to prevent hyperglycemia during the bypass. One trial had to be halted as the investigators realized that euglycemia was hard to achieve during bypass and a large number of patients became hypoglycemic after coming off the bypass. Acute mortality from stroke does not change dramatically with insulin administration. In the diabetes mellitus, insulin glucose infusion in acute myocardial infarction study (DIGAMI), long-term mortality benefit was seen with the intensive insulin regimen. However, most of the benefit in mortality was seen after hospital discharge. In the DIGAMI study, one subgroup, patients with diabetes who were not previously treated with insulin, did receive benefit acutely as well with reduced mortality prior to hospital discharge.

In one large study, referred to as the Leuven study, Van den Berghe et al reported that intensive therapy with insulin to tightly control blood glucose values between 80 to 110 mg/dL was associated with a significant reduction in mortality in a large group of patients, mainly postcardiac surgery and other surgical ICU patients. This study used a very different cut-off value to define hyperglycemia. The goal in the experimental group was to achieve euglycemia with insulin infusion. In the control group, the aim was to keep blood glucose levels less than 200 mg/dL and not to intervene if blood glucose levels were < 180mg/dL. All patients in the study received 9 g/h of glucose infusion until feeding was started or total parenteral nutrition was initiated. Van den Berghe et al used a protocol to regulate blood glucose level in the range of 80-110 mg/dL.

This study protocol resulted in a rather profound improvement in outcome. The study group experienced not only a reduction in ICU mortality but also reduced ICU length of stay, less ventilator days, less acute renal failure, less critical illness polyneuropathy, fewer infections, and fewer blood transfusions. In my opinion, these results seem almost too good to be true.

No study these days is without criticism. One can argue that 8% mortality in a peri-operative setting may be too high. Unless we have data from a similar group of patients from the same country or similar country, we cannot make that argument. The other problem that may not have been highlighted before was that all patients were given a fixed number for their neurological status on the APACHE II score. This resulted in a lower overall APACHE score. Van den Berghe et al argue that since both groups were given similar numbers, the APACHE II is comparable. It is possible that there was some imbalance in the 2 groups that went unrecognized. The intensive insulin infusion would have required more frequent blood glucose monitoring necessitating nurses to be at the bedside perhaps a greater number of times. That could have provided a difference in monitoring of the 2 groups. Closer monitoring may improve the outcome, at least intuitively speaking; don’t we all closely monitor some patients who are otherwise not critically ill to detect and treat adverse events early and improve the outcome?

Interestingly enough, the group of patients that benefited the most were those who required > 5 days of intensive care. The patients with higher glucose levels and receiving higher insulin dose experienced the highest mortality, further suggesting that perhaps sicker patients die regardless of whether they receive the intensive insulin infusion.

How Can We Get Glucose Under Control?

The protocol used to control blood glucose in the Leuven study is shown in Table 1.

There are similar protocols with different desired blood glucose ranges that are being used in cardiac surgery ICUs. The protocol used by Brown et al uses more elaborate scheme to control blood glucose.

How Easy is it to Tightly Control Blood Glucose?

Although the nomogram used in the Leuven study is arguably very simple, I am not sure how easy it is for bedside implementation. A study by Brown et al showed that although an algorithm may improve the blood glucose control, there is still a lot of variability among ICU patients. They used a slightly different protocol, the details of which are available in the article cited. Further attempts are being made to automate the process, where continuous blood glucose monitoring data is sent to a pump with preprogrammed algorithm that has the capacity to be manually altered. Such a technology may make it easier to implement the intensive insulin regimen.

Several obvious cautions are necessary when using an intensive insulin protocol. Whenever feedings are interrupted, there has to be a near-automatic adjustment or stopping rule for insulin infusion. I think it will be interesting to see how much of the time the patient actually stays under the goal blood glucose of 80-110 mg /dL on such a protocol. It would also be interesting to study how much increase in workforce is necessary since it seems that almost every single patient in ICU would be on an insulin infusion via central venous catheter, at least based on the Leuven study.

Protocol for Tight Glucose Control
Test Blood Glucose Action

Measure glucose on entry to ICU

> 220 mg/dL Start insulin 2-4 IU/h
220-119 mg/dL

Start insulin 1-2 IU/h

< 110 mg/dL

Do not start insulin but continue

BG monitoring every 4 h

Measure glucose every 1-2 h until > 140 mg/dL Increase insulin dose by
within normal range 1-2 IU/h
110-140 mg/dL Increase insulin dose by
0.5-1 IU/h
Approaching normal range Adjust insulin dose by
0.1-0.5 IU/h
Measure glucose every 4 h Approaching normal range Adjust insulin dose by
0.1-0.5 IU/h
Normal Insulin dose unchanged
Falling rapidly Reduce insulin dose by half
and check more frequently
60-80 mg/dL Reduce insulin dose and
check BG within 1 h


Stress hyperglycemia is variably defined. By and large, most studies suggest that stress hyperglycemia is a marker of increased morbidity and mortality in critically ill patients. There are limited data to suggest that tight control of blood glucose in normal range in non-diabetic critically ill patients will result in improved outcome. More studies are needed to confirm safety, efficacy and cost of strict glucose control in critically ill patients.

Suggested Reading

1. van den Berghe G, et al. Intensive insulin therapy in the critically ill patients. N Engl J Med. 2001;345(19): 1359-1367.

2. Martinez-Riquelme AE, Allison SP. Insulin revisited. Clin Nutr. 2003;22(1):7-15.

3. Van den Berghe G, et al. Outcome benefit of intensive insulin therapy in the critically ill: Insulin dose versus glycemic control. Crit Care Med. 2003;31(2):359-366.

4. Brown G, Dodek P. Intravenous insulin nomogram improves blood glucose control in the critically ill. Crit Care Med. 2001;29(9):1714-1719.

5. McCowen KC, et al. Stress-induced hyperglycemia. Crit Care Clin. 2001;17(1):107-124.

6. Umpierrez GE, et al. Hyperglycemia: An independent marker of in-hospital mortality in patients with undiagnosed diabetes. J Clin Endocrinol Metab. 2002;87(3):978-982.

7. Malmberg K. Prospective randomized study of intensive insulin treatment on long-term survival after acute myocardial infarction in patients with diabetes mellitus. DIGAMI (Diabetes Mellitus, Insulin Glucose Infusion in Acute Myocardial Infarction) Study Group. BMJ. 1997;314(7093):1512-1515.