Alcohol Intake and Risk of Dementia

Jose A. Luchsinger, MD; Ming-Xin Tang, PhD; Maliha Siddiqui, MPH; Steven Shea, MD; Richard Mayeux, MD


J Am Geriatr Soc. 2004;52(4) 

In This Article


Participants in the Washington Heights-Inwood Columbia Aging Project cohort were drawn by random sampling of 2,126 healthy Medicare beneficiaries aged 65 and older residing within a geographically defined area of northern Manhattan.[6] At entry, each subject underwent a structured in-person interview including an assessment of health and function, a standard medical history, a physical and neurological examination, and a neuropsychological battery.[7] Subjects were recruited between 1991 and 1996 and followed annually, repeating the baseline examination at each follow-up. A food frequency questionnaire was administered to 1,422 individuals between baseline and the first follow-up examination. Of these 1,422 individuals, 230 were excluded because of prevalent dementia, 210 because of loss to follow-up, and two because of missing dietary data. Thus, the analytic sample comprised 980 subjects. The 916 individuals excluded due to loss to follow-up or no dietary information were older than the final sample (mean age±standard deviation= 88.2±8.6 vs 73.3±5.8) and had a similar proportion of women (70% vs 67%).

A group of neurologists, psychiatrists, and neuropsychologists made a diagnosis of dementia and assignment of specific cause by consensus based on the information gathered at the initial and yearly follow-up visits. The diagnosis of dementia was based on Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition criteria[8] and required evidence of cognitive deficit on the neuropsychological test battery and evidence of impairment in social or occupational function; persons with a global summary score on the Clinical Dementia Rating (CDR) of 0.5 or more were considered to have dementia.[9] Diagnosis of Alzheimer's disease (AD) was based on the National Institute of Neurological and Cognitive Disorders and Stroke–Alzheimer's Disease and Related Disorders Association criteria.[10] Diagnosis of dementia associated with stroke (DAS) was made in all subjects with a history of stroke. Because moderate alcohol intake is related to a lower risk of stroke,[11] a less-conservative definition of DAS was used to avoid finding an association between alcohol and AD because of misclassification of DAS as AD. Brain imaging was available in 85% of cases of stroke; in the remainder, World Health Organization criteria were used to define stroke.[12] Subjects without dementia but with a history of stroke at the baseline examination were included in the analyses. These criteria and diagnostic methods have been used extensively in analysis of data in this cohort.

Dietary data were obtained using a 61-item version of Willett's semiquantitative food frequency questionnaire (SFFQ) (Channing Laboratory, Cambridge, MA).[13] Trained interviewers administered in English or Spanish the SFFQ by telephone between the baseline and first follow-up examinations. The questionnaire inquired about servings of beer, liquor, and wine by serving frequency; possible answers were one to three servings/month, one serving a week, two to four servings a week, five to six servings a week, one serving a day, two to three servings a day, four to five servings a day, and more than six servings a day. One serving of beer was equivalent to 12 oz (12.8 g of alcohol), one serving of liquor was equivalent to 1.5 oz (14 g of alcohol), and one wine serving was equivalent to 4 oz (11 g of alcohol). In light of the possibility of nonspecific alcohol effects and specific nonalcoholic effects of the different beverages, alcohol intake was examined in two ways: by examining the association between beer, liquor, and wine servings separately with incident dementia and by examining the association between total alcohol servings and dementia. Individuals were classified as nondrinkers (0 servings reported), light drinkers (1 serving a month to 6 servings a week), moderate drinkers (1 to 3 servings a day), and heavy drinkers (>3 servings a day); this classification was made trying to resemble previous publications in this field[2] for the sake of comparability and following the format of the questionnaire. Because of a low number of moderate drinkers, the light and moderate drinkers were aggregated in one category (1 serving a month to 3 servings a day). Heavy alcohol intake was included despite the low number of individuals in this category because heavy alcohol intake may increase the risk of dementia.[14] SFFQs have been used and validated for the determination of nutrient intake in the elderly.[15,16,17,18,19] The validity of the food frequency questionnaire used in the Washington Heights-Inwood Columbia Aging Project cohort was assessed in a subsample of 78 individuals using two 7-day food records as the criterion. The correlation for energy-adjusted alcohol intake was 0.44 (P<.01) (M. Siddiqui, personal communication, December 7, 2000). The reliability of the alcohol intake measure was compared between two SFFQs administered 2 months apart, and the measurements were not significantly different.

Ethnic group was based on self-report using the format of the 1990 census.[20] Individuals were also asked whether they were of Hispanic origin. Participants were then assigned to one of three groups: black (non-Hispanic), Hispanic, or white (non-Hispanic). Ethnic group was not found to significantly alter the models and was not included in the analyses. Data on years of education and heart disease were also obtained by self-report. Education was examined as a continuous variable (education in years) and as a categorical variable (≤ 6 years of education, 7-12 years of education, 13-16 years of education, and >16 years of education). Apolipoprotein (APO) genotype was determined using the method of Hixson and Vernier.[21] Participants were classified as positive for the APOE-ε4 allele genotype if they had one or two ε4 alleles.

Individuals were compared according to their beverage intake. Chi-square tests were used to compare sex, APOE-ε4 status, and the presence of heart disease. Analyses of variance and t tests were used to compare mean age and number of years of education. Cox proportional hazards regression was used for multivariate analyses, with the time-to-event variable in the models specified as time from baseline examination to onset of dementia. Individuals with dementia not caused by the subtype of interest were censored at the time of onset of dementia. The final model was stratified by education category using the STRATA statement in the SAS procedure PROC PHREG (SAS Institute, Inc., Cary, NC). Three types of analyses were conducted: one in which servings of beer, liquor, and wine were related separately to dementia subtypes; another in which all beverage types were included in one model; and another relating servings of total alcohol to dementia and dementia subtypes. Alcohol intake and other covariates were treated as baseline time-constant covariates. The multivariate model examining AD as an outcome included age, sex, APOE-ε4, and education as covariates, and the models examining dementia and DAS as outcomes also included the presence of heart disease as a covariate. Inclusion of diabetes mellitus, hypertension, lipid levels, and smoking did not add predictive ability and were not included in the final models. SAS version 7 for Windows was used for all analyses (SAS Institute, Inc.).


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: