Prediction Models for Maternal and Offspring Short- and Long-Term Outcomes Following Gestational Diabetes: A Systematic Review.

Abstract

Objectives: Gestational diabetes mellitus (GDM), affecting one in seven pregnant women worldwide, can have short- and long-term adverse outcomes for both the mother and her baby. Despite a raft of prognostic models aiming to predict adverse GDM outcomes, very few have impacted clinical practice. This systematic review summarizes and critically evaluates prediction models for GDM outcomes, to identify promising models for further evaluation.

Methods: We searched EMBASE, MEDLINE, Web of Science, CINAHL, and CENTRAL for studies that reported the development or validation of predictive models for GDM outcomes in mother or offspring (PROSPERO: CRD42023396697).

Results: Sixty-four articles detailing 103 developed and 12 validated models were included in this review. Of these, 45% predicted long term, 31% birth, and 23% pregnancy outcomes. Most models (87%) had a high risk of bias, lacking sufficient outcome events, internal validation, or proper calibration. Only eight models were found at low risk of bias.

Conclusions: Our findings highlight a gap in rigorously developed prediction models for adverse GDM outcomes. There is a need to further validate existing models and evaluate their clinical utility to generate risk prediction tools capable of improving clinical decision-making for women with GDM and their children.