Lc. Schall et al., Utilizing multiple vital status tracing services optimizes mortality follow-up in large cohort studies, ANN EPIDEMI, 11(5), 2001, pp. 292-296
PURPOSE: To compare the three national-scale death identification services
used in our two stage vital status tracing protocol, Pension Benefit Inform
ation Company (PBI), Social Security Administration (SSA), and the Health C
are Financing Administration (HCFA), with respect to death identification a
nd confirmation rate, and relevant demographic variables.
METHODS: information on 31,223 subjects with unconfirmed vital status in an
ongoing occupational cohort mortality study was simultaneously submitted t
o PBI, SSA, and HCFA to identify subjects deceased as of December 31., 1992
. Subjects whose dates of death were between 1979 and 1992 were then sent t
o the National Death Index (NDI) to obtain death certificate numbers and su
pplemental states of death.
RESULTS: PBI identified and confirmed the highest number deaths in this coh
ort. PBI and SSA identified a higher proportion of deaths for persons who d
ied in earlier years and/or who died at a younger age, for both confirmed a
nd unconfirmed deaths. HCFA identified fewer deaths overall and had a small
er proportion of unconfirmed deaths. These deaths occurred in later years a
mong older subjects and had the highest proportion of females. NDI provided
exact matches for 92-96% of deaths identified by each of the three service
s.
CONCLUSIONS: PBI was the most comprehensive service, especially for identif
ying younger subjects and those with an earlier date of death, while HCFA m
ay help to identify deceased female subjects. SSA data can be purchased and
used for periodic updates or interactively to identify deaths among subjec
ts with poor identifiers (such as incorrect or missing social security numb
ers or misspelled names). Because each service makes a valuable contributio
n to the identification of deceased cohort subjects, all three should be co
nsidered for optimal mortality follow-up.,Ann Epidemiol 2001;11:292-296. (C
) 2001 Elsevier Science Inc. All rights reserved.