Alcohol availability and homicide in New Orleans: Conceptual considerations for small area analysis of the effect of alcohol outlet density

Objective: To determine the geographic relation between homicide rate and two competing measures of exposure to alcohol outlets, alcohol outlets per square mile and alcohol outlets per person. Method: Homicides occurring in 1994 and 1995 and on-sale and off-sale alcohol outlets with active 1995 licenses were geocoded by address for aggregation at the census tract level. Ecologic analysis of the 155 urban residential census tracts in New Orleans was conducted with controls for potential sociodemographic confounders (% black, % adults unemployed, % unmarried households, and ratio males 15-24/males 35-44). Results: After logarithmic transformation of all study variables, sociodemographic confounders drone accounted for 58% (R-2 = .58) of the variance of homicide rates. Adding off-sale alcohol outlet density to the models, measured (beta +/- SE) either as outlets per square mile (beta = .211 +/- .062) or outlets per person (beta = .244 +/- .063), yielded strong geographic relations with homicide and increased the amount of variance explained (R-2 = .62). A 10% higher off-sale outlet density accounted for a 2.4% higher homicide rate. Conclusions: Both off-sale alcohol outlets per square mile and off-sale outlets per person demonstrate strong geographic associations with homicide rates among urban residential census tracts in New Orleans. These findings suggest that communities faced with high rates of assaultive violence might consider policy interventions that address alcohol outlet related factors.