(P-072) Deterministic Linkage of Pregnant Mothers and Their Newborns in Integrated US Insurance Claims and Electronic Health Record Data

Background: Studying the effects of maternal pregnancy related health events and exposures on newborns using insurance claims and Electronic Health Records (EHR) presents particular challenges because the newborn’s in utero exposure history is not available in the commonly used insurance claims and EHR databases, which start following patients at enrollment or birth.

Objectives: We developed a deterministic algorithm to address the absence of in utero exposure data by linking health records of singleton newborns to those of their mothers in Optum’s integrated insurance claims and EHR data.

Methods: Pregnant women were identified at delivery using data from January 1, 2015 to January 1, 2020. In the claims data, pregnant women with a record of a live singleton delivery were linked to newborns with an age of 0 years who shared the same family identifier as the woman. Live-born singleton newborns whose first records in the claims data occurred within 30 days of the birth were included. Calendar dates of the births were not retained, and only the timing of the birth relative to previous events for the pregnant mother and subsequent events for the neonate were used. After the linkage was created in the claims data, information from the integrated EHR was available for analysis.

Results: Initially, 79,045 women with live singleton birth were identified among 17,105,763 male and female patients over the study period. Of the women with live singleton births, 89.6% (n = 70,846) were linked to a newborn in the claims data. The maternal pregnancy related health events and exposures from both the mother’s EHR and claims data were then applied to the neonate.

Conclusions: EHR and claims data from nearly 9 out of 10 pregnant women were linked to records from their live-born singletons using a deterministic matching algorithm. With these linked data, the gestational age of the child at birth can be inferred and the mother’s pregnancy-related health events and exposures can inform the newborn’s in utero medical history. The method was limited to singleton births because, with multiple births (e.g. twins, triplets etc.), events around the delivery could not be clearly connected to the individual newborns. The Optum EHR does not have a unique family identifier, but integrating with Optum claims allowed the EHR data from newborns and their mothers to be linked, which facilitates future analyses of the combined mother and neonate data. Further work may include a validation of the linkage algorithm with chart review and the use of probabilistic matching techniques to link the remaining unmatched births.