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 t wo competing measures of exposure to alcohol outlets, alcohol outlets per s quare mile and alcohol outlets per person. Method: Homicides occurring in 1 994 and 1995 and on-sale and off-sale alcohol outlets with active 1995 lice nses were geocoded by address for aggregation at the census tract level. Ec ologic analysis of the 155 urban residential census tracts in New Orleans w as conducted with controls for potential sociodemographic confounders (% bl ack, % adults unemployed, % unmarried households, and ratio males 15-24/mal es 35-44). Results: After logarithmic transformation of all study variables , sociodemographic confounders drone accounted for 58% (R-2 = .58) of the v ariance of homicide rates. Adding off-sale alcohol outlet density to the mo dels, measured (beta +/- SE) either as outlets per square mile (beta = .211 +/- .062) or outlets per person (beta = .244 +/- .063), yielded strong geo graphic relations with homicide and increased the amount of variance explai ned (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 associa tions with homicide rates among urban residential census tracts in New Orle ans. These findings suggest that communities faced with high rates of assau ltive violence might consider policy interventions that address alcohol out let related factors.