Van den Berghe G et al. (2006) Intensive insulin therapy in mixed medical/surgical intensive care units. Diabetes 55: 3151–3159

Compared with conventional insulin therapy, intensive insulin therapy (IIT) improves the outcomes of critically ill patients, but concerns remained that IIT might be less beneficial for medical than surgical cases, and might have reduced efficacy in some subgroups of patients. Van den Berghe and colleagues, therefore, pooled datasets from two similar randomized, controlled trials of IIT in 1,200 medical and 1,548 surgical patients in critical care, respectively.

Overall mortality was lower in IIT-treated than conventionally treated patients (20.5% versus 23.6%). IIT had the greatest benefit in patients treated for ≥ 3 days in the critical-care unit (37.9% mortality with conventional therapy, versus 30.1% with IIT). IIT did not improve the survival of patients who remained <3 days in the critical-care unit, but caused no harm.

Maintenance of blood-glucose levels <150 mg/dl was crucial to reduce mortality, but the greatest benefit was obtained by blood-glucose levels <110 mg/dl, especially over several days. IIT strictly maintained blood glucose within 80–110 mg/dl from admission onwards, which protected the kidneys and peripheral nervous system, albeit with an increased risk of hypoglycemia that caused only transient morbidity (although the authors conceded that the survival benefit of IIT might be reduced). Only patients with diabetes experienced no survival benefit with IIT, and showed a trend towards increased mortality risk at blood-glucose levels <110 mg/dl. The authors postulate that diabetic patients might have adaptations to chronic hyperglycemia that render normalization of blood glucose potentially harmful; they suggest that diabetic patients' IIT blood-glucose targets should be based on the patient's usual values.