Objective: To determine rates of decline in Alzheimer's disease. Desig
n: A longitudinal review of patients diagnosed as having dementia duri
ng life, tested serially with the Extended Scale for Dementia, acid co
nfirmed by autopsy as having Alzheimer's disease. Subjects end Setting
: Twenty-nine dead patients with Alzheimer's disease from the particip
ants in the University of Western Ontario Dementia Study Project, conf
irmed at autopsy as having Alzheimer's disease. Methods: Analysis of t
he Extended Scale for Dementia data according to a trilinear model. Fi
ndings: In the middle phase of the trilinear model, there was a mean a
nnual change of 13% (range, 2.5% to 51.7%). Conclusions: It is likely
that the common method of averaging a group of different individual sc
ores from the initial and middle phases of observation of Alzheimer's
disease collapses together individuals at different stages of the diso
rder, some of whom are in the initial plateau phase and whose conditio
ns are not declining rapidly. The trilinear model of decline avoids th
is difficulty and the present study provides postmortem confirmed figu
res on rate of change.