Smoking and Cognitive Impairment: A Population‐Based Study

OBJECTIVE: To test the hypothesis that there is an inverse relationship between smoking and cognitive impairment by examining the prevalence and incidence of such impairment in a representative cohort of older urban residents. DESIGN: Survey follow‐up with three interviews over a 4‐year period. SETTING: City of Cleveland, Ohio. PARTICIPANTS: Six hundred forty‐seven survivors of a population‐based, multi‐stage probability sample of urban residents, aged 74 years and older in 1984. MEASUREMENTS: Past and present cigarette smoking was measured in 1984 by direct questioning. Cognitive impairment was measured in 1984, 1987, and 1988 by the 10‐item Short Portable Mental Status Questionnaire. Mortality was determined at each interview date, and death certificates were obtained. Data were analyzed by univariate analysis, stratified analysis, and multivariate logistic regression. RESULTS: There were 99 cases of cognitive impairment at baseline among 628 individuals for whom complete data were available. Crude data suggest that the prevalence of cognitive impairment was lower among smokers than among non‐smokers, but logistic regression, adjusted for age, income, and gender, showed this difference to be nonsignificant, OR 0.73; 95% CI (0.42,1.29). Fifty‐one new cases of cognitive impairment occurred during the 4‐year observation period. Neither the crude data nor logistic regression showed any significant relationship between smoking and the incidence of cognitive impairment, OR 1.03; 95% CI (0.54, 1.99). Four‐year mortality did not differ significantly between these older smokers and non‐smokers, but smokers died at a relatively younger age and were more likely to die of diseases associated with smoking. CONCLUSION: This study does not confirm the hypothesis that there is a negative relationship between smoking and cognitive impairment. The mortality data suggest that case‐control prevalence studies of the hypothesized relationship may be distorted by complex life‐long trends in behavior, morbidity, and mortality which could result in a misleading appearance of low prevalence of cognitive impairment among smokers. Additional large, prospective, population‐based studies of the incidence of cognitive impairment and related disorders and possible protective or risk factors other than smoking are needed. J Am Geriatr Soc 44:905–909, 1996.