Design: Systematic review and meta-analysis of 19 RCTs
Population: 6213 pregnant women; supervised exercise vs usual care
Primary outcome: Gestational diabetes mellitus
Effect: RR 0.74 (95% CI 0.57 to 0.96), I² = 47%

This systematic review pooled 19 randomised trials (6213 pregnant women) comparing supervised exercise with usual care. Gestational diabetes was the prespecified primary outcome. Certainty of evidence was rated moderate.

Supervised exercise was associated with a quarter less gestational diabetes (RR 0.74, 95% CI 0.57 to 0.96), and with less gestational hypertension (RR 0.55), excessive gestational weight gain (RR 0.75) and macrosomia (RR 0.61). Effects looked larger when exercise started early in pregnancy, ran beyond 20 weeks, reached at least 80% adherence, combined aerobic and resistance work, and totalled 150 minutes a week or more.

The operative word is supervised. These were structured programmes, not advice leaflets, and that is where referral effort belongs for women at high risk of gestational diabetes.

Identify high-risk women (previous GDM, raised BMI, family history, South Asian ethnicity) at booking.
Refer to a supervised programme early rather than waiting for the glucose tolerance test.
Aim for about 150 minutes a week, combining aerobic and resistance exercise.
Screen for obstetric contraindications to exercise before referral.

Why it matters

Prevention of gestational diabetes works best if it starts before the glucose test, not after.

The statistics, in plain English

An RR of 0.74 means about 26% fewer cases; the interval (0.57 to 0.96) stays below 1, but its upper end is close to it, so the true benefit could be small. I² of 47% means the trials disagreed moderately. The early-start and high-adherence advantages come from subgroup comparisons, which are weaker evidence than the pooled result.