Objective: To evaluate whether the excess mortality in men with AD can be e
xplained by a gender difference in the predictors of mortality. Methods: Th
e authors studied 2,838 men and 6,385 women over 65 years of age with AD ad
mitted, between 1992 and 1995, to 1 of nearly 1,500 nursing homes in five U
.S. states (Kansas, Maine, Mississippi, New York, and South Dakota). Reside
nt level data including sociodemographic characteristics, dementia severity
, measures of physical disability, comorbidity, and other clinical variable
s were collected with the Minimum Data Set. Information on death was derive
d through linkage to Medicare enrollment files; the median follow-up was 23
months. Baseline characteristics were used to predict age at time of death
in Cox proportional hazard models. Results: Men with AD had an increased r
isk of mortality relative to women, adjusted for differences in the distrib
ution of age and race. The most important predictors of death in men were t
hose related to the disease itself. These were the severity of dementia and
the occurrence of episodes of delirium. Instead, death among women was ass
ociated with measures of disability, namely, impairment in performing the a
ctivities of daily living, presence of pressure sores, malnutrition, and co
morbidity. Conclusion: These data suggest that the underlying mechanisms fo
r AD may be different in men and women. Future studies of survival and prog
ression of AD need to examine men and women separately.