Background: Pregnancy cohort studies in administrative claims data frequently require continuous enrollment before and after delivery to assess pregnancy characteristics before delivery and outcomes post-delivery. The impact of enrollment restrictions on pregnancy cohort characteristics that may affect internal or external validity has not been well described.
Objectives: To compare characteristics among pregnancy cohorts with or without minimum enrollment period requirement.
Methods: A retrospective cohort study was conducted in the Optum Clinformatics Data Mart claims data among persons with at least one pregnancy episode with an end-of-pregnancy date between 01 Jan 2017 and 31 December 2022. A published pregnancy algorithm was used to identify pregnancies and to link maternal to infant records (Ailes E.C. et al. 2023). We defined cohorts based on minimum enrollment period requirements: none (least restrictive), 3, 6, and 12 months (most restrictive) pre and post end-of-pregnancy date and summarized pregnancy cohort characteristics.
Results: Among 791,551persons with an end-of-pregnancy date between 01 Jan 2017 and 31 December 2022, we identified a total of 934,357 pregnancy episodes including 746,647 live births, 147,544 spontaneous abortions, 28,346 elective terminations, 5,660 ectopic pregnancies, 5,625 stillbirths, 434 multiple pregnancy episodes with at least one live birth and stillbirth, and 101 pregnancy episodes with unclassified pregnancy outcome. Requiring a minimum enrollment of 3, 6, or 12 months pre and post end-of-pregnancy date, 79.2%, 62.4%, and 41.2% of pregnancy episodes were retained in the pregnancy cohorts, respectively, compared to no minimum enrollment requirement. Comparing the most restrictive cohort with the least restrictive cohort, 41.6% of live births, 37.2% of stillbirths and 37.9% of elective terminations are retained. Further, pregnancies among females aged 20-24 years (37.0%) and 25-29 years (31.7%) are less likely to be retained in the most restrictive cohort compared to all other age groups (range 42.0 to 63.4% retained). Pregnancies with infant match are more likely to be retained (42.7%) compared to pregnancies without infant match (38.1%).
Conclusions: Pregnancy study designs that require extensive minimum enrollment time will have considerably reduced sample size and may suffer from selection bias related to type of pregnancy outcome or maternal age. Analyses of administrative claims databases should assess impact of enrollment requirements on pregnancy sample characteristics.
