Development and Application of an Algorithm to Identify the Primary Underlying Condition for Cases of Severe Maternal Morbidity.

objectiveThe Centers for Disease Control and Prevention's (CDC) severe maternal morbidity (SMM) index is a list of major complications, but it does not identify underlying causes. Our objectives were to create a hierarchical algorithm to identify a primary underlying condition for each SMM case and calculate the frequencies of underlying conditions that lead to SMM in large administrative datasets.

methodsA hierarchical algorithm using International Classification of Diseases, Tenth Revision codes was developed using a combination of medical record reviews and iterative analyses of large datasets over an 8-year period, 2016-2024. To assess validity, the algorithm's assignment of primary underlying conditions for SMM cases was compared with detailed medical record abstraction. The developed algorithm was then applied to 2016-2020 California and National Inpatient Sample (NIS) hospital discharge datasets. A nonhierarchical approach, which allowed the assignment of multiple diagnosis codes to identify comorbidities, was also evaluated. Frequencies of underlying conditions among SMM cases were compared with causes of pregnancy-related mortality by using the CDC's Pregnancy Mortality Surveillance System data (2017-2019). Total SMM and nontransfusion SMM were examined for each analysis.

resultsIn the 604 SMM cases used to assess validity, the primary underlying condition that resulted in SMM based on the algorithm had 94.5% concordance with a detailed medical record review. When applied to California discharge data (2016-2020) (n=43,897) hemorrhage (combined placental and other) was the most frequent primary underlying condition that resulted in SMM (50.5%) and nontransfusion SMM (38.3%). Severe hypertensive disorders and infection were also common-together accounting for 31.2% of SMM and 44.9% of nontransfusion SMM. Other medical conditions accounted for 12.9% of SMM and 19.8% of nontransfusion SMM. Rates of cardiovascular conditions as the primary underlying conditions were 2.4% of SMM and 4.3% of nontransfusion SMM. Results were similar for NIS data (n=63,880). Causes of maternal mortality in the United States (2017-2019) substantially differed from underlying conditions that resulted in SMM: rates for hemorrhage (12.1%), hypertensive disorders (6.3%), and infection (14.3%) were lower, and rates for cardiovascular conditions (26.6%) were higher.

conclusionThe hierarchical algorithm provides a method to assign a primary underlying condition to population SMM cases by using administrative codes. Hemorrhage, hypertensive disorders, and infection dominate underlying conditions for SMM, whereas cardiovascular disease, the most common cause of maternal death, is an uncommon SMM underlying condition.