A process evaluation of Community-led Distribution of HIV self-testing kits in rural communities in Zimbabwe

Community-led distribution of HIV self-test (HIVST) kits has potential to combine the benefits of community-based HIV testing with the strengths of community-led interventions. We report the process evaluation from the community-led arm of a cluster-randomized trial in rural Zimbabwe comparing community-led versus community-based HIVST distribution where distributors were paid. Twenty communities were supported to design HIVST distribution models that suited their context. We conducted community meetings where we introduced community-led HIVST, including messages on the benefits of viral load suppression (Undetectable=Untransmissible; "U=U"). Communities determined how, and by whom, distribution would be done. Communities implemented their distribution model over 4-6 weeks. We conducted in-depth interviews with community members (n=20), distributors (n=20) and health workers (n=20) as well as 12 focus group discussions (n=91) with community members. We triangulated findings with participant observations during community meetings and implementation and descriptively analysed program data on HIVST distribution. Qualitative data were analysed thematically. Participant observations found communities collaboratively developed and embraced HIVST distribution, demonstrating strong enthusiasm for participation. Messaging on U=U was met with interest and surprise, as many community members were unaware of this information, and became a key motivator for HIVST implementation and uptake. Across communities, 348 distributors distributed 27,812 kits door-to-door and/or from their homes or other community venues. Communities where model development was led by local leaders - rather than collective processes- expressed lower level of satisfaction and had slow implementation. Health workers supported HIVST distribution well, with opportunities for improvement noted. While distributors felt proud to deliver HIVST, many experienced logistical burdens, and need for material support from the programme and community. Communities designed HIVST delivery models that were acceptable and feasible, however efforts to make community-led models sustainable are critical. Delivering messages on U=U should continue to be prioritized and embedded within community-based HIVST distribution models.