Abstract
OBJECTIVE: To track cognitive change over time in dementia-free older adults and to examine terminal cognitive decline. METHODS: A total of 1,230 subjects who remained free from dementia over 14 years of follow-up were included in a population-based epidemiologic cohort study. First, we compared survivors and decedents on their trajectories of 5 cognitive functions (learning, memory, language, psychomotor speed, executive functions), dissociating practice effects which can mask clinically significant decline from age-associated cognitive decline. We used longitudinal mixed-effects models with penalized linear spline. Second, limiting the sample to 613 subjects who died during follow-up, we identified the inflection points at which the rate of cognitive decline accelerated, in relation to time of death, controlling for practice effects. We used mixed-effects model with a change point. RESULTS: Age-associated cognitive trajectories were similar between decedents and survivors without dementia. However, substantial differences were observed between the trajectories of practice effects of survivors and decedents, resembling those usually observed between normal and mildly cognitively impaired elderly. Executive and language functions showed the earliest terminal declines, more than 9 years prior to death, independent of practice effects. CONCLUSIONS: Terminal cognitive decline in older adults without dementia may reflect presymptomatic disease which does not cross the clinical threshold during life. Alternatively, cognitive decline attributed to normal aging may itself represent underlying neurodegenerative or vascular pathology. Although we cannot conclude definitively from this study, the separation of practice effects from age-associated decline could help identify preclinical dementia.