Validation of birth certificate data: A study of women in New Jersey's HealthStart Program

PURPOSE: This study assesses the accuracy of 1989-1992 birth certificate data from New Jersey for a group of high-risk women. METHODS: Birth records were linked to data an women who participated in HealthStart, a program of enriched prenatal care for pregnant women on Medicaid. Concordance was assessed for all variables common to the two data sets. RESULTS: The birth records had accurate reporting of birth-weight, demographic characteristics, and must: methods of delivery. Prenatal care use was over-reported, and alcohol, tobacco, transfer status, medical risk factors, obstetric procedures, as well as complications of labor and delivery mere underreported. CONCLUSIONS: While many variables are reported very accurately on birth certificates, other measures must be used cautiously. Analyses using birth certificate data, particularly those focusing on high-risk women, need to take the low levels of sensitivity for many risk factors into consideration. Ann Epidemiol 2001;11:186-193. (C) 2001 Elsevier Science Inc. All rights reserved.