Abstract
OBJECTIVE: Compare risk of intentional self-harm and overdose after visits for opioid use disorder (OUD) followed by starting vs. not starting buprenorphine. METHODS: Records from four health systems identified visits during 1/1/2012-12/31/2019 by health system members aged 13 or older with OUD diagnosis and no recent OUD medication. Following a target-trial emulation approach, visits followed by buprenorphine dispensing within 7 days were matched to unexposed visits. Analyses compared risk of diagnosed self-harm injury or poisoning (primary outcome) as well as opioid-involved poisoning and any injury or poisoning (secondary outcomes) within 90 days. RESULTS: Among 183,809 visits by 30,955 patients, 15,508 (8.4 %) had buprenorphine dispensing within 7 days, and 2260 (1.2 %) had self-harm diagnosis within 90 days. Average duration of buprenorphine treatment before interruption was 44.3 days (SD 32.1). In primary intention-to-treat analyses using logistic regression and adjusting for baseline risk of self-harm, starting buprenorphine was not associated with significant difference in self-harm (odds ratio [OR] 1.01, 95 % CI 0.81-1.24) or opioid-involved poisoning (OR 1.09, 95 % CI 0.86-1.38). In secondary as-treated analyses censoring outcomes after treatment change, buprenorphine initiation was associated with no significant difference in hazard of self-harm (Hazard Ratio [HR] 0.74, 95 % CI 0.53-1.02) and with significantly lower hazard of opioid-involved poisoning (HR 0.63, 95 % CI 0.43-0.94). CONCLUSIONS: Among people with OUD, starting buprenorphine was not followed by lower risk of self-harm, likely reflecting frequent discontinuation and high risk of self-harm or overdose shortly after discontinuation. These findings reinforce the need to improve treatment continuity among those starting buprenorphine.