Exploring Fit in a Mobile Health Intervention for Alcohol Use Disorder: Qualitative Study.

Background: Implementation frameworks such as the Exploration, Preparation, Implementation, Sustainment model emphasize the importance of the fit between an intervention and its context, which includes the needs of its target population, as well as the culture, resources, and capabilities of the implementing organization. Although lack of fit is a major barrier to implementation, fit has not often been a focus of implementation research. This paper uses fit as a lens to examine the implementation of Tula, a mobile health app aimed at reducing risky drinking days among individuals meeting the criteria for mild to moderate alcohol use disorder, in a 3-arm (app alone, app plus peer mentoring, and app plus health coaching) randomized controlled trial. Objective: We sought to better understand the trial results and to provide actionable guidance for future implementation of digital health interventions in health care organizations. Methods: Semistructured interviews with 18 trial participants and 7 Tula implementers were conducted. Trial participants were pulled equally from each arm of the trial and represented participants who demonstrated both high and low engagement with the app. Implementers consisted of a project manager, 4 peer mentors, and 2 health coaches. Interviews with participants focused on their motivations, opinions, and experiences of the intervention and their perception of their drinking behavior following the intervention, including how their use of the app worked to change that behavior. Interviews with implementers were centered on their roles, theories of change, perceptions of intervention, and areas for improvement. All interviews were analyzed using rapid qualitative analysis with deductive and inductive components. Results: We identified areas of both fit and misfit. For example, there was a good fit between implementers' theories of change and participants' description of how change occurred. Fit was improved by the versatility of the app, which allowed participants to customize their experiences. Conversely, misfit was noted in the app's inability to cultivate connection for many participants and a disjunction between the role of peer mentors in the intervention and their broader professional ethos. Conclusions: Focusing on fit provides a useful guide to enhance future iterations of the Tula app that lead to better sustainment of the intervention.