Community Based Screening for HIV Self Testing in FSW in 23 Priority Districts in Indonesia

1. Background
2. The HIV Epidemic in Indonesia Until recently, Indonesia was among the few countries in which annual numbers of new HIV infections continued to rise (MoH, 2016). The latest epidemic modeling update indicated that except among men who have sex with man (MSM), annual numbers of new HIV infections had stabilized and begun to decline. However, with the current epidemic trajectory there would still be over 40,000 new HIV infections in the year 2030 (MoH,2016). Therefore, Indonesia is not yet on course to end HIV and AIDS by 2030. Although there has been a significant increase in the number of persons being treated for HIV/AIDS, the 108,479 people receiving ART as of December 2018 amounts to only 17 percent of the estimated number of people live with HIV (PLHIV) in the country (MoH, 2018). This makes Indonesia a performance "outlier" when compared to countries at comparable levels of gross national income (GNI) and health system development. The lack of more rapid progress has recently led key development partners to question Government of Indonesia (GOI) commitment to meaningfully addressing HIV and AIDS. Insufficient HIV testing remains a barrier to increasing ART coverage. The number of HIV tests performed annually has risen steadily in recent years, reaching 3,077,653 in calendar year 2018 (MoH, 2018). However, pregnant women account for a sizeable proportion of the increased number of persons being tested. While commendable, the case detection "yield" from testing pregnant women is relatively low. More effective case finding strategies need to be implemented to reach population sub-groups with higher HIV incidence and prevalence, including key affected populations (KAPs) such as female sex workers (FSW). Unless Indonesia can significantly increase its volume and efficiency of HIV testing, it will not be able to reach the first "95" of the UNAIDS 95-95-95 framework - that is, 95% of PLHIV know their HIV status. The main rationale is Indonesia's concentrated epidemic transmission of HIV infection among KAPs, specifically FSWs. In 2016 it is estimated that there are 226,791 female sex workers and around 5,254,065 clients access their service per year (MoH, 2017). This mode of transmission continues to clients' sexual partner and moreover, their babies. Lowering the transmission of HIV infection from FSW to their clients would simultaneously lower its transmission to their sexual partners and furthermore their babies. FSW can be grouped by "direct" (i.e., brothel-based) and "indirect" status with regard to way of selling their sex services. In 2016, however, the estimated number of FSW was not grouped into direct and indirect because it was considered that most of FSWs had shifted to becoming indirect due to Indonesia's national policy to close brothels. Due to this policy, most FSWs have become hidden and hard to reach, thus increasing the challenge of increasing HIV testing uptake among this sub-population. Many commercial sex transactions have become underground, especially given the new popularity of digital platforms to sell sexual services. This phenomenon has created a new demand to identify alternative strategies for increasing HIV testing uptake among FSW.
3. HIV Testing Among FSW Community based HIV-testing has been implemented in some countries such as Vietnam (Nguyen et al., 2019), Uganda (Ortblad et al., 2018), Malawi and Zimbabwe (Napierala et al., 2019). Non-governmental organizations in Indonesia had conducted study or pilot project on HIV test in men who have sex with men (MSM) outside health facility (Hidayat et al., 2019). This study was done to find an alternative to facility-based HIV testing, however MoH has been rather conservative in supporting non-health facility-based approach outside of mobile clinic testing which involves trained healthcare professional to perform HIV testing procedure according to the algorithm developed by the government. Mobile clinic testing strategy gave out insufficient result, especially in terms of linking those with positive HIV result for further follow-up and initiation of treatment. Cost-effectiveness of this approach was also doubted, at least in Jakarta (Cantelmo et al., 2019). This study found that the cost needed for mobile testing to identify at least one HIV positive case among FSW was almost six-times fold higher compared to finding similar target among transgender and MSM, and seventeen-times fold higher than those for identifying one HIV positive case among people who inject drugs (PWID). For the effective use of resource, this study suggested modification of the test frequency, time, and location for FSWs in Jakarta. In addition, this study also recommended the need for alternative strategies to increase HIV test uptake among FSW. FSWs community are at greater risk of not only HIV infection, but also stigma, discrimination and violence. They first face stigma and discrimination due to engaging in sex work itself, or from HIV stigma, particularly in contexts of HIV burden, which later affect their access to HIV testing (UNAIDS, 2016). The latest World Health Organization (WHO) guidelines have highlighted HIV self-testing (HIVST) as important tool to identify more people with undiagnosed HIV and at high risk of HIV infection. Protection of privacy and confidentiality is one of the advantages of this modality which allows for removing stigma as barriers to access services. HIVST has been shown to be acceptable across varieties of population globally including for FSWs community (King et al, 2013). Oral fluid HIVST is an alternative to traditional HIV testing services in the facility or other healthcare provider testing (UNAIDS, 2016).
4. Role of Outreach Worker in HIV Self-testing Using Assisted and Unassisted Method In a study which assessed implementation and scale-up to HIV self-testing programs for female sex workers in Malawi and Zimbabwe, there were difference in preferences for how to access HIV self-testing, depended on how supportive the existing program infrastructure was (Napierala et al., 2019) In Zimbabwe, where there was a detailed understanding of the context of female sex workers and a ready framework to implement and evaluate HIV self-testing strategies, high acceptability (76%) and high accuracy of HIV self-testing was reported (Napierala et al., 2019). In contrast, peer-distribution models were favored by female sex workers in Malawi and female sex workers in Zimbabwe who were not engaged in the program (Napierala et al., 2019). Another study in Kampala (Uganda) evaluated HIV self-testing performance and results interpretation among FSWs who performed unassisted HIV self-testing, showed that misinterpretation of HIV self-test results were common among FSWs: 23% (12/56) of FSWs interpreted HIV-negative self-test results as HIV positive and 8% (3/37) of FSWs interpreted HIV-positive self-test results as HIV negative (Ortblad et al., 2018). The concordance between FSWs' instructions was 73% (95%CI 56% to 86%) for HIV-positive self-tests and 68% (95%CI 54% to 80%) for HIV-negative self-tests (Ortblad et al., 2018). This finding suggested training on use and interpretation of HIV self-test for the unassisted method might be necessary to prevent errors and to avoid the negative consequences of false-positive and false-negative HIV self-test results among FSWs.
5. Study Purpose

The study proposed in this protocol directly addresses the need to get more Indonesian FSW to "know their status" by providing an alternative, convenient HIV testing option in non-threatening community settings. In the Indonesian context where a reactive HIV test result using the MoH-mandated triple rapid test algorithm is needed to qualify for GoI-financed ART, community screening is seen as a mechanism for enabling FSW to conveniently determine their status and a facilitation mechanism for taking action based upon the community screening result, whether that entail going to a health facility for a confirmatory test in the case of a reactive screening result or adopting stronger prevention measures in the case of a non-reactive screening result, including Pre-Exposure Prophylactic (PrEP) as this prevention method is rolled out in Indonesia.

The study will produce scientifically strong evidence as to whether two alternative models of community HIV screening among FSW (assisted and unassisted) result in (1) increased rates of formal HIV testing at health facilities and (2) increased rates of treatment initiation in districts in which the community screening intervention is added to the existing FSW community outreach model.

Two alternative study protocols are described in this document. The bulk of this document presents a protocol for a community randomized controlled trial (cRCT), which is the preferred research design option. However, due to a delay in procurement of the OraQuick® rapid HIV test kits that are to be used in the study, there may be insufficient time to undertake a full cRCT. Accordingly, a contingency protocol that can be implemented in a shorter period of time is also presented.

1. Study Design
2. Design Summary The study will be undertaken in the 23 "acceleration" districts in the National AIDs Program. These are: Kota Medan, Deli Serdang, Kota Palembang, Kota Bandar Lampung, Kota Tangerang Selatan, Tangerang, Kota Jakarta Selatan, Kota Jakarta Timur, Kota Jakarta Pusat, Kota Jakarta Barat, Kota Jakarta Utara, Bogor, Kota Bekasi, Kota Bandung, Kota Depok, Kota Semarang, Kota Surakarta, Kota Malang, Kota Surabaya, Kota Denpasar, Kota Makassar, Kota Sorong and Kota Jayapura. These are districts with high HIV prevalence among HIV key populations, including FSW, and have comprehensive ongoing HIV-TB intervention packages consisting of both health services and community prevention and support programs. As for the FSWs program, these 23 priority districts already implementing a comprehensive outreach package (reach to test and simplified case management for FSWs living with HIV).
3. Randomization Stratified randomization was carried out for the 23 priority districts involved in this study. First, sampling strata were created by sorting the mean average value of achieved HIV testing target per semester from 2018-2019 in each district from the largest to the smallest. The second step was to sort the districts into eight groups of three districts, except for the last group which will only consist of two districts. Randomization for intervention and comparison group were then carried out within these groups of eight with 2:1 ratio (2 intervention: 1 comparison).
4. Recruitment and Sample Size

All FSW in intervention districts who meet study eligibility criteria will be offered community screening. FSW in comparison districts will continue to receive the current standard package of interventions.

In to be able to detect a 10 percentage point difference in the rate of HIV testing at health facilities between FSW in intervention vs. comparison districts and have 95% certainty that a difference of that magnitude would not have occurred by chance and 90% certainty of detecting a difference of this magnitude if the difference was real/"the truth," the following sample of FSW will be needed in intervention and in comparison districts:

n ≥ [Z1-α (2P(1-P))1/2 + Z1-β (P1(1-P1) + P2(1-P2)2 / (P1-P2)2] * deft

Where:

Z1-α = the Z score for the level of statistical confidence, or statistical precision, desired (for 95%, Z = 1.96) Z1-β = the Z score for the desired statistical power (for 90% one-sided test, Z = 1.282) P1 = the expected population proportion in the comparison group of districts (set = 0.5 - this is the worst-case scenario and will produced a sample size that is adequate irrespective of the actual proportion P2 = the expected population proportion in the intervention group of districts (set = 0.6 - assume minimum effect size to be realized of 10 percentages points) P = (P1-P2) / 2 (P1-P2) = the magnitude of comparison-group differences (or change over time) to be detected with the specified level of precision and power (assumed to be +/- 10 percentage points) deft = design effect to compensate for clustering at the district level (1.5 assumed).

The required sample size is thus n ≥ 635 per experimental group; ≥ 761 after 20% allowance for lost-to-follow up Interpretation: we would need to recruit 761 FSW who self-test for HIV testing in the intervention districts and 761 FSW for HIV testing at health facilities in comparison districts to be able to detect a 10 percentage point difference in the rate of HIV testing at health facilities (intervention vs. comparison districts) to have 95% certainty that a difference of that magnitude would not have occurred by chance and 90% certainty of detecting a difference of this magnitude if such a difference existed.

In order to assess whether "assisted" or "unassisted" community HIV screening among FSW results in a larger increase in the rate of HIV testing at health facilities, we would need samples of size n ≥ 761 each of FSW who received assisted and unassisted screening in intervention districts, a total of n ≥ 1,522 FSW in intervention districts, plus n ≥ 761 in comparison districts.

Sample size requirements for the third question measurement pertains to the number of FSW who test positive for HIV at a health facility and are thus eligible to initiate treatment. This will depend upon (1) the number of FSW presenting at health facilities for testing and (2) the positivity rate among those tested. Sample sizes should thus be calculated accordingly. If we assume a testing positivity rate of 3%, it is apparent that the expected number of FSWs who would be eligible for treatment will be small and we will lack sufficient statistical power to make meaningful comparisons with comparison districts. For this reason, it is recommended that the impact of community screening among FSW be measured using system information HIV and AIDS data (SIHA) for FSW in both intervention and comparison districts. This estimate of impact derived in this way will be confounded if it were to be the case that interventions other than HIV self-testing were to be better implemented in intervention vs. comparison districts. It might be possible to minimize this potential bias by including measures of intervention implementation performance in the intervention and comparison districts in multivariable analyses. Thus, a large enough sample size is required to ensure sufficient power to determine the difference in the intervention and control districts.

1. The Intervention
2. General Description of The Intervention

The community screening intervention will be implemented as an additional component to be added in intervention districts to the intervention package currently being implemented by UNFPA's Implementation Units (IUs). In "control districts" implementation of the UNFPA-supported GFATM will proceed without modification as concerns community screening.

1. Data Collection Protocols
2. Baseline survey An application will be developed so that the survey can be completed and submitted to a server using mobile phones Outreach worker (OW)/Peer educator (PE)/Female sex worker (FSW). The following basic information will be gathered (see draft questionnaires for the baseline and post-test surveys in the annex of this protocol).

• Name (full name/initial)
• Age
• Date of birth
• Test Kit ID Number
• Education
• Marital status
• Age at initiation of sex work
• Number of clients last seven (7) days
• Methods of clients' recruitment (fixed facility, street, internet, online platform)
• Consistency of condom use
• Ever been tested for HIV
• When last time tested
• Ever been diagnosed with an sexually transmitted infections/STI (other than HIV)

1. Post-test Survey The following inquiries will be made:

• The clarity of information being provided (flier or short video), on a scale of 1-5
• The easiness of test procedure, on a scale of 1-5
• Perceived accuracy of the test result, on a scale of 1-5
• The quality of response by the assigned contact person/hotline, on a scale of 1-5
• Clients' acceptance/perception towards the study procedure
• The likelihood of clients to recommend their peers to take the test, on a scale of 1-5
• Test result
• How likely is it that the client will get a HIV test at a health facility as a result of the community screening test, on a scale of 1-5
• Comments and suggestions.

1. Routine Recording

Because of limitations in the data recorded and reported by health facilities, the study will rely primarily upon data recorded by IU outreach staff. Modifications will be made to the data recording and reporting systems currently being used to facilitate collection of the data needed for the study in a manner that minimizes additional recording and reporting burden. The nature of the adjustments to be made will be different for "assisted" and "unassisted" community screening.

For unassisted approach, additional information will be gathered as indicated below:

1. Numbers of FSW who access the link online
2. Numbers of FSW who are deemed eligible
3. Numbers of FSW who fill in baseline survey and agree to participate
4. Numbers of FSW who agree to take HIV test at health facility
5. Numbers of FSW who request the kit being delivered vs picking-up at fixed location
6. Numbers of FSW who upload the test result
7. Numbers of reactive vs non-reactive test result
8. Numbers of FSW who complete the post-test survey
9. Among FSW with community screening tests, the numbers that were subsequently tested at a health facility (reactive OR non-reactive)
10. Number of FSW initiating ART.
11. Numbering System for Test Kit and Other Instruments
12. Recording Test Kit ID Every test kit will have a unique ID number, which will later be referred to as the Kit ID number. The Kit ID Number is to be filled in on all instruments and forms that are to be linked to the test results (e.g., baseline survey, test results form, post-test survey, routine data record in each IU, etc.). This number needs to be identified at all times to make sure that each participant is only assigned to one test kit, matching the kit with the results, and to track the stock of the test kit remained. The format of this ID kit will be "01-001", where the first 2 digits are to identify the number of the district in which participants take the test, and the last 3 digits are to identify which test kit the participant is using. This test kit ID will also be later used for each IU to track the distribution of the test kit. Every test kit being distributed, either in a fixed place or delivered through same-day delivery services, will be recorded and tracked by each IU. An MS Excel or MS Access form will be developed to record and track the Kit ID Number.
13. Recording temporary FSW ID On the ground of confidentiality, participants' full name will not be revealed and used within this study. A new system to ensure the reliable yet confidential identification of study participants needs to be established. Every participant in the study will be assigned to a temporary FSW ID. The format will be the first 4 letters of participants' names added by the date of birth (yy/mm/dd). For example, a participant named "INDRIYANTI" who was born on March 5th, 1976, her temporary FSW ID will be "INDR760305". For FSW whose name contains only 3 letters, the number zero will be added after the last letter. For example, an FSW whose name is "AYU" and was born on March 5th1976, her temporary will be "AYU0760305". This temporary FSW ID will be used to ensure that each participant is only assigned to one test kit, to track test results, and to be filled in the baseline and post-test survey.
14. Matching Test Kit ID and temporary FSW ID To match the test kit ID and temporary FSW ID, an automatic system needs to be developed to detect it precisely. This way, we can be sure that each participant only gets to participate in this study once (from receiving the kit, performing the test, and reporting the result).
15. Paper Based Data Collection

In the event that participants' data are collected by paper, a procedure to upload the data into the established online system will be developed. Each data manager in the IUs needs to conduct the procedure regularly. This is to ensure that all data are being captured thoroughly.

1. Research Ethics and Permission

Participation in the study will be entirely voluntary. For the assisted testing, formal signed consent or witnessed verbal informed consent will be obtained by outreach workers. Records of informed consent will be managed /safeguarded using sealed envelope with unique number. To ensure the confidentiality, in the base line data only unique number will be recorded. A draft of the informed consent form can be found in Annex-3 of this protocol. For unassisted screening, every study participant should fill in the check box of website or application to provide consent: "By signing this form/ticking this box, I understand that I am thereby agreeing to enroll in this study" and test kit is not sent out if the check box is not checked by the user.

This protocol will be reviewed by the Faculty of Medicine Udayana University Institutional Review Board (IRB) prior to study initiation. Approval of the protocol in either its original or modified form is required.

Research permission will be submitted to Department of Internal Affairs Republic of Indonesia, which will subsequently be referred to the licensing office in each provinces and priority districts.