Development and Validation of a Risk Prediction Model for 5-year Risk of Hypertension in Women of Reproductive Age

Abstract

Introduction: The rate of pregnancy complications has been steadily increasing over the past decade. These complications include hypertensive disorders during pregnancy (HDP), gestational diabetes, and preterm birth. Previous studies have shown an association between pregnancy complications and the development of cardiovascular disease (CVD). Hypertension has been established as a precursor to CVD, implying that predicting the incidence of hypertension could help reduce the overall prevalence of CVD. The objective of this study is to provide clinicians with a validated tool to identify women in the post-partum period with increased risk of hypertension. Methods: A risk prediction model was developed to estimate the 5-year risk of incident hypertension among nulliparous women, accounting for obstetrical history and complications of pregnancy. Variables with low prevalence (<1%) were excluded and lasso-regularized Cox regression was used to further exclude variables with negligible effects on the predictions. A baseline analysis was conducted using an extended Cox model. A proportional hazards assumption violation for HDP was dealt with by allowing the associated risk to vary over time. Harrell's concordance (C)-index was used to measure discrimination ability and internal validation was performed using bootstrap resampling (n=500 replicates). Results: Eleven variables were included in the final model, with an HDP diagnosis having the largest estimated effect on the risk of hypertension. The predicted 5-year probability of hypertension diagnosis for a subject with median attributes was 7% (95% CI: 6.5%, 7.5%). Discussion: HDP alongside other risk factors can provide insight into identifying women in the post-partum period who would benefit from the early initiation of cardiovascular prevention strategies and treatment.