Evidence of a structural effect for alcohol outlet density: A multilevel analysis

Background: Ecological studies reveal that alcohol-related outcomes tend to occur in high alcohol outlet density neighborhoods. The ecological design of these studies limits the interpretation of the findings in terms of the level of the effect. The effect of alcohol outlet density could be related to greater individual access to alcohol, an individual level effect, or to the grouping of drinkers by neighborhood, a structural effect at the neighb orhood level. Methods: To differentiate between individual and neighborhood level possibi lities, we conducted a multilevel study. Individual distance to the closest alcohol outlet was the individual level measure of the effect of alcohol o utlet density, whereas the mean distance to the closest alcohol outlet for all individuals within a census tract was the neighborhood level measure fo r the effect of alcohol outlet density. We analyzed telephone surveys of 26 04 telephone households within 24 census tracts stratified by poverty statu s and alcohol outlet density. Individual distance to alcohol outlets, age, sex, race/ethnicity, and level of education were entered as individual leve l covariates, and their corresponding aggregated means were entered as cens us tract level covariates (i.e., mean distance to outlets, mean age, percen tage male, percentage Black. mean education). Results: Analysis of variance revealed that 16.2% of the variance in drinki ng norms and 11.5% of the variance in alcohol consumption were accounted fo r at the census tract level. In multivariate hierarchical analysis, individ ual distance to the closest alcohol outlet was unrelated with drinking norm s and alcohol consumption, whereas mean distance to the closest alcohol out let demonstrated a negative relation with drinking norms (beta(e) = -5.50 /- 2.37) and with alcohol consumption (beta(e) = -0.477 +/- 0.195); that is , the higher the mean distance to the closest alcohol outlet, the lower the mean drinking norms score and mean level of alcohol consumption. Conclusions: The findings suggest that the effect of alcohol outlet density on alcohol-related outcomes functions through an effect at the neighborhoo d level rather than at the individual level. Problem drinkers tend to be gr ouped in neighborhoods, an effect predicted by alcohol outlet density.