Reaction Time Measures Predict Incident Dementia in Community-Living Older Adults: The Sydney Memory and Ageing Study

Objective To examine the utility of intraindividual variability of reaction times (IIVRT) and mean reaction time (RT) as behavioral markers of incident all-cause dementia. Methods A longitudinal cohort study followed biennially for 4 years, the community-based Sydney Memory and Ageing Study, 861 initially nondemented participants aged 70–90. Incident all-cause dementia determined by consensus, RT measures from simple and complex tasks, Mini-Mental State Exam and neuropsychological tests, Geriatric Depression Scale and Goldberg Anxiety Scale, cardiovascular risk score, apolipoprotein ε4 status, and the Bayer ADL Scale were used. Associations of baseline IIVRT and mean RT with time to dementia were evaluated with hazard ratios (HRs) using Cox proportional-hazards models with and without controlling for dementia risk factors. Results Forty-eight cases developed dementia. Greater complex IIVRT predicted a 40% (HR: 1.43) and mean RT a 50%–60% (simple RT: HR 1.53; complex RT: HR 1.59) per standard deviation increased risk of developing dementia, remaining significant after controlling for age, education, sex, general cognitive function, mood, cerebrovascular disease, and genetic susceptibility. Prediction of incident dementia using demographical information and RT measures combined was comparable with several traditional neuropsychological measures (AUC 0.75), although lower than a full neuropsychological battery (AUC 0.90). Prediction of functional decline by RT measures combined was equal to the neuropsychological battery (multiple Rs of .233 and .238, respectively). Conclusion Brief RT measures provided information on risk of imminent dementia and functional decline within 4 years in older adults at a population level, with mean RT the stronger predictor