Evaluating Effects of ECHO Training of Care Providers for Opioid Use Disorder: A Framework of Observational Studies Utilizing Medical Claims Data and Interrupted Time Series Models

• Activity Number: 247 - Causal Inference and Statistical Learning of Intervention and Policy Effects
• Type: Contributed
• Date/Time: Wednesday, August 11, 2021 : 10:00 AM to 11:50 AM
• Sponsor: Health Policy Statistics Section
• Author(s): Jessica Reno* and Yiliang Zhu and Laura Tomedi and Julio Facelli and Miriam Komaromy and Alica Ventura and Julie Donohue and Marisa Domino
• Companies: University of New Mexico and University of New Mexico and University of New Mexico and University of Utah and Boston Medical Center and Boston Medical Center and University of Pittsburgh and University of North Carolina
• Keywords: Substance use; substance abuse; opioid use disorder; evaluation; longitudinal analysis; medical claims data

Abstract

Project ECHO education programs for providers of patients with opioid use disorder (OUD) have been established across the globe. Robust observational studies integrating multiple administrative databases may provide timely clinical evidence on changes in outcomes at the provider level. We developed a framework of retrospective cohort study designs with matched intervention-control comparisons and interrupted time series models for aggregated patient outcomes. The framework allows for multi-level clustering, latency periods, random effects for clustering, general difference-in-difference evaluation, and propensity scoring for bias adjustment. We illustrate that this framework allows models to estimate intervention-related changes in outcome trends after intervention and in comparison with controls. For medication-assisted treatment, we find that providers with lower baseline rates had an accelerated uptake following ECHO training, while those with higher baseline rates had no change in medication prescription rates. We provide recommendations on study design, metrics, and statistical modeling to implement this framework.