The Coach Mpilo Study: Evaluation of a Peer-led Intervention to Promote Engagement in HIV Care for Men Living With HIV

Study setting The study will be conducted in KwaZulu-Natal (KZN), the province in South Africa with the highest adult HIV prevalence: 16%-24% (model estimates), 22% (survey estimates). The study will be conducted in about 13 primary healthcare facilities based in the eThekwini District.

Aim 1: Determine the effects of Coach Mpilo on retention in care among men living with HIV In partnership with the Department of Health in KwaZulu-Natal Province, eThekwini Municipality, The Health Systems Trust (HST) and Matchboxology (a South African organisation that designs health and social programs) a randomised controlled trial (RCT) will be conducted with men living with HIV who have dropped out of HIV care (i.e., having missed a clinic appointment by >28 days).

Recruitment: Study staff will be trained by HST as Linkage Officers. As Linkage Officers, they will contact men who have missed clinic appointments via telephone, encourage return to care, and ask to meet privately at local venues. At the in-person meeting the Linkage Officers will invite eligible participants to enrol in the study.

Consenting process: Written informed consent will be obtained.

Data collection Baseline survey: Study staff will administer a baseline questionnaire to obtain information on, among other things, participants' demographic characteristics, socio-economic status, health behaviour, future outlook, internalized stigma and mental health.

Follow-up survey: Study staff will contact participants at six months and administer an in-person follow-up survey; many of the questions administered at baseline will be assessed again.

Clinic records: Participants' health outcome data will be obtained from electronic clinic data via TIER.Net (South Africa's national electronic HIV/TB monitoring system), to assess individual- and facility-level outcomes related to HIV. Where necessary physical clinic files will be reviewed for confirmation. In addition, participants' data on laboratory tests relevant for HIV care (including viral load tests) will be requested from the National Health Laboratory Services (NHLS).

Coach Log Book: To collect data on the engagement between coaches and their clients, the coaches will maintain a log book of the dates, duration and mode (e.g., in-person, telephone call, WhatsApp messages) of contact with clients.

Randomization: Following the baseline survey, participants will be randomized (1:1) to a control group that receives the SOC and an intervention group that receives SOC plus the Coach Mpilo intervention.

Analyses: Primary analysis will be intention-to-treat using a logistic regression model that includes clinic fixed effects and month of enrollment, and controls for baseline participant characteristics that are significantly different between study groups. Similar analyses will be conducted for each secondary outcome.

Sample size and power calculations: With a sample size of 786 participants and 50% retention in care at 12 months, the study has 80% power to detect a difference of 10 percentage points between study groups. As a result, the study will aim to recruit about 800 participants in the Aim 1 trial.

Aim 2: Identify the populations of men who are not reached by and do not respond to Coach Mpilo In the intervention group, intervention uptake and amount of engagement participants have with a coach will be measured. Primary quantitative analysis will assess the determinants of uptake within the intervention group to advance the limited evidence on sub-groups that are typically missed by peer-led interventions and therefore require different types of support. Qualitative interviews will also be conducted to better understand program uptake and engagement with a coach.

Participants: Aim 2 will be conducted with data collected from coaches and from Aim 1 trial participants.

Follow-up procedures and data collection: In-depth interviews (IDIs) with participants in the intervention arm who decline a coach. The study will conduct about 15 IDIs with 1) men who decline the support of a coach to understand reasons they did not engage with a coach; 2) about 15 IDIs with men who, after initial engagement, decline the support of a coach, this will be in order to understand reasons they chose to end engagements with a coach; 3) about 15 IDIs with participants who receive support from a coach to understand engagement with coaches and elicit insights on reasons why the intervention may have worked for some participants but not others. IDIs will be conducted at about 6 months, when engagement with a coach will have ended.

IDIs with all coaches (about N=7) involved in the study will be conducted to gain insights on their engagement with participants.

Analyses: Primary quantitative analysis: Among Aim 1 intervention group participants analysis will evaluate factors associated with declining the offer of a coach using logistic regression, controlling for potential covariates. Qualitative data analysis: Data analysis will be iterative, and include both deductive and inductive approaches to thematic analysis.

Aim 3: Assess the cost-effectiveness of the Coach Mpilo program To inform strategic policy decisions about the scale-up of Coach Mpilo, and how it compares in cost and cost-effectiveness against current standard of care treatment, as well as other HIV interventions, we will conduct an economic evaluation. Data collection includes: cost-related data and resource utilization data to construct cost estimates for both arms. These data will be combined with retention in care data from Aim 1 to determine average costs per client retained in care and incremental cost per additional person retained. As part of the analysis, a sub-analysis will be conducted to adjust the cost estimates to reflect implementation in a routine care setting, excluding research-related expenses such as start-up and study-specific costs. This will provide more realistic national-level scale-up cost estimates for Coach Mpilo. Epidemiological modeling based on effectiveness estimates from Aim 1, will use the Thembisa model to estimate several long-term health outcomes, including life years saved and HIV infections averted due to onward transmission. These long-term health outcomes will form the basis to estimate the cost-effectiveness analysis of Coach Mpilo. To compare Coach Mpilo to other HIV prevention and treatment options in South Africa with the broader aim of informing strategic policy decisions, study estimates will be incorporated into the South Africa HIV Investment Case, a prominent initiative that assesses the relative cost-effectiveness of different HIV interventions and aims to inform government of the optimal mix of interventions that also accounts for allocative efficiency.

Analyses

The analyses will be conducted in three phases:

Cost analysis: a multiprong approach will be used that incorporates top-down and bottom-up costing methodologies to construct a cost for both standard of care and Coach Mpilo arms.

Cost-effectiveness analysis: both intermediate and long-term cost effectiveness analysis (using an HIV transmission model) will be conducted, comparing Coach Mpilo to standard of care.

Integration into the HIV Investment Case: the cost-effectiveness of Coach Mpilo to other HIV prevention and treatment options in South Africa will be compared to inform strategic policy decisions for the South African government.

Aim 4: Assess the acceptability, feasibility and fidelity of the Coach Mpilo program To inform intervention refinements and policy decisions on the scale-up of Coach Mpilo and peer-led interventions more generally, quantitative and qualitative assessment of implementation outcomes will be conducted alongside effectiveness and cost-effectiveness of the intervention. Implementation outcomes will be viewed as important for evaluation of complex health interventions in research and practice, providing indications into the implementation process as preconditions for attaining intended intervention engagement and health outcomes.

Sekhon's 2017 Theoretical Framework for Acceptability (TFA) will be used to assess the feasibility and acceptability of the intervention, informing interpretation of effectiveness overall (Aim 1), for sub-groups (Aim 2), and cost-effectiveness (Aim 3) by describing how acceptable the intervention is and how feasible the intervention is to deliver, from different stakeholder perspectives, and by describing the fidelity of the intervention.

The following data will be collected: i) primary quantitative survey data at baseline and six month follow-up (through Aim 1 survey instruments); ii) primary qualitative data from intervention arm participants that decline a coach, participants who are supported by a coach at six-month follow up, and coaches immediately after the intervention stops being implemented (through Aim 2 IDI instruments); iii) routine administrative data during intervention preparation and implementation.

Quantitative: At baseline, participants' affective attitude and values alignment will be assessed to understand prospective acceptability. In the follow up survey with intervention arm participants, the Acceptability of Intervention Measure (AIM)19, and the Client Satisfaction Questionnaire (CSQ-8) will be asked.

Qualitative: Follow-up in-depth-interviews described in Aim 2 include topics which answer our implementation research questions, focusing on retrospective acceptability

Coach log: an existing routine administrative data collection tool has been adapted to focus on implementation choices, their determinants, and coach-participant relationship quality.

Coach-to-Coach WhatsApp Group: the implementing partner routinely establishes a WhatsApp group used by Coaches and the Squad Manager (i.e. not including any clients/participants). Data from this group will include challenges Coaches are facing and solutions suggested by other coaches.

Monthly and quarterly implementation reports and follow-up discussions: these reports will be developed by Matchboxology, providing both quantitative data on reach and their own observations on fidelity, reach, mechanisms of change and implementation strategies and determinants.

Analysis: Quantitative analysis. Among Aim 1 intervention group participants, descriptive analysis of data from the AIM and CSQ-8 scales and linear regression analysis will be conducted to assess what factors at baseline were associated with these scales. Qualitative data analysis. Ongoing saturation analyses will be conducted, based on iterative coding during data collection. Data analysis will be iterative, and include both deductive and inductive approaches to thematic analysis.