Tailoring community antiretroviral therapy delivery to the needs and preferences of people living with HIV in refugee settlements in Uganda

Abstract

Ending the HIV epidemic as a public health threat by 2030 will require equitable progress across all subpopulations and contexts, including those affected by displacement and humanitarian instability. In refugee settlements in Uganda, engagement in care among people with HIV (PWH) remains below global targets, despite the availability of free HIV services, including antiretroviral therapy (ART). This shortfall reflects the numerous structural and social barriers that humanitarian populations must overcome to access care, including stigma, limited social support, long distances to clinics, inclement weather, underdeveloped road infrastructure, constrained livelihood opportunities to offset transportation costs, and competing survival needs. Community ART delivery, a differentiated service delivery model (DSDM) offered in Uganda, including in refugee settlements, moves HIV treatment out of health facilities and into the community with the goal of improving care engagement while reducing the burden on health systems. While community ART delivery has demonstrated effectiveness in improving retention and viral suppression in other settings across sub-Saharan Africa, refugee settlements represent a fundamentally different service environment. Evidence is lacking on how community ART delivery uptake and implementation are shaped by the unique context and what adaptations may be necessary to ensure its impact. The objective of this dissertation was to generate actionable evidence on what enables and motivates participation in community ART delivery among PWH in refugee settlements, and on the contextual factors shaping implementation feasibility for providers. The long-term goal of this research is to generate findings that will ultimately inform the design of effective, person-centered DSDMs for HIV in humanitarian settings. This work was carried out through three interconnected studies. In the first study, 34 in-depth interviews (IDIs) with PWH and seven focus group discussions (FGDs) with 51 HIV care providers were conducted across five refugee settlements in midwestern and southwestern Uganda (October 2024 – May 2025). Barriers to participation in and implementation of community ART delivery were explored within IDIs and FGDs, as well as strategies to address these barriers, and perspectives on ideal model design, guided by the Consolidated Framework for Implementation Research (CFIR) 2.0. Using rapid qualitative methods, barriers were mapped to CFIR constructs and proposed strategies to Expert Recommendations for Implementing Change (ERIC) strategy clusters, with findings synthesized into causal pathway diagrams. Barriers to community ART delivery participation and implementation were closely intertwined. PWH and providers described barriers related to both motivation and opportunity that were shaped by local conditions, local attitudes surrounding HIV, model characteristics, weak relational connections within ART groups, poor teaming, limited resources, lack of incentives, inadequate work infrastructure, and communication challenges. Key strategies to address these barriers that were proposed by participants included changing infrastructure, utilizing financial strategies, adapting and tailoring community ART delivery to the context, developing stakeholder interrelationships, and engaging consumers. Rather than endorsing a single model, participants highlighted the need for a flexible mix of community ART delivery options. Confidentiality, flexibility, and reliability were identified as hallmarks of ideal community ART delivery. In the second study, IDIs and FGD participants prioritized the barriers to community ART delivery participation and implementation that they had identified through structured ranking exercises. In IDIs, participants were asked to rank their first, second, and third most important barriers to community ART delivery participation. In FGDs, implementation barriers were prioritized using the nominal group technique, which consisted of individual rating of perceived barrier importance, frequency, duration, impact on equity, and addressability using Likert scales—followed by structured group discussion and voting. Participation and implementation barriers were prioritized using mean ranking and frequency of mention, and multi-dimensional priority was examined using go-zone analysis. Stigma and structural health system constraints were the highest-priority barriers to community ART delivery. Stigma-related concerns—including involuntary disclosure, anticipated stigma, and internalized stigma—were consistently ranked among the most important barriers to participation and were also identified as major challenges to implementation. Structural barriers included lack of suitable community ART delivery sites, geographic distance, transport limitations, and human resource shortages. Barriers related to community ART delivery site infrastructure were perceived as highly impactful and relatively addressable, whereas stigma and human resource constraints were considered less readily addressable. Transport and staffing limitations were described as more recent challenges in the context of funding and service delivery transitions. In the third study, 250 PWH receiving ART from 12 health centers in refugee settlements in midwestern and southwestern Uganda were offered participation in a discrete choice experiment (DCE) to quantify community ART delivery preferences that was designed using Sawtooth Lighthouse Studio and was informed by prior qualitative findings. Participants completed 8 choice tasks, each consisting of two hypothetical community ART delivery model alternatives and an opt-out option. Alternatives were defined by delivery location, delivery structure (group versus individual), provider type, dispensing interval, delivery timing, and service duration. Preferences were estimated using hierarchical Bayesian multinomial logit models. Additional non-DCE questions assessed willingness to pay and willingness to participate by group size. Findings revealed participant preferences were driven by delivery structure (relative importance 27.5%; 95% credible interval [CrI] 24.0%, 31.1%), delivery location (24.9%; 95% CrI 21.2%, 28.5%), provider type (22.0%; 95% CrI 17.5%, 26.3%), and the amount of ART dispensed at each refill (15.9%; 95% CrI 14.0%, 17.8%). Participants preferred group delivery with social support and income-generating activities (0.67; 95% CrI 0.48, 0.87), home delivery (0.41; 95% CrI 0.24, 0.60), delivery by health workers in uniform (0.32; 95% CrI 0.11, 0.53) and expert clients (0.22; 95% CrI 0.02, 0.42), and six-month dispensing (0.54; 95% CrI 0.43, 0.65). Individual delivery, delivery at non-healthcare community locations, and delivery by Village Health Team members were disliked. The opt-out alternative had strongly negative utility (−3.94; 95% CrI −4.52, −3.42) suggesting strong preference for community ART delivery. Limited evidence was found for clearly defined subgroups with distinct preference patterns. Willingness to participate was higher for community ART delivery models with small groups (82%) than large groups (46%), and 55% of participants reported willingness to pay for home delivery. Together, these three studies advance understanding of how community ART delivery functions in an understudied humanitarian context at a time when global displacement is rising and DSDMs are being scaled up across sub-Saharan Africa. By integrating qualitative inquiry, structured prioritization, and quantitative preference elicitation, this dissertation moves beyond cataloguing barriers to elucidate how structural, social, and organizational factors converge to shape the capacity of community ART delivery to reduce time and transport burdens for PWH in refugee settlements in Uganda. It distills priority targets for implementation optimization and identifies the model features most valued by participants. By centering the perspectives of PWH and applying rigorous implementation science methods, this research offers a stakeholder-informed framework for tailoring differentiated service delivery in resource-constrained humanitarian settings. In doing so, it provides actionable guidance to strengthen DSDM scale-up among displaced populations and advances equitable, person-centered HIV service delivery toward ending the HIV epidemic as a public health threat.