A dynamic model of the opioid drug epidemic with implications for policy

ABSTRACT

Background: The U.S. opioid epidemic has caused substantial harm for over 20 years. Policy interventions have had limited impact and sometimes backfired. Experts recommend a systems modeling approach to address the complexities of opioid policymaking.

Objectives: Develop a system dynamics simulation model that reflects the complexities and can anticipate intended and unintended intervention effects.

Methods: The model was developed from literature review and data gathering. Its outputs, starting in 1990, were compared against 12 historical time series. Four illustrative interventions were simulated for 2020–2030: reducing prescription dosage by 20%, cutting diversion by 30%, increasing addiction treatment from 45% to 65%, and increasing lay naloxone use from 4% to 20%. Sensitivity testing was performed to determine effects of uncertainties. No human subjects were studied.

Results: The model fits historical data well with error percentage averaging 9% across 201 data points. Interventions to reduce dosage and diversion reduce the number of persons with opioid use disorder (PWOUD) by 11% and 16%, respectively, but each of these interventions reduces overdoses by only 1%. Boosting treatment reduces overdoses by 3% but increases PWOUD by 1%. Expanding naloxone reduces overdose deaths by 12% but increases PWOUD by 2% and overdoses by 3%. Combining all four interventions reduces PWOUD by 24%, overdoses by 4%, and deaths by 18%. Uncertainties may affect these numerical results, but policy findings are unchanged.

Conclusion: No single intervention significantly reduces both PWOUD and overdose deaths, but a combination strategy can do so. Entering the 2020s, only protective measures like naloxone expansion could significantly reduce overdose deaths.

KEYWORDS:

• Opioid epidemic
• system dynamics
• simulation model
• historical time series
• policy intervention
• illicit market
• fentanyl
• naloxone

Acknowledgements

We would like to acknowledge the substantial contributions of our colleagues: data analyst Christina Ingersoll, researcher Elizabeth Etherton, methodological advisors John Sterman and Tom Fiddaman, and project supervisor Rayford Etherton.

Financial Disclosure and Funding Source

The initial version of the model described in this paper was developed under a contract funded by Herc Litigation Services LLC and the law firms of Levin Papantonio and Baron & Budd. Consulting fees were paid to the authors.

Supplementary material

Supplemental data for this article can be accessed on the publishers’s website.

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.