Computer program refines postsurgical insulin treatment

Numerous studies have demonstrated the benefit of "tight" control of hyperglycemia in the intensive care setting in a variety of clinical situations. The most impressive benefit has been in the reduction of in hospital infectious complications although many benefits have been reported including decreased mortality, improved surgical outcomes and decreased length of stay among others. Enthusiasm for decreasing or eliminating hyperglycemia has always been tempered by the potential risk of inducing hypoglycemia which can have catastrophic consequences especially in the very young and in patients with cardiovascular disease.

Fortunately, these events are uncommon. Most ICUs have developed paper based nursing protocols utilizing insulin drips and frequent serum glucose measurements. While effective, this approach is both labor intensive and costly. Computer based algorithms have been proven to be more effective than paper protocols in some studies in reducing the number of hyperglycemic episodes but have led to more frequent hypoglycemic events. This is the first study demonstrating a reduction in both hyper and hypoglycemic episodes. Perhaps equally impressive was the significant reduction in serum glucose determinations over time (nearly 50% in the study cohort) while reducing both hyper and hypoglycemic events by a similar rate.

If the results of this retrospective study using historical controls can be achieved in prospective randomized trials, computer based programs like this one may someday become the standard of care for glucose management in the ICU. Until that time, frequent glucose determinations and modifications of insulin dosing made on the basis of human experience and judgment will continue to be the preferred practice.

Dr. Frank Pomposelli is professor of surgery at Tufts University School of Medicine andchairman of surgery at St. Elizabeth’s Medical Center, Boston and an associate medical editor for Vascular Specialist.