Stability of alcohol consumption over time: evidence from three longitudinal surveys from the United States.

Abstract

To assess the stability of individual monthly alcohol consumption volume in three studies from the United States: The first National Health and Nutrition Examination Survey with three measurements, The Framingham Offspring and Spouse Cohort with four measurements, and the National Longitudinal Survey of Youth with six measurements. The results will aid in the interpretation of studies that assess consumption at one point only. Five analyses measure (1) the amount of change between measurements, (2) the correlation between reported consumption, (3) patterns of later drinking by the baseline group, (4) mean consumption of abstainers when they drink and heavy drinkers when not drinking heavily and (5) the strength of past consumption in predicting present consumption. Correlations between measurements are high for adult samples 5 years apart or less but low for longer follow-ups and younger samples. Heavy drinkers are much less stable than abstainers or moderate drinkers. Abstainers at one time who drink at another drink much less than the mean. One time heavy drinkers are close to the consumption mean of the entire sample when not heavy but two or more time heavy drinkers drink more than the mean when not heavy. An autoregressive, rather than Markov, model of consumption change is supported. Aspects of both stability and change are found. Baseline measures of drinking groups are especially unreliable for younger samples, longer follow-ups and heavier drinkers. There appear to be important subgroups moving between abstention and light drinking and moderate and heavy drinking that can be identified only by multiple measurements or retrospective measures.