In a randomized trial published in 2001, Belgian researchers demonstrated that intensive insulin therapy lowered mortality among hyperglycemic surgical intensive care unit patients, most of whom had undergone cardiac surgery. Now these researchers have conducted a similar randomized trial in their medical ICU.
Twelve hundred patients with anticipated ICU stays
3 days received
either "conventional" insulin treatment (continuous infusion
started when glucose exceeds 215 mg/dL and adjusted to maintain blood
glucose at 180–200 mg/dL) or intensive insulin treatment (infusion
started when glucose exceeds 110 mg/dL and adjusted to maintain
glucose at 80–110 mg/dL). About 17% had known diabetes.
The primary endpoint, in-hospital mortality, did not differ significantly
in the conventional and intensive groups (40% vs. 37%). However,
among the 433 patients whose ICU stays were <3 days, mortality was
higher with intensive than with conventional treatment. In
contrast, among the 767 patients who stayed
3 days,
mortality was lower with intensive than with conventional treatment.
Some morbidities (e.g., new kidney injury, time on mechanical
ventilation) were significantly less common in the intensive group
than in the conventional group, whereas other morbidities (e.g.,
bacteremia) were not.
In this study, intensive insulin therapy appeared to be beneficial for patients whose medical ICU stays extended beyond several days, but — for reasons which are unclear — there was a hint of harm for patients with short ICU stays. Unfortunately, patients destined to have shorter ICU stays were not identifiable prospectively. While we await the results of ongoing multicenter trials, an editorialist proposes the following: Attempt to keep blood glucose lower than 150 mg/dL during the first 3 days, and intensify therapy for patients whose critical illnesses extend beyond 3 days.
Reference:
1. Van den Berghe G et al. Intensive insulin therapy in the medical ICU. N Engl J Med 2006 Feb 2; 354:449-61.
2. Malhotra A. Intensive insulin in intensive care. N Engl J Med 2006 Feb 2; 354:516-8.