Evidence-based insulin treatment in type 1 diabetes mellitus

Abstract

Aim Evaluation of the evidence base for recommending different insulin treatment regimens in type 1 diabetes. Methods A computerised literature survey was conducted using The Cochrane Controlled Trials Register and the Pub Med database for the period of 1982–2007. Results A meta-analysis on only 49 out of 1295 references showed that CSII compared with conventional or multiple insulin injections therapy demonstrated a significant reduction in mean HbA1c (primary outcome) of 1.2% CI [0.73; 1.59] ( P < 0.001) without increasing the risk of hypoglycaemia. The evidence for using four versus two daily insulin injections was based on only one publication demonstrating an improved quality of life but no significant reduction in HbA1c or hypoglycaemia. A comparison of rapid-acting insulin analogues and human soluble insulin demonstrated a statistically significant reduction in HbA1c of 0.1% CI: [0.01; 0.16] ( P = 0.03) using rapid-acting insulin analogues. The mean frequency of hypoglycaemia was reduced with 14 ± 3.7% (<0.05). Conclusion The scientific evidence supporting the three common insulin regimens was rather sparse. Only five studies during the past 25 years fulfil the optimal criteria for a clinical trial, and only 5 trials on insulin analogues were performed as double-blinded. Current evidence suggests that CSII treatment results in a significant reduction in HbA1c without inducing more hypoglycaemia. Rapid-acting insulin analogues compared to human soluble insulin provide statistically significant but clinically minor improvement in HbA1c but seem to reduce the risk for hypoglycaemia.