Printed in U.S.A Smoking, Drinking, and Thinking The Zutphen Elderly Study

The authors examine the cross-sectional and longitudinal relation of smoking habits and current alcohol intake to cognitive status and decline over a 3-year period as well as the extent to which these relations are modified by the presence of clinical conditions indicating atherosclerosis (cardiovascular disease(CVD)/ diabetes). Data are from the cohort of men followed in the longitudinal Zutphen Elderly Study in 1990 (n = 489) and 1993 (n = 333). Cognitive function was measured in 1990 and 1993 with the 30-point Mini-Mental State Examination (MMSE). After adjustment for age, education, and alcohol intake, current smokers made 20% more errors on the MMSE than never smokers in the cross-sectional analyses. Cognitive decline was greatest in those with CVD/diabetes who currently smoked and never smoked (-1.9 and-1.3 points, respectively). After adjustment for age, education, and smoking status, men with CVD/diabetes and low-to-moderate alcohol intake had a significantly lower risk for poor cognitive function (MMSE £ 25) than abstainers (odds ratios of 0.3 for less than one drink and 0.2 for one to two drinks per day). Alcohol intake was not associated with cognitive decline. These findings do not support the hypothesis of a protective effect of smoking on cognitive function; they suggest that smoking may be harmful among those with CVD/diabetes. Alcohol may result in an acute beneficial effect on cognitive function among those with CVD/diabetes. However, selection bias and unmeasured confounding should be of concern when evaluating these results. Am J Epidemiol 1996