Age-period-cohort models in the US National Alcohol Surveys and recent heavy drinking cohorts

Trend analyses from an age-period-cohort perspective can potentially disentangle changes effecting the whole population from those driven by differences between sequential birth cohorts. These differences are important because they imply divergent future trajectories for alcohol consumption patterns by age group and the resulting impacts on health and social problems including alcohol dependence and treatment demand. Seven US National Alcohol Surveys utilizing comparable questions were conducted between 1979 and 2010. Previous analyses of the same series through 2005 suggested higher drinking volume and heavy occasion days among recent birth cohorts from 1976 to 1985. However, limited observation of these groups at different ages precluded strong conclusions. Age-period-cohort decomposition analyses, including demographic characteristics, were estimated for alcohol volume from beverage-specific graduated frequency question and for heavy occasion drinking days using both 5+ and 8+ drinks per day levels. Negative binomial models utilizing the natural logs of these variables were estimated for men and women separately. Results for alcohol volume indicate significant positive cohort effects for men in the 1976-1985 birth cohorts and for women in the 1981-1985 birth cohort relative to the reference 1956-60 birth cohort and others. Results for heavy occasion variables are consistent with these. Period effect estimates suggest increased volume and heavy occasions from 2000 to 2010. These recent birth cohorts, now aged 25-35, should be targeted in prevention efforts and studied further to improve our understanding of why these cohorts may have adopted heavier drinking patterns relative to those born both before and after.