Six RCTs (n=1,388 participants) were included in the review. Sample size ranged between 23 and 733 patients. The treatment and insulin groups were comparable in four trials, although in one trial body mass index was higher in the group that received oral hypoglycaemic agents and in another trial gestational age was more in the OHA group. Allocation concealment was adequately reported in two RCTs. There were no trials with assessor blinding. Doses of insulin and OHAs were not reported in most studies. Intention-to-treat analyses were used in four trials. Losses to follow-up were less than 10% or not significant in five trials.
There were no significant differences observed between insulin and OHAs for glycaemic control (WMD 1.31, 95% CI -0.81 to 3.43; five RCTs), neonatal hypoglycaemia (OR 1.59, 95% CI 0.70 to 3.62; five RCTs), birthweight (WMD 56.11, 95% CI -42.62 to 154.84; six RCTs), incidence of large-for-gestational age-babies (OR 1.01, 95% CI 0.61 to 1.68; four RCTs). Other outcomes in which no significant differences were found between treatments were admission to neonatal intensive care units, neonatal respiratory distress and birth injuries, incidence of small-for-gestational-age babies, preterm births, intrauterine foetal death, congenital abnormalities and maternal hypoglycaemia. The authors reported statistically significant heterogeneity across the results for all complications.
There was a statistically significant higher level of patients satisfaction found with OHA treatment with metformin compared to the group that received insulin (76.6% compared to 27.2%, one RCT) which was attributed to the greater ease of administration of the OHA.
The proportions of patients allocated to receive OHAs who also required insulin treatment ranged from "low" to 46.3%, with a higher conversion rate observed in the studies that used metformin.