The geospatial and temporal correlation of oxycodone and heroin deaths in Florida counties, 2011 to 2013

Abstract

Aims: There is growing concern that efforts to limit the availability of prescription opioids (e.g., oxycodone) has contributed to an emerging heroin epidemic. From 2010 to 2013, Florida reported a 66% decrease in the rate of oxycodone-caused deaths, consistent with a 24% reduction in oxycodone prescribing, but heroin-caused deaths increasedby293%over the same timeperiod. The aimof this study is to test the hypothesis that, at the county-level, (1) death rates for both drugs are correlated, (2) death rates exhibit similar patterns of geographic clustering, and (3) heroin-caused mortality is correlated to a time lag in oxycodone-caused mortality. Methods: We mapped county-level, mortality rates by quartile (determined from rates pooled over 3 years) for oxycodone and heroin in 2011 and 2013 using data from Florida’sMedical Examiners Commission.Wedescribe geographic clustering andused linear regression to test models for heroin-caused deaths as a function of a zero, one, or two-year lag of oxycodone-caused deaths at the county-level. Results: For both oxycodone and heroin, we identified several persistentlyhigh ratemortality clusters. In2013, countieswithhigh rates of oxycodone-caused deaths did not appear to overlap with countieswith high rates of death fromheroin. Linearmodels for the zero and one-year lagged models had r2 of 0.01 and 0.03, respectively, whereas the two-year lagged model had r2 =0.18. These models indicate that rates of oxycodone-caused deaths in 2011 were directly related to the rates heroin-related deaths in 2013 although several counties (e.g., Miami-Dade) had notable inverse relationships. Conclusions: This work has county, state and national implications for monitoring persistent population-level problems with oxycodone and heroin abuse and the ability of health agencies to anticipate emerging trends in heroin abuse. Financial support: Internal funding only.