Abstract
BACKGROUND: Electronic health records are increasingly used to conduct pregnancy-related research as pregnant women are under-represented in research. Creating a register of pregnancies by combining data from primary and secondary care will further facilitate research in pregnancy. This work describes the construction of an algorithm to create a unified pregnancy cohort in the QResearch database during the emergency phase of the COVID-19 pandemic. METHODS: National primary care records in the QResearch® database were linked to patient-level data from Hospital Episode Statistics (HES) datasets. Females aged 15-50 years with a pregnancy outcome recorded between 30 December 2020 and 30 September 2022 were included. Pregnancy (delivery/loss) episodes were identified and cohort demographics reported using a three-stage algorithm. Pregnancy start dates were derived using a combination of HES and primary care data, or individually estimated where no corresponding date could be identified. RESULTS: 266,758 women with 279,027 pregnancies are captured in the register. 232,673 pregnancies (83.4%) are deliveries (99.5% livebirths and 0.5% stillbirths) and 46,354 (16.6%) pregnancies are pregnancy losses. Pregnancy losses are highest amongst those of Caribbean (23.1%; n = 781) ethnicity and lowest in those of Pakistani ethnicity (13.9%, n = 1,579). 82.4% of pregnancies are derived from HES maternity records, 10.6% from primary care records, 3.4% from HES Admissions, and 3.6% from HES Procedures. CONCLUSION: The construction of a pregnancy register in QResearch® offers a valuable resource for future research. Its methodology can be adapted to construct new cohorts over any period, providing a comprehensive resource on pregnancy outcomes and events.