Development and validation of an algorithm to better identify pregnant women with inherited bleeding disorders within electronic health records.

Background: When using electronic health records (EHRs) to conduct population-based studies on inherited bleeding disorders (IBDs), using diagnosis codes alone results in a high number of false positive identifications.

Objective: The objective of this study was to develop and validate an algorithm that uses multiple data elements of EHRs to identify pregnant women with IBDs.

Methods: The population included pregnant women who had at least one live birth or fetal death (>20 weeks gestation) at our institution from 2016 to 2023. We iteratively developed the algorithm using a composite criteria of encounter diagnosis codes, laboratory and medications data. We assessed the performance of the algorithm for sensitivity and positive predictive value (PPV) using our local registry and manual chart review.

Results: Using the source population between 2016 and 2020, the initial algorithm identified 25 pregnant women with IBDs. Eight women with a known diagnosis of an IBD were missed resulting in a sensitivity of 75.8 % and a PPV of 100 %. We revised the algorithm to remove certain IBD diagnosis codes that resulted in contamination and added additional criteria to improve the sensitivity. The revised algorithm had a sensitivity of 97.0 % and a PPV of 91.4 %. The revised algorithm was validated using the source population between 2021 and 2023 and had a sensitivity of 97.1 % and a PPV of 91.7 %.

Conclusion: This study demonstrates the utility of an algorithm to better identify pregnant women with specific types of IBD, mainly hemophilia and hemophilia carriers, and von Willebrand disease, within EHRs.