Behavioral Risk Identification and Decision Guidance for HIV Engagement

Despite much progress towards reaching UNAIDS goals for HIV treatment, disengagement from care remains one of the biggest obstacles to HIV elimination. In many African countries, including South Africa, the early treatment period (first six months after starting or re-starting antiretroviral therapy (ART)) is the interval with the highest risk of treatment interruption and disengagement, but until recently, this period had received relatively little attention from researchers and policy makers. The Retain6 project was launched in 2021 to generate data about the early treatment period, propose potential improvements to guidelines and models of care, and conduct preliminary tests of some potential interventions. Since then, Retain6 has collected detailed data about, among other topics:

* Patient-level preferences, barriers, and facilitators of care during the first six months to help predict interruption risks and develop relevant interventions (PREFER survey, protocol M220440 (Wits HREC), H-42726 (BU IRB) * Predictors of risks of treatment interruption during the early treatment period and characteristics of patients most likely to experience interruptions, with an improved risk assessment approach for use at primary clinic level (PREDICT model, protocol M210472) * Interventions currently available at primary healthcare clinics to improve engagement in care, to match interventions to risks * Current facility compliance with South Africa's 2023 HIV treatment guidelines (FIRST-HIV, protocol 250409) * Provider views of retention in care and potential ways to improve it, to understand facility barriers to implementing guidelines (FIRST-HIV, protocol 250409) * Effectiveness of tracing interventions following an interruption, to optimize tracing procedures once disengagement occurs (GREAT-South Africa, protocol M2409113).

Under the proposed BRIDGE (Behavioral Risk Identification and Decision Guidance for Engagement) protocol, the investigators aim to synthesize the data listed above to create and test a client retention toolkit comprising a package of targeted, light-touch interventions that aim to improve outcomes during the early treatment period. The retention toolkit is expected to include:

* An updated adherence risk/vulnerability assessment tool for use by clients and providers during consultations to identify clients vulnerable to disengagement from care * A "menu" of available interventions that can be matched to identified vulnerability factors for disengagement and allow patients to choose what they believe will be most effective for them * A "treatment roadmap" to help new and re-starting ART patients understand and adhere to treatment schedules * A WhatsApp based tool, AI Coach, developed by external colleagues to provide confidential and empathetic support and information after treatment initiation. * An improved process for identifying patients eligible for tracing and monitoring tracing results. * A co-designed, user-friendly checklist/job aid to support providers in adhering to the 2023 ART guidelines during early treatment.

This package is currently being co-designed with key stakeholders, including the National Department of Health, and is intended to be implemented within the existing systems and require minimal additional resources. On completion of the co-design process, the investigators will conduct a pilot assessment of the retention toolkit to describe acceptability, feasibility, uptake, and preliminary impact on near-term outcomes (attendance at next scheduled visit) at a selected set of primary healthcare clinics in South Africa. The study procedures will combine secondary analysis of de-identified medical record data and qualitative data collection among patients and providers. Data collection will take place during the first half of 2026, with a waiver of consent requested for medical record data and written informed consent for qualitative data collection. The maximum total sample size for the medical record data will be 30,000 patients and for the qualitative data collection will be 180 patients and 90 providers.