Objectives, To estimate the prevalence and correlates of dementia at d
eath and to assess the usefulness of death certificate data in the rep
orting of dementia. Methods, The authors analyzed next-of-kin intervie
ws for 599 male and 628 female decedents using data from the National
Institute on Aging's Survey of the Last Days of Life. Results. Death c
ertificate data in this population show the prevalence of dementia to
be less than 1%, consistent with previous reports based on death certi
ficates but a substantial underestimate compared to the 11.9% reported
in a national survey. Using a dementia index based on the informant's
report of whether the decedent had been diagnosed with a dementing il
lness and the extent of her or his cognitive and functional limitation
s, this study found a prevalence of dementia of 8.5%. A high score on
the dementia index was significantly associated with older age, Parkin
son's disease, and incontinence. Lower relative odds for dementia at d
eath were found for people with either a lifetime history or a death c
ertificate report of cancer. Similarly, people with a lifetime history
of coronary heart disease were found to have lower relative odds for
dementia at death. Conclusion, These results suggest that informant in
terviews may be a useful source of data to examine factors associated
with dementia and to estimate the prevalence of dementi; in the last y
ear of life.