Project Summary: High quality evidence supports medication assisted treatment (MAT) as the first line treatment for opioid use disorder (OUD), especially opioid substitution therapy with buprenorphine or methadone as long-term maintenance of at least one year. Yet patient dropout (typically >50% at six months), and provider resistance remain major impediments to effective treatment. To date, few studies have examined patient characteristics that influence key steps in treatment initiation, stabilization, and long-term retention with buprenorphine. Most published studies assess outcomes for only 2-3 months of active treatment among highly selected patients in clinical trials. I will draw from the influential, sequential “cascade of care” framework from the HIV/AIDS field as a conceptual model to identify stage-specific barriers to continuity of buprenorphine treatment through an adapted OUD treatment cascade. A greater understanding of barriers at each sequential stage of the cascade including 1) initiation of buprenorphine, 2) stabilization of opioid use, and 3) retention with long-term treatment response, may inform clinical efforts to improve outcomes and mortality. Under healthcare reform and Medicaid redesign, increasing numbers of patients receive insurance reimbursement for substance abuse treatment. I propose to link clinical data from electronic health records (EHR) to insurance claims data to assess patient factors that promote continuity of care along the cascade. This strategy will capture critical information (ED visits, diagnoses, treatment services, mortality) that are outside of a given clinical site's records. Insurance claims data have the advantage of providing near complete follow up and objective outcomes that do not rely on self-report and reflect real-world clinical complexities. In order to accomplish these goals and achieve my long-term training objectives of developing expertise in quantitative methods, epidemiology, and the management and inferential analysis of large observational data sets, I have articulated five sequential training goals: 1) epidemiologic and statistical data analysis methods 2) services research, program evaluation, and model design 3) insurance claims data analysis 4) inferential methods with observational data, and 5) grant writing and administration. To fulfill my research aims, I will incorporate electronic records from a high-volume, multi-site MAT provider that has started 18,000 patients on buprenorphine since 2009 with frequent in-house urine drug testing. The research will identify patient characteristics associated with successful buprenorphine initiation, stabilization, and retention and will model barriers at each stage of the treatment cascade. With these findings, I also seek to evaluate the clinical utility of each stage by modeling associations between attaining each stage of treatment and improved clinical outcomes. The results, which will be based on a large, highly generalizable...