Does Opioid Use Deterrence Increase Heroin Deaths? Evidence from the Reformulation of OxyContin

Drug overdose deaths have risen steadily for the past two decades. In 2009, they became the leading cause of death from injuries in the United States, exceeding deaths from motor vehicle accidents. Death from prescription opioids has been the primary driver behind this upward trend, nearly quadrupling since 1999. Prescription opioids alone killed over 16,000 people in 2013.  The rise in opioid-related mortality slowed in 2011 and, for the first time since 1990, dropped in 2012.   This recent decline coincided with a significant rise in heroin overdoses.  Heroin-related overdoses tripled between 2010 and 2013, accounting for over 8,200 annual deaths.  The recent rise is drug overdoses has been deemed the “worst drug overdose epidemic in [U.S.] history” by the Center for Disease Control.

We examine the role of OxyContin reformulation on the recent dramatic rise in heroin-related mortality.  In August 2010, an abuse-deterrent version of OxyContin was introduced, representing a reduction in the supply of abusable opioids.  Time series evidence suggests a strong relationship between the OxyContin reformulation and the rise in heroin mortality, implying substitution towards heroin.  We study the relative change in heroin-related mortality in states with higher pre-2010 rates of OxyContin misuse relative to states with lower rates of OxyContin misuse using data from the Centers for Disease Control Vital Statistics database.  This approach allows us to condition on both state and time fixed effects instead of relying solely on overall trends.  We find that OxyContin misuse is a strong predictor of larger increases in both heroin substance abuse treatment admissions and heroin-related mortality.  There is little evidence that these results can be explained by differential pre-existing trends across states as both the trends and levels of heroin-related mortality were similar before 2010 across states with different rates of OxyContin misuse.  Moreover, we find that the rise in heroin can be attributed specifically to pre-2010 misuse of OxyContin and not misuse of pain relievers more generally.  While other changes in opioid supply occurred at approximately the same time as the OxyContin reformulation, these other factors appear to have little power in explaining recent trends in heroin use and mortality.  Our results also show that the reformulation led to substitution towards more powerful legal opioids, and we observe little evidence of differential reductions in opioid-related harms.