Background Earlier research has shown that residents of Dutch deprived neighbourhoods drink less alcohol than people in other areas. We aimed to assess the role of individual and neighbourhood characteristics in a cross-sectional, nationwide, multilevel study.

Methods Individual data of 30 117 Dutch adults, living in 1722 neighbourhoods across the Netherlands, were obtained from the 2004 to 2009 national health survey (POLS). Chronic heavy alcohol consumption was measured as ≥14 drinks/week for women and ≥21 for men, and episodic heavy drinking as ≥6 drinks/day at least once a week. Neighbourhood deprivation was dichotomous; deprived districts as selected by the Dutch government versus other areas. Multilevel logistic regression models of the association between deprivation and heavy drinking were corrected for age, gender, household composition, population density and potential predictors ethnicity, socioeconomic status (education, income), neighbourhood-level social cohesion and percentage Muslims.

Results The prevalence of heavy drinking was lower in deprived neighbourhoods than in the rest of the Netherlands. This association was found for both chronic and episodic heavy drinking (OR=0.58 (0.47 to 0.72) and OR=0.57 (0.45 to 0.72), respectively). Adding ethnicity to the model reduced these associations by approximately one half. Socioeconomic composition did not contribute to the relationship. The proportion of Muslims explained a small part, while social cohesion explained even less of the association. Stronger associations were observed for women and older adults than for men and younger adults.

Conclusions The lower prevalence of heavy drinking occurring in deprived areas is largely explained by the ethnicity of neighbourhood residents.