A modeling study of the opioid epidemic for vulnerable communities in Knoxville, Tennessee

Background: Over the past three decades, the opioid epidemic has emerged as one of the most pressing public health crises in the United States, disproportionately affecting vulnerable communities and minority populations. Mathematical modeling, widely applied in the study of infectious diseases, offers a rigorous framework to investigate the dynamics of opioid use disorder. Methods: We present a compartmental model to examine two primary risk factors for opioid misuse: opioid availability and access to preventive resources. The framework consists of two interacting systems of equations, one representing the general population and the other a subset community of interest. The general population model was calibrated using 2016-2019 data from the Knoxville, Tennessee Metropolitan Statistical Area, a region heavily impacted by the epidemic. Sensitivity analyses were conducted to quantify the influence of model inputs on outcomes. Results: Embedded within a data-calibrated general population, simulations of hypothetical communities were conducted under varying levels of treatment accessibility, recovery retention, and fatal overdose rates. We found that variations in recovery retention and fatal overdose risk exerted the strongest influence on total overdoses. Notably, increasing treatment access alone---without simultaneously addressing other factors such as relapse reduction---was associated with higher overdose rates, as individuals almost exclusively relapsed using fentanyl or heroin, rather than prescription opioids, all of which carry elevated fatal overdose risk. Conclusions: For Knoxville during 2016–2019, increasing resources aimed at preventing the initial development of opioid use disorder would likely have had minimal effect on the overall prevalence of opioid use disorder. Our analysis highlights the importance of model scope, as a related study, which was fitted to a broader geographic area and a longer time span, produced findings that differed from ours. While our conclusions are specific to this location and period, the modeling framework we created is readily adaptable to other regions and populations, provided appropriate data is available, and offers a valuable tool for studying vulnerable groups.