(735) Performance of diagnostic algorithms for identification of fetal malformations using perinatal claims data

Accurate identification of fetal anomalies (FAs) is essential for examining the effects of medication and environmental exposures in pregnancy. We aimed to determine the positive predictive value (PPV) of diagnostic algorithms using claims data for identifying births with common FAs.

Study Design:

Among births > 20 week’s gestation occurring between 1/1/2018 and 1/31/2025 at a single medical center, we identified potential FAs using an algorithm based on ICD10 coded diagnoses (Table 1) in either the maternal or infant record and included both inpatient and outpatient encounters. Multiple gestations were excluded. Among all records identified with potential FAs, we randomly selected up to 100 records from the 8 most common malformations for structured review. Two researchers with expertise in neonatology and obstetrics reviewed both maternal and infant medical records to determine if FAs were present. Potential FA cases with discordant reviews were reviewed by a third reviewer who made final adjudications. Using the medical records review-based decisions as reference, positive predictive values and 95% confidence intervals (CI) were calculated overall and for each malformation.

Results:

During the study period, 29,324 women experienced 33,939 singleton births including 4,426 (13.0%) infants with at least 1 potential FA based on the diagnostic algorithm. Of these infants, 1,179 (26.7%) infants had multiple potential FAs. We reviewed records from 727 infants and confirmed 560 malformations for a PPV of 77.2% for the diagnostic algorithm for the presence of any malformation (95% CI 74.1-80.1). The most common FAs were septal cardiac defects (51.9%) and severe cardiac defects (15.4%) followed by microcephaly (3.0%). Individual malformations with a PPV over 80% for the diagnostic algorithm included severe heart defects, cleft lip and palate, abdominal wall defects, and spina bifida (Table 2).

Conclusion:

Some major congenital malformations can be identified using diagnostic codes with reasonable accuracy. These estimates can inform subsequent studies of exposures or outcomes of FA.