Identifying deaths before 1979 using the social security administration death master file

For cohort studies, the Social Security Administration (SSA) traditionally has been the principal source of deaths that occurred before 1979. In 1988, the SSA abolished a system that provided a relatively complete accounting of deaths and replaced it with the Death Master File. We examined the completeness of the SSA Death Master File by comparing it with the U.S. Vital Statistics records and by searching the SSA Death Master File for known decedents from seven cohorts. Overall, only 53% of reported U.S. deaths and 75% of known deaths in our seven cohorts were included in the SSA Death Master File. Ascertainment was better after 1975 (89 - 95%). A re-analysis of two cohorts that excluded deaths before 1979 not found in the SSA Death Master File resulted in 20-35% decreases in both standardized mortality ratios and dose-response trends. Although the SSA system before 1988 provided relatively complete vital status information the SSA Death Master File is inadequate for vital status determination. New cohorts with a substantial number of deaths before the inception of the National Death Index in 1979 will be most seriously affected.