METHODS:
Cross-sectional study with data from 65,955 people aged ≥50 years from 18 countries (SHARE project, 2011-2013). The outcome variable, hazardous drinking, was calculated using an adaptation of the AUDIT-C test. Several individual (sociodemographic, life-style and health factors) and contextual variables (country socioeconomic indicators and alcohol policies) were analysed. The prevalence of hazardous drinkingwas estimated by each exposure variable. To estimate associations, multilevel Poisson regression models with robust variance were fit, yielding prevalence ratios and their 95% confidence intervals (95%CI).