OBJECTIVE A precise time-action profile of insulin lispro (Humalog) at mealtime may reduce the incidence of severe hypoglycemia. Because it is a rare complication, we performed a cumulative meta-analysis to compare the frequency of severe hypoglycemia during insulin lispro and human regular insulin therapy in type 1 diabetic patients.

RESEARCH DESIGN AND METHODS The analysis included eight large multicenter clinical trials, three with parallel and five with crossover designs. The studies included 2,576 type 1 diabetic patients in total, with 2,327 receiving insulin lispro and 2,339 receiving regular human insulin, representing >1,400 patient-years of insulin therapy. Severe hypoglycemia was defined as coma or requiring glucagon or intravenous glucose. The patients received either NPH or ultralente as their basal insulin and insulin lispro or regular human insulin before each meal.

RESULTS Seventy-two patients (3.1%) had a total of 102 severe hypoglycemic episodes during insulin lispro therapy, compared with 102 patients (4.4%) with a total of 131 episodes during regular human insulin therapy (P = 0.024).

CONCLUSIONS The results of this meta-analysis demonstrate that in type 1 diabetic patients, the frequency of severe hypoglycemia can be reduced by taking insulin lispro as compared with regular human insulin therapy.