Quality assessment of fetal death records in Georgia: a method for improvement

Objectives. Although more fetal deaths than neonatal deaths occur, routinely collected fetal death data are seldom used for perinatal epidemiologic research because of data quality concerns. We developed a strategy for identifying and correcting errors in birthweight and gestational age in fetal death records. Methods. Using data from Georgia for 1989 and 1990, we detected singleton fetal death records having improbable or missing birthweight or gestational age by comparing these values with referent values. To verify the questionable values, we contacted 100 reporting hospitals in 1992. Results. In 817 of 2226 records, values were either improbable (60.1%) or missing (39.9%). We were able to contact the hospitals to verify data for 716 (88%) of these records. Verification resulted in corrections to 405 (57%) records, and 48% of unreported birthweights were obtained. Conclusions. Many errors in recorded gestational age and birthweight were identified by this method. Rather than deleting or imputing problem data for analyses, researchers should consider efforts to verify them. Efforts to improve this information should include improved reporting, strict quality assurance, and procedures for routine verification and correction of records. (Am J Public Health. 1997:87:1323-1327)