1F-04 - Prevalence of Major Congenital Malformations Identified Using Diagnosis Codes Among Live-Born Infants in a US Commercial Claims Database

Background: Due to the lack of a validated algorithm, many US claims-based pregnancy safety studies use a two-step method to define major congenital malformations (MCMs), i.e., to first identify potential MCMs via diagnosis codes and then confirm cases by medical chart review per the Metropolitan Atlanta Congenital Defects Program (MACDP) classification. In such studies, ensuring a highly sensitive code list to identify MCMs is essential for the study validity and efficiency, however there is no consensus on which specific codes should be included.

Objectives: Describe the prevalence of MCMs in live-born infants in claims data, using three ICD-10-CM diagnosis code lists.

Methods: Optum’s de-identified Clinformatics® database was used to identify infants born from January 2017-June 2023. Infants were followed up until 1 year from the start of initial enrollment. Three diagnosis code lists were used to define MCMs: all codes that classify congenital defects except those considered minor by the New York State Department of Health Birth Defects Registry (NYS BDR; M1) or by the EUROCAT (M2), and codes compiled from published FDA Sentinel analyses in 2023-2024 (M3). Moreover, two algorithms were evaluated when using each code list: ≥1 claim regardless of setting (A1), and ≥ 1 inpatient or ≥ 2 outpatient claims (A2). The presence of MCMs defined using each criterion was described.

Results: A total of 912,118 infants were identified. With A1, the prevalence of MCMs was highest using M1 (15.76%), followed by M2 (13.17%) and M3 (12.21%). The prevalence decreased when A2 was used (M1, 11.18%; M2, 9.67%; M3, 8.95%). The total number of identified MCMs ranged from 228,040 (M2) to 274,162 (M1) using A1, and from 168,958 (M2) to 195,120 (M1) using A2. With either algorithm, the top diagnosis among all the identified MCMs was the same for each method – atrial septal defect (M1A1, 11.32%; M1A2, 13.01%; M2A1, 13.61%; M2A2, 15.03%; M3A1, 13.41%; M3A2, 14.74%). For the second leading MCM, M1 and M2 identified other specified congenital malformations of skin (M1A1, 7.57%; M1A2, 8.67%; M2A1, 9.10%; M2A2, 10.02%). However, this was not considered an MCM by M3.

Conclusions: Given the MACDP’s estimate of MCMs at 3% in births, all the diagnosis code lists lead to substantially elevated prevalence estimates in claims. The NYS BDR and the EUROCAT methods were largely consistent when identifying the MCMs, while the FDA Sentinel analyses tended to exclude more malformation codes than the others. Further research on the consistency between the code lists and the MACDP’s MCM definition and the validity of these methods against the confirmed cases are warranted. This will inform an efficient way to assess MCMs in claims-based studies augmented by medical chart validation.