Treatments for gestational diabetes: a systematic review and meta-analysis

Objective To investigate the effectiveness of different treatments for gestational diabetes mellitus (GDM). Design Systematic review, meta-analysis and network meta-analysis. Methods Data sources were searched up to July 2016 and included MEDLINE and Embase. Randomised trials comparing treatments for GDM (packages of care (dietary and lifestyle interventions with pharmacological treatments as required), insulin, metformin, glibenclamide (glyburide)) were selected by two authors and double checked for accuracy. Outcomes included large for gestational age, shoulder dystocia, neonatal hypoglycaemia, caesarean section and pre-eclampsia. We pooled data using randomeffects meta-analyses and used Bayesian network meta-analysis to compare pharmacological treatments (ie, including treatments not directly compared within a trial). Results Forty-two trials were included, the reporting of which was generally poor with unclear or high risk of bias. Packages of care varied in their composition and reduced the risk of most adverse perinatal outcomes compared with routine care (eg, large for gestational age: relative risk0.58 (95% CI 0.49 to 0.68; I-2 = 0%; trials 8; participants 3462). Network meta-analyses suggest that metformin had the highest probability of being the most effective treatment in reducing the risk of most outcomes compared with insulin or glibenclamide. Conclusions Evidence shows that packages of care are effective in reducing the risk of most adverse perinatal outcomes. However, trials often include few women, are poorly reported with unclear or high risk of bias and report few outcomes. The contribution of each treatment within the packages of care remains unclear. Large well-designed and well-conducted trials are urgently needed.