The aim of this study was to evaluate the longitudinal changes in alcohol consumption (total alcohol and types of alcoholic beverages) of the Greek EPIC cohort participants (28,572) during a 17-year period (1994-2011), with alcohol information being recorded repeatedly over time. Descriptive statistics were used to show crude trends in drinking behavior. Mixed-effects models were used to study the consumption of total alcohol, wine, beer and spirits/other alcoholic beverages in relation to birth cohort, socio-demographic, lifestyle and health factors. We observed a decreasing trend of alcohol intake as age increased, consistent for total alcohol consumption and the three types of beverages. Older birth cohorts had lower initial total alcohol consumption (8 vs. 10 g/day) and steeper decline in wine, spirits/other alcoholic beverages and total alcohol consumption compared to younger cohorts. Higher education and smoking at baseline had a positive association with longitudinal total alcohol consumption, up to +30% (vs. low education) and more than +25% (vs. non-smoking) respectively, whereas female gender, obesity, history of heart attack, diabetes, peptic ulcer and high blood pressure at baseline had a negative association of -85%, -25%, -16%, -37%, -22% and -24% respectively. Alcohol consumption changed over age with different trends among the studied subgroups and types of alcohol, suggesting targeted monitoring of alcohol consumption.
Keywords: EPIC; alcohol consumption; linear mixed models; longitudinal.