Association between opioid analgesic therapy and initiation of buprenorphine management: An analysis of prescription drug monitoring program data

Abstract

Background In the US, medication assisted treatment, particularly with office-based buprenorphine, has been an important component of opioid dependence treatment among patients with iatrogenic addiction to opioid analgesics. The predictors of initiating buprenorphine for addiction among opioid analgesic patients have not been well-described. Methods We conducted a time-to-event analysis using data from the North Carolina (NC) Prescription Drug Monitoring Program (PDMP). Our outcome of interest was time-to-initiation of sublingual buprenorphine. Our study population was a prospective cohort of all state residents receiving a full-agonist opioid analgesic between 2011 and 2015. Predictors of initiation of sublingual buprenorphine examined included: age, gender, cumulative pharmacies and prescribers utilized, cumulative opioid intensity (defined as cumulative opioid exposure divided by duration of opioid exposure), and benzodiazepine dispensing. Findings Of 4.3 million patients receiving opioid analgesics in NC between 2011 and 2015 (accumulated 8.30 million person-years of follow-up), and a total of 28,904 patients initiated buprenorphine formulations intended for addiction treatment (overall rate 3.48 per 1,000 personyears). In adjusted multivariate models, the utilization of 3 or more pharmacies (HR: 2.93; 95% CI: 2.82, 3.05) or 6 or more controlled substance prescribers (HR: 12.09; 95% CI: 10.76, 13.57) was associated with buprenorphine initiation. A dose-response relationship was observed for cumulative opioid intensity (HR in highest decile relative to lowest decile: 5.05; 95% CI: 4.70, 5.42). Benzodiazepine dispensing was negatively associated with buprenorphine initiation (HR: 0.63; 95% CI: 0.61, 0.65). Conclusions Opioid analgesic patients utilizing multiple prescribers or pharmacies are more likely to initiate sublingual buprenorphine. This finding suggests that patients with multiple healthcare interactions are more likely to be treated for high-risk opioid use, or may be more likely to be identified and treated for addiction. Future research should utilize prescription monitoring program data linked to electronic health records to include diagnosis information in analytic models.