Determinants of selection into buprenorphine/naloxone among people initiating opioid agonist treatment in British Columbia

Abstract Background Studies assessing the comparative effectiveness of methadone versus buprenorphine/naloxone for opioid use disorder in real-world settings are rare - challenged by structural differences in delivery across settings and factors influencing treatment selection. We identified determinants of selection into buprenorphine/naloxone and quantified contributions of individual and provider-level covariates in a setting delivering both medications within the same healthcare settings. Methods Utilizing linked health administrative datasets, we conducted a retrospective cohort study of people with opioid use disorder (PWOUD) receiving opioid agonist treatment (OAT) in British Columbia, Canada, from 2008-2017. Determinants of buprenorphine/naloxone selection were identified using a generalized linear mixed model with random intercept terms for providers and individuals. We determined the influence of individual demographics, clinical history, measures of provider experience and preference, and dates of key policy changes. Results A total of 39,605 individuals experienced 178,976 OAT episodes (methadone:139,439(77.9%);buprenorphine/naloxone:39,537(22.1%)). Male sex, less OAT experience, younger age, mental health conditions and chronic pain were associated with higher odds of buprenorphine/naloxone prescription. For providers, higher client-attachment, more complex OAT case-mixes, and higher buprenorphine/naloxone prescribing-preference were also associated with higher odds of buprenorphine/naloxone prescription. Observed individual-level covariates explained 9.7% of variance in odds of buprenorphine/naloxone selection, while observed provider-level covariates explained 20.0%. Controlling for covariates, residual unmeasured between-individual variance accounted for 18.5% of the explained variation in the odds of buprenorphine/naloxone selection, while unmeasured between-provider variance accounted for 28.4%. Conclusion Provider characteristics were more influential in selection of buprenorphine/naloxone over methadone informing subsequent analyses of comparative effectiveness of these regimens.