Retrospective Validation Of a Patient-Initiated Preconception Screener Against Obstetric Comorbidity Indices To Assess Pregnancy Complications

ObjectiveTo retrospectively validate an electronic health record (EHR) implementation of the patient-initiated PreMA screener and compare its association with severe maternal morbidity (SMM) outcomes against established obstetric comorbidity indices. MethodsWe conducted a retrospective observational study using UCSF (single center) and UC-wide (multi-center) de-identified EHR data, identifying live-birth deliveries with documented preconception data. PreMA and established comorbidity index (Bateman and Leonard) scores were computed from preconception diagnoses, standardized to z-scores, and modeled as continuous predictors of SMM and non-transfusion SMM (NT-SMM) using logistic and Poisson regression models, with stratified analyses by race, ethnicity, and neighborhood deprivation. To examine the relationship between individual PreMA questionnaire domains and outcomes, we used adjusted Poisson regression to estimate the association of each domain with SMM and NT-SMM. ResultsAcross both cohorts, higher standardized PreMA, Bateman, and Leonard scores were consistently significantly associated with increased risk of SMM and NT-SMM, with relative risk estimates generally in the [~]1.2-1.4 range per standard deviation (adj. p < 0.001), and similar magnitude across indices and cohorts. Significant associations persisted across racial, ethnic, and socioeconomic, and item-level analyses suggested heterogeneity across PreMA domains, with cardiovascular domains showing the strongest adjusted associations. ConclusionAn EHR-derived PreMA score demonstrated robust, generalizable associations with severe maternal morbidity outcomes comparable to established clinician-facing indices, supporting PreMAs validity as a scalable, patient-centered preconception risk assessment tool.