Progression to type 2 diabetes mellitus after gestational diabetes: An umbrella review of 15 systematic reviews and assessment of causal frameworks

Ballena-Caicedo J.; Valladolid-Sandoval L.A.M.; Zuzunaga-Montoya F.E.; Vera-Ponce V.J. · Endocrine and Metabolic Science (2026) · DOI: 10.1016/j.endmts.2025.100285

Resumen

Background Gestational diabetes mellitus (GDM) confers a markedly elevated risk of progression to type 2 diabetes mellitus (T2DM), although the precise magnitude of this risk and its temporal evolution remain inadequately characterized for clinical decision-making. Objective To systematically synthesize reviews on glycemic alterations following GDM, quantifying prevalence, incidence, and risk estimates, and to critically assess the use of causal frameworks for inference regarding risk factors. Methods We conducted an umbrella review with searches in PubMed/MEDLINE, Embase, Web of Science, and Scopus through September 2025. For reviews that synthesized risk factors, we assessed whether they employed directed acyclic graphs to guide adjustment for confounders. Results Fifteen reviews with data from more than 3 million women were included. Prior GDM conferred a 6- to 13-fold increased risk of T2DM compared with normoglycemic pregnancies, with cumulative incidence of 9–32 % depending on ethnicity, follow-up duration, and diagnostic criteria. The incidence rate was 26.20 per 1000 person-years, projecting to 20 % at 10 years. Twenty-seven percent presented glycemic alterations at 6–12 weeks postpartum. The most consistent predictors included glycemic markers during pregnancy, insulin requirement, elevated body mass index, family history, and non-White ethnicity. Critically, no review employed directed acyclic graphs to support confounder adjustment, introducing substantial risk of residual bias in the reported point estimates, although the direction of associations showed robust consistency across temporal windows and independent reviews. Conclusions GDM substantially increases T2DM risk, with glycemic burden evident from early postpartum. The universal absence of DAGs limits causal inference regarding point estimates, although the direction of associations is robust. Copyright

Instituciones: Universidad Nacional Toribio Rodríguez de Mendoza de Amazonas