Glycemic Control in Critically Ill Patients : Leuven and Beyond FREE TO VIEW

This editorial will start with a confession. Before November 8, 2001, glucose monitoring in critically ill patients was not “on my radar screen.” As an ICU director, I followed the general principles learned during my residency: hyperglycemia occurred commonly in this patient population, was very likely protective, and should not be treated unless it became excessive, perhaps above the level of 200 to 225 mg/dL. The experience that day of reading the first Leuven study¹ was, for me, startling and life changing. The following morning, I challenged the comprehensive database created in my ICU to identify a relationship between glycemic levels during ICU admission and mortality and found that even modest degrees of hyperglycemia were associated with increased risk.²

“Leuven I” was the study that launched a thousand protocols; tight glycemic control in the ICU (TGC) has been endorsed by professional societies³⁴ and is currently an emerging standard of care worldwide. Nevertheless, after nearly 6 years there has been a paucity of confirmatory evidence in the critical care literature. In 2004, the 1,600-patient “before and after” study performed by my ICU team was reported (the Stamford Study⁵), providing confirmation of the benefit of TGC in a mixed medical-surgical population. These observations have been updated recently in a cohort of 5,365 patients, half admitted before and half after the initiation of TGC, affirming a strong treatment effect.⁶ “Leuven II,” undertaken in a medical ICU setting, emerged 5 years after the first Belgian study, detailing more modest benefits of intensive insulin therapy.⁷ The article in this issue of CHEST (see page 268)⁸ summarizes the two Leuven studies and has provided an opportunity for their authors to mount a spirited defense to the methodologic questions that have been raised about their work.⁹¹⁰

How does one explain the divergent results of different interventional trials regarding TGC and mortality? Leuven I and the Stamford studies were strongly positive; Leuven II was positive when considering the targeted group of patients with ICU stays > 3 days but negative in the intention-to-treat analysis of the entire group. Moreover, the results of two trials have cast more doubt about the efficacy of TGC. The European investigators of the Efficacy of Volume Substitution and Insulin Therapy in Severe Sepsis trial¹¹ performed a multicenter, randomized, prospective study of intensive vs “conventional” glycemic control and two different fluid resuscitation strategies in a two-by-two design. The trial was closed prematurely because of a nearly sixfold increase in severe hypoglycemia among the patients in the intensively treated group. The Glucontrol Trial,¹² another multicenter European study of TGC in medical-surgical ICUs, also “crashed and burned” due to protocol violations and an unacceptable rate of hypoglycemia in the intensively treated group.

These two trials share one important attribute with Leuven II: a high rate of severe hypoglycemia (SH) [most typically defined as blood glucose < 40 mg/dL] and therefore, in my view, should not be considered indictments of TGC but rather as examples of failed trials. The rates of SH among the conventionally and intensively treated patients in Leuven II were 3.1% and 18.7%, respectively. In contrast, the corresponding rates in Leuven I were 0.8% and 5.1%, and the rate of SH in the Stamford study did not change with implementation of TGC. The Leuven II authors⁷ stated that “logistic-regression analysis identified hypoglycemia as an independent risk factor for death. Hence, it is possible that hypoglycemia induced by intensive insulin therapy may have reduced a portion of the potential benefit” (emphasis added). Indeed, even a single episode of SH conferred an increased risk of mortality in an analysis of the 5,365 patient cohort from my ICU.¹³

Why was the rate of hypoglycemia so different in the two Leuven studies? One important difference between Leuven I and Leuven II concerns the method of glycemic monitoring. The first study relied exclusively on arterial blood gas analyzers, while the second study used arterial blood in an unspecified smaller number of patients, with the remainder of samples obtained from capillary blood and measured on bedside glucometers. Another difference is the patient population. The patients in the medical ICU had much higher APACHE (acute physiology and chronic health evaluation) II scores and a higher prevalence of sepsis on admission, risk factors for the development of SH,¹³¹⁴ than did the surgical ICU patients.

The message, then, should be heard loud and clear: ICUs attempting to implement TGC protocols need to be able to do it safely. Avoidance of hypoglycemia requires several key components. The ICU culture must accept protocol-driven care; a glycemic target should be chosen that can be achieved safely; the strengths and weaknesses of the different available glucose monitoring technologies must be understood; and finally, clinicians need to have access to actionable, real-time data, not just relating to glucose control, but ideally also to relevant outcomes such as severity-adjusted mortality and measures of morbidity and resource utilization.

The ICU community eagerly awaits the publication of the Normoglycemia in Intensive Care Evaluation and Survival Using Glucose Algorithm Regulation study,¹⁵ a multicenter trial of TGC in mixed medical-surgical ICU populations. We can only hope that the investigators were able to achieve the following: the glycemic targets were met; that there was a clear separation of glycemic control between the conventionally and intensively treated groups; and, finally, that the intervention did not lead to an excessive rate of hypoglycemia. Meeting these three basic goals should provide important corroboration of the beneficial clinical impact of TGC.