A novel maternal prenatal risk index to predict mortality-weighted severe maternal morbidity at hospitalization: a retrospective cohort study.
The aim of this study is to develop and validate a maternal Prenatal Risk Index (m-PRI) comprised of preexisting maternal conditions and patient characteristics associated with increasing severity levels of severe maternal morbidity (SMM).Using administrative data from the Premier Healthcare Database, this cohort study analyzed inpatient deliveries between 2016 and 2023 across 864 hospitals and 49 U.S. states to develop the m-PRI. Multivariable ordinal logistic regression was used to quantify the association between 46 candidate conditions and an ordered set of SMM severity levels derived from an inpatient mortality-weighted SMM composite index. The final m-PRI is a composite risk score, calculated by summing the weights of existing patient conditions. The predictive performance of the m-PRI was evaluated against the Obstetric Comorbidity Scoring System (OCSS) using precision-recall area under the curve (PR-AUC) metrics overall and across subgroups.Across 7,174,412 maternal delivery hospital encounters, 56,612 patients experienced at least one SMM indicator. The final m-PRI incorporated 28 conditions. On the validation dataset, the m-PRI achieved a PR-AUC of 0.223 (95% CI: 0.194-0.254), modestly higher than the OCSS (0.173; 95% CI: 0.147-0.201), with a significant mean difference of 0.050 (95% CI: 0.036-0.064, p < 0.001). Subpopulation analysis demonstrated that PR-AUC values were consistently higher for the m-PRI than OCSS across all evaluated subgroups.The m-PRI improves upon existing methods by weighting maternal risk factors based on their association with mortality-informed SMM severity levels on an ordinal scale, rather than a dichotomous composite indicator. By incorporating a more expansive set of relevant maternal risk conditions, the m-PRI detects previously unmeasured risk, both overall and across clinical subgroups.The U.S. Department of Health and Human Services Office on Women's Health supported this project under contract 75P00120C00066.
