Perinatal & Pediatric
Development and Validation of the Adverse Pregnancy Outcomes Population Risk Tool (PregPoRT): A Population-Based Model for Predicting Maternal Risk in Canada Sabrina Chiodo* Sabrina Chiodo Chiodo Chiodo Chiodo Division of Epidemiology, Dalla Lana School of Public Health, University of Toronto, Toronto, Ontario, Canada
Introduction: Adverse pregnancy outcomes (APOs), including gestational diabetes, preeclampsia, and placental abruption, are major contributors to maternal and fetal morbidity. Existing prediction models rely on biomarkers and clinical records, limiting their utility for population health planning. We developed and validated the Adverse Pregnancy Outcomes Population Risk Tool (PregPoRT), a model integrating clinical, social, and environmental determinants to predict APO risk in Canada.
Methods: We conducted a retrospective cohort study using linked Canadian Community Health Survey (2000–2017) and Discharge Abstract Database records, with environmental exposures from national datasets. The cohort included females aged 15–49 who delivered within two years of survey participation. We compared three models: (1) full model (sociodemographic, behavioural, clinical, and environmental predictors); (2) survey-only model (self-reported sociodemographic, behavioural, and clinical predictors); and (3) clinical-only model (administrative and self-reported clinical predictors). Models were fit using a Weibull accelerated failure time framework with bootstrap validation. Performance was assessed via discrimination, calibration, and predictive accuracy.
Results: Among 13,603 pregnancies, 7.9% experienced ≥1 APO. The full model showed good discrimination (C-statistic: 0.72 development; 0.71 validation) and strong calibration (ICI: 0.03), The survey-only model demonstrated comparable discrimination (C-statistic: 0.70 validation), supporting population-level surveillance, while the clinical-only model performed less well (C-statistic: 0.68 validation).
Conclusions: PregPoRT is the first Canadian APO risk tool integrating routinely collected population-based data. It demonstrates robust calibration and good discrimination, outperforming models limited to clinical variables alone. PregPoRT enables identification of high-risk groups and advances precision public health in maternal care.