[Validation of birth certificates based on data from a case-control study].

The information recorded on birth certificates was validated with data from a perinatal mortality case-control study, obtained from home interviews of mothers and hospital records for cases (early neonatal deaths) and controls. Sensitivity, specificity, and concordance were calculated for all variables and their estimated and real prevalence. The completeness of birth certificates was lowest for mother's parity and presence of congenital anomalies (records without information range from 23% to 31% for cases and controls). Birth certificates correctly identified low birth weight and type of delivery for cases and controls. Birth certificates showed high sensitivity and specificity to detect preterm births within cases. The number of preterm births was underestimated at 30.8% of the controls and 2.9% of the cases. Low maternal education was two times greater on birth certificates than in the mother's interview, for cases and controls. Completeness of birth certificates was higher in controls, but data quality was better in cases.