Geographic modeling for children at risk for home fires and burns.

Abstract

This study developed a predictive model for fires and burns among parents and children in Jefferson County, Kentucky. Eight risk factors for pediatric burns with census tract level data available were identified. Risk factors were synthesized to develop a cartographic model with risk levels low, medium, high, and severe. Validation was performed with fire dispatch data. At-risk areas were concentrated in the county's northwest. Risk was correlated with fire incidence rate (ρ=0.67, p<0.001). Significant risk factors were race (β=0.54, p<0.001), education (β=0.38, p<0.001), and year home built (β=-0.17, p=0.005). Cartographic modeling is a underutilized tool to identify at-risk areas.