Presentation (CTI)

OBJECTIVES: Retrospective analyses to evaluate the safety of prenatal exposures may require validation of outcomes or exposures using electronic health records (EHR). When linkage to EHR records is not available the generation of synthetic EHR data may be used to augment existing data. This study’s objective was to assess the feasibility of generating synthetic EHR data by comparing infants in the MarketScan Database with and without linkage to the Veradigm Network EHR (VNEHR).
METHODS: Infants linked to a pregnancy with a live birth outcome in the MarketScan Commercial Database between January 1, 2018 and December 31, 2022 were identified. Infants were required to have health plan enrollment on date of birth (DOB) and DOB and at least one medical claim within -1 to 30 days of the pregnancy end date. Multiples and infants with a family member with a DOB within 365 days were excluded from the study. Infants with data in VNEHR were identified and patient characteristics and healthcare resource use were tabulated and compared in infants with and without VNEHR using standardized mean differences (SMD).
RESULTS: 484,707 infants linked to a live birth pregnancy were included in the study; 40,065 (8.3%) of infants linked to a pregnancy had data available in the VNEHR. The proportion of males in infants with (53.7%) and without (51.0%) VNEHR linkage was similar (SMD=0.05). Duration of follow-up was also similar in infants with (317 days, SD: 96) and without (306 days, SD: 106) EHR linkage (SMD=0.11). Most infants had at least 6 months of follow-up (83.2%) and proportion of infants with 6 months of follow-up was similar in infants with and without EHR linkage (86.3% vs. 82.9%, SMD=0.09).
CONCLUSIONS: No systematic differences in the characteristics of infants with and without VNEHR data were observed suggesting this database may be appropriate for the generation of synthetic data.