Prescription and Illicit Opioid Deaths and the Prescription Drug Monitoring Program in Florida

Kennedy-Hendricks et al. concluded that Florida state laws targeting “pill mills” were associated with a significant decline in prescription opioid overdose deaths and may have reduced heroin overdose deaths in the long-term.¹ These findings add to a growing body of evidence indicating that the opioid epidemic has changed significantly in Florida.^1–5 Undoubtedly, changes in policy and practice have shifted the opioid epidemic curve; however, scientific debate continues regarding which policy or enforcement contributed most to these changes. Additionally, the extent of unintended consequences—an increase in the demand for licit and illicit prescription opioid alternatives—remains unresolved.⁶

Our published research on this topic, using a quasi-experimental, time-series design, demonstrated that the prescription drug monitoring program (PDMP) in Florida reduced oxycodone-caused deaths by 25% immediately following implementation.² We concluded this after controlling for many of the same policies indicated by Kennedy-Hendricks et al.—US Drug Enforcement Administration operations, pill mill laws, and the reformulation of Oxycontin. Thus, the authors’ assertion that “research has not shown [PDMPs] to be associated with lower mortality from opioid overdose” is incorrect.^1(p6)

Our research provides evidence that heroin-caused deaths may have increased simultaneously.² For additional insight, we plotted quarterly counts of heroin, morphine, and fentanyl-caused deaths (and overlaid our previously studied oxycodone series). We also tested the change in monthly trend after Florida’s PDMP implementation (Figure 1). Note that the observation period shown extends 1.5 to two years beyond the Kennedy-Hendricks et al.¹ and Delcher et al.² articles, respectively. We used unadjusted, best-fitting ARIMA models for this purpose, and deaths were plotted quarterly to improve readability. PDMP implementation was associated with a statistically significant increase in the trends for heroin, morphine and fentanyl-caused deaths (P < .005).

FIGURE 1—

Modeled (Line) and Observed (Dot) Quarterly Counts of Heroin-, Morphine-, Fentanyl- and Oxycodone-Caused Deaths Before and After Implementation of Florida’s Prescription Drug Monitoring Program

Note. The best-fitting Autoregressive Integrated Moving Average (ARIMA) parameters (p,d,q) for each drug series are shown and post-PDMP trends are significant (P < .005).

Source. Florida Department of Law Enforcement, Medical Examiners Commission.

Three points are worth considering: (1) the magnitude of the heroin epidemic may be underestimated because, due to the rapid metabolism of heroin to morphine, many morphine deaths might be misclassified⁷; (2) in Florida, heroin and morphine deaths now exceed oxycodone-caused deaths; and (3) the increase in fentanyl-caused deaths (principally illicit) suggests that we should not focus solely on users’ shift to heroin.

Visually, one could argue that rates began increasing in late 2010, an observation that Kennedy-Hendricks et al. (2015) change-point analysis supports. Our approach tests a specific intervention at a specific point in time (October 2011) using four specific drug series, providing compelling evidence that PDMPs are affecting mortality. The scientific debate aside, Florida and the nation are clearly faced with a new and challenging course for the opioid epidemic.