Exploring classification of birth defects severity in national hospital discharge databases compared to an active surveillance program.

Abstract

To explore the extent to which the severity of birth defects could be differentiated using severity of illness (SOI) and risk of mortality (ROM) measures available in national discharge databases. Data from the 2012-14 National Inpatient Sample (NIS) was used to identify hospitalizations with one or more major birth defects reported annually to the National Birth Defects Prevention Network using the International Classification of Diseases, Ninth Edition, Clinical Modification (ICD-9-CM) diagnosis codes. Each hospitalization also contained a 4-level SOI and 4-level ROM classification measure. For each birth defect and for each individual birth defect-related ICD-9-CM code, we calculated mean and median SOI and ROM, the proportion of hospitalizations in each level of SOI and ROM, the inpatient mortality rate, and level of agreement between various existing or derived severity proxies in the NIS and the Texas Birth Defects Registry (TBDR). Mean SOI ranged from 1.5 (cleft lip alone) to 3.7 (single ventricle), and mean ROM ranged from 1.1 (cleft lip alone) to 3.9 (anencephaly). As a group, critical congenital heart defects had the highest average number of co-occurring defects, mean SOI, and ROM, whereas orofacial and genitourinary defects had the lowest SOI and ROM. We found strong levels of agreement between TBDR severity classifications and NIS severity classifications defined using Level 3 or 4 SOI or ROM Level 3 or 4. This preliminary investigation demonstrated how severity indices of birth defects could be differentiated and compared to a severity algorithm of an existing surveillance program.