RESUMO
BACKGROUND: The 2003 revision of the U.S. Birth Certificate was restricted to birth defects readily identifiable at birth. Despite being the lone source of birth defects cases in some studies, we lack population-based information on the quality of birth defects data from the most recent revision of the birth certificate. METHODS: We linked birth certificate data to confirmed cases from the Florida Birth Defects Registry (FBDR) to assess the sensitivity and positive predictive value (PPV) of birth defects indicators on the birth certificate. Descriptive statistics and log-binomial regression were used to examine variation in data quality measures by defect type and other characteristics. We also evaluated the contribution of birth certificates as a case ascertainment source for the FBDR. RESULTS: Sensitivity of the birth certificate was poor (19.1%) with variation across defects ranging from 55% for anencephaly and 54% for gastroschisis, to <10% for other defects. PPV was better (87.1%) and ranged from >93% for orofacial clefts and gastroschisis to <55% for anencephaly and limb reduction defects. We also observed variation in data quality across maternal, infant, and hospital characteristics. Of cases identified by the birth certificate and not any other FBDR data source, 54.9% were false-positive diagnoses. CONCLUSIONS: Efforts to restrict the 2003 revision of the birth certificate to defects identifiable at birth have not improved the likelihood that birth certificates will identify infants born with those defects. We do not recommend the use of birth certificates as a source of birth defects data without case verification strategies.