Congenital heart disease in pregnancy and severe maternal morbidity: A distributed data network study
Elizabeth B. Sherwin, Benjamin Martin, Xiao Xu, Anna Booman, Alyssa Howren, Kimberlee McKay, Sadaf Kazi, Sean O'Reilly, Katie S. Kolla, Khyzer Aziz, Katherine Bianco, Stephanie A. LeonardAbstract
Introduction
Pregnant people with congenital heart disease (CHD) are a growing patient population in obstetrics, yet evidence on the risk for severe maternal morbidity (SMM) has largely been limited to studies that lack specificity for CHD. We conducted this study to demonstrate the utility of distributed data networks for obstetric research and to characterize pregnant patients with CHD and their risk of SMM.
Methods
The study included female patients aged 12–55 years who delivered between 2015 and 2026 at three US academic healthcare organizations. We used a distributed data network study design by developing analytical code, executing the code locally at each organization using its electronic health record (EHR) data in a common data model (CDM), then aggregating the results. This analytical approach is rigorous, is reproducible, and protects patient confidentiality by not sharing patient‐level data. The Observational Medical Outcomes Partnership (OMOP) CDM was used by each organization, and we used OMOP standard concepts to identify CHD and SMM. We categorized outcomes as SMM, non‐transfusion SMM, and cardiac SMM based on the CDC SMM index. We analyzed the incidence of the outcomes and estimated risk ratios (RRs) and Wald 95% confidence intervals (CIs) using contingency tables.
Results
The prevalence of CHD in pregnancy was 2.20% among 166,033 deliveries across three healthcare organizations. A total of 19.30% of people with CHD had an acquired heart condition, compared to 2.14% of people without CHD. SMM occurred in 9.43% of deliveries to people with CHD compared to 4.53% among people without CHD (RR, 2.08; 95% CI, 1.88, 2.31), and non‐transfusion SMM occurred in 6.72% vs 2.55% of deliveries (RR, 2.63; 95% CI, 2.33, 2.98), respectively. Cardiac SMM represented 51.84% of non‐transfusion SMM among people with CHD compared to 16.18% among people without CHD.
Conclusions
Our distributed data network study using EHR data from three geographically diverse US healthcare organizations found that one in 15 people with CHD had non‐transfusion SMM compared with one in 39 people without CHD. This study demonstrates the feasibility and utility of distributed data networks for obstetric research and provides novel evidence on maternal outcomes for pregnant patients with CHD.