The aim of the study was to investigate the course of objective memory impairment in non-demented subjects who attended a memory clinic and to test predictors of outcome. Non-demented subjects (N=74) were included when they were older than 40 years and had a baseline score on the delayed recall of a word learning test below the tenth percentile. Subjects with memory impairment due to known somatic or neurological causes were excluded. The subjects were reassessed after 2 and 5 years. At the 5-year follow-up, 42% of the subjects had no memory impairment, 19% of the subjects had memory impairment without dementia, and 39% of the subjects had Alzheimer type dementia (AD). Predictors at baseline of reversible memory impairment in a multivariate analysis were age, scores on the MMSE and delayed recall, and the degree of functional impairment. Predictors at baseline of AD in a multivariate analysis were age and the score on the MMSE. The apolipoprotein E genotype and the presence of depression at baseline were not predictors of outcome. The positive predictive value was 72% for reversible memory impairment and 81% for AD. Memory impairment is often reversible and therefore its presence alone is not sufficient to consider subjects as preclinically demented. Predictive accuracy can be increased by including simple measures such as age, the scores on the MMSE and delayed recall, and the degree of functional impairment.
Copyright 2000 John Wiley & Sons, Ltd.