Longitudinal evaluation examining implementation and sustainment of an opioid overdose education and naloxone distribution among Veterans who are unstably housed
Abstract: Background: Rigorous implementation evaluations are needed to understand factors that influence implementation and sustainability of evidence-based interventions across contexts. In this study, we conducted a longitudinal, multi-methods, multi-site evaluation guided by the Dynamic Sustainability Framework (DSF). This evaluation focused on implementation of the Homeless Overdose Prevention-Expansion (HOPE), an opioid overdose education and naloxone distribution (OEND) trial in a permanent supportive housing program in the Veterans Health Administration (VA). Methods: We used a multi-methods study design comprised of qualitative interviews and completion of a three-item survey. Semi-structured interviews were completed with Department of Housing and Urban Development-VA Supportive Housing (HUD-VASH) staff, site leaders, and site prescribers at four VA healthcare systems in the Western United States. Interviews were conducted at three timepoints: pre-implementation, implementation, and sustainment. Site staff also completed the Provider REport of Sustainment Scale (PRESS) during sustainment to provide more context for our interpretation of results. We analyzed interview data using rapid directed content analysis guided by DSF constructs and analyzed PRESS using descriptive statistics. Results: We conducted 96 interviews with 67 unique individuals. Six determinants influenced the reciprocal fit of the intervention, practice setting, and ecological system across our study: (1) OEND for unstably housed veterans; (2) Staffing shortages and competing demands; (3) Training-related concerns; (4) Supervisor and leadership buy-in at site; (5) Social workers' scope of practice; and (6) Cultural climate and saliency of OEND initiatives. Both planned and unplanned implementation strategies were optimized by different actors (i.e., the implementation team, target practitioners) in response to evolving determinants at all three DSF levels to maintain overall fit. Approaches taken to optimize fit depended on longitudinal data collection and evaluation of determinants at each phase. Conclusions: Our approach is a theoretically driven example of capturing important determinants of implementation and sustainment of OEND in a high-risk setting. Despite the uniqueness of our study setting, our approach is generalizable and has the potential to promote sustainability of other public health interventions.