OBJECTIVES: Regulators are increasingly requiring retrospective studies of safety of prenatal exposures to medication. Though claims-based algorithms identifying major congenital malformations (MCM) are available, these were validated in Medicaid patients and their performance in commercially-insured populations is unknown. This study’s objective was to validate claims-based algorithms to identify MCM in a commercially-insured population.

METHODS: Infants (< 1 year) with an inpatient claim indicating livebirth and evidence of an MCM between 1/1/2016 and 12/31/2020 in the MarketScan Claims-EMR Database were included in this study. Infants were required to have at least six EMR encounters in the first year of life. Those with chromosomal, genetic, or other congenital anomalies with known etiology (e.g., congenital syphilis) were excluded. Evidence of MCM in the claims and EMR was flagged overall and for 11 categories of MCM. Percent positive agreement (PPA) was calculated as the percent of patients with an MCM in claims with evidence of the same MCM in EMR. Clinical characteristics and healthcare resource use (HCRU) were compared in infants with and without validated MCM.

RESULTS: In total 683 infants with MCM were included in this study; 42.9% were female with mean age of 16 days (SD: 29) at MCM diagnosis. Overall PPA between claims and EMR for any MCM was 80.4%; with highest PPA found in infants with oro-facial clefts (100%), eye anomalies (90%), and congenital heart defects (86.9%). Lowest PPA were found in infants with abdominal wall defects (56.3%) and digestive system malformations (62.7%). PPA for all other MCM types exceeded 70%. Clinical characteristics and HCRU in infants with and without validated MCM were similar though more infants with validated MCM had a physician specialist inpatient admission.

CONCLUSIONS: The PPA between the claims and EMR databases was high, indicating claims-based algorithms identify MCM in infants with a high degree of specificity.