Transporting treatment effects from clinical trials to real-world populations with co-occurring opioid and stimulant use disorders

PROJECT SUMMARY AND ABSTRACT Opioid-related overdoses are now responsible for twice as many deaths per year as car accidents in the United States. Encouraging downward trends in prescription opioid- and heroin-related mortality have been eclipsed by rises in overdoses and deaths attributable to synthetic opioids and concomitant use of opioids and stimulants. Nationally, the combined use of stimulants and opioids has nearly doubled in the past decade, and more than tripled in Western states. At least half of individuals with opioid use disorder (OUD) have another substance use disorder (SUD), and polysubstance use is the clinical norm. Although the efficacy of several treatments for SUDs, including OUD, has been established in rigorous randomized controlled trials (RCTs), the people enrolled in these RCTs often differ from the populations those treatments are intended to reach. In particular, little evidence supports the effectiveness of current SUD treatments for people who use multiple substances. Consequently, understanding and enhancing the effects of SUD treatments for people with polysubstance use remains a NIDA priority area (PAR-20-035). The proposed research addresses this critical knowledge gap by combining cutting-edge data science methods with an innovative data-fusion technique, transportability, to estimate what the results of SUD treatment trials would have been had they been conducted in representative, clinically-relevant populations of individuals with OUD and stimulant use disorders (OUD + StimUD). In Aim 1, I fuse data from National Drug Abuse Treatment Clinical Trials Network RCTs with the National Survey on Drug Use and Health. I estimate the effectiveness of medications, contingency management, motivational interviewing, counseling, and exercise to increase treatment initiation and retention and reduce opioid and stimulant use in the adult, civilian, noninstitutionalized U.S. population with OUD + StimUD. In my second Aim, I fuse the same clinical trial data with electronic health records to estimate the effectiveness of interventions to increase treatment engagement and retention among patients with OUD + StimUD in the Oregon Health & Science University health system. Through completing these aims, I estimate the real-world effectiveness of treatments for individuals with OUD + StimUD. Clinicians and policymakers may use the results to guide treatment selection to improve the health of these complex, vulnerable patients. Through this award, I build upon my background in epidemiology and statistics to gain expertise in cutting- edge data science and informatics methods. Combining causal inference, machine learning, and natural language processing techniques maximizes the innovation, validity, impact, and reach of my research. I also receive extensive instruction in clinical addiction medicine and build my leadership, grant writing, and scientific communication skills. Each of these new skills are crucial to completing my research aims and developing future high-impact, highly competitive research proposals leading to a successful independent research career in addiction science.