Background
Two oral hypoglycaemic agents, metformin and glibenclamide, have been compared with insulin in separate large randomised controlled trials and have been found to be as effective as insulin in gestational diabetes. However, very few trials have compared metformin with glibenclamide.

Materials and Methods
Of 159 South Indian women with fasting glucose ≥5.5 mmol/l and ≤7.2 mmol/l and/or 2‐h post‐prandial value ≥6.7 mmol/l and ≤13.9 mmol/l after medical nutritional therapy consented to be randomised to receive either glibenclamide or metformin. 80 women received glibenclamide and 79 received metformin. Neonatal outcomes were assessed by neonatologists who were unaware that the mother was part of a study and were recorded by assessors blinded to the medication the mother was given. The primary outcome was a composite of neonatal outcomes namely macrosomia, hypoglycaemia, need for phototherapy, respiratory distress, stillbirth or neonatal death and birth trauma. Secondary outcomes were birthweight, maternal glycaemic control, pregnancy induced hypertension, preterm birth, need for induction of labour, mode of delivery and complications of delivery.

Results
Baseline characteristics were similar but for the higher fasting triglyceride levels in women on metformin. The primary outcome was seen in 35% of the glibenclamide group and 18.9% of the metformin group [95% CI 16.1 (2.5, 29.7); P = 0.02]. The difference in outcome related to a higher rate of neonatal hypoglycaemia in the glibenclamide group (12.5%) versus none in the metformin group [95% CI 12.5(5.3, 19.7); P = 0.001]. Secondary outcomes in both groups were similar.

Conclusion
In a south Indian population with gestational diabetes, metformin was associated with better neonatal outcomes than glibenclamide.