Results.

<div><p>Objectives</p><p>To determine how pre-existing conditions contribute to racial disparities in adverse maternal outcomes and incorporate these conditions into models to improve risk prediction for racial minority subgroups.</p><p>Study design</p><p>We used data from the “Nulliparous Pregnancy Outcomes Study: Monitoring Mothers-to-be (nuMoM2b)" observational cohort study. We defined multimorbidity as the co-occurrence of two or more pre-pregnancy conditions. The primary outcomes of interest were severe preeclampsia, postpartum readmission, and blood transfusion during pregnancy or up to 14 days postpartum. We used weighted Poisson regression with robust variance to estimate adjusted risk ratios and 95% confidence intervals, and we used mediation analysis to evaluate the contribution of the combined effects of pre-pregnancy conditions to racial/ethnic disparities. We also evaluated the predictive performance of our regression models by racial subgroup using the area under the receiver operating characteristic curve (AUC) metric.</p><p>Results</p><p>In the nuMoM2b cohort (n = 8729), accounting for pre-existing conditions attenuated the association between non-Hispanic Black race/ethnicity and risk of severe preeclampsia. Cardiovascular and kidney conditions were associated with risk for severe preeclampsia among all women (aRR, 1.77; CI, 1.61–1.96, and aRR, 1.27; CI, 1.03–1.56 respectively). The mediation analysis results were not statistically significant; however, cardiovascular conditions explained 36.6% of the association between non-Hispanic Black race/ethnicity and severe preeclampsia (p = 0.07). The addition of pre-pregnancy conditions increased model performance for the prediction of severe preeclampsia.</p><p>Conclusions</p><p>Pre-existing conditions may explain some of the association between non-Hispanic Black race/ethnicity and severe preeclampsia. Specific pre-pregnancy conditions were associated with adverse maternal outcomes and the incorporation of comorbidities improved the performance of most risk prediction models.</p></div>