Duration of time on intensive insulin therapy predicts severe hypoglycemia in the surgically critically ill population.

Background: Hypoglycemia has emerged as a barrier to the practice of intensive insulin therapy. Current literature suggests that hypoglycemia occurs at variable rates and has different effects on outcomes in surgical and medical populations. We sought to determine the incidence, independent predictors, and effect on outcome of severe hypoglycemia (≤ 40 mg/dl) in a surgical population.

Methods: A retrospective analysis was performed on all critically ill surgical patients treated with IIT from October 2004 to February 2007. Euglycemia (goal 80-110 mg/dl) was maintained using automated computerized titration of an insulin infusion. The primary outcome of interest was any episode of severe hypoglycemia (≤ 40 mg/dl). Multivariate logistic regression was used to determine the independent predictors of developing severe hypoglycemia.

Results: A total of 60,298 data entries (1,118 patients) for glucose were analyzed. There were 64 severe hypoglycemic episodes in 52 patients (4.6% of the patients). There was a significant increase in deaths among patients who experienced at least one episode of hypoglycemia when compared with those who did not (26.9% vs. 15.3%, P = 0.03). Logistic regression revealed that the time spent on the protocol was the best predictor of developing a hypoglycemic event when controlling for other known risk factors of hypoglycemia.

Conclusions: Intensive insulin therapy can be implemented with a low percentage of patients (4.6%) experiencing severe hypoglycemia. Mortality rate was higher for patients experiencing hypoglycemia. The duration of the time spent on the protocol was the best predictor of hypoglycemia, suggesting that hypoglycemia is a mathematic probability of prolonged illness, not a reflection of illness severity or demographic features.