Classification of isolated versus multiple birth defects: An automated process for population-based registries

Epidemiologic studies of birth defects often conduct separate analyses for cases that have isolated defects (e.g., spina bifida only) and cases that have multiple defects (e.g., spina bifida and a congenital heart defect). However, in some instances, cases with additional defects (e.g., spina bifida and clubfoot) may be more appropriately considered as isolated because the co-occurring defect (clubfoot) is believed to be developmentally related to the defect of interest. Determining which combinations should be considered isolated can be challenging and potentially resource intensive for registries. Thus, we developed automated classification procedures for differentiating between isolated versus multiple defects, while accounting for developmentally related defects, and applied the approach to data from the Texas Birth Defects Registry (1999-2018 deliveries). Among 235,544 nonsyndromic cases in Texas, 89% of cases were classified as having isolated defects, with proportions ranging from 25% to 92% across 43 specific defects analyzed. A large proportion of isolated cases with spina bifida (44%), lower limb reduction defects (44%), and holoprosencephaly (32%) had developmentally related defects. Overall, our findings strongly support the need to account for isolated versus multiple defects in risk factor association analyses and to account for developmentally related defects when doing so, which has implications for interpreting prior studies.