Drug use patterns and factors related to the use and discontinuation of medications for opioid use disorder in the age of fentanyl: findings from a mixed-methods study of people who use drugs.

Background: Medications for opioid use disorder (MOUD; methadone, buprenorphine, naltrexone) are the most effective treatments for OUD, and MOUD is protective against fatal overdoses. However, continued illegal drug use can increase the risk of treatment discontinuation. Given the widespread presence of fentanyl in the drug supply, research is needed to understand who is at greatest risk for concurrent MOUD and drug use and the contexts shaping use and treatment discontinuation.

Methods: From 2017 to 2020, Massachusetts residents with past-30-day illegal drug use completed surveys (N = 284) and interviews (N = 99) about MOUD and drug use. An age-adjusted multinomial logistic regression model tested associations between past-30-day drug use and MOUD use (current/past/never). Among those on methadone or buprenorphine (N = 108), multivariable logistic regression models examined the association between socio-demographics, MOUD type; and past-30-day use of heroin/fentanyl; crack; benzodiazepines; and pain medications. Qualitative interviews explored drivers of concurrent drug and MOUD use.

Results: Most (79.9%) participants had used MOUD (38.7% currently; 41.2% past), and past 30-day drug use was high: 74.4% heroin/fentanyl; 51.4% crack cocaine; 31.3% benzodiazepines, and 18% pain medications. In exploring drug use by MOUD history, multinomial regression analyses found that crack use was positively associated with past and current MOUD use (outcome referent: never used MOUD); whereas benzodiazepine use was not associated with past MOUD use but was positively associated with current use. Conversely, pain medication use was associated with reduced odds of past and current MOUD use. Among those on methadone or buprenorphine, separate multivariable logistic regression models found that benzodiazepine and methadone use were positively associated with heroin/fentanyl use; living in a medium-sized city and sex work were positively associated with crack use; heroin/fentanyl use was positively associated with benzodiazepine use; and witnessing an overdose was inversely associated with pain medication use. Many participants qualitatively reported reducing illegal opioid use while on MOUD, yet inadequate dosage, trauma, psychological cravings, and environmental triggers drove their continued drug use, which increased their risk of treatment discontinuation and overdose.

Conclusions: Findings highlight variations in continued drug use by MOUD use history, reasons for concurrent use, and implications for MOUD treatment delivery and continuity.