Patterns and risk factors of opioid-suspected EMS overdose in Houston metropolitan area, 2015-2019: A Bayesian spatiotemporal analysis

Abstract

Background Opioid-related overdose deaths are the top accidental cause of death in the United States, and development of regional strategies to address this epidemic should begin with a better understanding of where and when overdoses are occurring. Methods and findings In this study, we relied on emergency medical services data to investigate the geographical and temporal patterns in opioid-suspected overdose incidents in one of the largest and most ethnically diverse metropolitan areas (Houston Texas). Using a cross sectional design and Bayesian spatiotemporal models, we identified zip code areas with excessive opioid-suspected incidents, and assessed how the incidence risks were associated with zip code level socioeconomic characteristics. Our analysis suggested that opioid-suspected overdose incidents were particularly high in multiple zip codes, primarily south and central within the city. Zip codes with high percentage of renters had higher overdose relative risk (RR = 1.03; 95% CI: [1.01, 1.04]), while crowded housing and larger proportion of white citizens had lower relative risks (RR = 0.9; 95% CI: [0.84, 0.96], RR = 0.97, 95% CI: [0.95, 0.99], respectively). Conclusions Our analysis illustrated the utility of Bayesian spatiotemporal models in assisting the development of targeted community strategies for local prevention and harm reduction efforts.