The following article requires a subscription:

(Format: HTML, PDF)

Background : Moderate alcohol consumption has been associated in several studies with decreased risk of cardiovascular and cerebrovascular events; however, available data on the effects of alcohol intake on cognitive functioning are conflicting. We assessed the association between alcohol consumption and cognitive impairment in a series of older subjects enrolled in a multicenter pharmacoepidemiology survey.

Methods : The association between average alcoholic intake and cognitive performance was assessed in 15,807 patients admitted to participating centers during the survey periods. Demographic variables, comorbid conditions, medications, and objective tests that were associated with cognitive impairment (as indicated by a Hodkinson Abbreviated Mental Test score <7) in separate logistical regression models were examined as potential confounders in a summary model.

Results : Cognitive impairment was detected in 1693 (19%) of 8755 drinkers and 2008 (29%) of 7052 nondrinkers (Fisher's exact test, p < 0.0001). After adjusting for potential confounders, alcohol consumption was associated with decreased probability of cognitive impairment (odds ratio, 0.75; 95% confidence interval, 0.66-0.85). The relationship between drinking level and cognitive dysfunction was nonlinear, because the probability of cognitive impairment was decreased for moderate alcohol use as compared with abstinence, but it was increased for daily consumption exceeding one wine-equivalent liter among men and 0.5 liter among women. This nonlinear association persisted when cerebrovascular and Alzheimer's disease were considered separately.

Conclusions : Alcohol abuse is associated with increased prevalence of cognitive dysfunction among older subjects; however, a daily alcohol consumption of less than 40 g for women and 80 g or less for men might be associated with a decreased probability of cognitive impairment. This possible protective effect of alcohol consumption should be further assessed by prospective studies.

(C) 2001 Lippincott Williams & Wilkins, Inc.