Metformin is offered in high-risk pregnancy — obesity, polycystic ovary syndrome, previous gestational diabetes — on the reasoning that improving insulin sensitivity should prevent gestational diabetes. This individual participant data meta-analysis tested that reasoning properly.
Ten double-blind placebo-controlled trials of metformin in pregnancies without diabetes were identified; seven supplied individual data, and 2,297 pregnancies were analysed after harmonisation — 1,159 on metformin and 1,138 on placebo — with adjustment for maternal age, body mass index, gestational age at commencement 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 it was identical. Only under IADPSG thresholds did an adjusted analysis suggest benefit (adjusted odds ratio 0.71, 95% CI 0.52 to 0.98). Fasting glucose was 0.06 mmol/L lower, which is meaningless clinically.
What metformin did do was prolong pregnancy: gestation 0.30 weeks longer (95% CI 0.06 to 0.54) and preterm birth less common (adjusted odds ratio 0.64, 95% CI 0.47 to 0.89). Neonatal head circumference was marginally larger. Gastrointestinal side effects were more frequent.
The counselling change is straightforward. A woman with obesity or polycystic ovary syndrome should not be told metformin will stop her developing gestational diabetes, because across seven trials it did not. The preterm birth signal is interesting and belongs in a trial designed to test it, not in a consultation as a reason to prescribe.
- Stop offering metformin as gestational diabetes prevention — the pooled adjusted odds ratio is exactly 1.00
- Continue metformin where it is already indicated for polycystic ovary syndrome, but do not extend it for glycaemic prevention
- Keep doing what works: preconception weight and activity support, early risk assessment and timely oral glucose tolerance testing
- Warn about gastrointestinal side effects, which were more common and drive discontinuation
- Treat the preterm birth finding as hypothesis-generating, not as a new indication
The statistics, in plain English
An adjusted odds ratio of exactly 1.00 with an interval of 0.71 to 1.41 is a clean null: the data are equally compatible with slight harm and slight benefit. The one positive result, adjusted odds ratio 0.71 under IADPSG criteria, has an upper limit of 0.98 — barely below 1.0 — and appears under only one of three diagnostic definitions tested. When a finding survives one definition and disappears under two others, the most likely explanation is the definition rather than the drug. Individual participant data pooling is the strongest form of meta-analysis because it reanalyses each woman's data rather than each trial's summary, which makes this null more persuasive than most.
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 obstetrics & gynaecology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free