Project Summary Addressing the rising trends in maternal mortality and severe maternal morbidity (SMM) is a critical priority in the United States. About half of adverse maternal health outcomes were found to be attributable to preventable harm or unintended consequences arising from clinical practice and the system of delivering perinatal care. Significant resources are currently being invested to implement quality improvement (QI) initiatives in birthing hospitals across the country. There is great need to evaluate these efforts and demonstrate their effectiveness to reducing the burden of preventable SMM and maternal deaths. Virtually all QI initiatives in birthing hospitals use SMM as an outcome measure, but their evaluation is hindered by the need to risk-adjust SMM rates to control for differences in patient composition within and between hospitals. To date, 3 different research groups proposed obstetric comorbidity indices, yet all have significant limitations. The overarching goal of this study is to develop and validate a refined comorbidity index for obstetric patients that allows SMM rate comparisons across hospitals and adequate monitoring of QI initiatives in obstetrics. We will use Maryland’s unique, gold- standard, hospital-based, state-representative SMM Surveillance and Review data to identify a comprehensive list of comorbidities in patients with SMM events. Using electronic health record data from the Johns Hopkins Health System, we will employ variable importance estimation with machine learning techniques to develop the comorbidity index. Subsequently, we will ascertain its accuracy using receiver operating characteristic (ROC)/precision-recall (PR) curves and areas under the curve (AUC) for outcome discrimination and lowess- smoothed calibration plots. Also, we will compare the performance of the refined comorbidity index to predict SMM against that of previously published comorbidity indices. To further validate our refined comorbidity index and assesses its performance consistency across various sociodemographic groups, we will use national hospital discharge data from the Healthcare Cost and Utilization Project’s National Inpatient Sample. A Technical Advisory Group comprised of clinicians, community partners, patient safety experts, and certified medical coders will meet quarterly for data interpretation sessions. At the end of the study, we expect to have a refined comorbidity index developed in gold-standard data, with superior psychometric properties than the previously published comorbidity indices and validated in both EHR and national hospital discharge data. Our results will be disseminated in the peer-reviewed literature and through presentations at scientific meetings.