Fentanyl related overdose in Indianapolis: Estimating trends using multilevel Bayesian models.

INTRODUCTION The opioid epidemic has been largely attributed to changes in prescribing practices over the past 20 years. Although current overdose trends appear driven by the opioid fentanyl, heroin has remained the focus of overdose fatality assessments. We obtained full toxicology screens on lethal overdose cases in a major US city, allowing more accurate assessment of the time-course of fentanyl-related deaths. METHODS We used coroner data from Marion County, Indiana comprising 1583 overdose deaths recorded between January 1, 2010 and April 30, 2017. Bayesian multilevel models were fitted to predict likelihood of lethal fentanyl-related overdose using information about the victim's age, race, sex, zip code, and date of death. RESULTS Three hundred and seventy-seven (23.8%) overdose deaths contained fentanyl across the seven-year period. Rates rose exponentially over time, beginning well below 15% from 2010 through 2013 before rising to approximately 50% by 2017. At the beginning of the study period, rates of fentanyl overdose were lowest among Black persons but increased more rapidly, eventually surpassing Whites. Currently, White females are at particularly low risk of fentanyl overdose whereas Black females are at high risk. Rates were highest for younger and middle-aged groups. Over time, fentanyl was more likely detected without the presence of other opioids. CONCLUSIONS Fentanyl has increasingly been detected in fatal overdose deaths in Marion County. Policy and program responses must focus on education for those at highest risk of fentanyl exposure and death. These responses should also be tailored to meet the unique needs of high-risk demographics.