Differentiated Prevention And Care To Reduce The Risk of HIV Acquisition And Transmission Among Female Sex Workers In Zimbabwe: Study Protocol For The ‘AMETHIST’ Cluster Randomised Trial

doi:10.21203/rs.3.rs-475222/v1

Background: Female sex workers (FSW) in sub-Saharan Africa are disproportionately affected by HIV and are critical to engage in both HIV prevention and care. Here we describe our approach to evaluation of the AMETHIST intervention for FSW, adapted to the African context and set in Zimbabwe where there is a nationally-scaled programme for sex workers (Sisters). We hypothesise this intervention will raise uptake and adherence to prevention and treatment among FSW resulting in a reduction in their risk of HIV acquisition/transmission.

Methods: The AMETHIST intervention (Adapted Microplanning to Eliminate Transmission of HIV in Sex Transactions) aims to provide risk differentiated prevention and care to FSW through microplanning and self help groups. Twenty two towns with the Sisters programme were randomised using restricted randomisation to either the Sisters programme or the Sister programme plus AMETHIST. The primary outcome is composite and is defined as the proportion of HIV positive FSW at risk of HIV transmission combined with the proportion of HIV negative FSW at risk of HIV acquisition. The outcome will be assessed after two years of intervention delivery in a respondent-driven sampling survey (total n=4400; n=200 FSW recruited at each site). Primary analysis will use the ‘RDS-2’ method to estimate cluster summaries and will adapt Hayes and Moulton’s ‘2-step’ method to adjust effect estimates for individual-level confounders and further adjust for cluster baseline prevalence. An in-depth process evaluation guided by our project trajectory will be undertaken.

Discussion: Innovative pragmatic trials are needed to generate evidence on effectiveness of combination interventions in HIV prevention and treatment in different contexts. We describe the design and analysis of such a study.

Trial Registration: The trial was registered at Pan African Clinical Trials Registry (PACTR202007818077777) on 2 July 2020.

Translational Medicine

Internal Medicine

Integrative & Complementary Medicine

Cluster randomised control trial

effectiveness

hidden population

pragmatic trials

respondent driven sampling

sex workers.

Effective tools for preventing and treating HIV are now widely available[1]. Yet globally gaps in implementation remain, especially among vulnerable populations[1, 2]. Eastern and Southern Africa is home to approximately half of all people living with HIV and just under half of all HIV-related deaths in 2019 [3]. Only 59% of those living with HIV in the region are virally suppressed [3] [4, 5]. Female sex workers (FSW) make up a small proportion of the population [6, 7] but bear a disproportionate burden of infection. The risk of HIV acquisition is up to 21 times higher for FSW than non sex working women aged 15–49 years [4, 8]. Further, sex work is directly or indirectly a cause of a high proportion (40%-80%) of all new infections [9] [4-9]. Implementing strategies that improve access, uptake and effective use over time of HIV prevention and treatment tools among FSW is needed urgently.

UNAIDS/WHO recommend targeted sexual and reproductive health services (including HIV related) for FSW, supported by peer-based community outreach. However, the optimal delivery model for such services in southern African contexts remains unclear. Effectiveness studies, especially randomised controlled trials (RCT), remain rare. In Zimbabwe, we previously conducted a cluster-RCT of an intervention comprising enhanced access to regular HIV testing, on-site initiation of ART, access to pre-exposure prophylaxis (PrEP), adherence interventions and intensified community mobilisation compared to the ‘standard’ Sisters programme. The intervention improved rates of HIV diagnosis and treatment initiation. Viral suppression improved in both arms of the trial, but there was no significant difference in the proportion of all FSW with unsuppressed HIV infection between the arms[10].

Building on our previous work we have developed “AMETHIST” (“Adapted Microplanning: Eliminating Transmissible HIV In Sex Transactions”), a new intervention package with a greater focus on HIV prevention to complement the existing focus on HIV testing and treatment, and sexual and reproductive health more broadly. We hypothesise that this theory-based intervention, combining microplanning and community empowerment enhanced by introduction of self-help groups (SHG), can raise uptake and adherence to HIV prevention and treatment tools among FSW and thereby reduce the proportion of FSW at risk of acquiring or transmitting HIV infection.

Here we formally describe the “AMETHIST” intervention and its intended mechanisms of action and describe the design and analysis strategy for a pragmatic cluster randomised trial to determine the impact of the intervention on HIV transmission risk in sex workers.

Interventions and their intended mechanisms of action

A structured description of the AMETHIST intervention is shown in Table 1. It incorporates key elements of both the TIDIER framework [11] (elements 1-9) for intervention description and replication, and the Proctor framework for specifying implementation strategies [12]. Elements 10 and 11 of TIDIER relate to post-hoc assessment of modifications and fidelity of intervention delivery in practice and are more suitably reported at the end of the trial.

Standard of Care: the Sisters with a Voice programme

The Sisters with a Voice (“Sisters”) programme was established in Zimbabwe in 2009 on behalf of the Ministry of Health and Child Care (MoHCC) and the National AIDS Council (Table 1).[13] It operates nationally within 57 primary care clinics and provides comprehensive services in line with WHO guidelines [14]. Sisters provides free condoms and contraception, provider-initiated HIV testing and counselling, HIV self-testing and counselling (and secondary distribution of self-test kits for partners), syndromic management of sexually transmitted infections (STIs), health education and legal advice supported by a network of peer educators. Additionally, clinics provide long-acting reversible contraception (implants), referral for cervical cancer screening and on-site access to PrEP. The programme is supported by over 300 peer educators who mobilise FSW to attend clinical services and encourage uptake of HIV testing and support referral of FSW for antiretroviral therapy (ART) for those who test positive for HIV to government services for HIV care/ART initiation. Programme data are collected electronically in real-time.

AMETHIST intervention

In the AMETHIST intervention communities we will introduce “microplanning” and establish SHG (Table 1). Through these components, the AMETHIST intervention seeks to strengthen FSW self-esteem, social support, and group capacity [15, 16] as well as their motivation and opportunity to access services [17]. AMETHIST focuses on peer empowerment and community-level self-efficacy to reduce entrenched barriers to the adoption of behaviours that decrease HIV transmission (Figure 2).

Peer-led microplanning aims to optimise programme coverage and support prevention and care uptake[18]. While it has been successful in India [19, 20] as a critical element of HIV prevention programming, it has not been widely implemented or evaluated in Africa. FSW peer educators called “empowerment workers” (EW), will be trained to identify geographic sex work ‘hotspots’, enumerate local FSW within those locations, assess their vulnerability, and tailor support to each FSW’s level of social and clinical vulnerability. Vulnerability to HIV acquisition/transmission risk will be assessed with a simple algorithm based on age, duration in sex work, client numbers, condom use consistency, alcohol use and experience of violence. EW will reassess their case load at 3 monthly intervals (Table 1).

FSW assessed as being at high risk will be followed up weekly, those at medium risk twice a month, and those at low risk monthly. Importantly, microplanning is “HIV status neutral”: EWs do not need to know or ask the HIV status of FSW. At each microplanning contact, EW provide support to individual FSW guided by data (e.g. revisiting previous discussions, reminding them about upcoming clinic appointments or ensuring they have an adequate condom supply) and remind women about clinic appointments. Microplanning generates data for subsequent contacts. EW update the data, help analyse it, and work with outreach worker (ORW) supervisors weekly to plan whom to see and topics for discussion. Each EW has a caseload of approximately 50-80 FSW she sees at least once a month (each ORW has a caseload of 7-10 EW). Geographic hotspots will be remapped every six months and the size of the sex work population re-estimated so that the rate of programme coverage can be monitored.

Self-help groups aim to build social cohesion and community empowerment. Each EW invites some of the FSW in her microplanning caseload to join a SHG comprised of 10-15 women who meet twice each month. FSW are invited to participate in SHG based on their level of vulnerability and their willingness/ability to attend meetings regularly. The EW facilitates the SHG, initially with the support of an ORW. FSW receive training on how to run a group, keep a register, manage money (including how to budget, save, open and manage a bank account), and keep minutes if required. Each SHG is assisted to identify priorities and take relevant action. Training is provided by the programme to support SHG with addressing their priorities. Over time, these groups strengthen programme ownership and community empowerment, to reduce HIV vulnerability and increase uptake of services among FSW[15, 21]. Although not all FSW will join the initial SHG, they provide a model of community leadership that goes beyond Sisters’ standard model of community mobilisation that relies on health-related workshops with a less interactive or sustained format.

The components of the AMETHIST intervention that occur outside of formal clinical settings are delivered in such a way as to be ‘HIV status neutral’. While the specific technologies and pathways of care for prevention and treatment differ, there are important similarities. Social and structural factors such as stigma, lack of social cohesion and the low socioeconomic status of female sex workers act as barriers to the uptake and effective use over time of both prevention and treatment tools. Condom use offers “dual protection” from both HIV transmission and acquisition, and consistent condom use is strongly recommended for all FSW. Antiretroviral medications prescribed to be taken daily, albeit in different formulations and under different clinical supervision approaches, form a component of both treatment (as ART) and prevention (in the form of PrEP). The AMETHIST intervention seeks to reduce the risk of both HIV acquisition and transmission among FSW by targeting these common drivers, thereby reducing the role of commercial sex work in HIV transmission more broadly. Furthermore, a “HIV status neutral” approach reduces a source of tension and mistrust between sex workers and retains a focus on the priorities and concerns they share as a community.

Process Evaluation and Costing

Guided by the intended mechanisms of action (illustrated in the Project Trajectory) and description of the intervention components provided here, we will undertake a nested process evaluation which will draw on both quantitative programme data and qualitative assessments to monitor delivery and uptake of each programme component over time, examining ease of delivery, varying rates of participation by FSW and their experiences, perceptions and satisfaction levels. We do not describe this data collection in any further detail here but provide details in the trial protocol (see Appendix).

We will also determine the cost of the intervention components and the cost–effectiveness of the AMETHIST intervention, taking into account the potential reduction in transmissions via selling sex due to the reduction in the proportion of FSW who are at risk of either HIV transmission or acquisition. For this, we will use an existing individual-based dynamic stochastic model to predict the potential impact of this intervention on HIV incidence in the population of Zimbabwe. These methods are not discussed further here.

Trial design

Cluster randomised controlled trial.

Cluster definition and selection

A cluster is defined as the FSW population working in the geographic location (usually a town or business centre) where there is a government health clinic providing dedicated FSW services through the Sisters programme. Trial sites were purposively selected to be reflective of a range of settings, of adequate size to ensure participation of between 175 and 475 FSW annually (based on 2017 programme data) (mean 314; 50% seen for first time) and located at geographic spacing sufficient to ensure that the risk of contamination/spill-over of intervention effect between study clusters through FSW mobility and migration will be minimised. Sites are named in the protocol (see Appendix).

Randomisation

Twenty-two sites have been randomised (1:1) to receive the AMETHIST intervention in addition to standard of care or to continue with the standard of care alone. Randomisation was conducted in January 2019 at a public meeting with key stakeholders, MoHCC, district representatives, and representatives of female sex worker community from the 22 sites.

To minimise imbalance between arms, restricted randomisation was used[22]. Restriction factors included province, number of FSW seen in the Sisters programme in 2017, mean age of first-time attenders, proportion of FSW under 20 years of age, proportion of all attendees aware of HIV status, proportion of all HIV positive attendees on ART, and mean number of visits by attendees. Restricted randomization generated a list of 705,432 potential combinations from which 999 with the closest balance between arms were selected (undertaken by trial statistician STC). Non-investigator attendees were invited to withdraw numbers from a bag to see which one of the 999 combinations was selected.

Primary endpoint

The primary end point of the trial is a composite endpoint intended to capture the proportion of all FSW who are at risk of either HIV acquisition or HIV transmission as measured in a cross-sectional survey after 2 years of implementation delivery. Table 2 below shows how FSW primary endpoint status will be assessed. At endline, all participating women will be identified with a binary endpoint measure: as either at risk of acquisition/transmission or not at risk of acquisition/transmission. This means that the primary endpoint will include FSW who are both HIV positive (‘not at risk’ left side of table) and HIV negative (‘ not at risk’ right side of table).

The determination of the primary endpoint incorporates data on four ‘domains’. First, is their HIV status. Second, HIV positive women are identified as either virally suppressed or not. Third, HIV negative women are identified as either being on oral PrEP, or not. This is done first through self-report (where possible cross verified with programme data), and then, for those who report that they are on PrEP, through requesting and analysing a sample for analysis of PrEP drug levels. Only those both reporting and having biological evidence of the presence of PrEP will be considered as ‘on PrEP’. Fourth and finally, among both HIV positive and HIV negative women, women will be classified as having had recent (in the past month) condomless sex unless she reports otherwise and there is not biological evidence to contradict this. This determination will incorporate three types of information: first, women will be asked to self-report if they have had any condomless sex in the past 1 month or not. For those who report no condomless sex, we will ask if they are willing to provide a sample for detection of Y chromosome, a biological marker of unprotected sexual activity with male partners[23]. In addition, all women will be offered testing for bacterial STIs. We will consider women as having been validated as having had no recent condomless sex if she reports no condomless sex and has no evidence of either bacterial STI or in her vaginal sample. Relevant data collection, questionnaire and laboratory methods are described in more detail below.

The trial protocol also pre-specifies a range of secondary outcomes (see Appendix).

Inclusion and exclusion criteria for outcome assessment

Women are eligible for outcome assessment if they are: aged 18 or older; currently working as a sex worker (has exchanged sex for money in the past 30 days); living or working in the study site (for at least 1 month).

Data collection: Population sampling

The trial outcome will be assessed after 24 months of intervention using a respondent driven sampling (RDS) survey in each of the trial clusters. Initial “seed” women who meet the study inclusion criteria will be selected after mapping sex work in that location [24]. Selection as a seed will not be dependent on having accessed the Sisters programme. Each seed will start a recruitment “chain” by recruiting up to two peers into the survey. Each subsequent respondent will be further provided with coupons to give to up to two peers to refer them into the study. Participants will receive financial compensation for completing the survey themselves, as well as for each of their recruits who participates. Recruitment will stop when approximately 200 women have been recruited in a given site (see sample size justification below).

Data collection: Questionnaires

Questionnaire data will be collected through self-administration directly onto tablet computers using an audio computer assisted survey instrument (ACASI). The questionnaire will include questions on demographics, sex work, sexual behaviour and condom use, HIV testing history, ART use (including specific drug regimen), stigma, experience of violence, quality of life, mental health, substance use, general health, relationships with other sex workers, and use of sexual and reproductive health services. We will collect data to determine personal network size for RDS adjustment: we will ask how many females sex workers the participant knows who are aged over 18, live at the site, who she has seen in the last month, and who she would consider recruiting to the study. In addition, we will collect information from consenting women to allow us to link questionnaire data to Sisters programme data.

Data collection: Sample collection and laboratory analysis All women will have a finger prick sample collected for testing for HIV infection, viral load testing, drug resistance testing, ART level testing, and syphilis (as appropriate). Samples will be tested according to the Zimbabwe National HIV testing algorithm with samples tested in series. The syphilis sample will be tested using DPP® Syphilis Screen & Confirm Assay (Chembio Diagnostic Systems, Inc. New York, USA), a near patient test that tests for both Rapid Plasma Reagin (RPR) and Treponema Pallidum Haemagglutination Assay (TPHA) in a single test in 20 minutes. Results of HIV and Syphilis rapid tests will be returned to all participants on site. All women will have two dried blood spot samples (DBS) collected for HIV viral load testing, LAg avidity assay testing (for incorporation into a recent infection testing algorithm)[25] at the Flow Cytometry Laboratory, Harare. Results of viral load tests will be made available to women within 4 weeks of the survey. DBS samples will be shipped to University of Cape Town for ART testing to assess presence of ART broadly in HIV positive FSW and of Tenofovir-diphosphate (TDF-DP) in HIV negative women reporting use of PrEP (PrEP adherence defined as ‘high’ if Tenofovir-diphosphate (TDF-DP) ≥ 700 fmol/dDBS punch, and ‘low’ if < 350 fmol/punch.))[26]

In addition, all participants will be asked to provide a self-administered vaginal swab to be tested for STIs. All STI samples will be transported to Flow Cytometry Laboratory (FCL) within 48 hours at 4^oC. Samples will be tested by the FCL using Xpert® TV and Xpert® CT/NG (Cepheid Inc Sunnyvale, California USA). Results of STI tests will be made available to women through the Sisters clinic within 4 weeks of the survey and free treatment will be made available and contact tracing of sexual partners offered.

Sample size and power considerations

The sample size of the trial was based on pragmatic considerations as well as statistical power. We chose to include 22 clusters and to recruit approximately 200 FSW in each cluster. It was not practically possible to include more than 22 clusters from across the country in the trial, and prior work has suggested that it should be possible to recruit 200 FSW in each included cluster, but in some clusters not more than this. We used restricted randomisation to minimise differences by arm. We used the approach of Hayes and Bennett for cluster randomised trials [27] to consider sample size and power. We estimated that 30% of FSW would meet our primary outcome definition. In a 2016 [10] survey, 22% of FSW were at risk of acquisition/transmission of HIV by a measure that did not include verification of self-reported consistent condom use, while in another 2017 survey 28% of FSW were at risk of acquisition/transmission of HIV using the same definition (PrEP was not available at these sites at the time of the survey). In our previous trial, k was 0.17 so we estimated sample size using both a k of 0.2 and 0.25. In most scenarios, we estimate 90% power to detect a 30% difference in the proportion of FSW who are at risk of HIV acquisition/transmission between the intervention and control arms (Table 3). If k is 0.25 we have 78% power to detect a 30% difference and over 99% power to detect a 50% difference.

Statistical Analysis

Our approach is based on our previously published approach to the analysis of data from the SAPPH-IRe trial[10, 28]. One substantive difference between the trials is that in the current AMETHIST trial we were unable to conduct RDS surveys at baseline in any of the 22 clusters because funding for the AMETHIST intervention and its evaluation were secured separately; the intervention was started prior to research funding being finalised.

Study profile

We will describe the characteristics of clusters recruited to each arm of the study using available programmatic data. Cluster drop-out is not expected but might occur if political or community acceptance for the research protocol is compromised during the trial duration. We will produce descriptive statistics to assess whether there was balance across the two arms in key socio-demographic and potentially confounding variables at the endline RDS survey to be conducted between July and October 2021.

For each arm of the study at endline, we will describe the range and mean size of the sample recruited through the RDS survey in each site, and construct participant recruitment trees. We will describe by arm, the range and cluster-mean of the number of women who do not forward recruit two participants and their understanding of the reasons for this. We will construct a trial profile diagram in line with CONSORT principles but adapted for our specific situation.

RDS diagnostics

We will conduct recommended RDS diagnostics for each site [29] and report our findings according to STROBE guidelines for the reporting of RDS surveys [30]. As in SAPPH-IRe, the community mobilisation activities might plausibly change the structure of the FSW social networks in the intervention sites. Consequently, the RDS sampling process that runs over these networks might be biased by trial arm. To investigate these biases we will: 1) compare self-reported network degree in the intervention versus usual care sites; 2) compare homophily/similarity between FSW and their personal networks, and between recruiters/recruitees; and 3) examine whether the place of recruitment, relationship between recruiter/recruitee, and motivation for recruitment differ between arms and compare time to recruit between arms.

Analysis principles

Our analysis will compare FSW recruited in each community without considering direct contact with the intervention components (i.e. an “intention to treat” approach). Data from individual FSW will be summarised for each cluster and we will express the intervention effect using a prevalence ratio, prevalence difference and associated 95% confidence intervals.

We will calculate cluster summaries using the RDS-2 methodology[31]. For unadjusted analysis, we will fit a linear regression model on the RDS-weighted cluster summaries, with a treatment dummy. We will use the p-value and confidence intervals for the coefficient of the treatment dummy.

Adjusted analysis will use the ‘two step’ method of Hayes and Moulton [32]. Age is specified a priori and will be included as a potential confounder that could affect the outcome, other factors will be included if they appeared imbalanced post-randomisation. To generate adjusted and RDS-2-weighted analysis, Step 1, we fit a logistic regression model for each site with the primary outcome as the dependent variable, age and other variables as independent variables. This model will be used to ‘predict’ outcomes for each woman. The predicted values will be multiplied by the inverse of the degree (RDS-2) to account for the sampling design, normalised for the cluster, and cluster summaries calculated. Step 2, we will divide the cluster summaries used in the unadjusted analysis (accounting for RDS only) by the adjusted cluster summaries from Step 1 to generate “residuals”. We will calculate the effects using the regression model described above with the residuals as the dependent variable. We will calculate the ‘residual’ risk difference (RD) by subtracting the predicted prevalence from the RDS-II weighted observed prevalence in each cluster. We will fit a linear regression model on the residuals to calculate the adjusted RD and 95% confidence intervals.

Sensitivity analysis

To investigate robustness of our primary effect estimate we will conduct sensitivity analyses in which we: 1) re-calculate the primary effect estimate using cluster-summaries unadjusted for RDS-2 methodology (without weighting). 2) re-calculate the estimated effect using the ‘Successive Sampling’ estimator (‘RDS-SS’), designed to avoid the with-replacement assumption in the random walk model. However, RDS-SS requires estimates of the population size of FSW to already be available [33]. We will also conduct RDS-SS analyses for a range of possible population sizes [33] and assess whether findings differ using these estimates.

We describe the intervention, rationale, trial design and analysis plan for a pragmatic cluster randomised trial, conducted within the Sisters programme in Zimbabwe, of a complex intervention combining microplanning and self-help groups to support the uptake of and adherence to HIV prevention and treatment interventions for female sex workers.

We have adapted approaches found to be effective in other contexts in building community capacity to reduce vulnerability to, and improve management of, HIV infection. First, microplanning formalises peer outreach activities into a series of specified activities that engage FSW at each stage, i.e. mapping hotspots, identifying local sex workers, conducting routine risk assessments, and collecting and analysing data with support for outreach workers[34]. This process allows for greater flexibility and local responsiveness of HIV programmes delivered at scale, and further consolidates FSW’s leadership and ownership over the programme [35]. Second, the microplanners, called Empowerment Workers in the AMETHIST intervention, are given further responsibility for bringing groups of 10-15 FSW together into self-help groups where they can discuss shared concerns, provide group support, and eventually take collective action to address identified priorities. Self-help groups have been shown to empower poor women from low and middle income countries financially, socially and politically [36]. Qualitative studies suggest that self-help groups work through building social cohesion and self-efficacy, which contribute to health-enhancing behavioural change at community level [37].

Both our trial design and analysis strategy have multiple strengths. AMETHIST is the first cluster randomised trial of a risk differentiated, need-based and targeted intervention for HIV prevention and treatment among female sex workers in Africa. The cluster randomised design facilitates the simultaneous estimation of the intervention impact on both HIV negative and positive women. The research participants contributing to primary endpoint analysis will be recruited through RDS surveys within trial clusters, as in our previous “SAPPH-IRe” trial. RDS reduces sampling bias and improves representativeness of “hard-to-reach” populations by limiting the number of referrals any one respondent can have creating “deep” rather than “wide” sample networks [31]. We have outlined our RDS implementation in advance, described potential biases related to the use of RDS design in practice, attempted to collect information vital to signal whether potential biases exist, and planned sensitivity analyses to quantify their impacts on our findings.

Our design has also limitations: a longer trial would have been preferred. To minimise contamination between FSW communities/spill-over effect we used 22 clusters, which might compromise the power of the study. However, our study is larger than previous clustered randomised trials of similar interventions in Africa[10]. We also sought to strengthen power by using restricted randomisation to maximise the chance of baseline balance.

Pragmatic trials that aim to evaluate the success and effectiveness of complex interventions pose particular challenges. One such challenge is how best to characterise the interventions under study and their hypothesised mechanism of action. This is essential so that interventions can be replicated, users of the research findings know what would be necessary to implement the intervention in other settings, and the generalisability of the findings can be considered. We combined aspects of the TIDIER and Proctor frameworks to provide a structured description of the intervention. In addition, we explicitly delineate each step through which we anticipate our intervention components will lead to measurable change, which we will test through a process evaluation conducted throughout the trial. Using mixed methods, we will formally document the intervention’s fidelity to design, feasibility of delivery, and acceptability to the intended target audience as well as contextual factors that may have affected its implementation and effectiveness.

Another challenge for our trial design has been in deciding on the most appropriate primary endpoint and how this will be measured. We have proposed a composite endpoint including data on four domains that together reflect the risk of a woman (either HIV positive or negative) being involved in HIV transmission/acquisition. The use of this composite endpoint poses both conceptual and methodological challenges. Composite endpoints are commonly used in clinical trials including in HIV,[38] with the most common reason being that this can increase statistical power. However, difficulties in interpretation are recognised. The multiple outcomes included within a composite endpoint may differ in their importance to clients and providers, may differ in their absolute frequency, and may experience different risk reductions as a result of the intervention, and all of these effects influence interpretation.[39] While we do not currently have available population level data among FSW data on the outcome as we will measure it in the trial, we do have similar data from previous RDS surveys, when either 22.0% [10] (no drug levels or verification of reported condom use) or 28.4% [7] (no drug levels but STI test results available) of FSW were positive for our primary outcome, which the intervention seeks to reduce. We anticipate the prevalence of the actual outcome will be higher if validation of self reports of PrEP and condom use are included. In both scenarios, approximately twice as many women who could be beneficially affected by the intervention are HIV negative than are HIV positive likely reflecting the stage of the epidemic in Zimbabwe where although there have been considerable gains in terms of treatment coverage, the rate of new infections has remained relatively static.[40] Optimising coverage and continuation of PrEP in AMETHIST sites has the potential to impact our primary outcome.

The key methodological challenge relates to the validity of self-reported data on sexual activity, condom use and uptake / adherence to medications. Our approach assumes that the likelihood that people will report not using PreP and/or not using condoms when they actually are doing so is low, and so we assign such people as at risk of HIV transmission on the basis of their self-reports. However, for those who report these protective behaviours, our approach is to collect additional information for these individuals to be classified as not at risk.

Trial Status

Protocol version 1.4 on 19 February 2020. Data collection from the sites will take place starting from July 2021 and will be completed 31 December 2021.

ACASI Audio Computer Assisted Survey Instrument

AMETHIST Adapted Microplanning to Eliminate Transmission of HIV in Sex Transactions

ART Antiretroviral Therapy

DBS Dried Blood Spot

EW Empowerment Workers

FSW Female Sex Workers

ORW Outreach Workers

MoHCC Ministry of Health and Child Care

PrEP Pre-exposure Prophylaxis

Sisters Nationally scaled programme for sex workers in Zimbabwe

SHG Self Help Group

RCT Randomised Control Trial

RDS Respondent Driven Sampling

RDS-SS Respondent Driven Sampling- Successive Sampling

Ethics approval and consent to participate

Ethical approval has been received for the trial from the Medical Research Council of Zimbabwe (MRCZ/A/2559), Liverpool School of Tropical Medicine (19-115RS), and London School of Hygiene and Tropical Medicine (19123).

Written, informed consent to participate will be obtained from all survey and process evaluation participants.

Consent for publication

Not applicable.

Availability of data and materials

The investigators and statistical team will have access to the final data set.

Competing interests

The authors declare that they have no competing interests.

Funding

The trial is funded through a Wellcome Trust Collaborative Award in Science (PI Cowan). The intervention is funded by the Elton John AIDS Foundation. The funding bodies are not involved in the design of the study and collection, analysis, and interpretation of data and in writing the manuscript.

Authors' contributions

FMC is the principal investigator and oversees all aspects of trial design and implementation, analysis, data interpretation and reporting.

FM is the Zimbabwe lead investigator and oversees day to day running of the AMETHIST trial and all field work in Zimbabwe and supports FMC to coordinate design, implement analyse & report on the trial.

STC Zimbabwe statistical lead and co designed analysis plan and with AS will undertake the analysis.

AS is the AMETHIST consortium statistical lead and co designed analysis plan and with STC will undertake the analysis.

JB Leads the process evaluation of the AMETHIST trial, with a focus on the design and application of qualitative methods.

RS Works with Sisters programme team to oversee the implementation of the AMETHIST intervention.

ND contributes to design, analysis and interpretation of the trial.

MS contributes to design, analysis and interpretation of the trial.

PR Oversees the design of cost effectiveness analysis and modelling.

AM contributes to design, analysis and interpretation of the trial.

YR contributes to design, analysis and interpretation of the trial.

OM contributes to design, analysis and interpretation of the trial.

ANP contributes to all aspects of design, analysis and interpretation of the trial and leads the modelling and cost effectiveness analysis, design of trial.

JRH Oversees design and conduct of statistical analysis and contributes to all aspects of design, analysis and interpretation of the trial.

All authors read and approved the final manuscript.

Acknowledgements

N/A – there are no additional people to be acknowledged who did not meet the criteria for authorship.

1. Geng, E.H., et al., Personalized public health: An implementation research agenda for the HIV response and beyond. PLOS Medicine, 2020. 16(12): p. e1003020.

2. Geng, E., et al., Implementation Research to Advance the Global HIV Response: Introduction to the JAIDS Supplement. JAIDS Journal of Acquired Immune Deficiency Syndromes, 2019. 82.

3. HIV/AIDS;, J.U.N.P.o., UNAIDS Global AIDS Report 2020 Seizing the Moment. 2020, Joint United Nations Programme on HIV/AIDS (UNAIDS): Geneva, Switzerland.

4. UNAIDS. Global HIV & AIDS statistics — 2019 fact sheet. Global HIV & AIDS statistics — 2019 fact sheet [webpage] 2019 2020 [cited 2020 1 June 2020]; Available from: https://www.unaids.org/en/resources/fact-sheet.

5. UNAIDS, UNAIDS Data, 2019, UNAIDS, Editor. 2019, UNAIDS: Geneva.

6. Vandepitte, J., et al., Estimates of the number of female sex workers in different regions of the world. . Sexually transmitted infections. , 2006. 82((Suppl III)): p. iii18- iii25.

7. Fearon, E., Estimating the population size of female sex workers in Zimbabwe: methods comparison across twenty sites and a national estimate. 2018: To be submitted.

8. UNAIDS, UNAIDS Guidance Note on HIV and Sex Work. 2012, Geneva.

9. HIV Prevention 2020 Road Map: Accelerating HIV prevention to reduce new infections by 75%. 2017, Geneva: Joint United Nations Programme on HIV/AIDS (UNAIDS).

10. Cowan, F.M., et al., Targeted combination prevention to support female sex workers in Zimbabwe accessing and adhering to antiretrovirals for treatment and prevention of HIV (SAPPH-IRe): a cluster-randomised trial. The Lancet HIV, 2018. 5(8): p. e417-e426.

11. Hoffmann, T.C., et al., Better reporting of interventions: template for intervention description and replication (TIDieR) checklist and guide. BMJ : British Medical Journal, 2014. 348: p. g1687.

12. Proctor, E.K., B.J. Powell, and J.C. McMillen, Implementation strategies: recommendations for specifying and reporting. Implementation Science, 2013. 8(1): p. 139.

13. Cowan, F.M., et al., Strengthening the scale-up and uptake of effective interventions for sex workers for population impact in Zimbabwe. J Int AIDS Soc, 2019. 22 Suppl 4: p. e25320.

14. Cowan, F.M., et al., The HIV care cascade among female sex workers in Zimbabwe: results of a population-based survey from the sisters antiretroviral therapy programme for prevention of HIV, an integrated response (SAPPH-IRe) trial. JAIDS Journal of Acquired Immune Deficiency Syndromes, 2017. 74(4): p. 375-382.

15. Kerrigan, D., et al., A community empowerment approach to the HIV response among sex workers: effectiveness, challenges, and considerations for implementation and scale-up. Lancet, 2015. 385(9963): p. 172-85.

16. Biradavolu, M.R., et al., Can sex workers regulate police? Learning from an HIV prevention project for sex workers in southern India. Social Science & Medicine, 2009. 68(8): p. 1541-1547.

17. Kerrigan, D., et al., Abriendo Puertas: Feasibility and Effectiveness a Multi-Level Intervention to Improve HIV Outcomes Among Female Sex Workers Living with HIV in the Dominican Republic. AIDS Behav, 2016. 20(9): p. 1919-27.

18. Blanchard, J.F., et al., Concepts and strategies for scaling up focused prevention for sex workers in India. Sex Transm Infect, 2008. 84 Suppl 2: p. ii19-23.

19. Blanchard, A., et al., Community mobilization, empowerment and HIV prevention among female sex workers in south India. BMC Public Health, 2013. 13(1): p. 234.

20. Beattie, T.S., et al., Community mobilization and empowerment of female sex workers in Karnataka State, South India: associations with HIV and sexually transmitted infection risk. American journal of public health, 2014. 104(8): p. 1516-1525.

21. Campbell, C., et al., Social capital and HIV Competent Communities: The role of community groups in managing HIV/AIDS in rural Zimbabwe. AIDS Care, 2013. 25(Suppl 1): p. S114-S122.

22. Li, F., et al., An evaluation of constrained randomization for the design and analysis of group-randomized trials. Statistics in Medicine, 2016. 35(10): p. 1565-1579.

23. Zenilman, J., et al., Detecting sperm DNA in vaginal swabs by PCR: a new bio-marker to validate recent sexual behavior and condom use. Int-J-STD-AIDS., 2001. 12 Suppl 2: p. 60.

24. Chiyaka, T., et al., Reaching young women who sell sex: Methods and results of social mapping to describe and identify young women for DREAMS impact evaluation in Zimbabwe. PLOS ONE, 2018. 13(3): p. e0194301.

25. Rice, B.D., et al., Can HIV recent infection surveillance help us better understand where primary prevention efforts should be targeted? Results of three pilots integrating a recent infection testing algorithm into routine programme activities in Kenya and Zimbabwe. Journal of the International AIDS Society, 2020. 23(S3): p. e25513.

26. Anderson, P.L., et al., Intracellular Tenofovir-Diphosphate and Emtricitabine-Triphosphate in Dried Blood Spots following Directly Observed Therapy. Antimicrobial agents and chemotherapy, 2017. 62(1): p. e01710-17.

27. Hayes, R. and S. Bennett, Simple sample size calculations for cluster randomized trials. International Journal of Epidemiology, 1999. 28: p. 319-326.

28. Hargreaves, J.R., et al., Statistical design and analysis plan for an impact evaluation of an HIV treatment and prevention intervention for female sex workers in Zimbabwe: a study protocol for a cluster randomised controlled trial. Trials, 2016. 17: p. 6.

29. Gile, K.J., L.G. Johnston, and M.J. Salganik, Diagnostics for Respondent-driven Sampling. J R Stat Soc Ser A Stat Soc, 2015. 178(1): p. 241-269.

30. White, R.G., et al., Strengthening the Reporting of Observational Studies in Epidemiology for respondent-driven sampling studies: "STROBE-RDS" statement. Journal of clinical epidemiology, 2015. 68(12): p. 1463-1471.

31. Volz, E. and D.D. Heckathorn, Probability Based Estimation Theory for Respondent Driven Sampling. Journal of Official Statistics, 2008. 24(1): p. 79-97.

32. Hayes, J.R. and L.H. Moulton, Cluster Randomised Trials. 2009: Chapman and Hall/CRC.

33. Gile, K.J., Improved Inference for Respondent-Driven Sampling Data With Application to HIV Prevalence Estimation. Journal of the American Statistical Association, 2011. 106(493): p. 135-146.

34. Reza-Paul, S., et al., Sex Worker Community-led Interventions Interrupt Sexually Transmitted Infection/Human Immunodeficiency Virus Transmission and Improve Human Immunodeficiency Virus Cascade Outcomes: A Program Review from South India. Sex Transm Dis, 2019. 46(8): p. 556-562.

35. McClarty, L.M., et al., Key Programme Science lessons from an HIV prevention 'Learning Site' for sex workers in Mombasa, Kenya. Sex Transm Infect, 2018. 94(5): p. 346-352.

36. Brody, C., et al., Economic Self-Help Group Programs for Improving Women’s Empowerment: A Systematic Review. 2015.

37. Morrison, J., et al., Exploring the equity impact of a maternal and newborn health intervention: a qualitative study of participatory women’s groups in rural South Asia and Africa. International Journal for Equity in Health, 2019. 18(1): p. 55.

38. Wittkop, L., et al., Methodological issues in the use of composite endpoints in clinical trials: examples from the HIV field. Clin Trials, 2010. 7(1): p. 19-35.

39. McCoy, C.E., Understanding the Use of Composite Endpoints in Clinical Trials. The western journal of emergency medicine, 2018. 19(4): p. 631-634.

40. Zimbabwe population-based HIV impact assessment ZIMPHIA 2020. 2020.

Table 1: Structured description of AMETHIST intervention and Standard of Care within which this intervention is set

Table 2: Measurement of the AMETHIST primary outcome

	HIV positive		HIV negative
	Not at risk	At risk of transmission	At risk of acquisition	Not at risk
Viral load	Suppressed	Unsuppressed	-	-
Oral pre-exposure prophylaxis	-	-	Self-report not on PreP OR (Report on PREP AND ((refuse a sample for PrEP testing) OR (PrEP adherence defined as ‘high’ if Tenofovir-diphosphate (TDF-DP) ≥ 700 fmol/dried blood spot (DBS) punch, and ‘low’ if < 350 fmol/punch.))	Report on PrEP, confirmed by drug levels indicating high PrEP adherence TFV-DP ≥ 700 fmol/dried blood spot (DBS) punch
	OR	AND	AND	OR
Evidence of condom-less sex	Self-report no condomless sex AND agree to a Y Chromosome test AND Y Chromosome negative AND Bacterial STI negative	Self-report any condomless sex in past 1 month OR (Self-report no condom-less sex AND ((Refuse a sample for Y chromosome testing) OR (Y Chromosome positive)) OR (Lab-positive for a bacterial STI))	Self-report any condomless sex in past 1 month OR (Self-report no condom-less sex AND ((Refuse a sample for Y Chromosome testing) OR (Y Chromosome positive)) OR (Lab-positive for a bacterial STI))	Self-report no condomless sex AND agree to a Y Chromosome test AND Y Chromosome negative AND Bacterial STI negative

Table 3: Power calculations

Prevalence of outcome (control)	% reduction	Prevalence of outcome (intervention)	Cluster size	K	# clusters per arm	# clusters, rounded up	Power
20%	20%	16%	200	0.2	10.8	11	0.5
20%	30%	14%	200	0.2	10.2	11	0.84
20%	50%	10%	200	0.2	10.8	11	0.9998
30%	20%	24%	200	0.2	10.3	11	0.54
30% *	30%	21%	200	0.2	10.4	11	0.9
30%	50%	15%	200	0.2	10.8	11	0.99997
40%	20%	32%	200	0.2	10.1	11	0.57
40%	30%	28%	200	0.2	10.2	11	0.92
40%	50%	20%	200	0.2	10.7	11	0.99999
20%	20%	16%	200	0.25	10.2	11	0.37
20%	30%	14%	200	0.25	10.4	11	0.73
20%	50%	10%	200	0.25	10.0	11	0.995
30%	20%	24%	200	0.25	10.1	11	0.4
30%	30%	21%	200	0.25	10.5	11	0.78
30%	50%	15%	200	0.25	10.9	11	0.999
40%	20%	32%	200	0.25	10.8	11	0.45
40%	30%	28%	200	0.25	10.3	11	0.8
40%	50%	20%	200	0.25	10.3	11	0.999

* most likely scenario

AMETHISTConsortExtchecklistforCRT.docx
AdditionalInformation1AMETHISTTrialProtocol.pdf
AMETHIST Trial Protocol - abstracted from AMETHIST Consortium Protocol is included as Additional Information.

Differentiated Prevention And Care To Reduce The Risk of HIV Acquisition And Transmission Among Female Sex Workers In Zimbabwe: Study Protocol For The ‘AMETHIST’ Cluster Randomised Trial

Status:

Journal Publication

Version 1

Abstract

Figures

Background

Methods

Discussion

List Of Abbreviations

Declarations

References

Tables

Supplementary Files

Status:

Journal Publication

Version 1