The aim of this study was to construct a prognostic model to predict the pr
ogression of Alzheimer's disease (AD). Prevalent and incident cases with AD
came from the Rotterdam Study, a population-based prospective cohort study
of persons aged 55 years and older, including those living in institutions
. Rate of cognitive decline, as measured by the Mini Mental State Examinati
on (MMSE score), was predicted by a random effects model. Risk of instituti
onalization and death were estimated with polytomous logistic regression an
alysis. At baseline, 306 subjects were diagnosed with prevalent AD and had
complete data on living conditions and cognitive function, After a mean fol
low-up of 2.1 years, 95 subjects with incident AD had been diagnosed. Preva
lent patients showed a slower decline in cognitive function than incident p
atients (p = 0.004). For prevalent and incident AD patients, high age and l
ow cognitive performance were the strongest predictors for institutionaliza
tion and death. These prognostic risk functions can provide information on
the decline of Alzheimer patients and might be used to better evaluate the
effect of treatments for AD. Copyright (C) 2001 S. Karger AG, Basel.