The specificity of the National Death Index and Social Security Administration death master file when information on Social Security Number is lacking.
Foster-SO; Schubauer-Berigan-MK; Waters-KM
Am J Epi 2000 Jun; 151(11)(Suppl):S43
Researchers conducting large cohort mortality studies often rely on the National Death Index (NDI) and the Social Security Administration (SSA) to obtain vital status information. Sensitivity and specificity are two performance criteria important in vital status database matching. Although the NDI has demonstrated high specificity using Social Security Number (SSN) and name as identifiers, the specificity of name and date of birth (without SSN) is not known for these national databases (Am J Epidemiol 1993;137:235-41). The present methodological study evaluated the specificity of different combinations of name and date of birth, in ascertaining deaths through the NDI and the SSA. The sensitivity of the SSA death master file (DMF) was also evaluated. The subjects for this study consist of 6,590 nuclear workers with valid SSN's from eight U.S. Department of Energy facilities, whose recent deaths were positively ascertained through the NDI. Their identifying information, without SSN, was submitted to the NDI for comparison to 15 years of records preceding their actual year of death, and to the entire SSA DMF. The sensitivity and specificity of the SSA DMF were 86.92% and 98.62%, respectively, using partial name and complete date of birth as identifiers. The specificity of full name and date of birth for the NDI was nearly 100% (99.95%). Specificity decreased when less-complete identifying information was used to ascertain deaths. The results of this research will provide guidance on the reliability of using identifiers other than SSN in ascertaining vital status from national databases for cohort studies.
Mortality-data; Mortality-rates; Mortality-surveys; Data-processing; Sociological-factors; Statistical-analysis
Conference/Symposia Proceedings; Abstract
American Journal of Epidemiology, Abstracts of the 33rd Annual Meeting of the Society for Epidemiologic Research, Seattle, Washington, June 15-17, 2000