RESUMO
INTRODUCTION: Some studies have been questioning the association between poverty and violence. This study's purpose is to assess the distribution of homicide indicators associated with living conditions in Salvador, Brazil. METHODS: A cluster study for the years 1991 to 1994 was carried out including the 75 data centers of the city of Salvador, BA, Brazil. Using death certificates for the study period, yearly mortality rates and mortality ratios were estimated. The 1991 census data of monthly wages and years of education for all family providers were used to define a four-category variable related to living conditions. Mortality rates due to homicide and the relative risk regarding the lowest living condition area were calculated for each social stratum. The 95% confidence intervals were calculated using the Confidence Interval Analysis software. RESULTS: The highest mortality rates due to homicide were seen in the poorest areas. The relative risk due to homicide for the lowest and the highest living condition areas was statistically significant at 5% level and ranged from 2.9 to 5.1. CONCLUSIONS: The data show a strong association between social inequalities and homicide in this urban area, emphasizing the importance of crime reduction programs.