Peer support lowered self-stigma, but it did not cleanly explain the trial’s effect
A mixed-method analysis from Zimbabwe has taken a closer look at HIV prevention and treatment delivery for female sex workers, asking a question that matters far beyond a single trial: through what social mechanisms do community-led programs actually work? The AMETHIST intervention combined peer-led microplanning with self-help groups, and the parent trial had already shown better HIV treatment outcomes. This analysis focused on two proposed drivers of change — self-stigma and social cohesion — to see whether either sat on the pathway between the intervention and improved service use.
The answer was mixed. Qualitative interviews and group discussions described a setting in which peer support could make clinic engagement feel less exposing and more possible. Women spoke in ways that linked group participation, contact with microplanners, and easier access to HIV services. Quantitative data, however, were less persuasive on causality. Lower self-stigma appeared in the intervention arm, while social cohesion was already high across both arms and did not separate neatly by study assignment or by participation in intervention activities.
A lower stigma signal emerged, but it was not enough to explain the whole intervention effect.
Why self-stigma matters in sex worker-focused HIV programs
For female sex workers, the barrier is rarely biomedical alone. Fear of judgment, prior mistreatment in health facilities, and the expectation of being blamed for one’s HIV risk can push care-seeking out of reach even when testing and treatment are available. In that context, self-stigma is not a vague psychological concept; it can shape whether a person walks into a clinic, discloses symptoms, returns for follow-up, or stays on antiretroviral therapy. Community-led programming aims to blunt those barriers by changing the social environment around care.
AMETHIST was built on that logic. Peer-led microplanning and self-help groups are designed to make the health system feel less hostile and more navigable. Peers can normalize care, translate clinic expectations into local realities, and reduce the isolation that often accompanies HIV risk and diagnosis. Self-help groups can do something similar by shifting experience from private burden to shared problem solving. The intervention’s success in the parent trial made the mechanism question especially relevant: if outcomes improved, what precisely changed in the lives of participants?
Qualitative accounts pointed to belonging and practical encouragement
Reanalysis of the qualitative material found narratives that connected intervention participation with reduced self-stigma and better uptake of services. Women described peer engagement and group membership as forms of support that could soften shame and make service use feel more acceptable. In practice, that sort of social movement matters because stigma often works silently. It does not always announce itself as a refusal of care; more often, it appears as delay, hesitancy, or withdrawal after a bad clinic experience.
The interviews and group discussions suggest that the intervention may have altered the emotional cost of seeking care. Peer supporters can help women anticipate what will happen at a visit, accompany them mentally if not physically through the process, and reinforce the idea that HIV services are for them as much as for anyone else. That social repair can be especially important in programs serving sex workers, where identity, labor, and health are tightly intertwined.
The self-stigma signal was real, but social cohesion was harder to move
Quantitative findings offered a sharper but narrower picture. Lower levels of self-stigma were seen in the intervention arm, with 58.6% versus 63.2% and an adjusted odds ratio of 0.77, with a 95% confidence interval of 0.61 to 0.86. The direction of effect fits the program’s theory: peer-led contact and organized support can reduce internalized shame enough to make service use more feasible.
Social cohesion told a different story. Levels exceeded 50% across the whole sample, but they were not associated with study arm and were not linked to engagement in intervention-related activities. In other words, cohesion was already present at a fairly high level, which blunts its ability to explain why the intervention improved HIV outcomes. If a construct is already elevated in both groups, there is less room for a detectable difference to emerge, even if the intervention is doing something important in daily practice.
Why the mechanism question remains partly unresolved
The central finding is not that community-led care failed to work through stigma or cohesion. It is that the measured variables did not map cleanly onto the trial’s clinical gains. Qualitative data suggested some women experienced the intervention as meaningful, supportive, and enabling. The quantitative analysis did not support either self-stigma or social cohesion as clear mediators of the intervention effect. For mixed-method research, that tension is familiar and informative: lived experience can point to a plausible mechanism even when survey instruments fail to capture it with enough sensitivity.
Measurement is likely part of the problem. Self-stigma and social cohesion are complex, relational constructs, and they may shift unevenly across time and setting. They can be real in clinical life yet difficult to capture in a survey snapshot. Endline respondent-driven sampling also samples a social world that may already be partially shaped by shared peer networks, which can compress differences between arms. In that setting, strong qualitative signal and modest quantitative separation can coexist without contradiction.
The intervention may have worked through social processes broader than the two variables tested here. Peer support can change expectations about care, improve navigation of services, and reduce the psychic burden of attending clinics repeatedly. Those effects can be clinically important even when they do not appear as large differences in a single scale score. In HIV care, adherence and retention often hinge on small changes in confidence and trust rather than on dramatic shifts in attitude.
What the AMETHIST data add to HIV service design
This analysis reinforces a practical point for programs serving female sex workers: community-led HIV care should not be judged only by service uptake numbers. The social architecture around care matters, and the program’s design likely helped create a safer route into prevention and treatment services. At the same time, the measured pathways were not strong enough to say that self-stigma or social cohesion alone accounted for the parent trial’s success.
The study was registered with the Pan African Clinical Trials Registry, and ethical approval came from the Medical Research Council of Zimbabwe, the London School of Hygiene & Tropical Medicine, and the Liverpool School of Tropical Medicine. The data are available on reasonable request. As a preprint, the findings remain provisional, and the unresolved question is not whether peer-led programs matter, but which social mechanisms can be measured well enough to show how they work in real-world HIV care.