Related Experiment Videos
Experiences with linked birth and infant death certificates from the NIMS project
Abstract:
The National Infant Mortality Surveillance (NIMS) project aggregated data provided by 53 vital statistics reporting areas--50 States, New York City, the District of Columbia, and Puerto Rico (subsequently called States)--from their files of linked birth and death certificates and compared individual States' total infant mortality experiences for the 1980 birth cohort by age at death, race, birth weight, and plurality. Therefore, it was essential to achieve maximum uniformity among the separate data sets and to specify when this uniformity could not be obtained. In working with these multiple sources, we identified five key issues that relate to data from linked birth and death certificates: Variations in definitions of variables are often embedded in data that have been gathered from several independent sources. (For NIMS, the sources were 53 reporting areas and the National Center for Health Statistics.) Variations in States' linking procedures--these are based on an individual State's primary purpose for linking the data--affect the completeness and comparability of the 1980 resident birth cohorts used for NIMS. Variations in the recording of some pregnancy outcomes as fetal deaths or live births are known to be a problem in vital statistics data that particularly affects data for events among infants weighing less than 500 g at birth. Ambiguities occur frequently in unknowns or zero values. For NIMS this effect was most pronounced for the pregnancy history variables. Examination of the values reported for unknown or zero categories helps in uncovering problems with and improving quality of data. (e) Analysis from a new perspective may reveal unexpected data problems. These problems tend to surface only during a reexamination of underlying data that is prompted by unusual findings.Continued alertness to these issues may improve further the quality of data in files of linked birth and death certificates and assure the integrity of analysis based on these data.