Background: The relationship between education and Alzheimer disease (AD) or dementia has been widely examined and the evidence obtained is mixed. Several hypotheses have been proposed to explain the observed association between them.

Objective: To further understand the relationship between education and incidence of clinically diagnosed AD or dementia.

Subjects and methods: A community-based, dementia-free cohort of 1296 aged 75 years and older was followed up to detect incident AD or dementia cases using Diagnostic and Statistical Manual of Mental Disorders, Revised Third Edition criteria. The vital status of all subjects who underwent the clinical examination at follow-up (n = 983) was ascertained for 5 years further. Data were analyzed with Cox proportional hazards model after adjustment for main potential confounders.

Results: Over an average (SD) of 2.8 (1.0) years of follow-up, 147 subjects were diagnosed as having dementia (109 subjects as having AD). Among those who were clinically examined at follow-up, 88 died with dementia (68 died with AD) within 5 years. Subjects with a low level of education (<8 vs > or =8 years) had a relative risk of 2.6 (95% confidence interval, 1.5-4.4) for AD and 1.7 (95% confidence interval, 1.1-2.6) for dementia. A low educational level was significantly related to all-cause mortality (relative risk, 1.3; 95% confidence interval, 1.0-1.7; P<.05), but not to the mortality of subjects with AD (relative risk, 1.1; 95% confidence interval, 0.5-2.2) or dementia (relative risk, 0.9; 95% confidence interval, 0.5-1.5).

Conclusions: A low level of education is related to an increased incidence of clinical AD or dementia, but not to the mortality of subjects with AD or dementia. These findings can be accounted for by the "cognitive reserve" hypothesis. Alternatively, the observed association between educational level and incidence of AD or dementia may partly reflect detection bias, by which subjects with a low level of education tend to be clinically diagnosed at an earlier point in time.