Utility of the National Death Index for Ascertainment of Mortality among Cancer Prevention Study II Participants

This paper reports the results of testing and using the National Death Index to ascertain vital status in the American Cancer Society's prospective cohort study, Cancer Prevention Study II. This cohort of over one million men and women, enrolled by volunteers in 1982, is one of the largest ever to be linked with the National Death Index. In a linkage of over 15,000 persons whose vital status through 1988 had been traced through manual follow-up, 93% of all known deaths were ascertained. Sensitivity varied by demographic factors (race, sex) and availability of identifying information (Social Security number, middle initial). When the Social Security number was available, 97% of known deaths were accurately identified. A computerized matching algorithm was used to minimize manual review of records. With this method, the authors were able efficiently and accurately to reject the many false-positive matches while maintaining a reasonable level of manual record review and death certificate acquisition, thus allowing for continued automated follow-up of this valuable cohort.