Crime and Features of the Built Environment Predicting Risk of Fatal Overdose: A Comparison of Rural and Urban Ohio Counties with Risk Terrain Modeling

Background: For nearly half of the period between 1999 and 2019, rates of rural overdose death surpassed those in urban areas. Despite this substantial increase, little attention has been given to rural overdose or the contextual factors that predict risk of fatal overdose in rural vs. urban communities. Methods: Risk terrain modeling was used to assess 2016–2017 overdose deaths in two urban and two rural Ohio counties. Spatial models incorporated criminal incidents and features of the built environment that have been previously associated with fatal overdose. The efficacy of spatial models was evaluated through the Predictive Accuracy Index (PAI) and Predictive Efficiency Index (PEI*). Results: Spatial models in rural counties were more influenced by past instances of crime, whereas risk in urban counties was determined by both crime and the built environment. Taken together, models accurately predicted 76% of 2018 overdoses. Rural models were overall more accurate, primarily in the areas predicted as having the highest risk of future overdose deaths. The predictive accuracy and efficiency of rural models varied more than those of urban models. Conclusions: It is feasible to apply risk terrain modeling to predict fatal overdose in rural areas. Though the underlying contextual risk factors and patterns of predicted risk differ between rural and urban areas, both can be utilized to place treatment and prevention resources more accurately for targeted intervention.