OBJECTIVES: Drug safety data in pregnancy prior to marketing approval and authorization is limited. Real-world data sources can support robust risk-benefit evaluation, as contraindicated drugs may be prescribed to unaware pregnant women or when no alternative exist for treating maternal conditions. Robust claims-based pregnancy safety analyses rely on valid code algorithms. We aimed to apply published algorithms in a large representative closed-claims database in the US.

METHODS: We applied an outcomes-based algorithm using diagnosis and procedure codes to identify pregnancy outcomes: live, full term, preterm and still birth (LB, FTB, PTB and SB), spontaneous and elective abortion (SA or AB), and ectopic pregnancy (ETC), between January 2022-January 2023, among women aged 15-49. Estimated date of conception and gestational age (GA) at outcome was assigned using a hierarchical code algorithm.

RESULTS: The outcomes algorithm assigned a pregnancy outcome to 842,583 (100%) pregnancy episodes: LB (62%) [of which 34% FTB, 4% PTB, 24% unassigned LB], SB (1%), SA (17%), AB (12%), ECT (3%) and unknown (5%).

CONCLUSIONS: Estimating date of conception and gestational age at outcome with code algorithms are crucial for defining pregnancy episode trimesters as exposure windows in claims data for inferential safety analyses. Linkage with electronic health record (EHR) data to validate a proportion of assigned outcomes and GA at outcome is needed.