Recurrent gestational diabetes and the tenfold later risk

Design

Retrospective cohort, time-varying exposure

Population

326,115 mothers, 442,225 singleton pregnancies, Southern California

Primary outcome

Incident diabetes outside pregnancy

Effect

GDM HR 5.87 (5.67–6.08); recurrent GDM HR 9.91 (9.14–10.75); Asian women with recurrence HR 11.74 (9.64–14.30)

A retrospective cohort from Kaiser Permanente Southern California followed 326,115 mothers across 442,225 singleton pregnancies from 2007 to 2022, with a median follow-up of 6.2 years. GDM was identified from glucose screening in each pregnancy and treated as a time-varying exposure.

Compared with no GDM history, one GDM pregnancy was associated with an HR of 5.87 for later diabetes, and recurrent GDM with an HR of 9.91. Asian women had the highest GDM rate (about 20%), the highest recurrence rate (about 53% of those with a prior GDM pregnancy) and the highest risk after recurrence (HR 11.74).

This is an association study in an insured US population, and the Asian group is heterogeneous. But the direction matters for South Asian practice, where GDM is common and postpartum follow-up is often lost. A second GDM pregnancy is a marker of a woman whose risk is running well ahead of her peers.

• Record the number of GDM-affected pregnancies, not just whether GDM ever occurred.
• Treat a woman with recurrent GDM as being at very high risk of type 2 diabetes after delivery.
• Check that a postpartum glucose test was actually done after each GDM pregnancy.
• Discuss weight, activity and interpregnancy planning before the next pregnancy, not during it.
• In South Asian women, expect higher recurrence and act earlier on abnormal postpartum results.

Why it matters

Recurrence is a risk marker in its own right, and most follow-up systems do not record it.

Don't overread it

This was observational — it shows who is at risk, not that any intervention lowers it.

The statistics, in plain English

A hazard ratio of 9.91 means women with recurrent GDM developed diabetes at roughly ten times the rate of women without GDM over the follow-up period. The narrow confidence intervals reflect the very large cohort, not the absence of bias — residual confounding by weight and family history is still possible.

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.