Senior Researcher Nuffield Department of Orthopaedics, Rheumatology and Musculoskeletal Sciences, University of Oxford, United Kingdom
Background: Many healthcare databases have detailed information related to the perinatal period. However, the current Observational Medical Outcomes Partnership common data model (OMOP CDM) lacks structure to easily define a pregnancy episode and its related clinical events, drug use, observations, or measurements.
Objectives: To develop and test the implementation of a perinatal extension for the OMOP CDM with the aim to facilitate future perinatal network studies.
Methods: We developed a perinatal extension for the OMOP CDM that includes a pregnancy and a fetus tables based on expert opinion with several rounds of review. Each table includes a set of required and optional variables corresponding to each pregnancy episode. The pregnancy table includes person and pregnancy id, pregnancy start and end dates, gestational age, mode of delivery, singleton pregnancy, and pregnancy outcomes as required fields. Optional elements include pregnancy-related data (e.g., number of fetuses, pre-pregnancy BMI), and data on previous pregnancies (e.g., parity, previous miscarriages). The fetus table includes required variables to link each fetus to a pregnancy (i.e., fetus and pregnancy id) and optional fields related to birth outcomes (e.g., birth weight, APGAR score). The extension allows interaction to existing tables including drug exposure and condition tables. We tested the developed extension tables in two databases with different source data: primary care electronic health records (SIDIAP; Spain) and nationwide health registries (UiO; Norway). We developed an R package to assess the feasibility and quality of its implementation and characterised the pregnancies identified.
Results: Feasibility and quality assessment of the developed extension tables was successful for both databases with complete data in all required variables. SIDIAP identified 672,058 pregnancies corresponding to 440,031 women during 2006-2022 and UiO 804,227 pregnancies from 491,944 women (2007-2021). Of all pregnancies, 0.4% ended in stillbirths (>20 weeks) in both databases. C-section was performed in 32% (SIDIAP) and 7% (UiO) of pregnancies. Women had a mean pre-pregnancy BMI of 24.8 (SD=4.9) in SIDIAP and 24.5 (SD=4.9) in UiO. Additionally, UiO identified 817,312 fetuses, with a median birth weight 3,520g (IQR=685) and APGAR score 10 (IQR=1). Data captured with the extension was consistent with original source data.
Conclusions: We developed a perinatal extension that will help standardise perinatal data with a high level of granularity in the OMOP CDM and will facilitate the conduction of network perinatal studies. These tables were successfully implemented in two large databases but need further testing and validation on a wider range of data types and populations.
