Individual and Contextual Factors Associated With Hazardous Drinking in Spain: Evidence From a National Population-Based Study

BACKGROUND: Heavy drinking (HD) and binge drinking (BD) exhibit marked differences in their relationships with contextual-level factors imbedded in geographical areas of residence. The objective is to identify sociodemographic factors, both at the individual and at the contextual level, associated with these 2 main hazardous consumption patterns.

METHODS: Cross-sectional study using data from the 2011 to 2012 National Health Survey in Spain. The sample included 21,007 individuals ≥15 years of age. HD was defined as an alcohol intake of ≥40 g/d in men and ≥24 g/d in women. BD was defined as the consumption in the previous month of ≥6 alcoholic drinks (men) or ≥5 drinks (women) within 4 to 6 hours. Individual-level variables included sociodemographic factors, urban/rural residence, smoking, and perceived social support. Contextual-level variables covered percentage of population with no schooling, unemployment rate, and hospitality industry-related economic activity, all at the census tract level. We analyzed data using multilevel logistic regression and calculated areas under the curve (AUC).

RESULTS: Being male, smoking, high-income, and low perceived social support were associated with both hazardous drinking patterns. Younger individuals were at higher risk for BD but at lower risk for HD. BD was more common among rural than urban dwellers (odds ratios [OR] = 1.35; 95% CI: 1.05 to 1.72), whereas HD was less likely in participants residing in areas with high unemployment rates (OR = 0.62; 95% CI: 0.41 to 0.93). HD was more likely in census tracts with higher levels of hospitality industry activity (OR = 1.74; 95% CI: 1.20 to 2.54). The AUC increased substantially for both HD and BD when the census tract variable was entered in the respective models (reaching 89.5 and 93.3%, respectively).

CONCLUSIONS: Except for age, both drinking patterns have similar associations with individual-level variables but disparate links to contextual-level indicators. In both cases, accounting for area of residence substantially increased the ability to discriminate between high-risk drinkers from nonhazardous alcohol consumers.