Who benefits from adding metformin in type 2 diabetic pregnancy

Design

Secondary effect-modification analysis of the MOMPOD randomised trial (metformin vs placebo)

Population

785 pregnant women with type 2 or early-pregnancy diabetes on insulin

Primary outcome

Composite adverse neonatal outcome, by baseline total daily insulin dose

Effect

Under 30 units/day: adjusted relative risk 0.76 (95% CI 0.59 to 0.98); no neonatal benefit above 60 units/day

The MOMPOD trial found that adding metformin to insulin in type 2 or early-pregnancy diabetes did not reduce composite adverse neonatal outcomes overall. This secondary analysis of 785 women asked whether that flat result hides a subgroup, using baseline total daily insulin dose as an effect modifier.

There was a borderline interaction (P = 0.089). Among women on low baseline insulin (under 30 units/day), adjunctive metformin was associated with a lower composite adverse neonatal outcome than placebo, 44% versus 55%, adjusted relative risk 0.76 (95% CI 0.59 to 0.98). Among those on higher doses (over 60 and over 90 units/day) there was no neonatal benefit, but metformin blunted the rise in insulin requirement during pregnancy (mean difference -13 units, 95% CI -26 to 1, and -27 units, 95% CI -50 to 5).

The practical reading is to individualise rather than switch everyone. In a woman on modest insulin, adjunctive metformin is worth discussing for neonatal benefit; in a woman already on large doses, its value is more about controlling the insulin climb than the neonatal outcome.

• Consider adjunctive metformin in type 2 diabetic pregnancy when baseline insulin is low (under 30 units/day), where it was linked to lower composite neonatal harm (44% vs 55%).
• Expect little neonatal benefit from adding metformin once baseline insulin is high (over 60 units/day).
• In high insulin users, the measurable effect was a smaller rise in insulin requirement, not better neonatal outcomes.
• Use this to individualise rather than as a blanket policy; the interaction was borderline.
• Keep glycaemic targets and insulin titration as the mainstay whether or not metformin is added.

Why it matters

A neutral parent trial may have averaged away a real benefit in the women who need least insulin, and a different benefit in those who need most.

Don't overread it

This is a secondary subgroup analysis with a borderline interaction; it generates a hypothesis and does not overturn the parent trial's overall neutral result.

The statistics, in plain English

The low-dose subgroup interval (0.59 to 0.98) just excludes 1.0; the insulin-sparing intervals cross zero, so that effect is suggestive rather than proven, and the interaction P of 0.089 would be non-significant at the usual 0.05 threshold.

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.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.