P2.A019: A Novel, Clinically Informative Classification System for Outcome Studies of Congenital Heart Defects in Birth Defect Registries

Background: Traditional congenital heart defect (CHD) classification systems in birth defect registries are useful for surveillance and etiologic studies, but are not well-suited for commonly co-occurring lesions in outcome studies. Current coding systems either: 1) evaluate cases with a given code regardless co-occurring lesions that may modify risk (defined as broad approach), or 2) assess cases with only the given code alone (defined as exclusive coding), the latter of which improves homogeneity but results in small counts. Our objective was to create a lesion-specific classification system for birth defect registry outcome studies that balances homogeneity in expected clinical consequences and maintain reasonably large group sample sizes.

Methods: For Core Cardiac Lesion Outcome Classifications (C-CLOC) groups, common CHDs expected to have clinical relevancy and homogeneity were defined by 4 clinical experts (3 pediatric cardiologists, 1 geneticist). Criteria based on Centers for Disease and Control-modified British Pediatric Association (BPA) CHD codes were created to define individual C-CLOC groups, while accounting for common associations and differential risk. Where appropriate, we further divided CHDs into subgroups representing distinct CHD combinations (e.g., double outlet right ventricle with and without malposition of the great arteries). At each level, codes were mutually exclusive. To demonstrate proof of concept and illustrate retention of reasonable case counts within C-CLOC groups, Texas Birth Defect Registry data (1999-2017 births) were used to compare case counts and neonatal mortality between traditional classifications versus C-CLOC.

Results: C-CLOC defined 59 groups. Among 62,262 children with CHDs in Texas, percent retained in a C-CLOC group of those in traditional “broad” CHD classification groups (counted if any CHD code present) ranged from 97.2% (double inlet left ventricle) to 9.8% (common atrium), with the majority of those not retained moved to a C-CLOC group of higher complexity. For most lesions, including hypoplastic left heart syndrome, mortality by C-CLOC group was similar to mortality using a traditional “exclusive” approach (counted if only 1 CHD code present), but C-CLOC case counts were higher (Figure 1). Further, C-CLOC enabled the study of clinically relevant CHD groups beyond singular BPA codes, which matters for differing surgical plans of care and expected mortality (Table 1). For example, among 1,079 cases with pulmonary atresia (PA, all with BPA code 746.000), PA could be distinguished among PA with intact ventricular septum (N=364, 16% mortality), PA with ventricular septal defect (N=574, 10% mortality), and complex PA (N=141, 16% mortality).

Conclusion: Our new birth defects registry-based CHD classification system, C-CLOC, increases clinical outcome homogeneity while maintaining sufficiently large case counts within categories. C-CLOC has the potential for increased clinical utility when using large-scale database outcome studies among people with CHD. Future validation using International Classification of Diseases codes and outcome studies utilizing C-CLOC-based classifications are planned.