Reproducibility of two approaches for assessing alcohol consumption among older adults

Objectives. In this study, we hypothesized that there is greater disclosure in self reports of alcohol intake when details of quantity-frequency measures of alcohol consumption are ascertained in the context of a general health and life style questionnaire as compared to a directed interview on usual drinking habits. Methods. Data are from the 1993 to 1994 follow-up of the Washington County cohort of men and women 65 years and older, participating in the Cardiovascular Health Study. A total of 918 subjects completed a questionnaire evaluation of their usual alcohol consumption by two separate approaches: (1) alcohol intake was derived from responses to questions contained within a medical and personal history questionnaire; (2) the same questions were asked in the context of a specific alcohol use in older adults questionnaire. Results. The mean alcohol intake for the entire cohort, and for drinkers alone were almost identical when assessed by either questionnaire, with high correlation between the two estimates, irrespective of beverage type. There was 89% agreement classifying drinkers versus nondrinkers by both approaches, with the strength of the agreement good (κ=0.76). This agreement became moderate if drinkers were further categorized into three levels of alcohol intake. Predictors of the differences in alcohol intake between the two questionnaires were explored by multiple regression. Differences were largest for those who stated that the reason they drank was because they were no longer working, and for those drinking on average more than 24 g (greater than approximately 2 drinks) of alcohol daily. Discussion. Although agreement between the two approaches was generally comparable, some findings may indicate that older adults who are problem drinkers or drink heavily report lower consumption patterns when administered a more directed questionnaire specifically focusing on drinking behavior. These findings have implications in the design of studies measuring alcohol consumption among elderly persons with a relatively low background alcohol intake.