Predicting cognitive impairment in high-functioning community-dwelling older persons: MacArthur studies of successful aging

OBJECTIVES: To examine whether simple cognitive tests, when applied to cognitively intact older persons, are useful predictors of cognitive impairment 7 years later. DESIGN: Cohort study. SETTING: Durham, North Carolina; East Boston, Massachusetts; and New Haven, Connecticut, areas that are part of the National Institute on Aging Established Populations for Epidemiological Studies of the Elderly. PARTICIPANTS: Participants, aged 70 to 79, from three community-based studies, who were in the top third of this age group, based on physical and cognitive functional status. MEASUREMENTS: New onset of cognitive impairment as defined by a score of less than 7 on the Short Portable Mental Status Questionnaire (SPMSQ) in 1995. RESULTS: At 7 years, 21.8% (149 of 684 subjects) scored lower than 7 on the SPMSQ. Using multivariate logistic regression, three baseline (1988) cognitive tests predicted impairment in 1995. These included two simple tests of delayed recallthe ability to remember up to six items from a short story and up to 18 words from recall of Boston Naming Test items. For each story item missed, the adjusted odds ratio (AOR) for cognitive impairment was 1.44 (95% confidence interval (CI) = 1.16-1.78, P < .001). For each missed item from the word list, the AOR was 1.20 (95% CI = 1.09-1.31, P < .001). The Delayed Recognition Span, which assesses nonverbal memory, also predicted cognitive impairment, albeit less strongly (odds ratio = 1.06 per each missed answer, 95% CI = 1,003-1.13, P = .04). CONCLUSIONS: This study identifies measures of delayed recall and recognition as significant early predictors of subsequent cognitive decline in high-functioning older persons. Future efforts to identify those at greatest risk of cognitive impairment may benefit by including these measures.