External Validation of the PeRSonal Gestational Diabetes Model for Predicting Adverse Pregnancy Outcomes in Women with Gestational Diabetes: A Retrospective Cohort Study in a Tertiary Hospital in Thailand

Abstract

Objective: This study aimed to validate the PeRSonal Gestational Diabetes (GDM) model with two-step glucose tolerance diagnostic criteria.Material and Methods: A retrospective cohort study was conducted on participants having delivered with GDM diagnosis in a tertiary hospital; from October 1, 2020, until September 30, 2022. The main outcome was a composite of maternal and perinatal adverse pregnancy complications. Model validation evaluated the predictors and calculated risk by using a two-step glucose tolerance test in the PeRSonal model formula. Model performance was analyzed for discrimination, calibration, and overall performance. Results: This study analyzed 685 from the initial 764 participants with GDM, with 218 (31.8%) developing adverse pregnancy outcomes. The most frequent adverse outcomes were hypertensive disorders in pregnancy 132 (19.3%) and neonatal hypoglycemia 91 (13.3%). This validation achieved an area under the curve (AUC) of 0.70 (95% confidence interval (CI) 0.65 to 0.74), calibration-in-the-large of 0.17, a calibration slope of 1.34, and a Brier score of 0.20, respectively. The cut-off clinical risk probability of 27.5% can predict adverse outcomes with a sensitivity of 67.3%, specificity of 63.8%, a positive predictive value (PPV) of 46.7%, and a negative predictive value (NPV) of 80.5%. Conclusion: The PeRSonal model maintains its predictive effectiveness in two-step glucose tolerance diagnostic criteria.