Early postpartum dysglycemia after gestational diabetes: β-cell dysfunction and risk stratification in a multiethnic cohort.

INTRODUCTION: Postpartum oral glucose tolerance test (OGTT) screening is recommended after gestational diabetes (GDM), but attendance is often incomplete and testing may not capture metabolic heterogeneity. Western-derived risk models may be less applicable to Asians, in whom dysglycemia may occur at lower body mass index (BMI) and reflect greater β-cell vulnerability. We examined the prevalence, clinical phenotype and predictors of early postpartum dysglycemia in women with GDM. METHODS: We retrospectively studied 575 women with GDM managed at a tertiary center in Singapore between 01/10/2023 to 30/04/2025. Early postpartum dysglycemia was defined by World Health Organization criteria using 6-8 weeks postpartum OGTT. Multivariate logistic regression identified predictors of postpartum dysglycemia. Missing data were handled by multiple imputation; candidate predictors were screened by univariate logistic regression and assessed for multicollinearity using variance inflation factors. Model performance was evaluated by area under the receiver operating characteristics curve (AUC), with internal validation using 1000 bootstrap resamples. Dominance analysis quantified relative predictor contributions. RESULTS: Postpartum OGTT completion rate was 71.1% (409/575). Among women tested, 34.2% (140/409) had early postpartum dysglycemia. Dysglycemia occurred in a relatively lean phenotype and was associated with lower homeostatic model assessment of β-cell function (HOMA-β). Independent predictors were family history of diabetes, prior GDM, elevated pregnancy 2-hours OGTT glucose, Chinese ethnicity, and higher HbA1c while in-vitro fertilization and higher pre-pregnancy BMI were inversely associated with dysglycemia. Discrimination was good with AUC 0.77 (95% CI 0.72 - 0.81). Dominance analysis ranked pregnancy 2-hour OGTT glucose (18.8%) and family history (17.1%) as top contributors. Threshold analyses suggested pregnancy 2-hours OGTT ≥8.5 mmol/L was specific but insensitive, whereas antepartum HbA1c ≥5.5% was sensitive but less specific. CONCLUSIONS: Early postpartum dysglycemia after GDM was common and appeared consistent with a relatively lean, β-cell-predominant phenotype. Pragmatic risk stratification may complement tiered postpartum surveillance and prevention.