This was a patient-level meta-analysis, the strongest form these syntheses take, pooling individual data rather than published summaries. Ten double-blind placebo-controlled trials of metformin in pregnancies without established diabetes were identified; seven supplied individual participant data, and after harmonisation 2,297 pregnancies were analysed, 1,159 on metformin and 1,138 on placebo, adjusted for maternal age, BMI, gestational age at starting and baseline glucose.
Metformin did not reduce gestational diabetes. By WHO 1999 criteria the adjusted odds ratio was 1.00 (95% CI 0.71 to 1.41), and by NICE 2015 criteria the same. Only the IADPSG-threshold analysis suggested a reduction, and only after adjustment (adjusted OR 0.71, 0.52 to 0.98). Fasting glucose fell by a trivial 0.06 mmol/L (-0.10 to -0.01) with no change in the two-hour post-load value.
What metformin did do sits elsewhere. Gestation was longer by 0.30 weeks (0.06 to 0.54), preterm birth was less likely (adjusted OR 0.64, 0.47 to 0.89), and neonatal head circumference was slightly larger (2.43 percentiles, 0.13 to 4.72). Gastrointestinal side effects were commoner, as always.
The practical reading is to stop offering metformin to high-risk women as gestational diabetes prevention. That indication is not supported. The preterm birth signal is interesting and biologically plausible, but it is a secondary outcome in a synthesis designed to answer a different question, and it needs its own trial before it changes anything. In Indian obstetric practice, where the threshold for starting metformin in women with PCOS or obesity entering pregnancy is often low, this is a reason to be clearer with women about what the drug is and is not expected to achieve.
- Do not start metformin in pregnancy for the purpose of preventing gestational diabetes.
- Where metformin is already in use for another indication, this evidence gives no reason to stop it.
- Keep screening for gestational diabetes on schedule regardless of metformin use — it does not shift the risk.
- Counsel on gastrointestinal side effects, which were consistently commoner on metformin.
- Treat the preterm birth finding as hypothesis-generating, not as a new indication.
The statistics, in plain English
An odds ratio of 1.00 means no difference at all, and the interval of 0.71 to 1.41 tells you the data are compatible with a modest benefit or a modest harm — this is a genuinely null result, not a near-miss. The one positive analysis, using IADPSG thresholds, gave 0.71 with an upper bound of 0.98: it only just clears 1.0, it emerged after adjustment when the unadjusted figure did not reach significance, and it is one of several diagnostic criteria tested. Testing several definitions of the same outcome makes one crossing the line by chance more likely, which is why the null results across the other criteria carry more weight.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for diabetes & endocrinology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free