Engagement of African Men and Transgender Women Who Have Sex with Men in HIV Research

Engagement of African men and transgender women who have sex with men in HIV research

Guest Editors: Trevor A Crowell, Patricia E Fast, Linda-Gail Bekker, Eduard J Sanders Supplement Editors: Marlène Bras, Elisa de Castro Alvarez

Volume 23, Supplement 6, October 2020 Acknowledgements The Guest Editors - Trevor A Crowell, Patricia E Fast, Linda-Gail Bekker and Eduard J Sanders - would like to thank all of the authors who submitted expressions of interest and full manuscripts. The authors worked hard throughout the rigorous selection process and we wish that more studies could have been included. We encourage investigators to continue to undertake research on this important topic and to disseminate critical ﬁ ndings. We are hopeful that the work presented in this supplement will inform the design of future HIV prevention and care delivery research with the needs of African men who have sex with men and other sexual and gender minorities in mind. We would like to thank Haoyu Qian for project management support provided throughout the preparation of this supplement. We also wish to thank the editors and staff of the Journal of the International AIDS Society for their professionalism, support, and thoughtful guidance throughout the editorial process.

Support This supplement was funded and supported by the US Military HIV Research Program (MHRP) through a cooperative agreement between the Henry M Jackson Foundation for the Advancement of Military Medicine, Inc., and the US Department of Defense [W81XWH-18-2-0040]; and the International AIDS Vaccine Initiative (IAVI), which receives generous support of the American people through the United States Agency for International Development (USAID).

Disclaimer The authors of the articles included in this supplement are solely responsible for the views expressed, which do not necessarily represent the views, decisions or policies of the institutions with which the authors are affi liated. The content of this supplement should not be construed to represent the positions of the US Army, US Department of Defense, USAID or the United States Government. Engagement of African men and transgender women who have sex with men in HIV research

Guest Editors: Trevor A Crowell, Patricia E Fast, Linda-Gail Bekker, Eduard J Sanders Supplement Editors: Marlène Bras, Elisa de Castro Alvarez

Contents Involvement of African men and transgender women who have sex with men in HIV research: progress, but much more must be done Trevor A Crowell, Patricia E Fast, Linda-Gail Bekker and Eduard J Sanders 1 HIV prevalence and incidence in a cohort of South African men and transgender women who have sex with men: the Sibanye Methods for Prevention Packages Programme (MP3) project Patrick S Sullivan, Nancy Phaswana-Mafuya, Stefan D Baral, Rachel Valencia, Ryan Zahn, Karen Dominguez, Clarence S Yah, Jeb Jones, Lesego B Kgatitswe, AD McNaghten, Aaron J Siegler, Travis H Sanchez and Linda-Gail Bekker 6 Sexually transmitted infection screening, prevalence and incidence among South African men and transgender women who have sex with men enrolled in a combination HIV prevention cohort study: the Sibanye Methods for Prevention Packages Programme (MP3) project Jeb Jones, Travis H Sanchez, Karen Dominguez, Linda-Gail Bekker, Nancy Phaswana-Mafuya, Stefan D Baral, AD McNaghten, Lesego B Kgatitswe, Rachel Valencia, Clarence S Yah, Ryan Zahn, Aaron J Siegler and Patrick S Sullivan 17 Risk factors for loss to follow-up among at-risk HIV negative men who have sex with men participating in a research cohort with access to pre-exposure prophylaxis in coastal Kenya Elizabeth W Wahome, Susan M Graham, Alexander N Thiong’o, Khamisi Mohamed, Tony Oduor, Evans Gichuru, John Mwambi, Elise M van der Elst and Eduard J Sanders 30 Retention of a cohort of men who have sex with men and transgender women at risk for and living with HIV in Abuja and Lagos, Nigeria: a longitudinal analysis Blessing O Kayode, Andrew Mitchell, Nicaise Ndembi, Afoke Kokogho, Habib O Ramadhani, Sylvia Adebajo, Merlin L Robb, Stefan D Baral, Julie A Ake, Manhattan E Charurat, Trevor A Crowell and Rebecca G Nowak 40 Enrolment characteristics associated with retention among HIV negative Kenyan gay, bisexual and other men who have sex with men enrolled in the Anza Mapema cohort study Colin Kunzweiler, Robert C Bailey, Duncan O Okall, Susan M Graham, Supriya D Mehta, Boaz Otieno-Nyunya, Gaston Djomand and Fredrick O Otieno 51 The feasibility of recruiting and retaining men who have sex with men and transgender women in a multinational prospective HIV prevention research cohort study in sub-Saharan Africa (HPTN 075) Theodorus GM Sandfort, Erica L Hamilton, Anita Marais, Xu Guo, Jeremy Sugarman, Ying Q Chen, Vanessa Cummings, Suﬁ a Dadabhai, Karen Dominguez, Ravindre Panchia, David Schnabel, Fatima Zulu, Doerieyah Reynolds, Oscar Radebe, Calvin Mbeda, Dunker Kamba, Brian Kanyemba, Arthur Ogendo, Michael Stirratt, Wairimu Chege, Jonathan Lucas, Maria Fawzy, Laura A McKinstry and Susan H Eshleman 59 Association of age with healthcare needs and engagement among Nigerian men who have sex with men and transgender women: cross-sectional and longitudinal analyses from an observational cohort Habib O Ramadhani, Trevor A Crowell, Rebecca G Nowak, Nicaise Ndembi, Blessing O Kayode, Afoke Kokogho, Uchenna Ononaku, Elizabeth Shoyemi, Charles Ekeh, Sylvia Adebajo, Stefan D Baral and Manhattan E Charurat 70 HIV infection and engagement in HIV care cascade among men who have sex with men and transgender women in Kigali, Rwanda: a cross-sectional study Jean Olivier Twahirwa Rwema, Carrie E Lyons, Sara Herbst, Benjamin Liestman, Julien Nyombayire, Sosthenes Ketende, Amelia Mazzei, Oluwasolape Olawore, Sabin Nsanzimana, Placidie Mugwaneza, Aﬂ odis Kagaba, Patrick S Sullivan, Susan Allen, Etienne Karita and Stefan D Baral 80 Online socializing among men who have sex with men and transgender people in Nairobi and Johannesburg and implications for public health-related research and health promotion: an analysis of qualitative and respondent-driven sampling survey data Elizabeth Fearon, Adam Bourne, Siyanda Tenza, Thesla Palanee-Phillips, Rhoda Kabuti, Peter Weatherburn, Will Nutland, Joshua Kimani and Adrian D Smith 91 Acute and early HIV infection screening among men who have sex with men, a systematic review and meta-analysis Shaun Palmer, Maartje Dijkstra, Johannes CF Ket, Elizabeth W Wahome, Jeffrey Walimbwa, Evanson Gichuru, Elise M van der Elst, Maarten F Schim van der Loeff, Godelieve J de Bree and Eduard J Sanders 106 From general to specific: moving past the general population in the HIV response across sub-Saharan Africa Keletso Makofane, Elise M van der Elst, Jeffrey Walimbwa, Steave Nemande and Stefan D Baral 120 A more responsive, multi-pronged strategy is needed to strengthen HIV healthcare for men who have sex with men in a decentralized health system: qualitative insights of a case study in the Kenyan coast Elise M van der Elst, Rita Mudza, Justus M Onguso, Leonard Kiirika, Bernadette Kombo, Nassim Jahangir, Susan M Graham, Don Operario and Eduard J Sanders 126

Volume 23, Supplement 6 October 2020

Crowell TA et al. Journal of the International AIDS Society 2020, 23(S6):e25596 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25596/full | https://doi.org/10.1002/jia2.25596

EDITORIAL Involvement of African men and transgender women who have sex with men in HIV research: progress, but much more must be done Trevor A Crowell1,2,§ , Patricia E Fast3,4 , Linda-Gail Bekker5 and Eduard J Sanders6,7 §Corresponding Author: Trevor A Crowell, U.S. Military HIV Research Program, 6720A Rockledge Drive, Suite 400, Bethesda, MD 20817, USA. Tel: +1 301 500 3990. ([email protected])

Keywords: Africa; men who have sex with men; transgender people; key and vulnerable populations; sexual and gender minorities; sexual behaviour; biomedical research; implementation science; population health management; transgender women

Received 16 July 2020; Accepted 20 July 2020

Copyright © 2020 The Authors. Journal of the International AIDS Society published by John Wiley & Sons Ltd on behalf of the International AIDS Society. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

1 | INTRODUCTION these studies has been dependent on collaboration with key opinion leaders [25], non-governmental organizations [26], In sub-Saharan Africa, men who have sex with men (MSM) religious leaders [27], healthcare providers [28] and other and transgender women (TGW) have been living with, and community stakeholders to facilitate engagement of hard-to- dying from, HIV since the start of the pandemic. However, the reach African MSM communities. However, recruitment and impact of the virus on these sexual and gender minority popu- retention of African MSM in HIV prevention and treatment lations was masked for decades by the perception of HIV as a research has proved challenging. The purpose of this supple- “generalized” epidemic in the African context [1]. While the ment is to present reports of HIV-related studies that Multicenter AIDS Cohort Study (MACS) began enrolling MSM recruited African MSM with a focus on challenges and suc- in Baltimore, Chicago, Pittsburgh and Los Angeles [2] just cesses related to engagement, recruitment and retention. three years after the first clinical cases of AIDS were reported Some studies also enrolled TGW, another key population that [3-6], another two decades passed before researchers began historically has often been conflated with MSM, but is deserv- HIV-focused studies enrolling African MSM [7]. Since then, ing of separate consideration and tailored, rights-affirming data have consistently shown that HIV prevalence and inci- engagement. Lessons learned from these studies will inform dence are far higher among African MSM and TGW than the development of strategies to test and, subsequently, deli- other reproductive age adults in their countries [8-13]. Miti- ver prevention and treatment interventions that are desper- gating and managing this high burden of disease is compli- ately needed to end the HIV pandemic. cated by barriers that African MSM and TGW face due to discriminatory policies, criminalizing legislation, stigmatizing healthcare systems and under-allocation of limited healthcare 2 | HIGH INCIDENCE OF HIV AND resources [14-20]. These barriers impede access and advance- OTHER SEXUALLY TRANSMITTED ment through each step of the HIV care cascade, including INFECTIONS: IMPLICATIONS FOR HIV testing, linkage to care and antiretroviral therapy to PREVENTION, TESTING AND achieve viral suppression [21]. As pre-exposure prophylaxis TREATMENT (PrEP) is becoming available to at-risk individuals in some Afri- can countries, there are already early signals that MSM and Two manuscripts in this supplement focus on MSM and TGW TGW face unique challenges accessing and adhering to this who enrolled in a one-year prospective cohort study called important biomedical HIV prevention strategy [22,23]. the Sibanye Methods for Prevention Packages Project in Cape Increasingly, MSM outreach, engagement and research par- Town and Port Elizabeth, South Africa. Sullivan et al.[29] ticipation have been recognized as vital to HIV prevention and describe a high prevalence of HIV in the study population with control efforts in sub-Saharan Africa. This has been evidenced only half of those living with HIV aware of their status. HIV by a proliferation of observational and interventional studies incidence during longitudinal follow-up was 6.3 cases per 100 engaging African MSM as one of several key populations person-years, and even higher among TGW (31.0 cases per affected by the HIV pandemic [24]. The success of many of 100 person-years) and MSM aged 18 to 19 years (21.8 cases

1 Crowell TA et al. Journal of the International AIDS Society 2020, 23(S6):e25596 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25596/full | https://doi.org/10.1002/jia2.25596 per 100 person-years). While PrEP is widely recommended Together, these three studies highlight tremendous chal- for use among adult African MSM and TGW, its uptake in this lenges in the retention of MSM and TGW in HIV-related study was low and did not significantly impact HIV incidence. observational studies. Common themes include the need for This report provides data on the importance of offering HIV early intervention with strategies to encourage continued prevention services before age 18 and suggests that imple- research participation, as loss to follow-up disproportionately mentation science studies should focus on decisions to start occurred early in studies and among participants who had and continue PrEP for those at highest risk, including young newly arrived in their communities. These studies also suggest MSM and TGW. the importance of identifying and addressing direct medical In their accompanying paper, Jones et al.[30] describe high needs as a motivator of continued research engagement, such prevalence and incidence of Chlamydia trachomatis (CT), Neis- as treatment for alcohol dependence or STIs. For MSM and seria gonorrhoeae (NG) and syphilis among South African MSM TGW who are otherwise healthy and without urgent medical and TGW. While acceptance of screening for syphilis and ure- needs – such as those who would be recruited for many HIV thral NG/CT was near universal, only about two-thirds of par- prevention studies – there may be limited incentives to risk ticipants accepted clinician-assisted screening for rectal NG/ the discrimination or human rights violations that can accom- CT. This discrepancy may reflect participant unfamiliarity with pany disclosure of same-sex sexual practices in order to par- the possibility of anorectal sexually transmitted infections ticipate in a research study, so researchers must work (STIs) [31], concern for physical discomfort with sampling at especially closely with community stakeholders to identify this anatomic site [32] and psychological discomfort address- strategies to recruit and retain these participants. ing anal sexual health within potentially stigmatizing health- In contrast to the other retention-focused studies included care systems [33]. Consistent with prior reports from other in this supplement, Sandfort et al.[41] report relative success parts of sub-Saharan Africa [34-37], the high burden of in retaining MSM and TGW throughout the HPTN 075 HIV asymptomatic infections in this cohort highlights the limita- research feasibility study in Kenya, Malawi and South Africa. tions of syndromic surveillance and suggests the need for pre- Over 90% of participants completed the final study visit and sumptive testing and/or treatment to address the STI 86% completed all of the quarterly visits. Participants epidemic among MSM and TGW in South Africa. reported strong motivation to participate, few participation barriers and rare social harms. Retention in this study was promoted by a wide variety of interventions, including visit 3 | RETENTION IN CARE: STRATEGIES reminders via multiple communication strategies, transporta- FOR TACKLING A PERENNIAL tion arrangement or reimbursement, home visits, free medical CHALLENGE services, creation of hospitable study site environments and community outreach events. These intensive efforts con- Four manuscripts in this supplement focus on retention of tributed to the successful enrolment of a multinational sample MSM and TGW who enrolled in cohort studies. Wahome et al. of MSM and TGW with high retention and at least some of [38] report loss to follow-up, defined as no contact for more these should be incorporated into future studies that require than 90 days after a missed visit, of approximately one-third of longitudinal participation from similar populations. participants in a study with monthly visits and access to daily PrEP in coastal Kenya. PrEP use did not significantly impact loss to follow-up. Loss to follow-up was more common among partic- 4 | THE HIV PREVENTION AND CARE ipants who lived further away from the research clinic, had alco- CASCADE: IDENTIFYING CRITICAL GAPS hol use disorder, joined the cohort recently, or had higher education level. Kayode et al.[39] also report high loss to fol- Two manuscripts in this supplement focus on identifying low-up in the TRUST/RV368 cohort study offering comprehen- opportunities for improving MSM and TGW engagement in sive and integrated prevention and treatment services for HIV the HIV prevention and care cascade. Ramadhani et al.[42] and other STIs at community venues in Abuja and Lagos Nigeria. describe the impact of age on healthcare engagement in the In this study with quarterly visits for up to 18 months, over TRUST/RV368 cohort in Abuja and Lagos, Nigeria. They found one-half of enrolled MSM and TGW were lost to follow-up, that participants aged 16 to 19 years had several markers of defined as not presenting for a scheduled visit in the past decreased healthcare engagement as compared to those aged 180 days and allowing for intermittent absences. Adherence to 25 or older, including decreased HIV testing uptake, individual visits was also low, with participants completing a decreased disclosure of same-sex sexual practices to health- median of only 71% of the expected visits. Retention was better care workers and increased avoidance of healthcare. After among participants living with HIV or diagnosed with other adjusting for other factors, the youngest age group had 3 to 4 STIs. Kunzweiler et al.[40] also report high loss to follow-up times higher incidence rates of HIV, anorectal gonorrhoea and among gay, bisexual and other MSM enrolled in a prospective anorectal chlamydia. MSM aged 16 to 19 years had HIV inci- cohort study in Kisumu, west Kenya. Their study also included dence of 20.9 cases per 100 person-years and TGW aged 16 quarterly visits and defined loss to follow-up as missing two to 19 years have HIV incidence of 43.8 cases per 100 per- consecutive visits after enrolment, which was observed in about son-years. These findings underscore the overwhelming need one out of every five participants. The odds of missing two fol- for tailored interventions to engage young people and TGW in low-up visits were higher for men who had resided in the study HIV prevention and care. area for less than one year at enrolment, who were not living Rwema et al.[43] also describe barriers to healthcare engage- with a male sexual partner and who engaged in transactional ment in their cross-sectional study of MSM and TGW in Kigali, sex during the last three months. Rwanda. They found a high burden of HIV and other STIs in

2 Crowell TA et al. Journal of the International AIDS Society 2020, 23(S6):e25596 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25596/full | https://doi.org/10.1002/jia2.25596 both populations. Importantly, they found that less than two- services. Authors observed power inequities between policy thirds of participants living with HIV were previously aware of leadership, healthcare providers and MSM, with MSM feeling their HIV status, but almost all who were aware of their status blamed for their sexual orientation. They argue that a more had started antiretroviral therapy, highlighting that HIV testing responsive, multi-pronged strategy adaptable and relevant to uptake is a critical and challenging first step in the care cascade. healthcare needs of MSM is required and can be facilitated through continued engagement.

5 | THE HIV PREVENTION AND CARE CASCADE: INNOVATIVE ENGAGEMENTS 7 | CONCLUSIONS TO TARGET GAPS The manuscripts included in this supplement illustrate that, as Two manuscripts in this supplement focus on strategies for for other populations, we need to tailor care delivery to individ- improving engagement in the HIV prevention and care casual needs of African MSM and TGW. Young people, in particular, cade. Fearon et al.[44] also found that uptake and frequency require differentiated services that facilitate continued engage- of HIV testing uptake were key challenges, with about 78% of ment in care, such as access to a variety of dosing regimens to participants living with HIV aware of their status and less than fit their diverse lifestyles, maintenance of contact through novel two-thirds of participants at risk for HIV having received an platforms such as social media, and creation of friendly and sen- HIV test in the six months prior to enrolment in Johannes- sitive clinical care spaces. Given very high HIV incidence rates in burg, South Africa and Nairobi, Kenya. They evaluated online MSM and TGW, participants should be targeted for acute and socialization venues as tools for engaging MSM, finding that early HIV evaluation using NAAT-based HIV-testing algorithms they had been used in the last month by 60% of participants – preferably point-of-care testing – and onwards linkage to pre- in Johannesburg and 71% in Nairobi. Wide engagement with vention and care [49]. In addition, targeted testing services for social media by African MSM may make these platforms useful STIs should be offered to MSM and TGW. tools for reaching, engaging and retaining research partici- Development of effective biomedical prevention strategies pants, including young MSM and TGW at risk for HIV who such as PrEP has changed the landscape of HIV prevention, but are not adequately reached by current strategies. PrEP is only effective if people at risk are able and willing to Deficiencies in HIV testing uptake and frequency MSM in adhere to it. Differentiated models are needed for delivering sub-Saharan Africa mean that HIV is often diagnosed late in the daily PrEP and future generations of biomedical prevention inter- disease course. Palmer et al.[45] conducted a systematic review ventions to MSM in ways that yield the necessary uptake and and meta-analysis of studies reporting strategies of mobilising adherence. Further engagement of end users of these interven- MSM for testing to identify acute and early HIV infection tions is needed in order to understand the factors that motivate (AEHI) and their yield of AEHI cases. Overall, AEHI was identi- uptake and adherence in order to tailor prevention interventions fied in 6.3% (95% CI: 2.1 to 12.4) and acute HIV infection in to the unique needs of African MSM and other key populations. 0.7% (95% CI: 0.4 to 1.2) of the visits at which it was assessed. Well-laid out multi-country and multicentre cohort studies Authors showed that these yields varied substantially between similar to MACS, although costly, will bring tremendous pro- studies using targeted strategies and those with universal test- gress to understanding HIV prevention and care uptake among ing, where targeted strategies employed symptoms and or risk MSM and TGW in sub-Saharan Africa. Such efforts should be scores to guide acute HIV infection evaluation. Sadly, the World planned in collaboration with MSM and TGW stakeholders via Health Organization has no recommendation for acute HIV community organizations, key opinion leaders and community infection testing in adult populations at substantial risk of acqui- advisory boards. Cross-learning between countries may be use- sition [46]. Given the high HIV incidence rates in MSM and ful to establish best practices for participant engagement and TGW in sub-Saharan Africa, the authors conclude that targeted promotion of African investigators as leaders of research in AEHI testing can be optimized using risk scores, especially if their own communities. Future research must also move on scores include symptoms, and that studies assessing AEHI yield from past conflation of MSM with other sexual and gender in sub-Saharan Africa are urgently needed. minorities through the design of service delivery and research studies that acknowledge and address the unique needs of transgender and non-binary individuals. Tremendous progress 6 | NEW NARRATIVES AND has been made over the last 15 years in engaging African MSM MULTIPRONGED ENGAGEMENTS in HIV research. However, much more must be done.

In their commentary, Makofane et al.[47] call for a focus on ’ the specific HIV prevention and care needs of specific popula- AUTHORS AFFILIATIONS tions, including MSM and TGW that bear a disproportionate 1U.S. Military HIV Research Program, Walter Reed Army Institute of Research, Silver Spring, MD, USA; 2Henry M. Jackson Foundation for the Advancement of burden of new HIV infections. They also discuss the limitations 3 “ ” “ Military Medicine, Bethesda, MD, USA; International AIDS Vaccine Initiative, of classifying entire populations as either general or high- New York, NY, USA; 4School of Medicine, Stanford University, Palo Alto, CA, risk,” which are categories that focus on behaviour, are shaped USA; 5Desmond Tutu HIV Centre, University of Cape Town, Cape Town, South by sexual morals, and do not consider reasons for or likelihood Africa; 6KEMRI-Wellcome Trust Research Programme, Kilifi, Kenya; 7Oxford of serodiscordant sexual contact. University, Oxford, United Kingdom Lastly, van der Elst et al.[48] present a case study of public healthcare facilities at the local level in Kenya, where key stake- COMPETING INTERESTS holders navigate diverse challenges to MSM healthcare The authors report no competing interests.

3 Crowell TA et al. Journal of the International AIDS Society 2020, 23(S6):e25596 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25596/full | https://doi.org/10.1002/jia2.25596

AUTHORS’ CONTRIBUTIONS 10. Kunzweiler CP, Bailey RC, Okall DO, Graham SM, Mehta SD, Otieno FO. TAC, PEF, LGB and EJS were involved in conceptualization and writing of the Factors associated with prevalent HIV infection among Kenyan MSM: the Anza – manuscript. Mapema study. J Acquir Immune Defic Syndr. 2017;76(3):241 9. 11. Ntale RS, Rutayisire G, Mujyarugamba P, Shema E, Greatorex J, Frost SDW, et al. HIV seroprevalence, self-reported STIs and associated risk factors among ACKNOWLEDGEMENTS men who have sex with men: a cross-sectional study in Rwanda, 2015. Sex – The Guest Editors wish to acknowledge the dedicated and courageous spirit of Transm Infect. 2019;95(1):71 4. the study participants, as well as their collaborative contributions to the 12. Hessou PHS, Glele-Ahanhanzo Y, Adekpedjou R, Ahouada C, Johnson RC, research projects described in this supplement. Likewise, the hard work, persis- Boko M, et al. Comparison of the prevalence rates of HIV infection between tence and creativity of the research staff in these many studies has made the men who have sex with men (MSM) and men in the general population in sub- insights possible. This paper is published with the permission of the Director, Saharan Africa: a systematic review and meta-analysis. BMC Public Health. Kenya Medical Research Institute. 2019;19(1):1634. 13. Couderc C, Dembele Keita B, Anoma C, Wade AS, Coulibaly A, Ehouman S, et al. Is PrEP needed for MSM in West Africa? HIV incidence in a prospective FUNDING multicountry cohort. J Acquir Immune Defic Syndr. 2017;75(3):e80–2. TAC is an employee of the Henry M. Jackson Foundation for the Advance- 14. Shangani S, Naanyu V, Operario D, Genberg B. Stigma and healthcare-seek- ment of Military Medicine, Inc., (HJF) in support of the U.S. Military HIV ing practices of men who have sex with men in western Kenya: a mixed-meth- – Research Program at the Walter Reed Army Institute of Research which is ods approach for scale validation. AIDS Patient Care STDS. 2018;32(11):477 funded by a cooperative agreement between HJF and the U.S. Department of 86. Defense (W81XWH-18-2-0040). PEF is an employee of IAVI. LGB has no 15. Kim HY, Grosso A, Ky-Zerbo O, Lougue M, Stahlman S, Samadoulougou C, specific funding in support of this editorial to disclose. EJS is supported by et al. Stigma as a barrier to health care utilization among female sex workers IAVI, receiving generous support of the American people through the United and men who have sex with men in Burkina Faso. Ann Epidemiol. 2018;28 – States Agency for International Development (USAID); a National Institutes of (1):13 9. Health (NIH)–funded programme (R01AI124968), and by the Wellcome Trust. 16. Ogunbajo A, Kershaw T, Kushwaha S, Boakye F, Wallace-Atiapah ND, Nel- The KEMRI-Wellcome Trust Research Programme at the Centre for Geo- son LE. Barriers, Motivators, And Facilitators To Engagement in HIV Care graphical Medicine Research–Kilifi is supported by core funding from the Well- Among HIV-infected ghanaian men who have sex with men (MSM). AIDS Behav. – come Trust (203077). This work was also supported in part through the sub- 2018;22(3):829 39. Saharan African Network for TB/HIV Research Excellence (SANTHE), a DEL- 17. Abara WE, Garba I. HIV epidemic and human rights among men who have TAS Africa Initiative (DEL-15-006). The DELTAS Africa Initiative is an indepen- sex with men in sub-Saharan Africa: Implications for HIV prevention, care, and – dent funding scheme of the African Academy of Sciences (AAS) Alliance for surveillance. Glob Public Health. 2017;12(4):469 82. Accelerating Excellence in Science in Africa (AESA) and is supported by the 18. Risher K, Adams D, Sithole B, Ketende S, Kennedy C, Mnisi Z, et al. Sexual New Partnership for Africa’s Development Planning and Coordinating Agency stigma and discrimination as barriers to seeking appropriate healthcare among (NEPAD Agency) with funding from the Wellcome Trust (107752) and the UK men who have sex with men in Swaziland. J Int AIDS Soc. 2013;16:18715. government. 19. Duvall S, Irani L, Compaore C, Sanon P, Bassonon D, Anato S, et al. Assess- ment of policy and access to HIV prevention, care, and treatment services for men who have sex with men and for sex workers in Burkina Faso and Togo. J DISCLAIMER Acquir Immune Defic Syndr. 2015;68 Suppl 2:S189–97. The views expressed are those of the authors and should not be construed to 20. Schwartz SR, Nowak RG, Orazulike I, Keshinro B, Ake J, Kennedy S, et al. represent the positions of the U.S. Army, U.S. Department of Defense, U.S. The immediate effect of the Same-Sex Marriage Prohibition Act on stigma, dis- Agency for International Development, U.S. Government, African Academy of crimination, and engagement on HIV prevention and treatment services in men Sciences, New Partnership for Africa’s Development, Wellcome Trust, or U.K. who have sex with men in Nigeria: analysis of prospective data from the TRUST – government. cohort. Lancet HIV. 2015;2(7):e299 306. 21. Stannah J, Dale E, Elmes J, Staunton R, Beyrer C, Mitchell KM, et al. HIV testing and engagement with the HIV treatment cascade among men who have sex with men in Africa: a systematic review and meta-analysis. Lancet HIV. REFERENCES 2019;6(11):e769–87. 1. Smith AD, Tapsoba P, Peshu N, Sanders EJ, Jaffe HW. Men who have sex 22. Pillay D, Stankevitz K, Lanham M, Ridgeway K, Murire M, Briedenhann E, with men and HIV/AIDS in sub-Saharan Africa. Lancet. 2009;374(9687): et al. Factors influencing uptake, continuation, and discontinuation of oral PrEP 416–22. among clients at sex worker and MSM facilities in South Africa. PLoS One. 2. Kingsley LA, Detels R, Kaslow R, Polk BF, Rinaldo CR Jr, Chmiel J, et al. Risk 2020;15:e0228620. factors for seroconversion to human immunodeficiency virus among male homo- 23. Emmanuel G, Folayan M, Undelikwe G, Ochonye B, Jayeoba T, Yusuf A, sexuals. Results from the Multicenter AIDS Cohort Study. Lancet. 1987;1 et al. Community perspectives on barriers and challenges to HIV pre-exposure (8529):345–9. prophylaxis access by men who have sex with men and female sex workers 3. Gottlieb MS, Schroff R, Schanker HM, Weisman JD, Fan PT, Wolf RA, et al. access in Nigeria. BMC Public Health. 2020;20(1):69. Pneumocystis carinii pneumonia and mucosal candidiasis in previously healthy 24. Sandfort TGM, Knox JR, Alcala C, El-Bassel N, Kuo I, Smith LR. Substance homosexual men: evidence of a new acquired cellular immunodeficiency. N Engl use and HIV risk among men who have sex with men in africa: a systematic J Med. 1981;305(24):1425–31. review. J Acquir Immune Defic Syndr. 2017;76(2):e34–46. 4. Siegal FP, Lopez C, Hammer GS, Brown AE, Kornfeld SJ, Gold J, et al. Severe 25. Tun W, Vu L, Dirisu O, Sekoni A, Shoyemi E, Njab J, et al. Uptake of HIV acquired immunodeficiency in male homosexuals, manifested by chronic perianal self-testing and linkage to treatment among men who have sex with men ulcerative herpes simplex lesions. N Engl J Med. 1981;305(24):1439–44. (MSM) in Nigeria: a pilot programme using key opinion leaders to reach MSM. J 5. Centers for Disease Control and Prevention. Pneumocystis pneumonia–Los Int AIDS Soc. 2018;21:e25124. Angeles. MMWR Morb Mortal Wkly Rep. 1981;30(21):250–2. 26. Wirtz AL, Trapence G, Jumbe V, Umar E, Ketende S, Kamba D, et al. Feasi- 6. Centers for Disease Control and Prevention. Kaposi’s sarcoma and Pneumo- bility of a combination HIV prevention program for men who have sex with men cystis pneumonia among homosexual men–New York City and California. in Blantyre, Malawi. J Acquir Immune Defic Syndr. 2015;70(2):155–62. MMWR Morb Mortal Wkly Rep. 1981;30(25):305–8. 27. Gichuru E, Kombo B, Mumba N, Sariola S, Sanders EJ, van der Elst EM. 7. Sanders EJ, Graham SM, Okuku HS, van der Elst EM, Muhaari A, Davies A, Engaging religious leaders to support HIV prevention and care for gays, bisexual et al. HIV-1 infection in high risk men who have sex with men in Mombasa, men, and other men who have sex with men in coastal Kenya. Crit Public Kenya. AIDS. 2007;21(18):2513–20. Health. 2018;28(3):294–305. 8. Nowak RG, Mitchell A, Crowell TA, Liu H, Ketende S, Ramadhani HO, et al. 28. van der Elst EM, Smith AD, Gichuru E, Wahome E, Musyoki H, Muraguri N, Individual and sexual network predictors of HIV incidence among men who have et al. Men who have sex with men sensitivity training reduces homoprejudice sex with men in Nigeria. J Acquir Immune Defic Syndr. 2019;80(4):444–53. and increases knowledge among Kenyan healthcare providers in coastal Kenya. 9. Kimani M, van der Elst EM, Chiro O, Oduor C, Wahome E, Kazungu W, et al. J Int AIDS Soc. 2013;16 4 Suppl 3:18748. PrEP interest and HIV-1 incidence among MSM and transgender women in 29. Sullivan PS, Phaswana-Mafuya N, Baral SD, Valencia R, Zahn R, Dominguez coastal Kenya. J Int AIDS Soc. 2019;22:e25323. K, et al. HIV prevalence and incidence in a cohort of South African men and

4 Crowell TA et al. Journal of the International AIDS Society 2020, 23(S6):e25596 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25596/full | https://doi.org/10.1002/jia2.25596

transgender women who have sex with men: the Sibanye Methods for Preven- 40. Kunzweiler C, Bailey RC, Okall DO, Graham SM, Mehta SD, Otieno-Nyunya tion Packages Programme (MP3) project. J Int AIDS Soc. 2020;23(S6):e25591. B, et al. Enrolment characteristics associated with retention among HIV negative 30. Jones J, Sanchez TH, Dominguez K, Bekker L-G, Phaswana-Mafuya N, Baral Kenyan gay, bisexual and other men who have sex with men enrolled in the SD, et al. Sexually transmitted infection screening, prevalence and incidence Anza Mapema cohort study. J Int AIDS Soc. 2020;23(S6):e25598. among South African men and transgender women who have sex with men 41. Sandfort TGM, Hamilton EL, Marais A, Guo X, Sugarman J, Chen YQ, et al. enrolled in a combination HIV prevention cohort study: the Sibanye Methods The feasibility of recruiting and retaining men who have sex with men and for Prevention Packages Programme (MP3) project. J Int AIDS Soc. 2020;23(S6): transgender women in a multinational prospective HIV prevention research e25594. cohort study in sub-Saharan Africa (HPTN 075). J Int AIDS Soc. 2020;23(S6): 31. Bell S, Wapling J, Ase S, Boli-Neo R, Vallely AJ, Kaldor JM, et al. Acceptabil- e25600. ity of testing for anorectal sexually transmitted infections and self-collected anal 42. Ramadhani HO, Crowell TA, Nowak RG, Ndembi N, Kayode BO, Kokogho swabs in female sex workers, men who have sex with men and transgender A, et al. Association of age with healthcare needs and engagement among Nige- women in Papua New Guinea. BMC Public Health. 2018;18(1):776. rian men who have sex with men and transgender women: cross-sectional and 32. Wayal S, Llewellyn C, Smith H, Hankins M, Phillips A, Richardson D, et al. longitudinal analyses from an observational cohort. J Int AIDS Soc. 2020;23(S6): Self-sampling for oropharyngeal and rectal specimens to screen for sexually e25599. transmitted infections: acceptability among men who have sex with men. Sex 43. Twahirwa Rwema JO, Lyons CE, Herbst S, Liestman B, Nyombayire J, Transm Infect. 2009;85(1):60–4. Ketende S, et al. HIV infection and engagement in HIV care cascade among 33. Chesang K, Hornston S, Muhenje O, Saliku T, Mirjahangir J, Viitanen A, men who have sex with men and transgender women in Kigali, Rwanda: a cross- et al. Healthcare provider perspectives on managing sexually transmitted infec- sectional study. J Int AIDS Soc. 2020;23(S6):e25604. tions in HIV care settings in Kenya: a qualitative thematic analysis. PLoS Medi- 44. Fearon E, Bourne A, Tenza S, Palanee-Phillips T, Kabuti R, Weatherburn P, cine. 2017;14:e1002480. et al. Online socializing among men who have sex with men and transgender 34. Keshinro B, Crowell TA, Nowak RG, Adebajo S, Peel S, Gaydos CA, et al. people in Nairobi and Johannesburg and implications for public health-related High prevalence of HIV, chlamydia and gonorrhoea among men who have sex research and health promotion: an analysis of qualitative and respondent-driven with men and transgender women attending trusted community centres in sampling survey data. J Int AIDS Soc. 2020;23(S6):e25603. Abuja and Lagos, Nigeria. J Int AIDS Soc. 2016;19:21270. 45. Palmer S, Dijkstra M, Ket JCF, Wahome EW, Walimbwa J, Gichuru E, et al. 35. Ross MW, Nyoni J, Ahaneku HO, Mbwambo J, McClelland RS, McCurdy SA. Acute and early HIV infection screening among men who have sex with men, a High HIV seroprevalence, rectal STIs and risky sexual behaviour in men who have systematic review and meta-analysis. J Int AIDS Soc. 2020;23(S6):e25590. sex with men in Dar es Salaam and Tanga, Tanzania. BMJ Open. 2014;4:e006175. 46. Elliott T, Sanders EJ, Doherty M, Ndung’u T, Cohen M, Patel P, et al. Chal- 36. De Baetselier I, Vuylsteke B, Yaya I, Dagnra A, Diande S, Mensah E, et al. lenges of HIV diagnosis and management in the context of pre-exposure pro- To pool or not to pool samples for Sexually Transmitted Infections detection in phylaxis (PrEP), post-exposure prophylaxis (PEP), test and start and acute HIV Men who have Sex with Men? An evaluation of a new pooling method using the infection: a scoping review. J Int AIDS Soc. 2019;22:e25419. GeneXpert instrument in West-Africa. Sex Transm Dis. 2020;47(8):556–561. 47. Makofane K, van der Elst EM, Walimbwa J, Nemande S, Baral SD. From 37. Crowell TA, Hardick J, Lombardi K, Parker Z, Kokogho A, Amusu S, et al. general to specific: moving past the general population in the HIV response Asymptomatic lymphogranuloma venereum among Nigerian men who have sex across sub-Saharan Africa. J Int AIDS Soc. 2020;23(S6):e25605. with men. Sex Transm Infect. 2018;94(8):578–81. 48. van der Elst EM, Mudza R, Onguso JM, Kiirika L, Kombo B, Jahangir N, 38. Wahome EW, Graham SM, Thiong’o AN, Mohamed K, Oduor T, Gichuru E, et al. A more responsive, multi-pronged strategy is needed to strengthen HIV et al. Risk factors for loss to follow-up among at-risk HIV negative men who healthcare for men who have sex with men in a decentralized health system: have sex with men participating in a research cohort with access to pre-expo- qualitative insights of a case study in the Kenyan coast. J Int AIDS Soc. 2020;23 sure prophylaxis in coastal Kenya. J Int AIDS Soc. 2020;23(S6):e25593. (S6):e25597. 39. Kayode BO, Mitchell A, Ndembi N, Kokogho A, Ramadhani HO, Adebajo S, 49. Agutu CA, Ngetsa CJ, Price MA, Rinke de Wit TF, Omosa-Manyonyi G, San- et al. Retention of a cohort of men who have sex with men and transgender ders EJ, et al. Systematic review of the performance and clinical utility of point of women at risk for and living with HIV in Abuja and Lagos, Nigeria: a longitudinal care HIV-1 RNA testing for diagnosis and care. PLoS One. 2019;14:e0218369. analysis. J Int AIDS Soc. 2020;23(S6):e25592.

5 Sullivan PS et al. Journal of the International AIDS Society 2020, 23(S6):e25591 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25591/full | https://doi.org/10.1002/jia2.25591

RESEARCH ARTICLE HIV prevalence and incidence in a cohort of South African men and transgender women who have sex with men: the Sibanye Methods for Prevention Packages Programme (MP3) project Patrick S Sullivan1,§ , Nancy Phaswana-Mafuya2, Stefan D Baral3 , Rachel Valencia1, Ryan Zahn1, Karen Dominguez1, Clarence S Yah4,5 , Jeb Jones1 , Lesego B Kgatitswe6, AD McNaghten1, Aaron J Siegler1 , Travis H Sanchez1 and Linda-Gail Bekker7 §Corresponding author: Patrick S Sullivan, 1518 Clifton Road, Atlanta, Georgia 30322, United States. Tel: +404 727 2038. ([email protected]) Clinical Trial Number: NCT02043015.

Abstract Introduction: Men who have sex with men (MSM) and transgender women (TGW) are at increased risk for acquiring HIV, but there are limited HIV incidence data for these key populations in Africa. Understanding HIV prevalence and incidence provides important context for designing HIV prevention strategies, including pre-exposure prophylaxis (PrEP) programmes. We describe HIV prevalence, awareness of HIV infection, HIV incidence and associated factors for a cohort of MSM and TGW in Cape Town and Port Elizabeth, South Africa. Methods: From 2015 to 2016, MSM and TGW in Cape Town and Port Elizabeth were enrolled and prospectively followed for 12 months, receiving a comprehensive package of HIV prevention services. HIV testing was conducted at baseline and at follow-up visits (targeted for three, six and twelve months). All HIV-negative PrEP-eligible participants were offered PrEP enrolment during the first four months of study participation. We determined HIV prevalence among participants at baseline, and incidence by repeat screening of initially HIV-negative participants with HIV tests at three, six and twelve months. Results: Among 292 participants enrolled, HIV prevalence was high (43%; 95% CI: 38 to 49) and awareness of HIV status was low (50%). The 167 HIV-negative participants who were followed prospectively for 144.7 person-years; nine incident HIV infections were documented. Overall annual incidence was 6.2% (CI: 2.8 to 11.8) and did not differ by city. Annual HIV incidence was significantly higher for younger (18 to 19 years) MSM and TGW (MSM: 21.8% (CI: 1.2 to 100); TGW: 31.0 (CI: 3.7, 111.2)). About half of participants started PrEP during the study; the annual incidence of HIV among 82 (49%) PrEP star- ters was 3.6% (CI: 0.4, 13.1) and among those who did not start PrEP was 7.8% (CI: 3.1, 16.1). Conclusions: HIV incidence was high among MSM and TGW in the context of receiving a comprehensive package of prevention interventions and offering of PrEP. PrEP uptake was high; the observed incidence of HIV in those who started PrEP was about half the incidence of HIV in those who did not. Future implementation-oriented studies should focus on decisions to start and continue PrEP for those at highest risk, including young MSM. Keywords: HIV; men who have sex with men; sexually transmitted infections; pre-exposure prophylaxis; HIV prevention; cohort studies

Additional Supporting information may be found online in the Supporting Information tab for this article.

Received 4 February 2020; Accepted 15 July 2020 Copyright © 2020 The Authors. Journal of the International AIDS Society published by John Wiley & Sons Ltd on behalf of the International AIDS Society. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

1 | INTRODUCTION MSM and TGW in sub-Saharan Africa requires additional data about HIV burden, inclusive of prevalence and incidence data HIV disproportionally impacts men who have sex with men and appropriately stratified to illustrate heterogeneity within (MSM) globally, but data about the impact of HIV in MSM [1] key populations. and transgender women (TGW) [2] in sub-Saharan Africa are Among MSM across sub-Saharan Africa, there has been limited. Most data about the HIV burden among MSM are observed heterogeneity in risk and needs for prevention ser- prevalence data [1]. Understanding HIV epidemics among vices [3]. Most available HIV incidence estimates emerged

6 Sullivan PS et al. Journal of the International AIDS Society 2020, 23(S6):e25591 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25591/full | https://doi.org/10.1002/jia2.25591 from biomedical efficacy trials and a few dedicated cohorts of Eligible participants self-reported being aged ≥18 years; MSM [4,5]. HIV incidence estimates have been consistently being assigned male sex at birth; having anal intercourse with a high, from 6% to 20% in control arms and the cohort studies man in the previous 12 months; being residents of the study city; [4,6-8]. Moreover, these data have been collected in settings planning to stay in the city for the next year; being able to answer where research capacity is higher and intervention trials have survey questions in English, Xhosa or Afrikaans; being willing to been conducted [9] and from cities where there are more provide contact information; and having a phone line. developed gay communities and gay-friendly HIV prevention The prospective cohort was designed to follow 80% HIV- and care resources. We need to understand how HIV epi- negative MSM and 20% MSM living with HIV, but all partici- demics in areas with more gay-friendly services differ from pants presenting for screening were provided informed con- and are similar to epidemics in places where healthcare ser- sent and completed a baseline study visit consisting of a self- vices are usually delivered through government health clinics. administered baseline questionnaire, HIV prevention coun- Outside sub-Saharan Africa, data have consistently demon- selling and testing and a clinical assessment. The clinical exam strated higher HIV incidence among younger MSM [10]. To assessed STI history, circumcision status and STI and liver dis- date, there are limited studies describing younger MSM given ease symptoms. Laboratory testing for syphilis by RPR testing the complex ethical and legal barriers across sub-Saharan and urethral and rectal chlamydia and gonorrhoea by PCR Africa. In most sub-Saharan African countries homosexuality is testing was offered to all participants. illegal [11], frustrating access to appropriate care [12]. Homo- All prospectively-enrolled participants were invited and sexuality is not criminalized in South Africa, but stigma is encouraged to attend follow-up visits scheduled at 3-, 6- and widespread, including in healthcare settings, and can discour- 12-month post-baseline, which included surveys, HIV preven- age disclosure of sexual orientation in healthcare settings [13]. tion counselling, HIV testing for participants who tested HIV- Characterizing appropriate HIV intervention strategies negative at their last study visit, a clinical exam assessing STI requires studying the differences in HIV risks across the lifes- symptoms and blood and urine collection. Repeat testing for pan for MSM by evaluating age-specific HIV prevalence and syphilis and urethral and rectal chlamydia and gonorrhoea were incidence. offered at 6- and 12-month visits. Participants who completed a To contribute data about age-diverse MSM, TGW and MSM study visit within six weeks of the target date were considered living in different urban settings, we analysed data from the to have attended that visit. Eight participants returned Sibanye Health Project [14,15], a pilot study of a combination >6 weeks after their target 12-month visit date. These partici- package programme of behavioural, biomedical and commu- pants are included in the primary incidence analysis. nity-led HIV interventions among MSM in Cape Town and All participants interested in PrEP were assessed for PrEP Port Elizabeth, South Africa from 2015 to 2016. For the study eligibility according to then-current South African national piloting of provision of combination HIV prevention services, guidelines [16]. Only daily oral PrEP with TDF/FTC was offered we screened MSM and TGW in Cape Town and Port Eliza- during the study. Participants interested in PrEP completed beth for HIV status, and prospectively followed a cohort of additional screening at either baseline or three months and 167 HIV-negative MSM and TGW for 12 months. returned one month later for PrEP initiation (month 1 or 4). Participants on PrEP returned for monitoring visits one month 2 | METHODS after PrEP initiation (month 2 or 5) and at month 9 to assess creatinine level, liver enzymes, phosphorus, proteinuria, glyco- suria, HIV status, medication adherence and to monitor side 2.1 | Study design and population effects. Data are from the Sibanye Health Project, a pilot study of the At baseline, and follow-up visits, participants completed a acceptability and uptake of a comprehensive combination questionnaire with domains including demographics, HIV-re- package of HIV prevention services from 2015 to 2016. The lated behaviours and knowledge, sexual history, condom and process of developing the package of services has been previ- lubricant use, use of health care services, outness to provi- ously described [14]. The final package included condoms with ders, alcohol and substance use, stigma and the participant’s condom-compatible lubricant choices, HIV testing with risk- sexual network. Retention in the cohort was 88% at reduction counselling, couples HIV testing and counselling, 12 months when follow-up ended in September 2016. This screening for sexually transmitted infections (STIs), pre-expo- study was approved by the IRBs of Emory University and sure prophylaxis (PrEP) for interested and eligible participants, University of Cape Town and the Research Ethics Committee and non-occupational post-exposure prophylaxis (PEP) for of the Human Sciences Research Council. those with an exposure at high risk for transmission of HIV. For the pilot study, MSM and TGW in Cape Town and Port 2.2 | Measures Elizabeth, South Africa were recruited from February to September 2015 through community outreach, MSM social There were two main outcomes: prevalent and incident HIV gatherings and events, MSM dialogues at homes of peers, infection. HIV testing was conducted on whole blood for anti- social media advertisements, contact of participants in previ- bodies to HIV with rapid HIV tests using a serial testing algo- ous prevention studies, walk in, utilization of MSM friendly rithm per South African guidelines [17]. In Cape Town, initial organizations/clinics, networking and education about the pro- screening was done with DetermineTM Rapid HIV-1/2 test kits ject with major community stakeholders and referral by peers. (Alere Medical Co. Ltd, Matsudo-Shi Chiba, Japan). Preliminary All participants provided a written informed consent and positive results were confirmed using a Uni-GoldTM HIV test agreed to participate in the research study and to have their (Trinity Biotech PLC., Bray, Ireland). In Port Elizabeth, initial samples used for research purposes. screening was done using Advanced QualityTM Rapid HIV-1/2

7 Sullivan PS et al. Journal of the International AIDS Society 2020, 23(S6):e25591 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25591/full | https://doi.org/10.1002/jia2.25591 test (InTec Products Inc., Xiamen, China) and confirmed with proportional hazards model was used to identify factors associ- ABONTM HIV 1/2/O Triline Rapid test (Abon Biopharm, Hang- ated with higher hazard of HIV. Age group, sexual orientation zhou, R.R China). Participants who were HIV negative were and gender identity were considered as possible predictors of followed prospectively for 12 months (or more if the nominal HIV risk. Site and characteristics determined to be associated 12-month visit was delayed), with retesting for HIV at regular with HIV incidence via log-rank tests at alpha = 0.20 were follow-up, PrEP initiation and PrEP monitoring visits [14]. included in the model. Direct age-standardized estimates of HIV prevalence was the proportion of all MSM screened HIV incidence and age-standardized incidence rate ratios were for HIV at enrolment who had a confirmed positive antibody calculated for Cape-Town and Port Elizabeth using PROC test. Among MSM living with HIV, awareness was defined STDRATE (SAS 9.4, Cary, NC, USA). PrEP was included as a as reporting themselves positive in the baseline survey, self- time-varying exposure. City-specific HIV incidence density was reporting HIV-negative to staff who recorded this report on a calculated for variables that were significantly associated with CRF, and not having a suppressed HIV viral load. Participants incidence in the Cox proportional hazard model and for PrEP. who had positive HIV confirmatory testing and undetectable Timelines for all participants with incident HIV infections were HIV viral load (<20 copies/mL by RT-PCR analysis) at baseline also constructed to identify timing of important prevention were considered to be aware of their HIV status. (PrEP start and stop) and risk events (new rectal STI diagnoses) Participants who seroconverted during the study had HIV relative to time of HIV diagnosis. viral load and CD4+ T-lymphocyte testing and were referred to a local clinic for antiretroviral treatment initiation. These persons continued to be followed through the end of the 3 | RESULTS study to assess other prevention outcomes (e.g. behaviours and condom use). A total of 292 MSM and TGW were enrolled, consented and Independent variables included age, race, gender identity tested for HIV: 115 were enrolled in Cape Town, and 177 (male vs transgender or other; transgender was defined as were enrolled in Port Elizabeth. The median age of partici- participants who reported male sex at birth and current gen- pants was 24 years (range, 18 to 69); a higher proportion of der identity as female or transgender), sexual orientation, edu- MSM were young (<20 years) in Cape Town (22%) than in cation level, work or student status, income and circumcision Port Elizabeth (10%). Most were Black African (87%). About status. The behavioural independent variables included self-re- half had at least a high school education; most (90%) identi- ported receptive condomless anal intercourse with a man in fied as male and 22 (8%) identified as transgender or other the last three months, number of male partners in the past non-male identified. Enrolment (and assessment of HIV preva- 12 months, having a female partner in the past 12 months, lence) occurred from 2015 to 2016; follow-up (for incidence use of recreational drugs (e.g. marijuana) in the past six assessment) occurred from 2015 to 2017. months and binge drinking. 3.1 | HIV prevalence 2.3 | Analyses HIV prevalence at enrolment was 43% overall; and was signifi- HIV prevalence and awareness were calculated overall and for cantly higher (PR = 1.7; 95% CI = 1.2, 2.3; Table 1) in Port Eliz- subgroups of participants by demographics, behaviours and rec- abeth (51%) than in Cape Town (30%; Table 1). Self-reported tal STI prevalence; 95% confidence intervals for prevalence awareness of HIV-positive status prior to the study-delivered were calculated using the Wilson procedure without continuity HIV testing was 50% overall and was higher in Port Elizabeth correction [18]. Trends in HIV prevalence and awareness of HIV (57%; 51/90) than in Cape Town (34%; 12/35). There were sig- infection by age were described using a Cochran-Armitage test nificant differences in HIV prevalence by participant demo- for trend. Bivariate prevalence ratios were calculated for factors graphic characteristics (Table 1), including age (Figure 1). associated with prevalent HIV infection. An adjusted multivari- In the multivariable model, HIV prevalence was higher in able logistic regression model included covariates significant at Port Elizabeth than Cape Town. HIV prevalence was signifi- p < 0.20 in bivariate analysis. Remaining covariates were con- cantly lower in participants aged 20 to 24 compared to ≤25, sidered for collinearity and reported as adjusted prevalence in non-black participants, and in participants who identified as ratios (APRs) with associated 95% CIs using a conditional pre- bisexual, straight or other. HIV prevalence was significantly dicted margins approach [19]. For covariates with p < 0.20 in higher in participants who self-reported having ≥3 male sex the prevalence model, site-specific bivariate prevalence ratios partners in the past 12 months. No other behaviour or rectal were calculated and site-specific multivariable logistic regres- STI diagnosis was associated with HIV prevalence. sion models including these covariates were run. There were substantial differences in factors that were sig- HIV incidence density was calculated overall and among sub- nificantly associated with HIV prevalence in the city-stratified groups using new HIV diagnoses as the numerator and person- analyses (Table 2). For Cape Town participants, there were no time from baseline to the last completed study visit, other date significant differences in HIV prevalence by any of the inde- of loss to follow-up, or mid-point between last negative and pendent measures. For Port Elizabeth participants, there were first-positive test as the denominator. Participants were allowed significant differences in HIV prevalence for participants like to attend follow-up visits outside the exact date of the three- what was observed in the overall sample. The exception was monthly intervals. For those who dropped out of follow-up, we that participants who self-reported receptive condomless anal censored their follow time at the time of the last attended intercourse in the prior three months were more likely to study visit. Confidence intervals (alpha = 0.05) for HIV inci- have prevalent HIV infection compared to those who did not dence density were calculated using Fisher’s exact test. A Cox report condomless anal intercourse.

8 Sullivan PS et al. Journal of the International AIDS Society 2020, 23(S6):e25591 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25591/full | https://doi.org/10.1002/jia2.25591

Table 1. HIV prevalence among enrolled South African MSM and transgender women, Sibanye Health Project, 2015 to 2017

Prevalence Prevalence APR (95% CI) Number/total proportion PR (95% CI) (N = 236)

Total 125/292 43% –– Site Cape Town 35/115 30% Ref Ref Port Elizabeth 90/177 51% 1.7 (1.2, 2.3) 2.3 (1.4, 3.7) Age ranges 18 to 19 15/43 35% 0.7 (0.5, 1.1) 0.6 (0.4, 1.1) 20 to 24 47/122 39% 0.8 (0.6, 1.0) 0.7 (0.6, 1.0) 25+ 63/127 50% Ref Ref Race Black 117/254 46% Ref Ref Coloured and other 8/38 21% 0.5 (0.2, 0.9) 0.5 (0.2, 0.9) Gender identitya Male 110/263 42% Ref Ref Transgender or other non-male identifiedb 13/22 59% 1.4 (1.0, 2.1) 1.3 (0.9, 1.8) Sexual orientationa Homosexual or gay 105/192 55% Ref Ref Bisexual, straight, other 19/95 20% 0.4 (0.2, 0.6) 0.6 (0.3, 0.9) Educationa Did not matric/less than high school 54/137 39% Ref – Matric/high school graduate 46/104 44% 1.1 (0.8, 1.5) Technical or university education 23/47 49% 1.2 (0.9, 1.8) Work or student statusa Full-time student or full-time job 50/107 47% Ref – Part-time student or part-time job 16/43 37% 0.8 (0.5, 1.2) Not a student and no job 58/137 42% 0.9 (0.7, 1.2) Incomea No income 61/141 43% Ref – R1 to R4,800 26/64 41% 0.9 (0.7, 1.3) R4,801 to R9,600 10/27 37% 0.9 (0.5, 1.5) R9,601 to R19,200 6/12 50% 1.2 (0.6, 2.1) R19,201 or more 12/27 44% 1.0 (0.7, 1.6) Circumcision status Full 44/89 49% 1.3 (0.9, 1.8) 0.7 (0.5, 1.1) Partial 12/36 33% 0.9 (0.5, 1.5) 0.9 (0.6, 1.4) Uncircumcised 29/74 39% Ref Ref No exam 40/93 43% 1.1 (0.8, 1.6) 0.7 (0.3, 1.2) Self-reported receptive condomless anal intercourse in last three monthsa Yes 59/93 63% 1.9 (1.4, 2.4) 1.3 (1.0, 1.7) No 56/165 34% Ref Ref Number of male partners in past 12 monthsa 0 to 2 56/167 34% Ref Ref 3+ 69/123 56% 1.7 (1.3, 2.2) 1.4 (1.0, 1.8) Any female partner in past 12 monthsa Yes 12/60 20% 0.4 (0.2, 0.7) 0.8 (0.5, 1.3) No 113/230 49% Ref Ref Transactional sex in past 12 monthsa Yes 16/43 37% 0.8 (0.6, 1.3) – No 101/225 45% Ref Any recreational drug use in past six monthsa Yes 22/79 28% 0.6 (0.4, 0.9) 0.8 (0.6, 1.1) No 102/211 48% Ref Ref (Continues) 9 Sullivan PS et al. Journal of the International AIDS Society 2020, 23(S6):e25591 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25591/full | https://doi.org/10.1002/jia2.25591

Table 1. (Continued)

Prevalence Prevalence APR (95% CI) Number/total proportion PR (95% CI) (N = 236)

Binge drinking (5 + drinks) on 5 or more days in past 30 daysa Yes 30/57 53% 1.3 (1.0, 1.7) 1.1 (0.8, 1.5) No 89/215 41% Ref Ref Rectal STI at baseline Yes 31/60 52% 1.3 (1.0, 1.8) 1.2 (0.8, 1.7) No 52/134 39% Ref Ref No assessment 42/98 43% 1.1 (0.8, 1.5) 1.1 (0.6, 2.0)

APR, adjusted prevalence ratio; PR, Prevalence ratio; R, rand; ref, reference category; STI, sexually transmitted infection. aExcludes participants with missing data; missing 7 from gender identity measure, missing 5 from sexual orientation, missing 4 from education, missing 5 from work or student status, missing 21 from income, missing 34 from self-reported receptive condomless anal intercourse in last three months, missing 2 from number of male partners in past 12 months, missing 2 from any female partner in past 12 months, missing 24 from transactional sex in past 12 months, missing 2 from any drug use in past six months, missing 20 from binge drinking. bincludes participants who reported current gender identity as female.

Figure 1. HIV prevalence and awareness of HIV positivity by age among 237 MSM and transgender women, Cape Town and Port Elizabeth, South Africa, 2015 to 2016. MSM, men who have sex with men.

(aHR = 18.5; 95% CI = 1.7 to 196.9) than for participants 3.2 | HIV incidence aged 25 and older (1.8/100 PY). Only gender identity (higher All 167 participants who were HIV negative at baseline were in TGW and other) and sexual orientation (higher in gay than followed prospectively and contributed a total of 145 person- other) were associated significant differences in HIV incidence. years (PY) of follow-up. There were nine incident HIV infec- City-stratified analyses were conducted because of the differ- tions (six in Cape Town; three in Port Elizabeth) during fol- ences in age structure in the two cities, and because of the low-up. The rate of HIV incidence was 5.3/100 PY (95% marked difference in availability of other prevention services CI = 2.1, 10.8) among MSM and 31.0/100 PY (95% CI = 3.7, in the two cities. In the city-stratified descriptive analyses, rate 111.2) among TGW. Log-rank tests were significant for age of HIV infection among those aged 18 or 19 was 25.0/100 group (p < 0.004), sexual orientation (p < 0.033) and gender PY in Cape Town and 14.5/100 PY in Port Elizabeth identity (p < 0.02). These variables were included in the CPH (Table S1). Because of the difference in age distribution model (Table 3). The rate of HIV incidence was 6.2/100 PY between cities, age-standardized HIV incidence was calculated overall, was 8.8% in Cape Town and was 3.9% in Port Eliza- for each city by standardizing to the South African age struc- beth. More than half (5/9) of incident HIV infections occurred ture; the age-standardized incidence ratio (CT/PE) was 0.96 in participants aged 18 or 19. The rate of HIV incidence in 18 (CI: 0.55 to 1.67). No participant utilized PEP during to 19 year olds was 21.8/100 PY, >18 times higher follow-up.

10 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25591/full ulvnP tal. et PS Sullivan ora fteItrainlAD Society AIDS International the of Journal

Table 2. HIV prevalence among enrolled south African MSM and transgender women, by city, Sibanye Health Project, 2015 to 2016

Cape Town Port Elizabeth

Prevalence Prevalence APR (95% CI) Prevalence APR (95% CI) number/total Proportion PR (95% CI) (N = 104) number/total Prevalence proportion PR (95% CI) (N = 151)

Total 35/115 30% 90/177 51% 2020,

Age | https://doi.org/10.1002/jia2.25591 18 to 19 6/25 24% 0.9 (0.4, 2.2) 0.7 (0.3, 1.8) 9/18 50% 0.8 (0.5, 1.3) 0.6 (0.4, 1.0) 20 to 24 16/40 40% 1.5 (0.8, 2.8) 1.4 (0.7, 2.8) 31/82 38% 0.6 (0.4, 0.8) 0.7 (0.5, 0.9) 23 (S6):e25591 25+ 13/50 26% Ref Ref 50/77 65% Ref Race Black 30/89 34% Ref Ref 87/165 53% Ref Ref Other 5/26 19% 0.6 (0.2, 1.3) 0.5 (0.2, 1.3) 3/12 25% 0.5 (0.2, 1.3) 0.6 (0.3, 1.1) Sexual orientation Gay/homosexual 27/80 34% Ref Ref 78/112 70% Ref Ref Bisexual and other 7/32 22% 0.7 (0.3, 1.4) 0.7 (0.3, 1.5) 12/63 19% 0.3 (0.2, 0.5) 0.4 (0.3, 0.7) Missing 0/2 0% Circumcision status Full 5/20 25% 0.8 (0.3, 1.8) 0.8 (0.3, 1.8) 39/69 57% 0.9 (0.6, 1.3) 0.9 (0.6, 1.3) Partial 10/32 31% 1.0 (0.5, 1.9) 0.8 (0.3, 1.7) 2/4 50% 0.8 (0.3, 2.2) 0.5 (0.2, 1.4) Uncircumcised 19/59 32% Ref Ref 10/15 67% Ref Ref No exam 1/4 25% 0.8 (0.1, 4.5) 1.1 (0.1, 8.2) 39/89 44% 0.7 (0.4, 1.0) 0.8 (0.5, 1.2) Reported receptive condomless anal intercourse in past three months Yes 15/36 42% 1.6 (0.9, 2.9) 1.2 (0.6, 2.3) 44/57 77% 1.9 (1.5, 2.6) 1.4 (1.0, 1.8) No 18/70 26% Ref Ref 38/95 40% Ref Ref Missing 2/9 22% 8/25 32% Number of male partners in past 12 months 0 to 2 16/60 27% Ref Ref 40/107 37% Ref Ref 3+ 19/53 36% 1.3 (0.8, 2.4) 1.4 (0.7, 2.6) 50/70 71% 1.9 (1.4, 2.6) 1.4 (1.1, 1.9)

APR, adjusted prevalence ratio; MSM, men who have sex with men; PR, prevalence ratio; Ref, reference category. 11 Sullivan PS et al. Journal of the International AIDS Society 2020, 23(S6):e25591 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25591/full | https://doi.org/10.1002/jia2.25591

Table 3. HIV incidence among baseline HIV-negative South African MSM and transgender women who were prospectively followed, Sibanye Health Project, 2015 to 2017

Unadjusted Participants HIV incident Susceptible HIV Infections per 100 hazard ratio Adjusted hazards ratio followed infections person-years person-years (95% CI) (95% CI) (95% CI) (N = 153)

Total 167 9 144.7 6.2 (2.8, 11.8) –– City Cape Town 80 6 68.5 8.8 (3.2, 19.1) 2.1 (0.5, 8.4) 1.4 (0.2, 8.2) Port Elizabeth 87 3 76.2 3.9 (0.8, 11.5) Ref Ref Age 18 to 19 28 5 22.9 21.8 (7.1, 51.0) 12.6 (1.5, 18.5 (1.7, 196.9) 108.5) 20 to 24 75 3 65.8 4.6 (0.9, 13.3) 2.6 (0.3, 24.7) 3.2 (0.2, 41.9) 25+ 64 1 56.0 1.8 (0.1, 10.0) Ref Ref Race Black 137 8 118.4 6.8 (2.9, 13.3) Ref – Other 30 1 26.3 3.8 (0.1, 21.2) 0.6 (0.1, 4.4) – Gender Identity Male 153 7 133.3 5.3 (2.1, 10.8) Ref Ref Transgender and 9 2 6.5 31.0 (3.7, 111.2) 5.5 (1.1, 26.3) 8.7 (1.3, 57.2) Other Sexual orientation Gay/homosexual 87 8 75.5 10.6 (4.6, 20.9) 7.0 (0.9, 55.8) 11.4 (1.0, 132.2) Bisexual, straight, 76 1 66.2 1.5 (0.0, 8.4) Ref Ref other Circumcision status Full 45 1 40.1 2.5 (0.1, 13.9) 0.2 (0.0, 1.5) – Partial 24 1 24.0 4.2 (0.1, 23.2) 0.3 (0.0, 2.4) – Uncircumcised 45 5 36.0 13.9 (4.5, 32.4) Ref – No exam 53 2 44.7 4.5 (0.5, 16.2) 0.3 (0.1, 1.8) – Education Did not matric/less 83 9 69.6 12.9 (6.0, 24.6) –– than high school Matric/high school 58 0 50.9 0.0 –– Technical or 24 0 22.2 0.0 –– university education Combined work/student Full-time student or 57 4 49.1 8.1 (2.2, 20.9) Ref – full-time job Part-time student 27 0 24.2 0.0 –– or part-time job Not a student and 79 5 67.5 7.4 (2.4, 17.3) 0.9 (0.2, 3.4) – no job Income No income 80 5 66.6 7.5 (2.4, 17.5) Ref – R1 to R4,800 38 1 32.8 3.1 (0.1, 17.0) 0.4 (0.0, 3.6) – R4,801 to R9,600 17 1 15.6 6.4 (0.2, 35.7) 0.9 (0.1, 7.5) – R9,601 to R19,200 6 0 5.6 0.0 (–) –– R19,201 or more 15 0 14.4 0.0 (–) –– Missing 11 2 9.8 20.3 (2.5, 73.7) 2.7 (0.5, 13.9) –

(Continues)

12 Sullivan PS et al. Journal of the International AIDS Society 2020, 23(S6):e25591 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25591/full | https://doi.org/10.1002/jia2.25591

Table 3. (Continued)

Unadjusted Participants HIV incident Susceptible HIV Infections per 100 hazard ratio Adjusted hazards ratio followed infections person-years person-years (95% CI) (95% CI) (95% CI) (N = 153)

On PrEP Yes 82 2 55.3 3.6 (0.4, 13.1) 0.7 (0.2, 3.0) 0.6 (0.1, 2.6) No 85 7 89.4 7.8 (3.1, 16.1) Ref Ref

MSM, men who have sex with men. TGW, transgender women; PrEP, pre-exposure prophylaxis; STI, sexually transmitted infections Timeline depic- tion of key events in the study participation of nine MSM and TGW who seroconverted for HIV during study follow-up. Participants were enrolled for a period of 12 months with visits and HIV and STI screening at 3, 6, 9 and 12 months, although the actual dates of study visits varied. Events depicted on the timeline show PrEP-related events, participant reported sexual risks, STI diagnoses and HIV diagnoses.

Figure 2. Participant timelines for nine participants who seroconverted for HIV during follow-up, Cape Town and Port Elizabeth, 2015 to 2017.

The study timelines and relevant clinical events for the nine example, 7% in Kenya; 12.5% in South Africa) [5,7]. In other participants who seroconverted during the study are shown in settings, comprehensive prevention programmes including Figure 2. Only five of the nine used PrEP during the study PrEP have resulted in lowered risk of infection for MSM period; of those, two used PrEP for less than a week (due to [20,21]. The Sibanye study included the delivery of culturally side effects), and three reported using PrEP for longer peri- competent and tailored HIV prevention services for MSM; ods, albeit reporting limited adherence. Of the four serocon- thus, participants in this study are likely at lower risk for HIV verting participants who did not use PrEP, one was not infection than other MSM. Our data reinforce the importance interested in PrEP and two were interested in PrEP but did of PrEP to complement existing HIV prevention strategies. not return for a follow-up PrEP initiation visit. All but two of They also provide a window to expected challenges in PrEP the seroconverters had one or more rectal STI diagnoses dur- implementation for South African MSM and TGW. ing the follow-up period. Most HIV diagnoses were made at The data on HIV prevalence and HIV incidence suggest an the final (12-month) study visit. Overall, 57% to 72% of partic- intense HIV epidemic among MSM in these cities, calling for ipants on PrEP reported adherence across PrEP study visits. sustained and comprehensive HIV prevention programmes, with appropriate prioritization of MSM and TGW. In this study, over one-third of participants aged 18 to 19 years were 4 | DISCUSSION already living with HIV, suggesting that many new HIV infections in these men and TGW occur during early adolescence HIV incidence in MSM from two South African cities are con- [22]. Alternatively, these might have represented perinatal sistent with incidence data from elsewhere in Africa (for infections that were not previously diagnosed. Therefore, it is

13 Sullivan PS et al. Journal of the International AIDS Society 2020, 23(S6):e25591 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25591/full | https://doi.org/10.1002/jia2.25591 important to reach these men and women with information Finally, because PrEP use was not assigned to participants about HIV and MSM- and trans-friendly prevention, HIV randomly, our data might be subject to confounding by indica- screening and health services during their teenage years. HIV tion, such that the participants at highest risk were most likely prevalence peaked by age 25 to 29, reinforcing the idea that to use PrEP. However, we did not observe any differences in opportunities for prevention are most critical earlier in the life PrEP uptake based on measured risk behaviours including course. Intervening with primary prevention at younger ages receptive anal intercourse in the past three months, partici- also results in lower lifetime transmission risk, providing good pating in exchange sex, or having multiple partners. value for prevention investment [23]. Similarly, only about one A contextual interpretation of the results of our study in seven participants who were <20 years old and living with requires an understanding that the Sibanye Health Project HIV were aware of their HIV infection, and half of participants was primarily designed to show the acceptability and uptake living with HIV aged 20 to 29 were unaware of their infec- of HIV prevention services for MSM and TGW in these two tions. The major change in the prevalence epidemiology for South African cities [14]. As such, participants were offered a MSM aged ≥30 years was higher awareness of infec- broad range of prevention services, including HIV and STI tion (76%). This suggests that developing and promoting cul- testing, couples risk reduction counselling, free condoms and turally relevant HIV testing programmes for MSM and TGW condom-compatible lubricant [33] and PrEP services – all pro- aged 18 to 29 would be especially important in helping those vided in culturally competent service settings. Therefore, the living with HIV to become aware of their infections and gain finding of over 6% annual incidence in the context of a com- the benefits of medical care and treatment [24]. prehensive package of prevention offerings delivered by provi- HIV incidence data show clear disparities in risk among the ders trained in serving MSM and TGW should be interpreted MSM and TGW and should inform prevention programmes. as an underestimate of the true risk in these populations, Consistent with the interpretation of age-specific HIV preva- many of whom do not have access to this broad range of pre- lence data, the HIV incidence among MSM and TGW aged 18 vention services and providers trained to provide culturally to 19 was over 21% per year – a staggering statistic, but one congruent care. For example in other settings, incidence in that is consistent with other data from across sub-Saharan MSM in Africa has been estimated to be as high as 8.6% [4]. Africa and beyond that show much higher incidence among This suggests that the components of a comprehensive pre- younger than in older MSM [4,6-8]. The concentration of HIV vention package including PrEP are crucial, but that delivering risk among young MSM identified in South Africa is part of a these tools in diverse service settings, implementation and larger global vulnerability of young MSM [10,25-27]. There getting to scale are outstanding challenges to achieving popu- are unique challenges to providing HIV prevention services to lation-level reductions in HIV incidence among MSM and MSM and TGW. Ages of sexual debut range widely [28], and TGW [34]. many young men engage in sex with male partners for a per- Our study has important limitations, primarily related to the iod of time before identifying as gay or bisexual or accessing context of the Sibanye Health Project. Our data were col- gay-serving or culturally competent health services [29]. Inci- lected in 2015 to 2017; even though our data were collected dence was also higher for MSM who identified as gay and for several years ago, recent publications suggest that the high TGW (versus participants with male gender identity). These incidence observed in our study might be ongoing in African findings are consistent with other data from the United MSM [5,7,9]. Our convenience sample is not representative of States, Asia [30] and Africa [31]. HIV incidence was lower for all MSM and TGW in the two study cities. We took steps to MSM who identified as straight; for MSM, straight identity minimize selection bias (with respect, for example to knowl- might be confounded with sexual positioning (i.e. participants edge of HIV status) by recruiting all MSM and TGW, regard- who identified as straight might have been more likely to have less of perceived HIV status, to the baseline study visit; all an insertive role for anal sex, which is associated with lower participants who screened HIV negative were invited to par- per-act risk of HIV acquisition) [32]. ticipate in the incidence cohort. Our data are subject to social Our data did not document a significant reduction in HIV desirability or obsequiousness bias [35]. Exposures to risk or incidence associated with PrEP use, despite nearly half of the protective behaviours were not assigned; therefore confound- MSM in the cohort engaging in the PrEP programme at some ing by indication or by other measured or unmeasured cofac- time during their year of prospective follow time. The point tors might affect our measures of association. Our estimates estimate suggested a 40% reduction in HIV incidence, but we of HIV incidence were likely artificially lowered by the provi- had a small absolute number of seroconversions and limited sion of free high-quality prevention services, and the higher power to detect an association of PrEP with incidence. Inter- frequency of interactions with healthcare providers associated preting our PrEP findings, it is also important to note that no with PrEP use. We had 12% loss to follow-up; it is possible HIV infections occurred in 37 participants who were pre- that these participants dropped out of care because they scribed PrEP in Port Elizabeth; all new infections among MSM tested positive for HIV elsewhere. One participant who sero- on PrEP were observed in Cape Town. There are several pos- converted for HIV was tested after the 12-month follow-up sible reasons that we failed to find a significant protective timepoint; his testing was in line with the IRB protocol, and effect of PrEP in our data. First, our sample size might have the follow time for his seroconversion appropriately accounted been inadequate. The Sibanye study was developed to under- for in the survival analyses. stand the uptake of specific prevention services, not to iden- The results presented here demonstrate two key concepts: tify the effectiveness of PrEP. The current analysis does not the overall high impact of HIV among MSM and TGW globally account for PrEP adherence; analyses of additional data on and the heterogeneity of risks within these groups. The epi- self-reported and laboratory-measure PrEP adherence might demiology of HIV is similar among MSM and TGW in two help provide context to our data on PrEP effectiveness. diverse South African cities and among MSM and TGW in

14 Sullivan PS et al. Journal of the International AIDS Society 2020, 23(S6):e25591 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25591/full | https://doi.org/10.1002/jia2.25591

Asia, Europe and North America. In addition, these data high- ACKNOWLEDGEMENT light the heterogeneity in HIV risks within and across popula- FUNDING tions even in the context of a general HIV epidemic in South Africa. The current stated goal of the global HIV response is This work was supported by the Center for AIDS Research at Emory University to end new HIV infections by 2030, but the data presented (P30AI050409) and by the National Institute for Allergy and Infectious Diseases (R01AI094575). The authors acknowledge the Centers for Disease Control and here highlight high HIV incidence, particularly among youth, Prevention for support of clinical prevention services for participants. which challenges the likelihood of achieving this goal. Specifi- cally, the findings that by the age of 19, over a third of young REFERENCES participants were living with HIV and by age 25 about half of 1. Beyrer C, Baral SD, van Griensven F, Goodreau SM, Chariyalertsak S, Wirtz participants were living with HIV suggest the importance of AL, et al. Global epidemiology of HIV infection in men who have sex with men. tailored responses to address these unmet needs. Traditionally, Lancet. 2012;380(9839):367–77. MSM and TGW have been considered less relevant to HIV 2. Poteat T, Ackerman B, Diouf D, Ceesay N, Mothopeng T, Odette K-Z, et al. responses in more generalized HIV epidemic settings. How- HIV prevalence and behavioral and psychosocial factors among transgender ever, when studied empirically, the HIV transmission dynamics women and cisgender men who have sex with men in 8 African countries: A cross-sectional analysis. PLoS Medicine. 2017;14:e1002422. among MSM and TGW in South Africa are like other settings, 3. Baral S, Rao A, Sullivan P, Phaswana-Mafuya N, Diouf D, Millett G, et al. The suggesting that dedicated services for MSM and TGW are disconnect between individual-level and population-level HIV prevention bene- fundamental to a comprehensive HIV response. fits of antiretroviral treatment. Lancet HIV. 2019;6(9):e632–8. 4. Sanders EJ, Okuku HS, Smith AD, Mwangome M, Wahome E, Fegan G, et al. High HIV-1 incidence, correlates of HIV-1 acquisition, and high viral loads fol- | lowing seroconversion among men who have sex with men in Coastal Kenya. 5 CONCLUSIONS AIDS. 2013;27(3):437–46. 5. Price MA, Rida W, Mwangome M, Mutua G, Middelkoop K, Roux S, et al. Services for MSM and TGW should be scaled up urgently, Identifying at-risk populations in Kenya and South Africa: HIV incidence in consistent with the goals of the Ethekwini Declaration cohorts of men who report sex with men, sex workers, and youth. J Acquir – launched in 2019 calling for a renewed focus on HIV in South Immune Defic Syndr. 2012;59(2):185 93. 6. Couderc C, Keita BD, Anoma C, Wade AS, Coulibaly A, Ehouman S, et al. Is Africa with specific attention to those most traditionally PrEP needed for MSM in West Africa? HIV incidence in a prospective multi- marginalized. country cohort. J Acquir Immune Defic Syndr. 2017;75(3):e80–2. 7. Lane T, Osmand T, Marr A, Struthers H, McIntyre JA, Shade SB. Brief report: high HIV incidence in a South African community of men who have sex COMPETING INTERESTS with men: results from the Mpumalanga men’s study, 2012–2015. J Acquir No competing interests are declared. Immune Defic Syndr. 2016;73(5):609–11. 8. Sandfort T, Mbilizi Y, Sanders EJ, Guo X, Cummings V, Hamilton EL, et al.HIV incidence and correlates of HIV acquisition in a multinational cohort of AUTHORS’ AFFILIATIONS men and transgender women who have sex with men in sub-Saharan Africa: 1Rollins School of Public Health, Emory University, Atlanta, GA, USA; 2DVC findings from HPTN 075. 2019. Research and Innovation Office, North West University, Potchefstroom, South 9. Sanders EJ, Okuku HS, Smith AD, Mwangome M, Wahome E, Fegan G, et al. Africa; 3Bloomberg School of Public Health, Johns Hopkins University, Balti- High HIV-1 incidence, correlates of HIV-1 acquisition, and high viral loads fol- more, MD, USA; 4Wits Reproductive Health and HIV Institute, Faculty of Health lowing seroconversion among MSM. AIDS. 2013;27(3):437–46. Sciences, University of the Witwatersrand, Johannesburg, South Africa; 5School 10. Holtz T, Wimonsate W, Mock P, Pattanasin S, Chonwattana W, Thienkrua of Health Systems and Public Health, University of Pretoria, Pretoria, South W, et al. Why we need pre-exposure prophylaxis: incident HIV and syphilis Africa; 6Human Sciences Research Council of South Africa, Port Elizabeth, South among men, and transgender women, who have sex with men, Bangkok, Thai- Africa; 7Desmond Tutu HIV Center, University of Cape Town, Cape Town, South land, 2005–2015. Int J STD AIDS. 2019;30(5):430–9. Africa 11. Alozie NO, Thomas K, Akpan-Obong P. Global liberalization on homosexuality: explaining the African gap. Soc Sci J. 2017;54(2):120–31. 12. Stromdahl€ S, Williams AO, Eziefule B, Emmanuel G, Iwuagwu S, Anene O, AUTHORS’ CONTRIBUTIONS et al. An assessment of stigma and human right violations among men who have PSS conceived of the study, obtained funding, wrote the protocol, participated sex with men in Abuja, Nigeria. BMC Int Health Hum Rights. 2019;19(1):7. in site development, analysed data and wrote the manuscript. RPM participated 13. Stahlman S, Sanchez TH, Sullivan PS, Ketende S, Lyons C, Charurat ME, in writing the protocol, oversaw the Port Elizabeth site activities, and edited the et al. The prevalence of sexual behavior stigma affecting gay men and other manuscript. SDB participated in writing the protocol, site development, led pro- men who have sex with men across sub-Saharan Africa and in the United States. vider trainings and edited the manuscript. RV participated in site development JMIR Public Health Surveill. 2016;2(2):e35. and management, analysed data and edited the manuscript. RZ participated in 14. McNaghten A, Kearns R, Siegler AJ, Phaswana-Mafuya N, Bekker LG, site development and management, analysed data and edited the manuscript. Stephenson R, et al. Sibanye methods for prevention packages program project KD participated in site development and management for the Cape Town site protocol: pilot study of HIV prevention interventions for men who have sex with and edited the manuscript. CY participated in site development and manage- men in South Africa. JMIR Res Protoc. 2014;3(4):e55. ment for the Port Elizabeth site and edited the manuscript. JJ analysed data 15. Sullivan PS. Comprehensive HIV prevention package for MSM in Southern and edited the manuscript. LK participated in protocol implementation and edi- Africa: pilot study (Sibanye Health Project). 2014. ted the manuscript. AM wrote the protocol, provided administrative oversight 16. The Consensus Committee of the Southern African HIV Clinicians Society. to the study and edited the manuscript. AJS participated in writing the protocol Southern African guidelines for the safe use of pre-exposure prophylaxis in men and edited the manuscript. THS participated in writing the protocol and pro- who have sex with men who are at risk for HIV infection. S Afr J HIV Med. vided overall management for study sites and data activities, and edited the 2012;13(2):40–55. manuscript. LGB participated in writing the grant and protocol, oversaw the 17. South Africa Department of Health. Updated HIV Counseling and Testing Cape Town site and edited the manuscript. Guidelines 2015 [online public health guideline]. [cited 2019 Jun 18]. Available from: https://www.health-e.org.za/wp-content/uploads/2015/07/HCT-Guidelines- 2015.pdf ABBREVIATIONS 18. Newcombe RG. Two-sided confidence intervals for the single proportion: comparison of seven methods. Stat Med. 1998;17(8):857–72. APR, adjusted prevalence ratio; MSM, men who have sex with men; PrEP, pre- 19. Muller CJ, MacLehose RF. Estimating predicted probabilities from logistic exposure prophylaxis; STI, sexually transmitted infections; TGW, transgender regression: different methods correspond to different target populations. Int J women. Epidemiol. 2014;43(3):962–70.

15 Sullivan PS et al. Journal of the International AIDS Society 2020, 23(S6):e25591 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25591/full | https://doi.org/10.1002/jia2.25591

20. Brown AE, Mohammed H, Ogaz D, Kirwan PD, Yung M, Nash SG, et al. Fall 29. McCabe J, Brewster KL, Tillman KH. Patterns and correlates of same-sex in new HIV diagnoses among men who have sex with men (MSM) at selected sexual activity among U.S. teenagers and young adults. Perspect Sex Reprod London sexual health clinics since early 2015: testing or treatment or pre-expo- Health. 2011;43(3):142–50. sure prophylaxis (PrEP)? Euro Surveill. 2017;22(25):30553. 30. Chariyalertsak S, Kosachunhanan N, Saokhieo P, Songsupa R, Wongthanee 21. Grulich AE, Guy R, Amin J, Jin F, Selvey C, Holden J, et al. Population-level A, Chariyalertsak C, et al. HIV incidence, risk factors, and motivation for effectiveness of rapid, targeted, high-coverage roll-out of HIV pre-exposure pro- biomedical intervention among gay, bisexual men, and transgender persons in phylaxis in men who have sex with men: the EPIC-NSW prospective cohort Northern Thailand. PLoS One. 2011;6:e24295. study. Lancet HIV. 2018;5(11):e629–37. 31. Baral SD, Grosso A, Holland C, Papworth E. The epidemiology of HIV 22. Gray GE. Adolescent HIV—cause for concern in Southern Africa. PLoS among men who have sex with men in countries with generalized HIV epi- Medicine. 2010;7(2):e1000227. demics. Curr Opin HIV AIDS. 2014;9(2):156–67. 23. Galarraga O, Colchero MA, Wamai RG, Bertozzi SM. HIV prevention cost- 32. Patel P, Borkowf CB, Brooks JT, Lasry A, Lansky A, Mermin J. Estimating per- effectiveness: a systematic review. BMC Public Health. 2009;9(S1):S5. act HIV transmission risk: a systematic review. AIDS. 2014;28(10):1509–19. 24. Trickey A, May MT, Vehreschild J-J, Obel N, Gill MJ, Crane HM, et al. Sur- 33. Siegler AJ, Voux A, Phaswana-Mafuya N, Bekker LG, Sullivan PS, Baral SD, vival of HIV-positive patients starting antiretroviral therapy between 1996 and et al. Elements of condom-use decision making among south african men who 2013: a collaborative analysis of cohort studies. Lancet HIV. 2017;4(8):e349– have sex with men. J Int Assoc Provid AIDS Care. 2014;13(5):414–23. 56. 34. Caceres CF, Borquez A, Klausner JD, Baggaley R, Beyrer C. Implementation 25. Thienkrua W, van Griensven F, Mock PA, Dunne EF, Raengsakulrach B, of pre-exposure prophylaxis for human immunodeficiency virus infection: progress Wimonsate W, et al. Young men who have sex with men at high risk for HIV, and emerging issues in research and policy. J Int AIDS Soc. 2016;19:21108. Bangkok MSM Cohort Study, Thailand 2006–2014. AIDS Behav. 2018;22 35. Sackett DL. Bias in analytic research. J ChronicDis. 1979;32(1–2):51–63. (7):2137–46. 26. van Griensven F, Mock PA, Benjarattanaporn P, Premsri N, Thienkrua W, Sabin K, et al. Estimating recent HIV incidence among young men who have sex with men: reinvigorating, validating and implementing Osmond’s algorithm for behavioral imputation. PLoS One. 2018;13:e0204793. SUPPORTING INFORMATION 27. Chapin-Bardales J, Schmidt AJ, Guy RJ, Kaldor JM, McGregor S, Sasse A, et al. Trends in human immunodeficiency virus diagnoses among men who have Additional Supporting Information may be found in the online sex with men in North America, Western Europe, and Australia, 2000–2014. version of this article: – Ann Epidemiol. 2018;28(12):874 80. Table S1. HIV incidence among baseline HIV-negative South 28. Brink J. Considerations for South African higher education: a ‘National student men who have sex with men’sexual behaviour survey. S Afr J High Educ. African MSM and transgender women who were prospectively 2017;31(4):184–207. followed, by city, Sibanye Health Project, 2015 to 2017

16 Jones J et al. Journal of the International AIDS Society 2020, 23(S6):e25594 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25594/full | https://doi.org/10.1002/jia2.25594

RESEARCH ARTICLE Sexually transmitted infection screening, prevalence and incidence among South African men and transgender women who have sex with men enrolled in a combination HIV prevention cohort study: the Sibanye Methods for Prevention Packages Programme (MP3) project Jeb Jones1,§ , Travis H Sanchez1, Karen Dominguez1,2, Linda-Gail Bekker2 , Nancy Phaswana-Mafuya3, Stefan D Baral5, AD McNaghten1, Lesego B Kgatitswe4, Rachel Valencia1, Clarence S Yah6,7 , Ryan Zahn1, Aaron J Siegler1 and Patrick S Sullivan1 §Corresponding author: Jeb Jones, 1518 Clifton Road, Atlanta, Georgia 30322, USA. Tel: +404 712 2275. ([email protected])

Abstract Introduction: Men who have sex with men (MSM) and transgender women (TGW) experience high incidence and prevalence of sexually transmitted infections (STI), and data are needed to understand risk factors for STIs in these populations. The Siba- nye Health Project was conducted in Cape Town and Port Elizabeth, South Africa from 2015 to 2016 to develop and test a package of HIV prevention interventions for MSM and TGW. We describe the incidence, prevalence and symptoms of Chlamy- dia trachomatis (CT), Neisseria gonorrhea (NG) and syphilis observed during the study. Methods: Participants completed HIV testing at baseline. All participants who were HIV negative were followed prospectively. Additionally, a sample of participants identified as living with HIV at baseline was selected to be followed prospectively so that the prospective cohort was approximately 20% HIV positive; the remaining participants identified as HIV positive at baseline were not followed prospectively. Prospective participants were followed for 12 months and returned for clinic-based STI/HIV testing and assessment of STI symptoms at months 6 and 12. Additional HIV/STI testing visits could be scheduled at participant request. Results: Following consent, a total of 292 participants attended a baseline visit (mean age = 26 years), and 201 were enrolled for the 12-month prospective study. Acceptance of screening for syphilis and urethral NG/CT was near universal, though acceptance of screening for rectal NG/CT was lower (194/292; 66%). Prevalence of urethral CT and NG at baseline was 10% (29/289) and 3% (8/288) respectively; incidence of urethral CT and NG was 12.8/100 person-years (PY) and 7.1/100 PY respectively. Prevalence of rectal CT and NG at baseline was 25% (47/189) and 16% (30/189) respectively; incidence of rectal CT and NG was 33.4/100 PY and 26.8/100 PY respectively. Prevalence of syphilis at baseline was 17% (45/258) and incidence was 8.2/100 PY. 91%, 95% and 97% of diagnosed rectal NG/CT, urethral NG/CT and syphilis infections, respectively, were clinically asymptomatic. Conclusions: Prevalence and incidence of urethral and rectal STIs were high among these South African MSM and TGW, and were similar to rates in other settings in the world. Clinical symptoms from these infections were rare, highlighting limitations of syndromic surveillance and suggesting the need for presumptive testing and/or treatment to address the STI epidemic among MSM/TGW in South Africa. Keywords: men who have sex with men; chlamydia; gonorrhoea; syphilis; STI testing; STI incidence; HIV; MSM/TGW; sexually transmitted infections; chlamydia; gonorrhoea; syphilis

Additional Supporting information may be found online in the Supporting Information tab for this article.

Received 28 January 2020; Accepted 15 July 2020 Copyright © 2020 The Authors. Journal of the International AIDS Society published by John Wiley & Sons Ltd on behalf of the International AIDS Society. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

17 Jones J et al. Journal of the International AIDS Society 2020, 23(S6):e25594 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25594/full | https://doi.org/10.1002/jia2.25594

1 | INTRODUCTION All eligible participants completed a baseline visit and were included in the baseline cohort. All HIV negative and a sample There is evidence worldwide that men who have sex with of HIV-positive participants were then enrolled into a men (MSM) and transgender women (TGW) experience high prospective cohort. The prospective cohort was designed to rates of sexually transmitted infections (STIs), such as syphilis, be 20% MSM and TGW currently living with HIV, with the Chlamydia trachomatis (CT) and Neisseria gonorrhea (NG) [1,2]. remainder at risk of HIV. The remaining participants who were The South African National Strategic Plan on HIV/AIDS, HIV positive at the baseline visit were not enrolled in the Tuberculosis and STIs identifies MSM and TGW as key popu- prospective cohort. Prospective participants were followed for lations [3]. Many STIs – especially rectal STIs in MSM and one year and completed STI screening at 6- and 12-month TGW – are asymptomatic [4], and timely diagnosis and treat- timepoints. Participants were compensated R65 for each of ment for individual and public health benefits requires routine these study visits and up to R60 for transport to attend study screening [5]. Current STI screening recommendations in visits. Additional ad hoc visits also occurred for patients who South Africa are based on syndromic surveillance and manage- initiated pre-exposure prophylaxis (PrEP) or who requested ment [6]. Much of the STI screening that is conducted with STI testing and/or treatment. Tests for syphilis and rectal and MSM and TGW is by blood or urine specimens, but in some urethral CT and NG were conducted at baseline, month 6 and studies of MSM and TGW, there is considerably higher preva- month 12. Participants could choose to opt out of testing. lence of rectal STIs than urethral STIs or syphilis [7-9] and This study was approved by the Institutional Review Board multi-site screening has been shown to substantially increase of Emory University, the University of Cape Town Institutional the yield of positive tests [10]. Review Board and the Research Ethics Committee of the Despite a well-described body of research on STI preva- Human Sciences Research Council. Informed consent was lence among MSM [11-13], there are few published studies of obtained from participants at the beginning of the baseline incident STI infection among MSM and TGW that examine study visit. factors associated with STI acquisition. These data are needed to draw stronger inferences about risk factors for STI acquisi- 2.2 | Measures tion and potential intervention targets for STI prevention. A recent study in the Netherlands identified partner age, HIV STI testing was conducted at baseline, month 6, and month infection and sex following alcohol consumption as risk factors 12. Syphilis testing was performed using the syphilis rapid for incident STIs among MSM [14]. The dearth of data may be plasma reagin (RPR) test. Positive test results were confirmed due to the complexity in differentiating persistent STI infec- with titres and T pallidum particle agglutination (TPPA). Urine tion from true STI incidence, which is likely only feasible in a was self-collected and rectal swabs were taken by clinician prospective research design in which treatment can be veri- direct swabbing to obtain samples for CT and NG testing. CT fied. These studies have not been commonly undertaken solely and NG were diagnosed using the Cepheid GeneXpert NG/ for examining STI incidence, but have been conducted within CT test in Cape Town and Gen-Probe Aptima Assay in Port HIV prevention or epidemiological research [12,15]. Elizabeth. A clinical exam and patient history were also con- The Sibanye Health Project, a pilot study of a comprehen- ducted at all visits to assess the extent to which STIs were sive HIV prevention programme for MSM and TGW, was con- symptomatic. Visual genital inspections were conducted to ducted in Cape Town and Port Elizabeth, South Africa. The note the presence of urethral or perianal STI signs/symptoms project enrolled and prospectively followed a cohort of MSM (urethral symptoms: urethral discharge and painful/burning and TGW who selected from a suite of HIV prevention ser- sensation during urination; rectal symptoms: rectal discharge, vices that included STI screening and treatment [16]. In this anal itching and painful bowel movements). Syphilis signs/ study, we examined screening acceptance, STI prevalence and symptoms included ulcers on the genitals, rectum or buttocks incidence and treatment of diagnosed STIs. We also assessed and vesicles in the rectal or groin area. Diagnoses were made factors associated with STI prevalence and predictive of inci- based on laboratory results; participants who received an STI dent STI diagnosis. diagnosis were provided appropriate treatment or referred to a local clinic. 2 | METHODS Demographic and behavioural data were collected via self- administered surveys at all study visits. Participants reported age, race, gender and sexual identity, highest educational 2.1 | Study population and procedures attainment, work/student status, and income. Relevant beha- The Sibanye Health Project was conducted in Cape Town and vioural variables included sexual risk factors (receptive con- Port Elizabeth, South Africa from 2015 to 2016 to develop a domless anal intercourse, number of male and female package of HIV prevention interventions for MSM and TGW partners, transactional sex) and substance use. in South Africa and to conduct a pilot study to test the package of interventions [16]. Eligible participants were at least 2.3 | Analyses 18 years old, had anal sex with a man in the previous 12 months, resided in Cape Town or Port Elizabeth with plans Acceptance of NG/CT screening was defined as agreement to stay in the city for the next year, could complete surveys in for specimen collection for screening by anatomical site. Syphi- English, Xhosa or Afrikaans, were assigned male sex at birth, lis testing was conducted routinely as part of the blood collec- were willing to provide contact information, and had a phone tion performed at the scheduled visits; thus, acceptance of to facilitate scheduling study visits. Participants who identified syphilis testing was defined as agreement for blood collection. as any gender other than male were classified as TGW. Acceptance of urethral and rectal STI specimen collection for

18 Jones J et al. Journal of the International AIDS Society 2020, 23(S6):e25594 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25594/full | https://doi.org/10.1002/jia2.25594 screening is reported at baseline (all enrolled participants), at Table 1. Demographic characteristics of men who have sex any point during the twelve-month follow-up and at the six- with men and transgender women enrolled for baseline and twelve-month visits specifically (prospectively enrolled (N = 292) and prospective follow-up (N = 201) in Cape Town participants only). We present uptake at any point during the and Port Elizabeth, South Africa 12-month period because screening could occur at ad hoc visits outside the 6-monthly visit schedule. Baseline Only Prospective STIs detected at the baseline study visit were considered to Total Participants Participants be prevalent infections. If treatment of diagnosed STIs was N N (%) N (%) confirmed, subsequent STIs were considered to be incident infections. Concurrent STIs were identified if a participant was Site infected with more than one organism at the same time point Cape Town 115 15 (16.5) 100 (49.8) or infection with the same organism at more than one Port Elizabeth 177 76 (83.5) 101 (50.3) anatomical site. STI prevalence for urethral and rectal CT, urethral and rectal NG, and syphilis are reported at baseline for Age all enrolled participants, regardless of whether they were 18 to 24 165 43 (47.3) 122 (60.7) enrolled in the prospective cohort. 25+ 127 48 (52.8) 79 (39.3) Unadjusted associations between STI screening acceptance Race and prevalence and demographic, clinical and behavioural fac- Black 254 89 (87.0) 165 (82.1) tors were assessed via chi square tests except when expected Other 38 2 (13.0) 36 (17.9) cell values were small and Fisher exact tests were used. We Gender identity used Poisson regression with robust variance [17] to estimate Male 263 83 (91.2) 180 (88.6) prevalence ratios (PR) comparing acceptance adjusted for Transgender or other 22 6 (8.8) 16 (11.4) study site and other factors found to be statistically significant non-male identified (p < 0.05) in bivariate analyses. Sexual identity STI incidence for urethral and rectal chlamydia, urethral and rectal gonorrhoea, and syphilis are reported for prospectively Gay/homosexual 192 77 (85.6) 115 (57.9) enrolled participants. STI incidence rates are expressed as Bisexual, heterosexual, 95 13 (14.4) 82 (42.1) number of incident infections per 100 person-years (PY) at or other risk. Person-years of follow-up were determined by totalling Educationa the number of days of observation for those who were at risk Did not matriculate 137 36 (40.0) 101 (51.0) of STI infection for each anatomical site and STI combination. Matriculate or higher 151 54 (60.0) 97 (49.0) We considered participants at risk for a given combination if Work/student status they had no evidence of prevalent infection during the follow- Part/full-time student 150 49 (54.4) 101 (51.3) up period (e.g. untreated infection). The time period a partici- or part/full-time job pant was on treatment was excluded from the at-risk period. Not a student and no 137 41 (45.6) 96 (48.7) Rates and rate ratios of incident NG, CT and syphilis infec- job tions were modelled using Poisson regression. Because MSM and TGW are heterogenous populations and the study popula- Income tion was predominantly composed of MSM, we present a sen- No income 141 43 (50.6) 98 (52.7) sitivity analysis of incidence rates and rate ratios restricted to Any income 130 42 (49.4) 88 (47.3) MSM in Table S2. Baseline HIV statusb Additional descriptive analyses are reported to describe the pro- Negative 167 – 167 (83.1) portion of laboratory diagnosed urethral and rectal chlamydia, ure- Positive 125 91 (100.0) 34 (16.9) thral and rectal gonorrhoea, and syphilis infections that were Initiated PrEP During follow-up symptomatic, concurrent at the baseline visit, and successfully trea- No 85 – 85 (50.9) ted. All analyses were conducted using SAS 9.4 (SAS Institute, Cary, Yes 82 – 82 (49.1) NC, USA). Statistical significance was determined at p < 0.05. Receptive condomless anal intercourse, past three months No 165 39 (47.0) 126 (72.0) 3 | RESULTS Yes 93 44 (53.0) 49 (28.0) Number of male partners, past three months A total of 292 (115 in Cape Town, 177 in Port Elizabeth) par- 0 to 2 212 72 (81.9) 140 (81.9) ticipants were enrolled, 201 (100 in Cape Town, 101 in Port 3+ 47 16 (18.2) 31 (18.1) Elizabeth) of whom were followed prospectively (Table 1). The Any female partners, past 12 months prospective participants were composed of all HIV-negative No 230 81 (89.0) 149 (73.4) participants and a sample of HIV-positive participants such Yes 60 10 (11.0) 50 (26.6) that the HIV prevalence in the prospective cohort was Transactional sex, past 12 months approximately 20% at the beginning of follow-up. Most partici- No 225 76 (90.5) 149 (76.4) pants identified as Black (254/292; 87%), male (263/285; Yes 43 8 (9.5) 35 (23.6) 92%) and gay (192/287; 67%); a total of 22 (7.7%) participants were TGW. The prevalence of HIV was 31% (91/292)

19 Jones J et al. Journal of the International AIDS Society 2020, 23(S6):e25594 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25594/full | https://doi.org/10.1002/jia2.25594

Table 1. (Continued) infection. Among MSM, the prevalence of rectal CT and GC were 24.0% (95% CI: 17.6, 30.4) and 15.2% (95% CI: 9.8, Baseline Only Prospective 20.6) respectively. Among TGW, the prevalence of rectal CT Total Participants Participants and GC were 28.6% (95% CI: 4.9, 52.2) and 21.4% (95% CI: N N (%) N (%) 0.0, 42.9) respectively. Prevalent rectal CT infection was associated in crude estimates with age, sexual identity, receptive Injection drug use, past six months condomless anal intercourse in the previous three months, No 72 16 (94.1) 56 (90.3) and transactional sex (Table 3). In adjusted models, only age Yes 7 1 (5.9) 6 (9.7) group remained statistically significant, with 18 to 24 year Any drug use, past six months olds having 2.4 (95% CI: 1.1, 5.1) times higher prevalence of rectal CT compared to those 25 and older. Prevalent rectal No 211 74 (81.3) 137 (68.8) NG infection was also associated with age in crude and Yes 79 17 (18.7) 62 (31.2) adjusted analyses. Controlling for study site and baseline HIV + Binge drinking (5 drinks) on 5 or more days, past 30 days status, participants age 18 to 24 experienced an incidence No 215 65 (75.6) 150 (80.7) rate of rectal NG 3.1 times higher (95% CI: 1.3, 7.1) than Yes 57 21 (24.4) 36 (19.4) those age 25 and older. Among the 278 participants screened for syphilis at base- All participants completed a baseline visit. All HIV-negative partici- line, 50 (18%) had prevalent syphilis infection (21.6% among pants and a sample of HIV-positive participants were followed prospectively. PrEP, pre-exposure prophylaxis. MSM and 18.2% among TGW). Prevalent syphilis was associ- aDid not matriculate indicates not completing high school; Matriculate ated in crude analyses with older age, identifying as gay, being or higher indicates high school graduate or above HIV positive and receptive condomless anal intercourse in the bBaseline-only participants were all HIV positive. past three months. at baseline. Overall, 11% of participants had two or more con- 3.2 | STI testing and incidence over 12 months of current STIs at baseline. Among prospective participants, 86% follow-up (172/201) and 87% (174/201) completed study visits at months 6 and 12 respectively. Nearly all (193/201; 96%) participants enrolled in the follow-up procedures had at least 1 visit where follow-up STI screening 3.1 | Baseline screening and STI prevalence was offered. Of the 193, 144 (75%) accepted rectal screening at least once in follow-up. Acceptance of at least one urethral (182/ Of 292 participants enrolled in baseline procedures, there 193; 94%) and syphilis (181/189; 94%) screening was high dur- was universal acceptance of urethral (292/292; 100%) screening follow-up. Follow-up rectal screening was more likely to be ing, near-universal acceptance of syphilis (289/292; 99%) accepted among participants who identified as gay compared to screening, and 189 (64.7%) accepted rectal STI screening some other sexual identity (85.2% vs. 62.8%, p < 0.01) and (Table 2). Baseline rectal screening was more likely to be among participants with no female partners in the past accepted among participants in Cape Town compared to Port 12 months (82.0%) compared to those who did (58.3%, Elizabeth (93.9% vs. 48.6%, p < 0.01) and those who identified p < 0.01). No associations with demographic characteristics as gay compared to some other sexual identity (71.4% vs. remained statistically significant in adjusted models. Urethral and 55.8%, p = 0.01). In adjusted models, only age group was sig- syphilis screening did not significantly differ by study site, partici- nificantly associated with acceptance: acceptance of rectal pant characteristics or behaviours (Table S1). screening was higher among 18- to 24-year-old participants The rate of incident urethral CT was 12.8/100 PY and the [PR = 1.1, 95% confidence interval (CI): 1.0, 1.2] compared to rate of incident urethral NG was 7.1/ 100 PY. No incident participants age 25 and over. Baseline urethral and syphilis urethral infections were observed among TGW. The incidence screening did not significantly differ by site, participant charac- of urethral CT was greater among participants in Port Eliza- teristics or behaviours (Table S1). beth (Table 4). This difference persisted in models adjusting Among 289 participants screened for urethral STI at base- for baseline HIV status and age group; the rate of urethral CT line, 29 (10%) had urethral CT infection and 8 (3%) had ure- was 3.1 (95% CI: 1.2, 8.1) times higher in Port Elizabeth com- thral NG infection (Table 3). Among MSM, the prevalence of pared to Cape Town. Controlling for study site, age group and urethral CT was 10.8% (95% CI: 7.0, 14.5) and urethral GC baseline HIV status, the incidence rate of urethral NG was 5.1 was 2.3% (95% CI: 0.5, 4.1). Among TGW, the prevalence of times higher (95% CI: 1.6, 16.0) among participants reporting urethral CT was 4.5% (95% CI: 0.0, 13.2) and urethral GC transactional sex in the past 12 months. was 4.8% (95% CI: 0.0, 13.9). Prevalent urethral CT infection The rate of incident rectal CT was 33.4/100 PY and the was associated in crude analyses with baseline HIV status and rate of incident rectal NG was 26.8/100 PY. Rates of rectal receptive condomless anal intercourse in the past three CT were similar among MSM (29.7/100 PY) and TGW (30.3/ months; prevalent urethral NG infection was associated in 100 PY), but rates of rectal GC were lower among MSM crude analyses with sexual identity and having any female sex (19.1/100 PY) compared to TGW (65.0/100 PY). The inci- partners in the past 12 months. None of the observed associ- dence of rectal CT was greater among participants in Cape ations with prevalent urethral NG or CT remained statistically Town, and those who were aged 18 to 24, identified as gay, significant in adjusted models. reported no income and reported receptive condomless anal Among 189 participants screened for rectal STI at baseline, intercourse in the past three months. Controlling for study 47 (25%) had rectal CT infection and 30 (16%) had rectal NG site, baseline HIV status, sexual identity and receptive anal sex

20 Jones J et al. Journal of the International AIDS Society 2020, 23(S6):e25594 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25594/full | https://doi.org/10.1002/jia2.25594

Table 2. Acceptance of Rectal STI screening at baseline and over 12 months of follow-up among men who have sex with men and transgender women in Cape Town and Port Elizabeth, South Africa

Rectal STI screening acceptance

Baseline (N = 292) Follow-upa (N = 189) 6 Month Visit (N = 172) 12 Month Visit (N = 174)

Prevalence Prevalence Prevalence Prevalence (95% CI) p-value (95% CI) p-value (95% CI) p-value (95% CI) p-value

Site Cape Town 93.9 (89.5, 98.3) <0.01 83.0 (75.4, 90.6) 0.04 82.1 (74.0, 90.3) 0.08 64.4 (54.3, 74.4) 0.12 Port Elizabeth 48.6 (41.2, 56.0) 69.5 (60.2, 78.7) 70.5 (60.9, 80.0) 51.7 (41.2, 62.2) Age ranges 18 to 24 67.3 (60.1, 74.4) 0.73 77.0 (67.4, 86.6) 0.86 76.4 (68.3, 84.5) 1.00 57.0 (47.6, 66.4) 0.75 25+ 65.4 (57.1, 73.6) 75.7 (67.8, 83.5) 75.8 (65.4, 86.1) 59.7 (48.0, 71.4) Race Black 66.1 (60.3, 72.0) 0.78 75.3 (68.5, 82.1) 0.66 77.9 (71.0, 84.7) 0.36 56.7 (48.6, 64.9) 0.56 Other 68.4 (53.6, 83.2) 80.0 (66.7, 93.3) 68.8 (52.7, 84.8) 63.6 (47.2, 80.0) Gender identity Male 66.9 (61.2, 72.6) 0.75 74.6 (68.0, 81.1) 0.12 75.0 (68.1, 81.9) 0.53 55.2 (47.3, 63.0) 0.10 Other 63.6 (43.5, 83.7) 93.3 (80.7, 100.0) 86.7 (69.5, 100.0) 80.0 (59.8, 100.0) Sexual identity Gay/homosexual 71.4 (65.0, 77.7) 0.01 85.2 (78.5, 91.9) <0.01 85.0 (78.0, 92.0) <0.01 62.0 (52.5, 71.5) 0.16 Bisexual or other 55.8 (45.8, 65.8) 62.8 (52.1, 73.5) 62.3 (50.9, 73.8) 50.7 (39.1, 62.3) Educationb Did not matriculate 63.5 (55.4, 71.6) 0.40 80.6 (72.6, 88.7) 0.17 80.2 (71.8, 88.6) 0.21 56.8 (46.5, 67.2) 1.00 Matriculate or higher 68.2 (60.8, 75.6) 71.0 (61.7, 80.2) 71.1 (61.3, 80.8) 57.8 (47.2, 68.5) Combined work/student Part/full-time student 71.3 (64.1, 78.6) 0.07 80.2 (72.2, 88.2) 0.23 80.5 (72.1, 88.8) 0.21 63.6 (53.6, 73.7) 0.12 or part/full-time job Not a student and 61.3 (53.2, 69.5) 71.9 (62.6, 81.2) 71.6 (61.8, 81.4) 51.2 (40.4, 62.0) no job Income No income 61.0 (52.9, 69.0) 0.02 75.6 (66.7, 84.4) 0.86 77.5 (68.3, 86.7) 0.85 61.7 (51.1, 72.3) 0.42 Any income 73.8 (66.3, 81.4) 77.6 (68.8, 86.5) 75.6 (66.1, 85.2) 54.4 (43.4, 65.4) Baseline HIV status Negative 66.5 (59.3, 73.6) 0.99 75.6 (68.9, 82.4) 0.82 75.5 (68.5, 82.6) 0.81 56.6 (48.5, 64.8) 0.55 Positive 66.4 (58.1, 74.7) 78.8 (64.8, 92.7) 79.3 (64.6, 94.1) 64.5 (47.7, 81.4) Initiated PrEP during follow-up No 60.0 (49.6, 70.4) 0.07 73.7 (63.8, 83.6) 0.71 78.8 (68.9, 88.7) 0.44 58.2 (46.4, 70.0) 0.74 Yes 73.2 (63.6, 82.8) 77.5 (68.3, 86.7) 72.7 (62.8, 82.7) 55.3 (44.1, 66.4) Receptive condomless anal intercourse, past three months No 64.2 (56.9, 71.6) 0.27 74.1 (66.2, 82.1) 0.42 74.0 (65.6, 82.5) 0.28 57.5 (48.1, 67.0) 0.36 Yes 71.0 (61.7, 80.2) 81.3 (70.2, 92.3) 83.7 (72.7, 94.8) 66.7 (52.9, 80.4) Number of male partners in past three months 0 to 2 64.6 (58.2, 71.1) 0.11 76.7 (69.5, 83.9) 0.62 75.8 (68.3, 83.3) 0.60 56.5 (47.7, 65.2) 0.28 3+ 76.6 (64.5, 88.7) 82.1 (68.0, 96.3) 83.3 (68.4, 98.2) 69.2 (51.5, 87.0) Any female partners, past 12 months No 68.7 (62.7, 74.7) 0.08 82.0 (75.6, 88.4) <0.01 80.6 (73.8, 87.4) 0.02 64.1 (55.9, 72.3) 0.01 Yes 56.7 (44.1, 69.2) 58.3 (44.4, 72.3) 61.0 (46.0, 75.9) 39.0 (24.1, 54.0) Transactional sex, past 12 months No 64.0 (57.7, 70.3) 0.19 76.1 (69.0, 83.2) 1.00 75.4 (67.9, 82.9) 1.00 59.7 (51.2, 68.2) 0.30 Yes 74.4 (61.4, 87.5) 76.5 (62.2, 90.7) 76.7 (61.5, 91.8) 48.3 (30.1, 66.5)

21 Jones J et al. Journal of the International AIDS Society 2020, 23(S6):e25594 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25594/full | https://doi.org/10.1002/jia2.25594

Table 2. (Continued)

Rectal STI screening acceptance

Baseline (N = 292) Follow-upa (N = 189) 6 Month Visit (N = 172) 12 Month Visit (N = 174)

Prevalence Prevalence Prevalence Prevalence (95% CI) p-value (95% CI) p-value (95% CI) p-value (95% CI) p-value

Injection drug use, past six months No 59.7 (48.4, 71.1) 1.00 72.7 (61.0, 84.5) 0.32 72.0 (59.6, 84.4) 0.31 52.9 (39.2, 66.6) 0.68 Yes 57.1 (20.5, 93.8) 100.0 (100.0, 100.0) 100.0 (100.0, 100.0) 66.7 (28.9, 100.0) Any drug use, past six months No 69.2 (63.0, 75.4) 0.12 76.2 (68.8, 83.6) 1.00 76.5 (68.8, 84.3) 0.85 59.1 (50.1, 68.1) 0.62 Yes 59.5 (48.7, 70.3) 75.4 (64.6, 86.2) 74.5 (63.0, 86.1) 54.4 (41.5, 67.3) Binge drinking (5 + drinks) on 5 or more days, past 30 days No 66.0 (59.7, 72.4) 0.74 76.4 (69.4, 83.5) 0.83 76.0 (68.6, 83.3) 1.00 54.5 (46.1, 63.0) 0.23 Yes 68.4 (56.4, 80.5) 74.3 (59.8, 88.8) 77.4 (62.7, 92.1) 67.7 (51.3, 84.2)

All participants are included in the baseline estimates; only prospective participants are included in the follow-up estimates. Prospective participants are all HIV-negative participants and a sample of HIV-positive participants. CI, confidence interval; PrEP, pre-exposure prophylaxis; STI, sexually transmitted infections. aFollow-up prevalence column indicates any uptake over 12 months of follow-up, columns for month 6 and month 12 indicate uptake at those visits specifically bDid not matriculate indicates not completing high school; Matriculate or higher indicates high school graduate or above. in the past three months, being age 18 to 24 (rate ratio 4 | DISCUSSION (RR) = 2.9, 95% CI: 1.1, 7.7) and reporting no income (RR = 2.5, 95% CI: 1.1, 5.8) were associated with increased We implemented a comprehensive package of HIV/STI screen- rectal CT incidence. The crude rate of rectal NG was greater ing and treatment with high acceptance among MSM and among participants who were aged 18 to 24, identified as gay, TGW in South Africa. Our study population was comprised of and reported receptive condomless anal intercourse in the a baseline cohort of whom all HIV negative and a sample of past three months. The crude rate of rectal NG was lower HIV-positive participants were prospectively followed for one among participants who identified as male compared to those year. Urethral STI and syphilis screening were high overall, with another gender identity. Controlling for study site, base- but rectal screening acceptance was substantially lower in line HIV status, sexual identity, gender identity and condom- Port Elizabeth compared to Cape Town at baseline and during less anal sex in the past three months, participants age 18 to follow-up. We observed exceptionally high prevalence and inci- 24 experienced a rate of rectal gonorrhoea incidence 5.3 dence of rectal STIs, the vast majority of which were asymp- (95% CI: 1.2, 23.7) times higher than those over age 25. tomatic, consistent with previous findings among MSM [18]. The rate of incident syphilis infection was 8.2/100 PY. Because the prevalence of rectal infections was higher than Syphilis incidence was higher among TGW (14.6/100 PY) com- urethral infections, this difference in willingness to screen has pared to MSM (6.4/100 PY). Syphilis incidence was associated important implications for the STI epidemics in each city. The with having 3 or more male partners in the previous three current STI management guidelines in South Africa, adapted months in crude analyses. This association was no longer sta- from the World Health Organization, call for syndromic man- tistically significant in a model controlling for study site, age agement of STIs [6]. Given the high prevalence of asymp- and baseline HIV status. tomatic STIs in our study population, it is likely that a syndromic approach is inadequate to detect STIs among MSM and TGW. This study was conducted from 2015 to 2016; 3.3 | Symptomatic and concurrent infections however, the environment with respect to STI incidence and The identification of STI symptoms for infections observed at prevalence has been stable for decades [19], and we believe baseline and follow-up visits was low. Overall, 91%, 95% and these data remain relevant. The continuing reliance on syn- 97% of rectal, urethral and syphilis infections were clinically dromic management will result in many missed opportunities asymptomatic (Table 5). Of those who received STI testing, to identify and treat infections compared to screening. 10% had more than one infection concurrently, either one The prevalence of CT, NG and syphilis was high in both organism at multiple sites or multiple organisms. study sites. The prevalence of rectal CT and NG was substantially higher than the prevalence of urethral infection, similar to previous studies [18,20]. In both cities, more than one-fifth 3.4 | PrEP use of participants had a rectal STI at baseline. Approximately There were no differences in the incidence of CT, NG or 20% of the study population had syphilis at the baseline visit. syphilis among participants who initiated PrEP during study These findings represent substantial unmet needs for STI follow-up compared to those who did not. screening and treatment among MSM and TGW in these

22 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25594/full oe tal. et J Jones Table 3. Prevalence and 95% confidence intervals of urethral and rectal chlamydia, urethral and rectal gonorrhoea, and syphilis among 292 men who have sex with men and transgender women in Cape Town and Port Elizabeth, South Africa

Chlamydia Gonorrhoea Syphilis Society AIDS International the of Journal

Rectal (N = 189) Urethral (N = 270) Rectal (N = 189) Urethral (N = 288) (N = 288)

Prevalence (95% CI) p-value Prevalence (95% CI) p-value Prevalence (95% CI) p-value Prevalence (95% CI) p-value Prevalence (95% CI) p-value

Site Cape Town 26.4 (18.0, 34.8) 0.61 9.6 (4.2, 15.1) 1.00 18.9 (11.4, 26.3) 0.23 5.3 (1.2, 9.4) 0.06 17.4 (10.5, 24.3) 0.19 Port Elizabeth 22.9 (13.9, 31.9) 10.3 (5.8, 14.8) 12.0 (5.0, 19.1) 1.1 (0.0, 2.7) 24.3 (17.9, 30.7) Age 18 to 24 34.3 (25.3, 43.2) <0.01 7.9 (3.8, 12.0) 0.17 22.2 (14.4, 30.1) 0.01 3.6 (0.8, 6.5) 0.47 15.3 (9.8, 20.9) <0.01

25+ 12.3 (5.2, 19.5) 12.9 (7.0, 18.8) 7.4 (1.7, 13.1) 1.6 (0.0, 3.9) 29.6 (21.6, 37.6) 2020, |

Race https://doi.org/10.1002/jia2.25594 23

Black 26.7 (19.9, 33.4) 0.20 9.9 (6.2, 13.6) 0.77 15.8 (10.2, 21.3) 1.00 2.8 (0.8, 4.8) 1.00 22.0 (16.9, 27.1) 0.83 (S6):e25594 Other 12.5 (0.0, 25.7) 10.8 (0.8, 20.8) 16.7 (1.8, 31.6) 2.7 (0.0, 7.9) 18.4 (6.1, 30.7) Gender identity Male 24.0 (17.6, 30.4) 0.75 10.8 (7.0, 14.5) 0.71 15.2 (9.8, 20.6) 0.46 2.3 (0.5, 4.1) 0.42 21.6 (16.6, 26.6) 1.00 Transgender or other 28.6 (4.9, 52.2) 4.5 (0.0, 13.2) 21.4 (0.0, 42.9) 4.8 (0.0, 13.9) 18.2 (2.1, 34.3) non-male identified Sexual identity Gay/homosexual 29.3 (21.6, 37.1) 0.02 7.9 (4.1, 11.7) 0.09 18.0 (11.5, 24.6) 0.18 1.1 (0.0, 2.5) 0.04 25.4 (19.2, 31.6) 0.03 Bisexual, heterosexual, 13.5 (4.2, 22.7) 14.9 (7.7, 22.1) 9.6 (1.6, 17.6) 5.3 (0.8, 9.9) 13.7 (6.8, 20.6) or other Educationa Did notmatriculate 20.9 (12.3, 29.5) 0.39 9.6 (4.7, 14.6) 0.85 14.0 (6.6, 21.3) 0.84 1.5 (0.0, 3.5) 0.45 22.1 (15.1, 29.0) 0.89 Matriculate or Higher 27.3 (18.5, 36.0) 10.7 (5.7, 15.6) 16.2 (8.9, 23.4) 3.3 (0.5, 6.2) 20.9 (14.4, 27.5) Work/student status Part/full-time student 27.5 (18.8, 36.1) 0.40 8.8 (4.2, 13.3) 0.44 12.7 (6.3, 19.2) 0.31 2.7 (0.1, 5.4) 1.00 20.1 (13.7, 26.6) 0.57 or part/full-time job Not a student and 21.4 (12.7, 30.2) 11.8 (6.3, 17.2) 19.0 (10.7, 27.4) 2.2 (0.0, 4.7) 23.1 (16.0, 30.3) no job Income No income 22.1 (13.3, 30.9) 0.86 10.9 (5.5, 16.2) 0.84 19.8 (11.4, 28.2) 0.14 2.2 (0.0, 4.6) 0.71 24.5 (17.3, 31.6) 0.38 Any income 24.2 (15.4, 33.0) 10.1 (5.1, 15.1) 11.0 (4.6, 17.4) 3.1 (0.1, 6.1) 19.5 (12.7, 26.4) Baseline HIV status Negative 30.1 (20.3, 40.0) 0.18 14.5 (9.2, 19.9) <0.01 13.2 (6.8, 19.7) 0.32 3.6 (0.8, 6.5) 0.47 9.6 (5.1, 14.1) <0.01 Positive 20.8 (13.0, 28.5) 4.0 (0.6, 7.5) 19.3 (10.8, 27.8) 1.6 (0.0, 3.9) 37.7 (29.1, 46.3) Initiated PrEP during follow-up

23 No 20.4 (9.1, 31.7) 1.00 14.1 (6.7, 21.5) 1.00 14.3 (4.5, 24.1) 0.78 1.2 (0.0, 3.5) 0.11 7.1 (1.6, 12.7) 0.30 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25594/full oe tal. et J Jones ora fteItrainlAD Society AIDS International the of Journal

Table 3. (Continued)

Chlamydia Gonorrhoea Syphilis

Rectal (N = 189) Urethral (N = 270) Rectal (N = 189) Urethral (N = 288) (N = 288)

Prevalence (95% CI) p-value Prevalence (95% CI) p-value Prevalence (95% CI) p-value Prevalence (95% CI) p-value Prevalence (95% CI) p-value

Yes 21.1 (10.5, 31.6) 15.0 (7.2, 22.8) 12.3 (3.8, 20.8) 6.3 (0.9, 11.6) 12.2 (5.1, 19.3) Receptive condomless anal intercourse, past three months 2020, No 17.6 (10.2, 25.0) 0.02 12.3 (7.2, 17.3) 0.02 12.7 (6.3, 19.2) 0.09 3.1 (0.4, 5.7) 0.42 17.1 (11.3, 22.8) <0.01 | https://doi.org/10.1002/jia2.25594

Yes 33.8 (22.3, 45.3) 3.2 (0.0, 6.8) 23.1 (12.8, 33.3) 1.1 (0.0, 3.2) 33.3 (23.6, 43.1) 23 Number of male partners in past three months (S6):e25594 0 to 2 27.8 (20.2, 35.4) 0.67 10.5 (6.3, 14.6) 0.59 14.3 (8.3, 20.2) 0.30 1.4 (0.0, 3.0) 0.21 20.0 (14.6, 25.4) 0.17 3+ 22.9 (8.9, 36.8) 6.5 (0.0, 13.7) 22.9 (8.9, 36.8) 4.4 (0.0, 10.5) 30.4 (17.1, 43.7) Any female partners, past 12 months No 27.5 (20.4, 34.5) 0.08 9.3 (5.5, 13.0) 0.62 17.0 (11.0, 22.9) 0.30 0.9 (0.0, 2.1) 0.02 23.0 (17.5, 28.5) 0.38 Yes 11.8 (0.9, 22.6) 11.7 (3.5, 19.8) 8.8 (0.0, 18.4) 6.7 (0.4, 13.0) 16.7 (7.2, 26.1) Transactional sex, past 12 months No 26.6 (19.3, 34.0) 0.01 9.0 (5.2, 12.8) 0.17 17.3 (11.0, 23.5) 0.17 2.3 (0.3, 4.2) 1.00 21.4 (16.1, 26.8) 0.68 Yes 6.3 (0.0, 14.6) 16.3 (5.2, 27.3) 6.3 (0.0, 14.6) 2.3 (0.0, 6.8) 17.1 (5.6, 28.6) Injection drug use, past six months No 19.0 (7.2, 30.9) 1.00 13.9 (5.9, 21.9) 1.00 11.9 (2.1, 21.7) 1.00 2.8 (0.0, 6.6) 0.25 15.3 (7.0, 23.6) 0.58 Yes 0.0 (0.0, 0.0) 14.3 (0.0, 40.2) 0.0 (0.0, 0.0) 14.3 (0.0, 40.2) 0.0 (0.0, 0.0) Any drug use, past six months No 26.8 (19.5, 34.0) 0.24 8.7 (4.8, 12.5) 0.19 16.9 (10.7, 23.1) 0.48 1.9 (0.1, 3.8) 0.40 24.6 (18.8, 30.5) 0.05 Yes 17.4 (6.4, 28.3) 13.9 (6.3, 21.6) 10.9 (1.9, 19.9) 3.8 (0.0, 8.0) 13.9 (6.3, 21.6) Binge drinking (5 + drinks) on 5 or more days, past 30 days No 22.6 (15.6, 29.6) 0.52 12.2 (7.8, 16.6) 0.15 14.6 (8.7, 20.5) 1.00 1.4 (0.0, 3.0) 0.11 20.9 (15.4, 26.3) 0.72 Yes 28.2 (14.1, 42.3) 5.3 (0.0, 11.1) 15.4 (4.1, 26.7) 5.3 (0.0, 11.1) 22.8 (11.9, 33.7)

Different samples sizes reflect differences in acceptance of screening and missing data. CI, confidence interval; PrEP, pre-exposure prophylaxis. a Did not matriculate indicates not completing high school; Matriculate or higher indicates high school graduate or above. 24 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25594/full Table 4. Rate (per 100 person years), unadjusted rate ratios (RR), and 95% confidence intervals of urethral and rectal chlamydia, urethral and rectal gonorrhoea, and syphilis al. et J Jones among men who have sex with men and transgender women in Cape Town and Port Elizabeth, South Africa

Chlamydia Gonorrhea Syphilis ora fteItrainlAD Society AIDS International the of Journal

Rectal (N = 127) Urethral (N = 178) Rectal (N = 126) Urethral (N = 179) (N = 172)

Rate Rate ratio Rate Rate ratio Rate Rate ratio Rate Rate ratio Rate Rate ratio (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) (95% CI)

Site CapeTown 44.4 (30.0, 65.7) 3.7 (1.3, 10.7) 6.9 (3.1, 15.4) 0.4 (0.1, 0.9) 31.9 (20.1, 50.7) 1.8 (0.7, 4.4) 3.4 (1.1, 10.5) 0.3 (0.1, 1.1) 12.2 (6.6, 22.7) 3.2 (0.9, 11.5) Port Elizabeth 12.0 (4.5, 31.9) Ref 19.6 (11.8, 32.5) Ref 18.1 (8.1, 40.2) Ref 11.1 (5.8, 21.3) Ref 3.9 (1.3, 12.0) Ref Age ranges 18-24 46.0 (30.8, 68.6) 3.4 (1.3, 9.0) 16.1 (9.9, 26.2) 2.1 (0.8, 5.6) 38.3 (24.7, 59.4) 3.6 (1.2, 10.5) 8.8 (4.6, 16.9) 2.0 (0.5, 7.3) 5.9 (2.7, 13.2) 0.5 (0.2, 1.5) 2020, 25+ 13.3 (5.6, 32.0) Ref 7.8 (3.3, 18.8) Ref 10.7 (4.0, 28.5) Ref 4.5 (1.4, 13.8) Ref 12.0 (5.7, 25.2) Ref |

Race https://doi.org/10.1002/jia2.25594 23

Black 32.0 (21.3, 48.2) 1.0 (0.4, 2.4) 12.8 (8.0, 20.6) 1.0 (0.3, 2.9) 28.0 (18.1, 43.4) 1.3 (0.4, 3.7) 8.0 (4.4, 14.4) 2.5 (0.3, 19.7) 7.7 (4.1, 14.3) 0.7 (0.2, 2.7) (S6):e25594 Other 33.4 (15.0, 74.3) Ref 13.0 (4.9, 34.6) Ref 22.1 (8.3, 58.8) Ref 3.1 (0.4, 22.3) Ref 10.3 (3.3, 31.9) Ref Gender identity Male 29.7 (19.7, 44.7) 1.0 (0.3, 3.3) 14.6 (9.5, 22.4) – 19.1 (11.5, 31.7) 0.3 (0.1, 0.8) 8.0 (4.5, 14.0) – 6.4 (3.3, 12.2) 0.4 (0.1, 2.0) Transgender 30.3 (9.8, 94.1) Ref 0.0 (0.0, 0.0) Ref 65.0 (29.2, 144.7) Ref 0.0 (0.0, 0.0) Ref 14.6 (3.6, 58.2) Ref or other non-male identified Sexual identity Gay/homosexual 41.2 (27.4, 62.1) 3.3 (1.1, 9.4) 9.3 (4.8, 17.8) 0.5 (0.2, 1.2) 36.2 (23.4, 56.1) 11.9 (1.6, 88.6) 5.0 (2.1, 11.9) 0.5 (0.1, 1.5) 6.3 (2.9, 14.1) 0.8 (0.2, 2.6) Bisexual, hetero 12.7 (4.8, 33.8) Ref 19.0 (10.8, 33.5) Ref 3.0 (0.4, 21.6) Ref 10.6 (5.1, 22.3) Ref 8.0 (3.3, 19.3) Ref sexual, or other Education Did not 36.6 (22.5, 59.8) 1.6 (0.7, 3.6) 14.7 (8.3, 25.8) 1.3 (0.5, 3.0) 22.1 (11.9, 41.1) 0.8 (0.3, 1.8) 7.0 (3.1, 15.2) 0.9 (0.3, 2.9) 7.5 (3.4, 16.8) 1.0 (0.3, 3.0) matriculatea Matriculate or 22.7 (12.2, 42.1) Ref 11.4 (6.0, 22.0) Ref 28.2 (16.0, 49.6) Ref 7.4 (3.3, 16.5) Ref 7.8 (3.5, 17.3) Ref higher Combined work/ student Part/full-time 31.4 (19.1, 50.8) 1.0 (0.5, 2.2) 11.7 (6.3, 21.7) 0.8 (0.3, 1.8) 26.8 (15.9, 45.3) 1.2 (0.5, 2.9) 7.9 (3.8, 16.6) 1.2 (0.4, 3.9) 8.3 (4.0, 17.5) 1.5 (0.4, 5.1) student or part/full-time job Not a student 30.0 (16.6, 54.2) Ref 14.9 (8.2, 26.9) Ref 22.2 (11.1, 44.4) Ref 6.5 (2.7, 15.5) Ref 5.5 (2.1, 14.7) Ref and no job Income No income 45.8 (28.9, 72.7) 2.3 (1.0, 5.0) 14.3 (7.9, 25.9) 1.0 (0.4, 2.5) 24.9 (13.4, 46.2) 1.2 (0.5, 3.0) 6.3 (2.6, 15.1) 0.7 (0.2, 2.2) 6.8 (2.8, 16.3) 1.0 (0.3, 6.4) Any income 20.2 (10.5, 38.8) Ref 13.7 (7.4, 25.5) Ref 20.2 (10.5, 38.8) Ref 9.2 (4.4, 19.3) Ref 6.8 (2.9, 16.4) Ref 25 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25594/full oe tal. et J Jones Table 4. (Continued)

Chlamydia Gonorrhea Syphilis ora fteItrainlAD Society AIDS International the of Journal Rectal (N = 127) Urethral (N = 178) Rectal (N = 126) Urethral (N = 179) (N = 172)

Rate Rate ratio Rate Rate ratio Rate Rate ratio Rate Rate ratio Rate Rate ratio (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) (95% CI)

Baseline HIV status Negative 34.0 (22.8, 50.8) Ref 12.0 (7.4, 19.6) Ref 26.8 (17.1, 42.0) Ref 7.3 (3.9, 13.5) Ref 7.5 (4.0, 14.0) Ref Positive 26.0 (10.8, 62.6) 0.8 (0.3, 2.0) 16.7 (6.9, 40.1) 1.4 (0.5, 3.8) 26.9 (11.2, 64.5) 1.0 (0.4, 2.7) 6.3 (1.6, 25.1) 0.9 (0.2, 3.9) 11.4 (3.7, 35.4) 1.5 (0.4, 5.5) Initiated PrEP During Follow-up No 30.6 (16.4, 56.8) Ref 9.5 (4.3, 21.2) Ref 18.1 (8.2, 40.4) Ref 9.3 (4.2, 20.7) Ref 8.0 (3.3, 19.2) Ref 2020, Yes 37.0 (21.9, 62.5) 1.2 (0.5, 2.7) 14.2 (7.6, 26.4) 1.5 (0.5, 4.1) 34.3 (19.9, 59.0) 1.9 (0.7, 5.0) 5.5 (2.1, 14.5) 0.6 (0.2, 2.1) 7.1 (3.0, 17.0) 0.9 (0.3, 3.1) |

Receptive condomless anal intercourse, past three months https://doi.org/10.1002/jia2.25594 23

No 21.9 (12.4, 38.5) Ref 13.3 (7.7, 22.9) Ref 16.7 (8.7, 32.0) Ref 7.9 (4.0, 15.8) Ref 9.5 (5.0, 18.3) Ref (S6):e25594 Yes 57.2 (33.9, 96.6) 2.6 (1.2, 5.6) 8.9 (3.4, 23.8) 0.7 (0.2, 2.1) 47.3 (26.9, 83.4) 2.8 (1.2, 6.7) 2.2 (0.3, 15.5) 0.3 (0.0, 2.2) 7.1 (2.3, 21.9) 0.7 (0.2, 2.7) Number of male partners in past three months 0 to 2 31.7 (20.2, 49.7) Ref 13.9 (8.5, 22.7) Ref 27.8 (17.3, 44.7) Ref 7.4 (3.9, 14.3) Ref 5.2 (2.3, 11.5) Ref 3+ 35.4 (15.9, 78.8) 1.1 (0.4, 2.8) 11.9 (3.8, 36.8) 0.9 (0.2, 2.9) 25.2 (9.4, 67.0) 0.9 (0.3, 2.7) 11.7 (3.8, 36.4) 1.6 (0.2, 5.8) 22.8 (9.5, 54.7) 4.4 (1.4, 14.5) Any female partners, past 12 months No 36.1 (24.5, 52.9) Ref 12.8 (7.8, 20.9) Ref 30.6 (20.1, 46.4) Ref 8.5 (4.7, 15.3) Ref 8.1 (4.4, 15.1) Ref Yes 6.0 (0.8, 42.4) 0.2 (0.0, 1.2) 13.7 (5.7, 32.9) 1.1 (0.4, 2.9) 6.0 (0.9, 42.6) 0.2 (0.0, 1.5) 2.6 (0.4, 18.7) 0.3 (0.0, 2.4) 5.8 (1.5, 23.2) 0.7 (0.2, 3.3) Transactional sex, past 12 months No 35.9 (23.9, 54.0) Ref 11.5 (6.8, 19.3) Ref 31.5 (20.3, 48.8) – 4.7 (2.1, 10.5) Ref 5.9 (2.8, 12.3) Ref Yes 11.9 (3.0, 47.6) 0.3 (0.1, 1.4) 26.4 (12.6, 55.4) 2.3 (0.9, 5.7) 0.0 (0.0, 0.0) – 21.6 (9.7, 48.1) 4.6 (1.5, 14.2) 14.7 (5.5, 39.1) 2.5 (0.7, 8.5) Injection drug use, past six months No 12.7 (4.1, 39.4) Ref 17.6 (8.8, 35.1) Ref 23.0 (9.6, 55.3) – 10.6 (4.4, 25.5) Ref 8.5 (3.2, 22.5) Ref Yes 26.0 (3.7, 184.3) 2.0 (0.2, 19.6) 19.0 (2.7, 134.6) 1.1 (0.1, 8.6) 0.0 (0.0, 0.0) – 0.0 (0.0, 0.0) – 20.6 (2.9, 146.4) 2.4 (0.3, 21.8) Any drug use, past six months No 37.7 (25.1, 56.7) Ref 10.8 (6.2, 19.1) Ref 25.4 (15.6, 41.5) Ref 6.1 (2.9, 12.7) Ref 6.6 (3.2, 13.9) Ref Yes 14.6 (5.5, 38.8) 0.4 (0.1, 1.1) 17.7 (9.2, 34.0) 1.6 (0.7, 3.9) 19.4 (8.1, 46.6) 0.8 (0.3, 2.1) 9.5 (4.0, 22.9) 1.6 (0.5, 4.9) 9.6 (4.0, 23.0) 1.4 (0.5, 4.5) Binge drinking (5 + drinks) on 5 or more days, past 30 days No 27.9 (17.6, 44.3) Ref 14.8 (9.3, 23.5) Ref 28.3 (17.8, 44.8) Ref 7.8 (4.2, 14.5) Ref 7.5 (3.9, 14.5) Ref Yes 31.8 (14.3, 70.7) 1.1 (0.5, 2.9) 6.7 (1.7, 26.8) 0.5 (0.1, 2.0) 20.7 (7.8, 55.1) 0.7 (0.2, 2.2) 3.3 (0.5, 23.6) 0.4 (0.1, 3.3) 7.0 (1.8, 28.1) 0.9 (0.2, 4.3)

CI, confidence interval; PrEP, pre-exposure prophylaxis. a Did not matriculate indicates not completing high school; Matriculate or higher indicates high school graduate or above. 26 Jones J et al. Journal of the International AIDS Society 2020, 23(S6):e25594 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25594/full | https://doi.org/10.1002/jia2.25594

Table 5. Frequency of urethral NG/CT, rectal NG/CT and syphilis symptoms at baseline, month 6 and month 12 overall and among those with diagnosed STI in a cohort of men who have sex with men and transgender women in Cape Town and Port Elizabeth, South Africa

Baseline – all Baseline – STI + a Month 6 Month 6 – STI + a Month 12 Month 12 – STI + a n/N (%) n/N (%) n/N (%) n/N (%) n/N (%) n/N (%)

Urethral STI symptoms 3/292 (1.0%) 1/34 (2.9%) 1/172 (0.6%) 0/12 (0.0%) 2/174 (1.2%) 1/15 (6.7%) Rectal STI symptoms 6/292 (2.0%) 5/60 (8.3%) 0/172 (0.0%) 0/24 (0.0%) 3/174 (1.7%) 3/26 (11.5%) Syphilis symptoms 4/292 (1.4%) 1/50 (2.0%) 5/172 (2.9%) 1/16 (6.3%) 1/174 (0.6%) 1/22 (4.6%)

All enrolled participants contributed to baseline data; all HIV negative and a sample of HIV-positive participants were prospectively followed and contributed data at months 6 and 12. CT, Chlamydia trachomatis; NG, Neisseria gonorrhea; STI, sexually transmitted infections. aThe denominator for each cell is the number of participants diagnosed with a relevant STI (e.g. urethral NG or CT for those with urethral symptoms).

South African cities. Younger participants were more likely to Age, sexual identity and condomless receptive anal sex were have rectal NG/CT, and participants who identified as gay all associated with incident infection, consistent with the asso- were more likely to have rectal CT at the baseline visit. Partic- ciations observed for prevalent infections at baseline. These ipants who reported receptive condomless anal sex in the pre- characteristics might be useful in identifying MSM and TGW vious three months had higher prevalence of rectal and other in need of more frequent STI screening due to increased risk, STIs. Previous studies have found similar characteristics to be and align with findings from other studies of incident STI associated with asymptomatic NG/CT infection including among MSM [14]. Indeed, the WHO guidelines for prevention transgender identity, multiple male sex partners in the previ- and treatment of STIs among MSM and TGW [23] call for ous 12 months and transactional sex [13]. presumptive treatment of STIs among MSM and TGW who Incident STIs followed a similar pattern. The incidence of rec- report receptive anal intercourse and either multiple partners tal infection was higher than urethral infection at both study or a partner with a STI in the past six months; unfortunately, sites for both CT and NG. Higher incidence rates of rectal these guidelines do not include recommendations on how fre- infections and syphilis were observed in Cape Town compared quently presumptive treatment should occur. Our data sup- to Port Elizabeth; however, the difference was only statistically port the WHO guidelines, however, implementation of these significant for rectal CT. The lower acceptance of rectal STI guidelines in the absence of screening will still result in miss- screening in Port Elizabeth compared to Cape Town (72.5% vs. ing substantial numbers of asymptomatic infections [24]. 83.0%) might at least partially account for the difference in rec- Based on our data, a large proportion of STIs are asymp- tal CT incidence. The acceptance of syphilis screening and ure- tomatic, a phenomenon observed elsewhere [9]. In the thral STI screening was near universal at both sites, so we absence of screening, individuals would not be able to report believe that all or most incident urethral and syphilis infections a partner with an asymptomatic STI. It is unlikely that pre- were detected; however, some rectal STIs might have been sumptive treatment will be sufficient to meaningfully reduce missed due to lower acceptance of rectal screening. It remains the STI burden in these key populations. Rather, incorporation unclear what led some participants to refuse rectal screening. of point-of-care screening [25] to diagnose both symptomatic It might be the case that those at the highest risk of rectal and asymptomatic STIs will likely have a greater effect on the infections were more likely to accept rectal screening; however, STI epidemic among MSM and TGW. there were no differences in rectal screening acceptance based There is growing interest in the intersection of HIV and on reporting anal intercourse in the past three months. A other STIs [26]. A recent modelling study estimated that recent study of Thai TGW found that rectal screening pro- approximately 10% of HIV incidence among MSM might be duced the highest yield of positive NG/CT infections [10], attributable to prevalent NG and CT [27], suggesting that STI implying that rectal screening will be vitally important to reduce detection and treatment might lead to meaningful reductions NG/CT incidence and prevalence. Future studies should assess in HIV incidence. Additionally, there are concerns that MSM reasons for refusal of rectal screening. We did not observe dif- and TGW who use HIV pre-exposure prophylaxis (PrEP) may ferences in STI incidence based on gender identity. However, continue to have (or increase frequency of) condomless anal we did not observe any incident urethral infections among sex once PrEP has been started, a phenomenon known as risk TGW. The rate of rectal GC and syphilis were much higher compensation [28]. Although some studies have observed little among TGW compared to MSM; however, the CI for these or no behavioural risk compensation [15,28,29], a recent rates were very wide due to the small sample size of TGW and review found an increased risk of rectal CT among PrEP-using the differences were not statistically significant. We did not MSM and TGW [30]. Condomless anal sex may lead to STI screen for pharyngeal infection. However, there is evidence acquisition, which could also undermine the HIV prevention that pharyngeal NG/CT infections can cause urethral [21] and benefits of PrEP by increasing biological risks for HIV infec- rectal [22] infections in sexual partners. Therefore, it remains tion. We did not observe differences in STI incidence between necessary to characterize the burden of pharyngeal infections PrEP users and non-users in this study. Surveillance estimates among MSM and TGW in South Africa and pharyngeal screen- indicate HIV prevalence is higher than 18% among MSM in ing should be part of all STI screening programmes. South Africa [31], yet STI prevalence among MSM is

27 Jones J et al. Journal of the International AIDS Society 2020, 23(S6):e25594 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25594/full | https://doi.org/10.1002/jia2.25594 unreported and there are no previous studies examining STI ACKNOWLEDGEMENT incidence and rectal STI screening in this group. We are grateful to Charlotte Rolle, MD, MPH who provided invaluable assis- This study has a number of limitations. First, these data tance adjudicating incident infections for our incidence analyses. were generated as part of a pilot study of a combination HIV prevention package that was not specifically powered to FUNDING examine STI prevalence and incidence and associated risk fac- The authors acknowledge funding from the National Institutes of Health tors. Thus, our estimates are imprecise; the direction and rela- (R01AI094575). tive strength of the observed associations should be used to generate hypotheses that can be tested in larger studies. As described earlier, we conducted RPR and confirmatory TPPA REFERENCES syphilis testing and monitored titres over the course of the 1. World Health Organization. Report on global sexually transmitted infection study; however, it is possible that some of the prevalent syphi- surveillance 2015. Geneva, Switzerland: World Health Organization; 2016. lis infections at baseline had previously been treated. Some 2. Centers for Disease Control and Prevention. Sexually transmitted disease surveillance 2016. Atlanta, GA: U.S. Department of Health and Human Services; participants were referred to community clinics for STI treat- 2017. ment, and completion of a treatment regimen could not be 3. The South African National AIDS Council. South Africa’s National strategic verified. Low acceptance of rectal STI screening indicates that plan for HIV, TB, and STIs 2017-2022. 2017. our estimates of rectal STI prevalence and incidence are 4. Farley TA, Cohen DA, Elkins W. Asymptomatic sexually transmitted diseases: the case for screening. Prev Med. 2003;36(4):502–9. underestimates and could be prone to information bias. 5. Marcus JL, Bernstein KT, Kohn RP, Liska S, Philip SS. Infections missed by A major limitation of this study is the small sample of TGW urethral-only screening for chlamydia or gonorrhea detection among men who who were enrolled: We are limited in our ability to make have sex with men. Sex Transm Dis. 2011;38(10):922–4. inferences about predictors of prevalence and incidence 6. Department of Health Republic of South Africa. Sexually transmitted infec- among TGW specifically and about differences between MSM tions management guidelines 2015. 2015. 7. Mustanski B, Feinstein BA, Madkins K, Sullivan P, Swann G. Prevalence and and TGW due to the small sample of TGW enrolled in the risk factors for rectal and urethral sexually transmitted infections from self-col- study. We are also unable to assess differences between TGW lected samples among young men who have sex with men participating in the who choose to participate in a study that primarily comprises keep it up! 2.0 randomized controlled trial. Sex Transm Dis. 2017;44(8):483–88. MSM. MSM and TGW are both marginalized populations who 8. Yang LG, Zhang XH, Zhao PZ, Chen Z-Y, Ke W-J, Ren X-Q, et al. Gonorrhea and chlamydia prevalence in different anatomical sites among men who have frequently experience stigma in healthcare settings [32]; how- sex with men: a cross-sectional study in Guangzhou, China. BMC Infect Dis. ever, they are also unique identities and each group has 2018;18:675. unique needs. Future studies should make efforts to focus 9. Keshinro B, Crowell TA, Nowak RG, Adebajo S, Peel S, Gaydos CA, et al. specifically and more robustly on the needs and experiences High prevalence of HIV, chlamydia and gonorrhoea among men who have sex of TGW with respect to STI screening, incidence and preva- with men and transgender women attending trusted community centres in Abuja and Lagos, Nigeria. J Int AIDS Soc. 2016;19:21270. lence. 10. Hiransuthikul A, Janamnuaysook R, Sungsing T, Callander D, Fairley CK, Grulich AE, et al. High burden of chlamydia and gonorrhoea in pharyngeal, rectal and urethral sites among Thai transgender women: implications for anatomi- 5 | CONCLUSIONS cal site selection for the screening of STI. Sex Transm Infect. 2019;95(7):534–9. 11. Comninos NB, Garton L, Guy R, Callander D, Fairley CK, Grulich AE, et al. Increases in pharyngeal Neisseria gonorrhoeae positivity in men who have sex STIs are highly prevalent among MSM and TGW in South with men, 2011-2015: observational study. Sex Transm Infect. 2019. Africa, and rectal STIs are more common than urethral infec- 12. Sullivan PS, Peterson J, Rosenberg ES, Kelley CF, Cooper H, Vaughan A, tions. High incidence rates indicate ongoing STI risk even fol- et al. Understanding racial HIV/STI disparities in black and white men who have lowing successful treatment. Because the vast majority of STIs sex with men: a multilevel approach. PLoS One. 2014;9:e90514. 13. Rebe K, Lewis D, Myer L, de Swardt G, Struthers H, Kamkuemah M, et al. A in our study were asymptomatic, multi-site STI screening and cross sectional analysis of gonococcal and chlamydial infections among men-who- treatment among MSM and TGW are of paramount impor- have-sex-with-men in Cape Town, South Africa. PLoS One. 2015;10:e0138315. tance in combating the STI epidemic. 14. Ramadhani HO, Liu H, Nowak RG, Crowell TA, Ndomb T, Gaydos C, et al. Sexual partner characteristics and incident rectal Neisseria gonorrhoeae and Chlamydia trachomatis infections among gay men and other men who have sex AUTHORS’ AFFILIATIONS with men (MSM): a prospective cohort in Abuja and Lagos, Nigeria. Sex Transm Infect. 2017;93(5):348–55. 1 2 Emory University, Atlanta, GA, USA; Desmond Tutu HIV Centre, University of 15. Hightow-Weidman LB, Magnus M, Beauchamp G, Hurt CB, Shoptaw S, Emel 3 Cape Town, Observatory, South Africa; North West University, Potchefstroom, L, et al. Incidence and correlates of STIs among black men who have sex with men 4 South Africa; Human Sciences Research Council of South Africa, Pretoria, participating in the HPTN 073 PrEP study. Clin Infect Dis. 2019;69(9):1597-604. 5 South Africa; Johns Hopkins University School of Public Health, Baltimore, MD, 16. McNaghten A, Kearns R, Siegler AJ, Phaswana-Mafuya N, Bekker L-G, 6 USA; Wits Reproductive Health and HIV Institute, Faculty of Health Sciences, Stephenson R, et al. Sibanye methods for prevention packages program project 7 University of the Witwatersrand, Johannesburg, South Africa; School of Heath protocol: pilot study of HIV prevention interventions for men who have sex with Systems and Public Health, University of Pretoria, Pretoria, South Africa men in South Africa. JMIR Res Protoc. 2014;3(4):e55. 17. Spiegelman D, Hertzmark E. Easy SAS calculations for risk or prevalence – COMPETING INTERESTS ratios and differences. Am J Epidemiol. 2005;162(3):199 200. 18. Annan NT, Sullivan AK, Nori A, Naydenova P, Alexander S, McKenna A, The authors have no conflicts to disclose. et al. Rectal chlamydia—a reservoir of undiagnosed infection in men who have sex with men. Sex Transm Infect. 2009;85(3):176–9. 19. Kularatne RS, Niit R, Rowley J, Kufa-Chakezha T, Peters RPH, Taylor MM, ’ AUTHORS CONTRIBUTIONS et al. Adult gonorrhea, chlamydia and syphilis prevalence, incidence, treatment TS, LGB, NPM, SB, AM, AJS and PSS contributed to study design; KD, LK, CY and syndromic case reporting in South Africa: estimates using the spectrum-STI and RZ contributed to data collection; JJ, TS and RV conducted the analyses; JJ model, 1990–2017. PLoS One. 2018;13:e0205863. and TS drafted the manuscript; all authors read and approved the final version 20. Kent CK, Chaw JK, Wong W, Liska S, Gibson S, Hubbard G, et al. Preva- of the manuscript. lence of rectal, urethral, and pharyngeal chlamydia and gonorrhea detected in 2

28 Jones J et al. Journal of the International AIDS Society 2020, 23(S6):e25594 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25594/full | https://doi.org/10.1002/jia2.25594

clinical settings among men who have sex with men: San Francisco, California, 30. Traeger MW, Schroeder SE, Wright EJ, Hellard ME, Cornelisse VJ, Doyle 2003. Clin Infect Dis. 2005;41(1):67–74. JS, et al. Effects of pre-exposure prophylaxis for the prevention of human 21. Bernstein KT, Stephens SC, Barry PM, Kohn R, Philip SS, Liska S, et al. immunodeficiency virus infection on sexual risk behavior in men who have sex Chlamydia trachomatis and Neisseria gonorrhoeae transmission from the with men: a systematic review and meta-analysis. Clin Infect Dis. 2018;67 oropharynx to the urethra among men who have sex with men. Clin Infect Dis. (5):676–86. 2009;49(12):1793–7. 31. UNAIDS. AIDSinfo. UNAIDS. 2019 [cited 2019 Sep 2]. Available from: 22. McMillan A, Young H, Moyes A. Rectal gonorrhoea in homosexual men: http://aidsinfo.unaids.org/ source of infection. Int J STD AIDS. 2000;11(5):284–7. 32. Lyons C, Stahlman S, Holland C, Ketende S, Van Lith L, Kochelani D, et al. 23. World Health Organization. Guidelines: prevention and treatment of HIV Stigma and outness about sexual behaviors among cisgender men who have sex and other sexually transmitted infections among men who have sex with men with men and transgender women in Eswatini: a latent class analysis. BMC and transgender people. Geneva: World Health Organization; 2011. Infect Dis. 2019;19(1):211. 24. Sanders EJ, Wahome E, Okuku HS, Thiong’o AN, Smith AD, Duncan S, et al. Evaluation of WHO screening algorithm for the presumptive treatment of asymptomatic rectal gonorrhoea and chlamydia infections in at-risk MSM in Kenya. Sex Transm Infect. 2014;90(2):94–9. 25. Tucker JD, Bien CH, Peeling RW. Point-of-care testing for sexually transmit- SUPPORTING INFORMATION ted infections: recent advances and implications for disease control. Curr Opin Infect Dis. 2013;26(1):73–9. Additional Supporting Information may be found in the online 26. Cohen MS, Council OD, Chen JS. Sexually transmitted infections and HIV in the era of antiretroviral treatment and prevention: the biologic basis for epi- version of this article: demiologic synergy. J Int AIDS Soc. 2019;22:e25355. Table S1. Acceptance of urethral and syphilis STI screening at 27. Jones J, Weiss K, Mermin J, Dietz P, Rosenberg ES, Gift TL, et al. Proportion baseline and over 12 months of follow-up among men who of incident HIV cases among men who have sex with men attributable to gonor- have sex with men and transgender women in Cape Town – rhea and chlamydia: a modeling analysis. Sex Transm Dis. 2019;46(6):357 63. and Port Elizabeth, South Africa 28. Freeborn K, Portillo CJ. Does pre-exposure prophylaxis for HIV prevention in men who have sex with men change risk behaviour? A systematic review. J Table S2. Rate (per 100 person years), unadjusted rate ratios Clin Nurs. 2018;27(17–18):3254–65. (RR), and 95% confidence intervals of urethral and rectal 29. Milam J, Jain S, Dube MP, Daar ES, Sun X, Corado K, et al. Sexual risk com- chlamydia, urethral and rectal gonorrhea, and syphilis among pensation in a pre-exposure prophylaxis demonstration study among individuals men who have sex with men (MSM) in Cape Town and Port at risk for HIV. J Acquir Immune Defic Syndr. 2019;80(1):e9–e13. Elizabeth, South Africa

29 Wahome EW et al. Journal of the International AIDS Society 2020, 23(S6):e25593 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25593/full | https://doi.org/10.1002/jia2.25593

SHORT REPORT Risk factors for loss to follow-up among at-risk HIV negative men who have sex with men participating in a research cohort with access to pre-exposure prophylaxis in coastal Kenya Elizabeth W Wahome1,§ , Susan M Graham1,2 , Alexander N Thiong’o1, Khamisi Mohamed1, Tony Oduor1, Evans Gichuru1, John Mwambi1, Elise M van der Elst1,3 and Eduard J Sanders1,3,4 §Corresponding author: Elizabeth W Wahome, KEMRI-Wellcome Trust Research Programme, P.O. Box 230-80108, Kilifi, Kenya. Tel: (+254) 41 522535. ([email protected])

Abstract Introduction: Retention in preventive care among at-risk men who have sex with men (MSM) is critical for successful prevention of HIV acquisition in Africa. We assessed loss to follow-up (LTFU) rates and factors associated with LTFU in an HIV vaccine feasibility cohort study following MSM with access to pre-exposure prophylaxis (PrEP) in coastal Kenya. Methods: Between June 2017 and June 2019, MSM cohort participants attending a research clinic 20 km north of Mombasa were offered daily PrEP and followed monthly for risk assessment, risk reduction counselling and HIV testing. Participants were defined as LTFU if they were late by >90 days for their scheduled appointment. Participants who acquired HIV were censored at diagnosis. Cox proportional hazards models were used to estimate adjusted Hazard Ratio (aHR) of risk factors for LTFU. Results and discussion: A total of 179 participants with a median age of 25.0 years (interquartile range [IQR]: 23.0 to 30.0) contributed a median follow-up time of 21.2 months (IQR: 6.5 to 22.1). Of these, 143 (79.9%) participants started PrEP and 76 (42.5%) MSM were LTFU, for an incidence rate of 33.7 (95% confidence interval [CI], 26.9 to 42.2) per 100 person-years. Disordered alcohol use (aHR: 2.3, 95% CI, 1.5 to 3.7), residence outside the immediate clinic catchment area (aHR: 2.5, 95% CI, 1.3 to 4.6 for Mombasa Island; aHR: 1.8, 95% CI, 1.0 to 3.3 for south coast), tertiary education level or higher (aHR: 2.3, 95% CI, 1.1 to 4.8) and less lead-in time in the cohort prior to 19 June 2017 (aHR: 3.1, 95% CI, 1.8 to 5.6 for zero to three months; aHR: 2.4, 95% CI, 1.2 to 4.7 for four to six months) were independent predictors of LTFU. PrEP use did not differ by LTFU status (HR: 1.0, 95% CI, 0.6 to 1.5). Psychosocial support for men reporting disordered alcohol use, strengthened engagement of recently enrolled participants and focusing recruitment on areas close to the research clinic may improve retention in HIV prevention studies involving MSM in coastal Kenya. Conclusions: About one in three participants became LTFU after one year of follow-up, irrespective of PrEP use. Research preparedness involving MSM should be strengthened for HIV prevention intervention evaluations in coastal Kenya. Keywords: MSM; attrition; pre-exposure prophylaxis; Kenya; LGBTQ; sub-Saharan Africa; HIV

Received 14 January 2020; Accepted 15 July 2020 Copyright © 2020 The Authors. Journal of the International AIDS Society published by John Wiley & Sons Ltd on behalf of the International AIDS Society. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

1 | INTRODUCTION other settings where same-sex behaviour is criminalized [8,9]. Men who have sex with men (MSM) are among the popula- HIV vaccine feasibility cohorts offering PrEP and other pre- tions at highest risk of HIV acquisition globally [1]. Incidence vention services to MSM provide helpful information on HIV estimates of HIV in MSM in sub-Saharan Africa (SSA) are incidence and retention. Among HIV-negative MSM followed 10 to 15 fold higher than in general populations in Africa: in such cohorts in Nairobi and coastal Kenya before PrEP roll- ranging from 5.1/100 person-years (PY) (95% confidence out, high rates of loss to follow-up (LTFU) have been docu- interval [CI], 2.6 to 9.8) in Kenya to 15.4/100 PY (95% CI, mented [10,11], with LTFU estimates as high as 42.2 [95% CI, 8.1 to 19.2) in Nigeria [2-4]. Pre-exposure prophylaxis (PrEP) 29.5 to 60.4] per 100 PY among men who have sex with men is effective for HIV prevention if adhered to [5,6]. While only [11]. With PrEP added as a biomedical prevention option PrEP provision to MSM is ongoing through research and offered at our clinic in 2017, we hypothesized that retention programmes in some settings in SSA [7], retaining MSM in among MSM would be better than previously recorded in the HIV prevention programmes is challenging in Kenya and same cohort. For this study, our objective was to estimate

30 Wahome EW et al. Journal of the International AIDS Society 2020, 23(S6):e25593 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25593/full | https://doi.org/10.1002/jia2.25593

LTFU rates and assess risk factors of LTFU among at-risk 2.3 | Retention and tracing activities MSM participating in a cohort on the Kenyan coast after PrEP became available in June 2017. At each visit, locator and contact information were verified and updated. Participants received 400 Kenyan shillings (KSh, 2 | METHODS $3.87) for transportation and time at each monthly visit, per local research guidelines. An additional 100 KSh ($0.98) was provided when they reported on the scheduled appointment 2.1 | Study setting and population date. Participants were reminded of scheduled visits one week Since July 2005, at-risk individuals have been recruited for an in advance and contacted by telephone within one day after a open HIV vaccine feasibility cohort at the Kenya Medical missed appointment. Those who could not be contacted by Research Institute (KEMRI) clinic in Mtwapa, coastal Kenya. phone were physically traced, with up to three attempts made Located 20 km north of Mombasa, Mtwapa is known for its within a 14-day window after the missed appointment. busy night life [12]. Participants were identified for recruitment by 10 to 15 trained peer mobilizers who approached 2.4 | Measures individuals through personal networks and at venues where sex workers meet clients. While any man aged 18 to 49 years 2.4.1 | Loss to follow-up who reported anal sex in the three months before screening was eligible [11], peer mobilizers were encouraged to mobilize Participants were defined as LTFU if they were late by participants at elevated risk for HIV acquisition, including >90 days for their scheduled appointment date. Participants younger men (18 to 24 years of age) and those who reported (N = 10) who re-engaged before the censoring date after they receptive anal intercourse (RAI) or sex with men exclusively were late by >90 days were defined as LTFU. [11,13]. This study includes all follow-up visits between 19 June 2017 (when PrEP became freely available to partici- 2.4.2 | Predictors of loss to follow-up pants) and 30 June 2019. Sociodemographic characteristics The following characteristics reported at enrolment were eval- 2.2 | Cohort procedures uated as potential risk factors: marital status, education, reli- Procedures have been described elsewhere [11,13,14]. Briefly, gion, employment status, earnings per month and years lived all monthly visits included a face-to-face interview to assess within the region. The study region was divided into four risk behaviour, HIV counselling and testing using rapid anti- areas; north coast (closest to KEMRI clinic), Mombasa island body tests, medical history and physical examination. All par- (20 km distant), south coast (24 km distant requiring a ticipants were provided syndromic treatment for symptoms ferry crossing) and Mombasa mainland and other more suggestive of sexually transmitted infections (STI), received remote areas (≥26 km distant, Figure 1). care for minor illnesses as indicated and were vaccinated against hepatitis B. Time-varying risk and mental health characteristics Beginning in 2016, participants completed a yearly assess- The following time-varying characteristics were evaluated as ment for depressive symptoms (Patient Health Questionnaire potential risk factors: age (18 to 24 vs. 25 + years), sex of sex 9 [PHQ-9]), alcohol use (Alcohol Use Disorder Identification partners in past three months (men and women or men only), Test [AUDIT]), use of substances other than alcohol and anal sex role in past three months (receptive, versatile, inser- tobacco (Drug Abuse Screening Test 10 [DAST-10]), sexual tive, no anal sex), condom use for anal sex in past three months stigma (abridged China MSM Stigma Scale) and recent trauma (yes/no), number of sex partners in past week (none, one, ≥2), via audio computer-assisted self-interview (ACASI) in English sexual behaviour in past week (no sexual activity, 100% condom or Swahili [15]. Following self-assessment, participants use, <100% condom use for reported sex acts), receiving pay- debriefed with a counsellor. Participants who reported moder- ment for sex in past three months (yes/no), paying for sex in ate or severe depressive symptoms, hazardous or harmful past three months (yes/no), group sex in past three months drinking, or moderate to severe abuse of other substances (yes/no), moderate to severe depressive symptoms in past two were engaged for supportive counselling at the KEMRI clinic weeks (PHQ-9 score 10 to 27), disordered alcohol use in past or referred to local services. year (AUDIT score ≥ 8), problematic substance use in past year Starting on 19 June 2017, free daily PrEP was provided (DAST-6 score ≥ 1), sexual stigma score (score 0 to 33), recent according to Kenyan Ministry of Health (MoH) guidelines to trauma (score ≥ 1) in past one year, travel in past three months participants eligible by MoH criteria or a cohort-derived (yes/no) and follow-up time in the cohort as of the June 2017 HIV risk score calculated at each visit [11,13]. Participants baseline (zero to three months, four to six months, >6 months). eligible for PrEP and interested in taking it were provided PrEP use was defined as receiving a PrEP refill at the previous with a 30-day PrEP supply. During monthly visits, PrEP visit and reporting continuation at the current visit (yes/no). adherence and adverse effects, HIV status and syndromic STIs were assessed and refills provided. Participants who 2.5 | Statistical analysis tested HIV-positive discontinued PrEP and were engaged for HIV care and treatment. Those not taking PrEP were Descriptive statistics were used to summarize demographic, re-assessed at each visit for eligibility and offered PrEP behavioural and mental health characteristics of participants when found eligible [16]. at first visit after June 2017 in the study. For mental health

31 Wahome EW et al. Journal of the International AIDS Society 2020, 23(S6):e25593 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25593/full | https://doi.org/10.1002/jia2.25593

Figure 1. Map of study area in coastal Kenya, 2017 to 2019.

32 Wahome EW et al. Journal of the International AIDS Society 2020, 23(S6):e25593 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25593/full | https://doi.org/10.1002/jia2.25593 data collected annually, the last observations were carried for- 33 (43.4%) could not be contacted through telephone calls or ward to monthly visits. Data for each participant were cen- physical tracing, 31 (40.8%) relocated to other towns outside sored at their last monthly visit before the censoring date (30 the study area and 9 (11.8%) withdrew from the study. Ten June 2019) or LTFU. Participants (N = 10) who re-engaged (13.2%) of 76 participants who were LTFU re-engaged in after being LTFU were censored at the last visit before being follow-up before study censoring. Of 143 participants who LTFU. Participants who acquired HIV infection during follow- started PrEP, 18 reported PrEP side effects: four (7%) among up were censored at the HIV diagnosis visit. Participants who those who were LTFU and 14 (16%) among those who were attended only one visit (N = 9) were included in the analysis retained (p = 0.09). and assigned a follow-up time of one day. Individual follow-up In bivariable analysis, LTFU was associated at p < 0.1 with, time was calculated from 19 June 2017 until the last visit age (18 to 29 years), education, earnings, years lived within before 30 June 2019. Attrition rates were calculated as the the area, area of residence, having sex with men only, sexual number of LTFU cases divided by total PY of follow-up and role taking, AUDIT score ≥8 and prior follow-up time in expressed as incidence per 100 PY. Cox proportional hazards cohort. Participants who became LTFU had similar reported models were used to assess risk factors for LTFU. Variables PrEP use as those retained (IRR 1.0, 95% CI, 0.6 to 1.5). significant in bivariable analysis at p < 0.1 and age a priori Receipt of incentive among participants who reported on the were included in the initial multivariable model of potential appointment date did not impact retention (IRR 1.0, 95% CI, predictors of LTFU. To reduce the number of predictors, only 0.7 to 1.6). In the final multivariable analysis, participants who variables with p < 0.1 were retained in the final multivariable were LTFU were more likely to have a tertiary education level model. Tests of collinearity were run on the model and predic- or higher, reside further away from the research clinic, report tors with a variance-covariance correlation of ≥0.5 were con- disordered alcohol use (AUDIT ≥ 8) and have participated in sidered collinear and not included in the final multivariable the cohort for a shorter time before PrEP became available model. p values were two-sided, and significance was set at than participants who were retained (Table 2). p < 0.05. Data were cleaned, recoded and analysed using In this study, we document a higher LTFU rate than previ- Stata 15.0 (StataCorp LLC, College Station, TX, USA). ously reported in the pre-PrEP period (33.7 vs. 23.9 per 100 PY) [11]. While PrEP interest [2,17-19], and uptake [16] among MSM in SSA is high, PrEP concentrations among 34 2.6 | Ethical considerations MSM and 8 TGW assessed six months following PrEP initia- The KEMRI Ethics Review Committee approved the study. All tion was low [20], suggesting that MSM face challenges in participants provided written informed consent. sustained daily PrEP taking. Therefore, alternative strategies are needed to support daily PrEP taking and to retain partici- 3 | RESULTS AND DISCUSSION pants in prevention services [21]. Consistent with other studies among MSM in Africa [22], we previously reported a high proportion of hazardous alco- 3.1 | Participants characteristics hol use (44% to 45%) among MSM in Kenya [15,23]. In this Out of 179 participants followed during the study period, 177 study, men reporting disordered alcohol use (AUDIT (98.9%) were eligible for PrEP and 143 (79.9%) started it. At score ≥ 8) were more than twice as likely to become LTFU the first visit after PrEP became available, the median age than those with lower AUDIT scores. These findings support was 25.0 years, interquartile range [IQR]: (23.0 to 30.0), more the need to screen participants for harmful substance use than half (52.0%) had a secondary education or higher. More and to facilitate individual or group supportive counselling than three-quarters (76.0%) had lived in the study area for services at the research clinic or referral to alcohol and >1 year, nearly half (49.2%) resided in north coast (closest to substance use harm reduction, treatment and support orga- research clinic), half (50.3%) qualified for the incentive for pre- nizations [24]. senting on the scheduled appointment date and nearly two- Men who had joined the cohort recently were more likely thirds (62.0%) had contributed >6 months of cohort follow-up to be LTFU. Similarly, in a study conducted among HIV-nega- before PrEP became available (Table 1). tive MSM in Brazil, the rate of LTFU was greatest in the first year of follow-up [25]. While men without an altruistic reason to join a longitudinal study may soon lose interest in partici- 3.2 | Loss to follow-up and risk factors pating after their immediate needs (e.g. STI treatment, HIV Overall, participants were followed for a median of testing) are met, our study suggests that participants need 21.2 months (IQR: 6.5 to 22.1), contributing 225.3 PY. Partici- stronger support or access to additional services during their pants made a total of 2,822 scheduled follow-up visits, of first few months of participation in order to be retained. which 1,404 (49.8%) occurred on the scheduled appointment Participants residing further away from the KEMRI clinic date. Nine participants acquired HIV during follow-up, of were more likely to be LTFU. Distance and time taken to tra- whom 7 (66.7%) had contributed >6 months of cohort follow- vel to the clinic likely contributed to missed study visits and up before PrEP became available (HIV incidence: 3.8 per 100 LTFU [26]. Of note, 42% of participants who became LTFU PY (95% CI: 2.0 to 7.3)). reported that they had relocated outside the study area. Seventy-six (42.5%) participants became LTFU, for a crude While transactional sex was not associated with LTFU in our incidence of 33.7 (95% CI, 26.9 to 42.2) per 100 PY. While study, it was a predictor of missed visits among 609 HIV- MSM who were LTFU contributed a median follow-up time of negative gay, bisexual and other MSM followed in a cohort 4.5 months (IQR: 1.6 to 13.8), those retained contributed study in Kisumu, Kenya [27]. MSM and male sex workers are 22.1 months (IQR: 21.2 to 22.8). Of the 76 LTFU participants, often mobile, which may impact LTFU [10,12]. Providing

33 Wahome EW et al. Journal of the International AIDS Society 2020, 23(S6):e25593 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25593/full | https://doi.org/10.1002/jia2.25593

Table 1. Characteristics of 179 HIV negative Kenyan MSM at first visit following PrEP availability, June 2017–June 2019

All (n = 179) LTFU (n = 76) Not LTFU (n = 103) Characteristics n (%) n (%) n (%)

Age group (years) 18 to 24 72 (40.2) 34 (44.7) 38 (36.9) 25+ 107 (59.8) 42 (55.3) 65 (63.1) Education Primary/none 70 (39.1) 24 (31.6) 46 (44.7) Secondary 93 (52.0) 43 (56.6) 50 (48.5) Higher/tertiary 16 (8.9) 9 (11.8) 7 (6.8) Marital status Never married 158 (88.3) 68 (89.5) 90 (87.4) Ever married 21 (11.7) 8 (10.5) 13 (12.6) Religion Christian 92 (51.4) 39 (51.3) 53 (51.5) Muslim 45 (25.1) 18 (23.7) 27 (26.2) Other/none 42 (23.5) 19 (25.0) 23 (22.3) Employment None 30 (16.8) 12 (15.8) 18 (17.5) Self 120 (67.0) 51 (67.1) 69 (67.0) Formal 29 (16.2) 13 (17.1) 16 (15.5) Earnings per month (100 KSh$0.98) ≥10,000 ($98.0) 44 (24.6) 23 (30.3) 21 (20.4) 5,000 to 9,000 91 (50.8) 41 (53.9) 50 (48.5) ($49.0-$88.2) <5,000 ($49.0) 44 (24.6) 12 (15.8) 32 (31.1) Years lived in area ≤1 43 (24.0) 22 (28.9) 21 (20.4) >1 136 (76.0) 54 (71.1) 82 (79.6) Area of residence North coast 88 (49.2) 27 (35.5) 61 (59.2) Mombasa island 21 (11.7) 15 (19.7) 6 (5.8) South coast 46 (25.7) 26 (34.2) 20 (19.4) Mombasa mainland/other 24 (13.4) 8 (10.5) 16 (15.5) areas Gender identityb Male 144 (80.4) 59 (77.6) 85 (82.5) Transgender woman/other 32 (17.9) 15 (19.7) 17 (16.5) Sex of sex partners, past three months Men and women 77 (43.0) 24 (31.6) 53 (51.5) Men only 102 (57.0) 52 (68.4) 50 (48.5) Anal sex role, past three months Receptive 51 (28.5) 24 (31.6) 27 (26.2) Versatile 83 (46.4) 39 (51.3) 44 (42.7) Insertive 38 (21.2) 12 (15.8) 26 (25.2) None 7 (3.9) 1 (1.3) 6 (5.8) Condom use for anal sex, past 73 (40.8) 30 (39.5) 43 (41.7) three monthsa Sex partners, past week None 71 (39.7) 27 (35.5) 44 (42.7) One 39 (21.8) 15 (19.7) 24 (23.3) ≥Two 69 (38.5) 34 (44.7) 35 (34.0) Sexual behaviour, past week No activity 71 (39.7) 27 (35.5) 44 (42.7) 100% condom use 72 (40.2) 34 (44.7) 38 (36.9)

34 Wahome EW et al. Journal of the International AIDS Society 2020, 23(S6):e25593 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25593/full | https://doi.org/10.1002/jia2.25593

Table 1. (Continued)

All (n = 179) LTFU (n = 76) Not LTFU (n = 103) Characteristics n (%) n (%) n (%)

<100% condom use 36 (20.1) 15 (19.7) 21 (20.4) Paid for sex with cash, living 31 (17.3) 14 (18.4) 17 (16.5) expenses or goods, past three months Received payment for sex with 114 (63.7) 56 (73.7) 58 (56.3) cash, living expenses or goods, past three months Group sex, past three months 3 (1.7) 1 (1.3) 2 (1.9) Depressive symptoms (PHQ-9), past 2 weeksb Minimal to mild (0 to 9) 113 (63.1) 45 (59.2) 68 (66.0) Moderate to severe 64 (35.8) 30 (39.5) 34 (33.0) (10 to 27) Disordered alcohol use (AUDIT), past yearb Low (0 to 7) 112 (62.6) 39 (51.3) 73 (70.9) Hazardous (8 to 40) 65 (36.3) 36 (47.4) 29 (28.2) Problematic substance use (DAST-6), past yearb Yes (≥1) 138 (77.1) 66 (86.8) 72 (69.9) Sexual stigma score (0 to 33) 6 (3 to 13) 6 (3 to 14) 6 (2 to 11) [Median (IQR)]b Recent trauma, past yearb None 67 (37.4) 27 (35.5) 40 (38.8) Any 108 (60.3) 46 (60.5) 62 (60.2) Travelled out of county, past 3 105 (58.7) 44 (57.9) 61 (59.2) monthsb Follow-up time in cohort as of June 2017 0 to 3 months 40 (22.3) 28 (36.8) 12 (11.7) 4 to 6 months 28 (15.6) 17 (22.4) 11 (10.7) >6 months 111 (62.0) 31 (40.8) 80 (77.7) Reported on appointment date 90 (50.3) 38 (50.0) 52 (50.5) Recent sexually transmitted 1 (0.6) 0 (0.0) 1 (1.0) infectionc

AUDIT, Alcohol Use Disorder Identification; DAST-6, Drug Abuse Screening Test 6; IQR, interquartile ranges; LFTU, loss to follow-up; PHQ-9, Patient Health Questionnaire 9. aAmong 172 MSM reporting anal sex in the past three months. bMissing 3 values for gender identity, PHQ-9, AUDIT, DAST to 6 and travelled out of county; missing 4 values for recent trauma; missing 9 values for sexual stigma. cDefined as detection of Gram-negative, intracellular diplococci in urethral or rectal secretions or rectal secretions or a new syphilis diagnosis within six months.

information about LGBTQ-friendly PrEP-providing facilities In this study, having sex with men only was not associated through peers or eHealth support [28], may improve continu- with LTFU. Previously, we documented that the most vulnera- ity of care when participants move or decide to change clinics. ble participants for HIV acquisition were also those who Compared to primary or no education, the rate of LTFU reported having sex with men only (HIV incidence rates: 35.2 was two times greater for participants with tertiary education per 100 PY) [11]. Same-sex behaviour is criminalized in most or higher. Why educated participants in our study were more of SSA [8,9], and MSM, especially those with no female part- likely to be LTFU is not clear. It is possible that they were ners, face stigma and discrimination [30-32]. While we did not more likely to have a formal employment, and their work collect data on sexual orientation for this study, gay men and schedule may have restricted them from keeping scheduled men who are more publicly visible such as sex workers likely clinic visits. Flexible visit schedules and access to prevention face greater stigma and discrimination in Kenya, and may services outside formal working hours may improve retention require additional support to remain in PrEP care. Facilitating for more educated participants [29]. linkage of MSM participants to local LGBTQ organizations for

35 Wahome EW et al. Journal of the International AIDS Society 2020, 23(S6):e25593 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25593/full | https://doi.org/10.1002/jia2.25593

Table 2. Risk factors for loss to follow-up among 179 Kenyan MSM, June 2017–June 2019

LTFU, n = 76 Bivariable analysis Multivariable analysis Characteristics n/100PY [rate] HR (95% CI)p value aHR (95% CI) p value

Age group (years)a 18 to 24 31/63.0 [49.2] 1.5 (0.9 to 2.4) 0.085 1.3 (0.8 to 2.2) 0.338 25+ 45/162.4 [27.7] Reference Reference Educationa Primary/none 24/95.1 [25.2] Reference Reference Secondary 43/114.1 [37.7] 1.4 (0.9 to 2.4) 0.156 1.3 (0.7 to 2.1) 0.391 Higher/tertiary 9/16.2 [55.5] 2.0 (0.9 to 4.4) 0.090 2.3 (1.1 to 4.8) 0.024 Marital status Never married 68/198.7 [34.7] Reference Ever married 8/26.6 [30.1] 0.9 (0.4 to 1.8) 0.692 –– Religion Christian 39/116.7 [33.4] Reference Muslim 18/56.7 [31.8] 1.0 (0.6 to 1.7) 0.879 –– Other/none 19/51.9 [36.6] 1.1 (0.6 to 1.9) 0.750 –– Employment None 12/39.8 [30.1] Reference Self 51/149.4 [34.1] 1.1 (0.6 to 2.1) 0.700 –– Formal 13/36.1 [36.0] 1.2 (0.5 to 2.7) 0.646 –– Earnings per month (100 KSh$0.98) ≥10,000 23/51.6 [44.6] 2.2 (1.1 to 4.4) 0.028 –– ($98.0) 5,000 to 9,000 41/109.6 [37.4] 1.9 (1.0 to 3.6) 0.057 –– ($49.0 to $88.2) <5,000 12/64.1 [18.7] Reference ($49.0) Years lived in area ≤1 22/44.9 [48.9] 1.6 (1.0 to 2.6) 0.070 –– >1 54/180.4 [29.9] Reference Area of residencea North coast 27/123.0 [22.0] Reference Reference Mombasa 15/18.2 [82.4] 3.4 (1.8 to 6.4) <0.001 2.5 (1.3 to 4.6) 0.004 island South coast 26/50.6 [51.4] 2.2 (1.3 to 3.7) 0.003 1.8 (1.0 to 3.3) 0.042 Mombasa 8/33.5 [23.9] 1.1 (0.5 to 2.4) 0.851 0.9 (0.4 to 2.0) 0.709 mainland/ other areas Gender identity Male 60/177.3 [33.8] Reference Transgender 15/46.4 [32.3] 1.0 (0.6 to 1.8) 0.984 –– woman/ other Sex partners, past week None 25/72.6 [34.5] Reference One 18/60.2 [29.9] 0.9 (0.5 to 1.7) 0.799 –– ≥Two 33/92.6 [35.6] 1.0 (0.6 to 1.8) 0.855 –– Sexual behaviour, past week No activity 25/73.1 [34.2] Reference 100% condom 39/94.8 [41.1] 1.2 (0.7 to 2.0) 0.470 –– use

36 Wahome EW et al. Journal of the International AIDS Society 2020, 23(S6):e25593 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25593/full | https://doi.org/10.1002/jia2.25593

Table 2. (Continued)

LTFU, n = 76 Bivariable analysis Multivariable analysis Characteristics n/100PY [rate] HR (95% CI)p value aHR (95% CI) p value

<100% 12/57.4 [20.9] 0.7 (0.3 to 1.3) 0.250 –– condom use Sex of sex partners, past three months Men and 26/113.3 [22.9] Reference women Men only 50/112.0 [44.6] 1.8 (1.1 to 2.9) 0.015 –– Anal sex role, past three months Receptive 23/50.6 [45.4] 1.2 (0.7 to 2.0) 0.487 –– Versatile 37/101.5 [36.4] Reference Insertive 11/58.9 [18.7] 0.5 (0.3 to 1.1) 0.075 –– None 5/14.2 [35.2] 1.0 (0.4 to 2.5) 0.962 –– Condom use for anal sex, past three months No 36/116.0 [31.0] Reference Yes 35/94.6 [37.0] 1.1 (0.7 to 1.8) 0.630 –– Paid for sex with cash, living expenses or goods, past three months No 64/196.1 [32.6] Reference Yes 12/29.2 [41.0] 1.1 (0.6 to 2.0) 0.785 –– Received payment for sex with cash, living expenses or goods, past three months No 30/108.2 [27.7] Reference Yes 46/117.1 [39.3] 1.3 (0.8 to 2.0) 0.245 –– Group sex, past three months No 75/219.0 [34.3] Reference Yes 1/6.3 [15.8] 0.5 (0.1 to 3.7) 0.524 –– Depressive symptoms (PHQ-9), past two weeks Minimal to 48/154.2 [31.1] Reference mild (0 to 9) Moderate to 27/69.5 [38.9] 1.1 (0.7 to 1.8) 0.586 –– severe (10 to 27) Disordered alcohol use (AUDIT), past yeara Low (0 to 7) 35/150.1 [23.3] Reference Reference Hazardous (8 40/73.6 [54.4] 2.2 (1.4 to 3.5) <0.001 2.3 (1.5 to 3.7) <0.001 to 40) Problematic substance use (DAST-6), past year No (0) 14/62.1 [22.5] Reference Yes (≥1) 61/161.6 [37.8] 1.5 (0.8 to 2.8) 0.161 –– Sexual stigma – 1.0 (1.0 to 1.0) 0.489 –– score (0 to 33) Recent trauma, past year None 28/107.7 [26.0] Reference Any 46/115.9 [39.7] 1.4 (0.9 to 2.2) 0.181 –– Travelled out of county, past three months No 46/135.5 [33.9] Reference –– Yes 29/88.2 [32.9] 1.1 (0.7 to 1.7) 0.818 –– Follow-up time in cohort as of June 2017a 0 to 3 months 28/35.9 [78.0] 3.7 (2.2 to 6.1) <0.001 3.1 (1.8 to 5.6) <0.001 4 to 6 months 17/26.7 [63.7] 3.0 (1.7 to 5.3) <0.001 2.4 (1.2 to 4.7) 0.009 >6 months 31/162.7 [19.1] Reference Reference Reported PrEP use No 36/90.8 [39.6] Reference Yes 40/134.5 [29.7] 1.0 (0.6 to 1.5) 0.855 ––

37 Wahome EW et al. Journal of the International AIDS Society 2020, 23(S6):e25593 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25593/full | https://doi.org/10.1002/jia2.25593

Table 2. (Continued)

LTFU, n = 76 Bivariable analysis Multivariable analysis Characteristics n/100PY [rate] HR (95% CI)p value aHR (95% CI) p value

Reported on appointment date No 42/125.2 [32.7] Reference Yes 35/100.1 [35.0] 1.0 (0.7 to 1.6) 0.885 –– Recent sexually transmitted infectionb No 76/222.9 [34.1] –– Yes 0/2.4 [0.0] –– aHR, adjusted hazard ratio; AUDIT, Alcohol Use Disorder Identification; CI, confidence intervals; DAST-6, Drug Abuse Screening Test 6; HR, hazard ratio; LTFU, loss to follow-up; PHQ-9, Patient Health Questionnaire 9; PrEP, pre-exposure prophylaxis; PY, person-years. aFactors significant at p < 0.1 in initial multivariable model (data not shown) and age a priori were included in the multivariable model. bDefined as detection of Gram-negative, intracellular diplococci in urethral or rectal secretions or rectal secretions or a new syphilis diagnosis within six months. support services may improve retention of study participants, newly enrolled participants. Further research is needed to especially when those organizations are included in research assess if peer outreach in collaboration with local LGBTQ planning and implementation [33]. organizations will improve research retention. To mitigate LTFU, we have adopted a peer mobilization model, using trained peer educators to act as a link between AUTHORS’ AFFILIATIONS participants and the research clinic. PrEP educational sessions 1KEMRI/Wellcome Trust Research Programme Centre for Geographic Medicine led by peers and study staff are provided to improve PrEP Research–Coast, Kilifi, Kenya; 2Departments of Medicine, Epidemiology and Glo- knowledge and discuss barriers. A collaboration with a local bal Health, University of Washington, Seattle, WA, USA; 3Department of Global 4 LGBTQ organization was strengthened to facilitate retention Health, University of Amsterdam, Amsterdam, the Netherlands; Nuffield and promote continuity of services for participants who move Department of Medicine, University of Oxford, Headington, United Kingdom or disengage from research follow-up. Strengthening engagement with participants during the study and targeted mobiliza- COMPETING INTERESTS tions of participants who reside in areas closer to the No competing interests were disclosed. research clinic may also improve retention. Our study had several limitations. First, the face-to-face AUTHORS’ CONTRIBUTIONS behavioural interview may have been subject to social desir- ES and SG designed the research study. AT, KM and JM collected the data. TO ability bias. However, the questionnaire on mental health and and EW managed data. EW analysed data and wrote the original draft of the substance use was conducted via ACASI and resulted in iden- manuscript. EW, SG, AT, KM, TO, JM, EG, JM, EMvdE and ES reviewed and edi- tification of an important factor (i.e. disordered alcohol use) ted the manuscript. All authors have read and approved the final manuscript. associated with LTFU. Second, the last observations for mental health variables collected at yearly time points were carried ACKNOWLEDGEMENTS forward to subsequent monthly visits and may have intro- We thank the participants and research team for their contributions to the duced misclassification bias. Third, we analysed LTFU as a bin- study. We thank Christopher Nyundo for generating a map of the study areas. ary outcome and did not assess number or patterns of missed visits. Of note, all participants made monthly visits, and the FUNDING LTFU rate might be lower if only quarterly visits were This work was supported by the International AIDS Vaccine Initiative (IAVI) and required. Fourth, assessments of sexual behaviour at monthly the University of Washington Center for AIDS Research, a National Institutes visits used a 3-month recall period, which may have led to of Health (NIH)-funded programme [R01AI124968], which is supported by the over-reporting of some exposures. Lastly, we could not deter- following NIH institutes and centres (NIAID, NCI, NIMH, NIDA, NICHD, NHLBI, mine reasons for LTFU or ascertain reengagement in PrEP NCCAM). SMG was also supported by the University of Washington/ Fred Hutch Center for AIDS Research, an NIH-funded programme under award num- care and other services among participants who were LTFU. ber AI027757 which is supported by the following NIH Institutes and Centers: NIAID, NCI, NIMH, NIDA, NICHD, NHLBI, NIA, NIGMS, NIDDK. The KEMRI Wellcome Trust Research Programme at the Centre for Geographical Medicine 4 | CONCLUSIONS Research–Kilifi is supported by core funding from the Wellcome Trust [203077]. This study was made possible by the generous support of the Ameri- can people through the United States Agency for International Development Our study documented a substantial LTFU from research par- (USAID). This work was also supported in part through the sub-Saharan African ticipation. We identified disordered alcohol use, distance to Network for TB/HIV Research Excellence (SANTHE), a DELTAS Africa Initiative research clinic, education level and prior follow-up time in the [DEL-15-006]. The DELTAS Africa Initiative is an independent funding scheme cohort as risk factors for LTFU. These factors suggest that of the African Academy of Sciences (AAS) Alliance for Accelerating Excellence in Science in Africa (AESA) and is supported by the New Partnership for Africa’s additional interventions to strengthen research participation Development Planning and Coordinating Agency (NEPAD Agency) with funding are needed, including screening and services for alcohol use, from the Wellcome Trust [107752] and the UK government. The views and that greater engagement and support may be needed by expressed in this publication are those of the authors and not necessarily those

38 Wahome EW et al. Journal of the International AIDS Society 2020, 23(S6):e25593 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25593/full | https://doi.org/10.1002/jia2.25593

of AAS, NEPAD Agency, Wellcome Trust or the UK government. The contents among men who have sex with men (MSM) in Ghana. AIDS Care. 2020;32 are the responsibility of the study authors and do not necessarily reflect the (3):330–6. views of USAID, NIH, the United States government, or the Wellcome Trust. 18. Karuga RN, Njenga SN, Mulwa R, Kilonzo N, Bahati P, O’Reilley K, et al. This report was published with permission from the director of KEMRI. "How I Wish This Thing Was Initiated 100 Years Ago!" willingness to take daily oral pre-exposure prophylaxis among men who have sex with men in Kenya. PLoS One. 2016;11:e0151716. 19. Emmanuel G, Folayan M, Undelikwe G, Ochonye B, Jayeoba T, Yusuf A, et al. Community perspectives on barriers and challenges to HIV pre-exposure REFERENCES prophylaxis access by men who have sex with men and female sex workers 1. UNAIDS. UNAIDS DATA. 2019 [cited 2019 Oct 23]. Available from: https:// access in Nigeria. BMC Public Health. 2020;20(1):69. www.unaids.org/sites/default/files/media_asset/2019-UNAIDS-data_en.pdf 20. EvdE MK, Chirro O, Ibrahim F, Mukuria N, Rinke T, et al.Pre-exposure pro- 2. Kimani M, van der Elst EM, Chiro O, Oduor C, Wahome E, Kazungu W, et al. phylaxis adherence and persistence in Kenyan MSM and transgender women. PrEP interest and HIV-1 incidence among MSM and transgender women in 2020 Conference on Retroviruses and Opportunistic Infections (CROI); Boston, coastal Kenya. J Int AIDS Soc. 2019;22:e25323. USA; Mar 08–11, 2020; Abstract 0878. 3. Lane T, Osmand T, Marr A, Struthers H, McIntyre JA, Shade SB. Brief 21. Pillay D, Stankevitz K, Lanham M, Ridgeway K, Murire M, Briedenhann E, report. J Acquir Immune Defic Syndr. 2016;73(5):609–11. et al. Factors influencing uptake, continuation, and discontinuation of oral PrEP 4. Nowak RG, Mitchell A, Crowell TA, Liu H, Ketende S, Ramadhani HO, et al. among clients at sex worker and MSM facilities in South Africa. PLoS One. Individual and sexual network predictors of hiv incidence among men who have 2020;15:e0228620. sex with men in Nigeria. J Acquir Immune Defic Syndr. 2019;80(4):444–53. 22. Sandfort TG, Knox JR, Alcala C, El-Bassel N, Kuo I, Smith LR. Substance 5. Grant RM, Lama JR, Anderson PL, McMahan V, Liu AY, Vargas L, et al. Pre- use and HIV risk among men who have sex with men in Africa: a systematic exposure chemoprophylaxis for HIV prevention in men who have sex with men. review. J Acquir Immune Defic Syndr. 2017;76(2):e34–e46. N Engl J Med. 2010;363(27):2587–99. 23. Korhonen C, Kimani M, Wahome E, Otieno F, Okall D, Bailey RC, et al. 6. Fonner VA, Dalglish SL, Kennedy CE, Baggaley R, O’Reilly KR, Koechlin FM, Depressive symptoms and problematic alcohol and other substance use in 1476 et al. Effectiveness and safety of oral HIV preexposure prophylaxis for all popu- gay, bisexual, and other MSM at three research sites in Kenya. AIDS. 2018;32 lations. AIDS. 2016;30(12):1973–83. (11):1507–15. 7. AVAC. Country updates. 2019 [cited 2019 Oct 28]. Available from: https:// 24. Alcoholics anonymous [cited 2019 Nov 11]. Available from: https://www.aa- www.prepwatch.org/country-updates/ kenya.or.ke/meetings.html 8. Clark F. Discrimination against LGBT people triggers health concerns. Lancet. 25. Silva AP, Greco M, Fausto MA, Carneiro M. Loss to follow-up in a cohort of 2014;383(9916):500–2. HIV-negative men who have sex with men: project horizonte. Rev Saude Publica. 9. Congress LLo. Criminal laws on homosexuality in African nations. 2014 [cited 2017;51:60. 2019 Oct 28]. Available from: https://www.loc.gov/law/help/criminal-laws-on-ho 26. Mayer CM, Owaraganise A, Kabami J, Kwarisiima D, Koss CA, Charlebois mosexuality/homosexuality-laws-in-african-nations.pdf ED, et al. Distance to clinic is a barrier to PrEP uptake and visit attendance in a 10. Price MA, Rida W, Mwangome M, Mutua G, Middelkoop K, Roux S, et al. community in rural Uganda. J Int AIDS Soc. 2019;22:e25276. Identifying at-risk populations in Kenya and South Africa: HIV incidence in 27. Kunzweiler C, Bailey RC, Okall DO, Graham SM, Mehta SD, Otieno‐Nyu- cohorts of men who report sex with men, sex workers, and youth. J Acquir nya B, Djomand G, et al. Enrolment characteristics associated with retention Immune Defic Syndr. 2012;59(2):185–93. among HIV negative Kenyan gay, bisexual and other men who have sex with 11. Sanders EJ, Okuku HS, Smith AD, Mwangome M, Wahome E, Fegan G, men enrolled in the Anza Mapema cohort study. J Int AIDS Soc. 2020;23(S6): et al. High HIV-1 incidence, correlates of HIV-1 acquisition, and high viral loads e25598. following seroconversion among MSM. AIDS. 2013;27(3):437–46. 28. Maloney KM, Bratcher A, Wilkerson R, Sullivan PS. Electronic and other 12. Geibel S, van der Elst EM, King’ola N, Luchters S, Davies A, Getambu new media technology interventions for HIV care and prevention: a systematic EM, et al. ’Are you on the market?’: a capture-recapture enumeration of men review. J Int AIDS Society. 2020;23:e25439. who sell sex to men in and around Mombasa, Kenya. AIDS. 2007;21 29. Sahay S, Gupte N, Brahme RG, Nirmalkar A, Bembalkar S, Bollinger RC, (10):1349–54. et al. Predictors of retention among men attending STI clinics in HIV prevention 13. Wahome E, Thiong’o AN, Mwashigadi G, Chirro O, Mohamed K, Gichuru E, programs and research: a case control study in Pune, India. PLoS One. 2011;6: et al. An Empiric risk score to guide PrEP targeting among MSM in coastal e17448. Kenya. AIDS Behav. 2018;22 Suppl 1:35–44. 30. Smith AD, Tapsoba P, Peshu N, Sanders EJ, Jaffe HW. Men who have sex 14. Wahome E, Ngetsa C, Mwambi J, Gelderblom HC, Manyonyi GO, Micheni with men and HIV/AIDS in sub-Saharan Africa. Lancet. 2009;374(9687):416–22. M, et al. Hepatitis B virus incidence and risk factors among Human immunodefi- 31. Sharma A, Bukusi E, Gorbach P, Cohen CR, Muga C, Kwena Z, et al. Sexual ciency virus-1 negative men who have sex with men In Kenya. Open Forum identity and risk of HIV/STI among men who have sex with men in Nairobi. Sex Infect Dis. 2016;4:ofw253. Transm Dis. 2008;35(4):352–4. 15. Secor AM, Wahome E, Micheni M, Rao D, Simoni JM, Sanders EJ, et al. 32. Schwartz SR, Nowak RG, Orazulike I, Keshinro B, Ake J, Kennedy S, et al. Depression, substance abuse and stigma among men who have sex with men in The immediate effect of the Same-Sex Marriage Prohibition Act on stigma, dis- coastal Kenya. AIDS. 2015;29:S251–9 crimination, and engagement on HIV prevention and treatment services in men 16. Wahome E, Graham S, Thiong’o A, Chirro O, Mohamed K, Gichuru E,, et al. who have sex with men in Nigeria: analysis of prospective data from the TRUST Assessment of PrEP eligibility and uptake among at-risk MSM participating in a cohort. Lancet HIV. 2015;2(7):e299–306. HIV-1 vaccine feasibility cohort in coastal Kenya. Wellcome Open Res. 33. Graham SM, Micheni M, Secor A, van der Elst EM, Kombo B, Operario D, 2019;4:138 et al. HIV care engagement and ART adherence among Kenyan gay, bisexual, 17. Ogunbajo A, Leblanc NM, Kushwaha S, Boakye F, Hanson S, Smith MDR, and other men who have sex with men: a multi-level model informed by qualita- et al. Knowledge and Acceptability of HIV pre-exposure prophylaxis (PrEP) tive research. AIDS Care. 2018;30 Sup5:S97–105.

39 Kayode BO et al. Journal of the International AIDS Society 2020, 23(S6):e25592 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25592/full | https://doi.org/10.1002/jia2.25592

RESEARCH ARTICLE Retention of a cohort of men who have sex with men and transgender women at risk for and living with HIV in Abuja and Lagos, Nigeria: a longitudinal analysis Blessing O Kayode1,*, Andrew Mitchell2,*, Nicaise Ndembi1, Afoke Kokogho3,4, Habib O Ramadhani2 , Sylvia Adebajo5, Merlin L Robb4,6, Stefan D Baral7 , Julie A Ake4, Manhattan E Charurat2, Trevor A Crowell4,6 , Rebecca G Nowak2,§ and on behalf of the TRUST/RV368 Study Group §Corresponding author: Rebecca G Nowak, 725 W. Lombard Street, Baltimore, Maryland 21201. Tel: +1 410 706 4642. ([email protected]) *Blessing O Kayode and Andrew Mitchell contributed equally to this work. Members of the The TRUST/RV368 Study Group are listed in the Acknowledgements section. This work was presented, in part, at the International Conference on AIDS and STIs in Africa (ICASA) in Abidjan, Cote^ d’Ivoire, 4 to 9 December 2017.

Abstract Introduction: Men who have sex with men (MSM), and transgender women (TGW), face specific obstacles to retention in care, particularly in settings with stigmatization such as sub-Saharan Africa. We evaluated the impacts of HIV status and other factors on loss-to-follow-up (LTFU) and visit adherence among MSM and TGW in Abuja and Lagos, Nigeria. Methods: TRUST/RV368 is an open cohort that provides comprehensive and integrated prevention and treatment services for HIV and sexually transmitted infections (STIs) at community venues supportive of sexual and gender minorities. Recruit- ment began in March 2013 and participants were followed every three months for up to 18 months. LTFU was defined as not presenting for an expected visit in the past 180 days. Visit adherence was calculated as a rate of completed visits adjusted by the number of three-month intervals elapsed since enrolment. HIV and other factors predictive of LTFU and visit adherence were evaluated using Cox proportional hazards and Poisson regression models, respectively. Results: A total of 1447 participants who completed enrolment evaluations over two visits as of November 2018 were included in these analyses. Their median age was 24 years (interquartile range [IQR]: 21 to 28) and 53% (n = 766) were living with HIV. LTFU occurred in 56% (n = 808) and visit adherence was 0.62 (95% confidence interval: 0.61 to 0.64) visits per three-month interval. Participants at risk and living with HIV had median follow-up times of 12 months (IQR: 6 to 22), and 21 months (IQR: 12 to 30), respectively (p < 0.01). After controlling for other factors, LTFU was less common among participants living with HIV or other STIs and more common among those who did not own a cell phone, sold sex and had never undergone HIV testing prior to enrolment. These factors had parallel associations with visit adherence. Conclusions: Retention was suboptimal in Nigerian clinics designed to serve MSM and TGW. Particularly high LTFU and low visit adherence among participants at risk for HIV could complicate deployment of HIV prevention interventions. Marketing the benefits of testing, improving access to cell phones and nurturing more trust with clients may improve retention among marginalized communities in Nigeria. Keywords: sub-Saharan Africa; retention in care; treatment adherence and compliance; sexual and gender minorities; HIV

Additional Supporting information may be found online in the Supporting Information tab for this article.

Received 10 January 2020; Accepted 15 July 2020 Copyright © 2020 The Authors. Journal of the International AIDS Society published by John Wiley & Sons Ltd on behalf of the International AIDS Society. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

1 | INTRODUCTION linkage to treatment in Nigeria [1,2]. In Nigeria and several other African countries, the criminalization of same-sex sexual Retention of people at risk for and living with HIV (PLWH) in practices further hinders linkage and retention for sexual and evidence-based programmes has its challenges in countries gender minorities [3]. A recent meta-analysis of African men across sub-Saharan Africa where loss to follow-up (LTFU) esti- who have sex with men (MSM) living with HIV showed that mates range from 3% to 45% and are as high as 75% after only 37% to 53% were taking antiretroviral therapy (ART) and

40 Kayode BO et al. Journal of the International AIDS Society 2020, 23(S6):e25592 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25592/full | https://doi.org/10.1002/jia2.25592

34% achieved viral suppression [4]. Sub-optimal engagement 2 | METHODS and retention in care undermine the continent-wide successes of expanded ART access for key populations. 2.1 | Study design and population The broad range of reported engagement and retention among MSM in sub-Saharan Africa can be explained by both TRUST/RV368 recruited biological males who reported sex study-level differences in assessing LTFU [1] as well as individ- with men in Abuja and Lagos, Nigeria, two urban centres ual factors that predict LTFU such as younger age, transporta- approximately 530 km apart, into an open prospective HIV tion difficulties, lack of social support and overall perception treatment-as-prevention study as previously described of feeling healthy [5]. Men are also more difficult to retain [22,23]. In brief, respondent-driven sampling was used by initi- than women due to a historical focus on HIV testing and ating well-networked gender and sexual minorities, termed treatment as a maternal and child health issue [6]. For sexual “seeds,” to provide referral coupons to three eligible peers, and gender minorities, sexual behaviour stigma further who similarly received three referral coupons upon enrolment. impedes access and retention in healthcare facilities [7]. Sex- Participants were compensated with 1000 Naira (approxi- ual behaviour stigma as a multifaceted construct includes mately $3 USD) for each peer referral. enacted stigma (behavioural expressions including physical vio- Enrolment criteria included assigned male sex at birth, anal lence); internalized stigma (feelings of stigma) and anticipated sex with a male partner in the past year, a valid referral cou- stigma (expectations of stigma) [7,8]. pon, and written informed consent in English or Hausa. Age Community-based models integrate members of distinct inclusion criteria differed between sites based on Institutional sub-populations to mobilize care outside of traditional models Review Board (IRB) recommendations (≥16 years in Abuja or or facilities of healthcare [9]. Such models circumvent many of ≥18 years in Lagos). Participants were expected to present the contextual challenges stigmatized individuals face. Merging for a total of eight visits, beginning with enrolment evaluations of HIV prevention with MSM peer educators in Malawi spread over two visits approximately two weeks apart (visits 0 enabled retention of 81% of 106 at-risk MSM through three and 1). Subsequent visits (2 to 7) were scheduled at three- quarterly follow-up visits, although many had previously month intervals for a total of 18 months. Participants received demonstrated high adherence [10]. In Senegal, a stigma miti- a monetary incentive for visit completion, starting at 1000 gation study sensitizing peer educators and healthcare work- Naira upon enrolment and increasing by 200 Naira for each ers at government health facilities reported a lower six-month subsequent visit. Participants completed a structured ques- retention of 14% (102/724) [11]. In South Africa, two public tionnaire that captured demographic, behavioural and clinical health facilities designated for men-only incorporated sensi- characteristics, and provided urine samples, anal swabs and tized training on the sexual health needs of MSM and blood samples for HIV/STI diagnostics. Only those who com- reported a two-year retention of 82%, although less than 20% pleted the enrolment evaluations at least 180 days before were MSM and retention estimates were restricted to those data censoring, underwent an HIV test with a valid result, and on ART [12]. Community-based clinics for MSM, TGW and did not relocate, die, or voluntarily withdraw from the study other sexual and gender minorities may mitigate stigma and were included in these analyses. social barriers, but it remains unclear whether they can retain a large cohort in HIV prevention and clinical care. 2.2 | Ethical considerations The TRUST/RV368 study, in conjunction with non-governmental organizations, tailored their clinical care to thousands IRBs at the Nigerian Federal Capital Territory Health of Nigerian MSM and TGW. These facilities synergized health Research Ethics Committee, the Nigerian Ministry of Defense and human rights by housing an advocacy group and a health in Nigeria, the University of Maryland Baltimore and the Wal- clinic with staff inclusive of sexual and gender minorities [13]. ter Reed Army Institute of Research approved the research Together, facilities focused on the social, legal and sexual protocol. All participants provided informed consent. health needs of study participants and sensitization training was provided to promote integration of the two entities. Ser- 2.3 | Retention strategies vices included education about safer sex practices, distribution of condoms and condom-compatible lubricants and diagnosis Prior to study implementation, staff underwent sensitization and treatment of HIV and other sexually transmitted infec- training that included a week-long session with educators tions (STIs) [14-19]. Despite these services, we have previ- from the Fenway Institute, an internationally recognized inter- ously reported a high HIV incidence of 15 infections per 100 disciplinary centre focused on the delivery of destigmatized person-years in the cohort [20]. Incidence was highest among medical care to MSM, TGW and PLWH. Ethnographic assess- participants under 19 years of age, a group that has poor gen- ments, focus groups and meetings with grassroots organiza- eral engagement in the HIV prevention and care cascade [21]. tions were conducted to educate staff on community needs. We hypothesized that retention would be lower for those Phone numbers, emails and residential addresses were used who were not living with HIV and characteristics more com- to contact participants two weeks prior to each appointment. mon among young MSM and TGW would be independently If an appointment was missed, staff attempted to reschedule associated with a decrease in retention. Our objective was to with repeated phone calls or through social networks. During identify factors associated with LTFU and visit adherence – study visits, participants were escorted to each point of care two complementary measures of retention – among sexual to maximize comfort. To promote retention and minimize and gender minorities in a community-based HIV prevention stigmatization, a Community Advisory Board comprised of and treatment study. sexual and gender minorities as well as other stakeholders,

41 Kayode BO et al. Journal of the International AIDS Society 2020, 23(S6):e25592 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25592/full | https://doi.org/10.1002/jia2.25592 totalling 20 to 25 members, met as frequently as bi-monthly receptive sexual practices (IAI, RAI), condom use with a male to discuss the study procedures and environment at each partner at last anal sex, buying and/or selling of sex, prior HIV clinic. Regular community social events, such as beauty testing and worry about HIV infection. For gender, partici- pageants, candlelight processions, film screenings and panel pants were asked “What do you consider your gender to be?”. discussions were organized to promote the clinics as safe Options included man, woman, other or both man and woman, spaces. and participants were categorized as cisgender men, TGW or other/unknown gender. Characteristics related to stigma included disclosure of MSM status to healthcare workers and 2.4 | Laboratory procedures sexual behaviour stigma (self-reported experience of verbal At enrolment and subsequent visits, participants who were at harassment as a result of being MSM and fear of accessing risk for HIV underwent HIV testing using fingerstick collection healthcare services because of worry someone may learn of whole blood and Determine (Alere, Waltham, MA, USA) and MSM status). Social support characteristics were captured on Uni-gold (Trinity biotech, Co-Wicklow, Ireland) test kits as out- a 4-point Likert scale (1 = strongly disagree to 4 = strongly lined by the parallel testing algorithm [24]. A third rapid test, agree) and dichotomized as any agreement vs. any disagree- HIV-1/2 Stat-Pak (Chembio Diagnostics, Medford, NY, USA) ment to the following statements: “The group of friends with was used for discordant results. For all participants, voided whom you socialize is a mix of straight people and MSM,” and urine and anal swabs were tested for Neisseria gonorrhoeae “You can trust the majority of MSM you know.” HIV, CT and (NG) and Chlamydia trachomatis (CT) using the Aptima Combo NG status at enrolment were based on study-provided labora- 2 CT/NG Assay (Hologic, San Diego, CA, USA). Participants tory testing. found to be living with HIV underwent ART preparation and initiation per treatment-as-prevention guidelines [22]. Bacte- 2.7 | Statistical analysis rial STIs were treated with antibiotic therapy provided at the clinics. LTFU rates were calculated for all participants and within groups of each characteristic. For our primary analyses, the youngest age group considered as an independent variable in 2.5 | Outcomes models was 16 to 19 years; since only the Abuja site enrolled 2.5.1 | Loss to follow-up participants aged 16 to 17 years, we compared characteristics of the 16 to 17 year and 18 to 19 year groups from Abuja Under a prospective definition of LTFU, participants who do not using Chi-square tests. For LTFU, Cox proportional hazards present for a minimum number of days or study visits are cate- regression models were used to identify predictive character- gorized as LTFU, regardless of whether they ultimately re-en- istics in bivariate analyses (p < 0.05), and a multivariable gage. This method often results in misclassification of model was built using forward stepwise selection with priority individuals who return to the facility after an extended absence given to factors previously associated with LTFU. Remaining [1]. To avoid misclassification, we categorized participants based factors were entered according to the magnitude of their on number of days since their most recent visit, regardless of crude association with LTFU, and all that remained significant intermittent gaps in care. These gaps may have exceeded were retained in the model. 180 days, but participants were not considered LTFU if they Poisson regression models of visit adherence were offset by eventually returned to clinic. Participants who completed all the log of expected number of visits according to time elapsed scheduled visits were censored at their last visit and those in since date of enrolment, and resulting coefficients were expo- ongoing follow-up were censored on 2 November 2018. nentiated into rate ratios. Similar to LTFU analyses, candidate variables identified in bivariate analyses were entered into a 2.5.2 | Visit adherence multivariable model via forward stepwise selection. Crude and adjusted rate ratios with 95% confidence intervals (CIs) were To account for varying periods of study observation, visit calculated for each characteristic. adherence was calculated by dividing the number of com- Analyses were repeated with stratification by gender to pleted visits by the number of expected visits. This was evaluate differences in correlates of retention among MSM expressed as a rate of visits completed per three-month inter- and TGW. Data were analysed using Statistical Analysis Soft- val, which was the expected interval between scheduled visits. ware (SAS) version 9.4 (SAS Institute, Cary, NC).

| 2.6 Independent variables 3 | RESULTS HIV status was the main exposure of interest. Demographic and behavioural characteristics were also explored as indepen- A total of 2,386 participants enrolled in the TRUST/RV368 dent predictors of LTFU and visit adherence. All predictors cohort from March 2013 to November 2018. Twenty-one per- were assessed at enrolment in order to replicate the risk cent (n = 491) were excluded because of a missing HIV test stratification and profiling that occur at initial entry into care. result, 14% (n = 323) did not complete both enrolment visits, This did not allow for covariates that changed during follow- 4% (n = 107) enrolled less than 180 days before data censor- up, such as in the case of HIV seroconversion. Covariates ing and <1% relocated (n = 9), withdrew (n = 5) or died included age, study site, education level, employment status, (n = 4). sexual orientation, gender identity, cell phone ownership, num- A total of 1447 participants were included in these analyses ber of male sexual partners in the past year, insertive and/or with median age 24 years (interquartile range [IQR]:21 to 28).

42 Kayode BO et al. Journal of the International AIDS Society 2020, 23(S6):e25592 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25592/full | https://doi.org/10.1002/jia2.25592

Table 1. Distribution of demographic, behavioural and clinical characteristics of participants enrolled in TRUST/RV368 overall and by HIV status

Total At risk for HIV PLWH N = 1447 N = 681 N = 766 Characteristic n (%) n (%) n (%) p value

Age (years) <0.01 16 to 19 197 (13.6) 135 (19.8) 62 (8.1) 20 to 24 603 (41.7) 294 (43.2) 309 (40.3) 25+ 647 (44.7) 252 (37.0) 395 (51.6) Study site <0.01 Abuja 986 (68.1) 522 (76.7) 464 (60.6) Lagos 461 (31.9) 159 (23.3) 302 (39.4) Education 0.11 ≤ High school 914 (63.2) 445 (65.3) 469 (61.3) > High school 532 (36.8) 236 (34.7) 296 (38.7) Employment status 0.62 Unemployed 352 (24.8) 168 (25.4) 184 (24.2) Employed/student 1069 (75.2) 494 (74.6) 575 (75.8) Sexual orientation 0.02 Homosexual 453 (31.4) 192 (28.3) 261 (34.2) Bisexual 988 (68.6) 486 (71.7) 502 (65.8) Gender identity <0.01 Cisgender man 1153 (79.9) 571 (84.1) 582 (76.2) Transgender woman 158 (10.9) 60 (8.8) 98 (12.8) Other/unknown 132 (9.1) 48 (7.1) 84 (11.0) Owns a cell phone <0.01 No 69 (4.8) 49 (7.2) 20 (2.6) Yes 1371 (95.2) 629 (92.8) 742 (97.4) Number of male sexual partners in past year <0.01 0 to 4 691 (48.3) 354 (52.8) 337 (44.2) 5 to 9 379 (26.5) 175 (26.1) 204 (26.8) 10+ 362 (25.3) 141 (21.0) 221 (29.0) Receptive and/or insertive anal sexual practices in the <0.01 past year RAI 309 (21.7) 123 (18.5) 186 (24.5) IAI 351 (24.6) 247 (37.1) 104 (13.7) IAI and RAI 766 (53.7) 296 (44.4) 470 (61.8) Condom used at last anal sex with male partner 0.11 No 498 (34.6) 249 (36.7) 249 (32.7) Yes 943 (65.4) 430 (63.3) 513 (67.3) Sold sex in past year 0.09 No 818 (57.0) 368 (54.6) 450 (59.1) Yes 618 (43.0) 306 (45.4) 312 (40.9) Bought sex in past year 0.27 No 1032 (71.9) 493 (73.3) 539 (70.6) Yes 404 (28.1) 180 (26.7) 224 (29.4) Ever test for HIV <0.01 Yes 1165 (80.7) 502 (73.9) 663 (86.7) At least somewhat worried about HIV in the past year 0.04 No 917 (63.6) 449 (66.4) 468 (61.1) Yes 525 (36.4) 227 (33.6) 298 (38.9) Ever been verbally harassed for being MSM <0.01 No 1000 (69.2) 515 (75.7) 485 (63.3) Yes 446 (30.8) 165 (24.3) 281 (36.7) (Continued)

43 Kayode BO et al. Journal of the International AIDS Society 2020, 23(S6):e25592 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25592/full | https://doi.org/10.1002/jia2.25592

Table 1. (Continued)

Total At risk for HIV PLWH N = 1447 N = 681 N = 766 Characteristic n (%) n (%) n (%) p value

Ever disclosed MSM status to healthcare worker <0.01 No 923 (64.0) 500 (73.9) 423 (55.3) Yes 519 (36.0) 177 (26.1) 342 (44.7) Ever been afraid to access health services <0.01 No 936 (64.7) 474 (69.6) 462 (60.4) Yes 510 (35.3) 207 (30.4) 303 (39.6) Friends with whom socialize are MSM and heterosexual 0.95 No 90 (6.2) 42 (6.2) 48 (6.3) Yes 1351 (93.8) 635 (93.8) 716 (93.7) Trusts the majority of other MSM they know <0.01 No 686 (47.6) 274 (40.5) 412 (53.9) Yes 755 (52.4) 402 (59.5) 353 (46.1) Chlamydia trachomatis 0.34 Negative 1196 (83.6) 555 (82.6) 641 (84.5) Positive 235 (16.4) 117 (17.4) 118 (15.5) Neisseria gonorrhoeae <0.01 Negative 1127 (78.8) 552 (82.1) 575 (75.8) Positive 304 (21.2) 120 (17.9) 184 (24.2)

IAI, insertive anal intercourse; MSM, men who have sex with men; PLWH, people living with HIV; RAI, receptive anal intercourse.

PLWH had median age 26 years (IQR: 23 to 29), whereas older. Exploratory predictors of retention consistent across participants at risk for HIV had median age 23 years (IQR: 20 models for both LTFU and visit adherence included not own- to 27; p < 0.01). Eighty percent (n = 1153) identified as cis- ing a cell phone, selling sex within the past year, no prior test- gender men, and 43% (n = 618) sold sex within the past year ing for HIV, trusting the majority of MSM acquaintances, and (Table 1). Enrolment prevalence of HIV, CT and NG was 53% not presenting with NG at enrolment (Tables 2 and 3). (n = 766), 16% (n = 235) and 21% (n = 304) respectively. A Belonging to a social group comprised of MSM and heterosex- significantly lower proportion of 16 to 17 year-olds owned a uals was significantly associated with increased visit adherence cell phone, used a condom at last anal sex, ever tested for but not LTFU. After stratification by gender, participants at HIV, and were afraid to access health services as compared to risk for HIV had lower retention among cisgender MSM and 18 to 19 year olds in Abuja (all p < 0.05) (Table S1). participants with other/unknown gender, but not TGW. All Median follow-up time in the cohort was 17.3 months (IQR: exploratory predictors identified earlier were also significant 8 to 26 months) and participants completed a median 71% of in analyses restricted to cisgender MSM (Tables S1 and S2). their expected visits (IQR: 25 to 100%). Participants at risk For TGW, owning a cell phone or prior testing for HIV were for HIV had a median follow-up time of 12.4 months (IQR: 6 the only significant predictors of retention (Tables S4 and S5). to 22 months), compared to 20.5 months (IQR: 12 to For participants with other/unknown gender, living with HIV 30 months) for PLWH (p < 0.01). Participants at risk for HIV was the only independent predictor of retention (Tables S6 had a median proportion of visit adherence of 42.9% (IQR: and S7). 14% to 86%), compared to 85.7% (IQR: 40% to 100%) for PLWH (p < 0.01). Over the course of the study 56% (n = 808) of participants 4 | DISCUSSION were LTFU. This included 66% (n = 449) of all participants at risk for HIV and 47% (n = 359) of all PLWH. Being at risk for HIV status was a significant predictor of both LTFU and visit HIV was independently associated with increased risk of LTFU adherence in our study. While PLWH had better retention (HR: 1.72, 95% CI: 1.49 to 2.00; Table 2) and lower visit outcomes, overall retention for all groups evaluated in this adherence (RR: 0.80, 95% CI: 0.75 to 0.85; Table 3). Younger study was suboptimal. Prior retention estimates for MSM in age was not predictive of LTFU or visit adherence in the mul- Malawi and Senegal were 81% and 14% over six or nine tivariable models, though participants in the 20- to 24-year- months, respectively [10,11], a wide range that brackets our old group had lower visit adherence (RR: 0.92, 95% CI: 0.87, estimate of 44% despite shorter follow-up. Our annualized 0.98) and a non-significant increase in LTFU (HR: 1.14, 95% LTFU rate of 32.7 per 100 person-years was lower than the CI: 0.97 to 1.34) as compared to participants 25 years or 59.5 per 100 person-years observed in a retrospective cohort

44 Kayode BO et al. Journal of the International AIDS Society 2020, 23(S6):e25592 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25592/full | https://doi.org/10.1002/jia2.25592

Table 2. Loss-to-follow-up among participants enrolled in TRUST/RV368 by demographic, behavioural and clinical characteristics

Characteristic LTFU n PY LTFU rate Crude HR (95% CI) p-value Adjusted HR (95% CI) p-value

Overall 808 2471 32.70 –– HIV status At risk 449 975 46.03 1.95 (1.69 to 2.24) <.01 1.72 (1.49 to 2.00) <.01 Living with HIV 359 1495 24.01 ref –– – Age (years) 16 to 19 125 298 41.90 1.51 (1.23 to 1.86) <0.01 1.09 (0.86 to 1.37) 0.48 20 to 24 353 993 35.55 1.31 (1.12 to 1.52) <0.01 1.14 (0.97 to 1.34) 0.10 25+ 330 1179 27.98 ref –– – Study site Abuja 538 1639 32.82 ref – Lagos 270 831 32.47 0.99 (0.85 to 1.15) 0.87 Education ≤High school 536 1491 35.94 1.32 (1.14 to 1.52) <0.01 >High school 271 977 27.75 ref – Employment status Unemployed 180 592 30.40 ref – Employed/student 602 1863 32.31 1.04 (0.88 to 1.23) 0.62 Sexual orientation Homosexual 246 795 30.94 ref – Bisexual 557 1668 33.40 1.08 (0.93 to 1.25) 0.32 Gender identity Cisgender man 663 1966 33.73 ref – Transgender woman 81 275 29.45 0.88 (0.70 to 1.11) 0.29 Other/unknown 61 222 27.44 0.84 (0.64 to 1.09) 0.18 Owns a cell phone No 51 73 69.88 2.30 (1.73 to 3.06) <0.01 1.82 (1.35 to 2.45) <0.01 Yes 752 2384 31.55 ref –– – Number of male sexual partners in past year 0 to 4 367 1147 32.00 ref – 5 to 9 214 640 33.46 1.07 (0.90 to 1.26) 0.45 10+ 212 672 31.56 0.99 (0.84 to 1.17) 0.91 Receptive and/or insertive anal sexual practices in the past year RAI only 155 546 28.38 ref IAI only 216 591 36.54 1.28 (1.04 to 1.57) 0.02 IAI and RAI 418 1312 31.87 1.13 (0.94 to 1.36) 0.18 Condom used at last anal sex with male partner No 285 855 33.32 1.04 (0.90 to 1.20) 0.59 Yes 517 1605 32.22 ref – Sold sex in the past year No 415 1467 28.29 ref –– – Yes 383 995 38.50 1.38 (1.20 to 1.58) <0.01 1.34 (1.15 to 1.55) <0.01 Bought sex in the past year No 574 1773 32.37 ref Yes 224 691 32.42 1.01 (0.86 to 1.18) 0.93 Ever tested for HIV No 206 419 49.21 1.72 (1.46 to 2.01) <0.01 1.46 (1.23 to 1.73) <0.01 Yes 600 2048 29.29 ref –– – At least somewhat worried about HIV in past year No 525 1557 33.72 1.11 (0.96 to 1.23) 0.16 Yes 278 910 30.55 ref – Ever been verbally harassed for being MSM No 540 1652 32.69 ref – (Continued)

45 Kayode BO et al. Journal of the International AIDS Society 2020, 23(S6):e25592 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25592/full | https://doi.org/10.1002/jia2.25592

Table 2. (Continued)

Characteristic LTFU n PY LTFU rate Crude HR (95% CI) p-value Adjusted HR (95% CI) p-value

Yes 267 819 32.62 0.98 (0.85 to 1.14) 0.83 Ever disclosed MSM status to healthcare worker No 516 1503 34.34 1.18 (1.02 to 1.37) 0.02 Yes 287 965 29.75 ref – Ever been afraid to access health services No 529 1589 33.28 1.06 (0.92 to 1.23) 0.40 Yes 279 880 31.69 ref – Friends with whom socialize are MSM and heterosexual No 63 162 38.86 ref – Yes 739 2304 32.08 0.83 (0.64 to 1.08) 0.17 Trusts the majority of other MSM they know No 344 1257 27.38 ref –– – Yes 458 1210 37.86 1.38 (1.20 to 1.58) <0.01 1.30 (1.13 to 1.51) <0.01 Chlamydia trachomatis Negative 667 2048 32.56 1.05 (0.87 to 1.27) 0.59 Positive 128 413 31.00 ref – Neisseria gonorrhoeae Negative 647 1849 34.98 1.45 (1.21 to 1.73) <0.01 1.47 (1.22 to 1.77) <0.01 Positive 148 612 24.19 ref –– –

Bolding indicates p < 0.05. CI, confidence interval; HR, hazard ratio; IAI, insertive anal intercourse; LTFU, loss-to-follow-up; MSM, men who have sex with men; PY, person- years; RAI, receptive anal intercourse. study of reproductive aged adults attending PEPFAR-sup- surrounding disclosure of same sex sexual practices by partici- ported pre-ART programmes in Nigeria [2]. Variable measures pants in our cohort [7,33-35]. Despite the goal of non-stigma- and a paucity of data on retention in sub-Saharan Africa com- tizing care delivery in our clinics, it is possible that plicate contextualization of our findings, but our study shows participants who had not previously been tested for HIV had clear room for improvement in retaining MSM and TGW in experiences that affirmed their anticipated stigma and HIV prevention and care. reduced likelihood of retention. Alternatively, non-stigma-re- Younger age did not predict retention, but there were some lated characteristics such as being young and having a lower downstream factors indicative of youth that were associated risk perception, which may have been confirmed with negative with poor retention. For example among our youngest partici- HIV/STI test results, could have impeded retention [5]. Fur- pants, 16 to 17 year-olds were significantly less likely to own ther qualitative evaluation of factors that influence retention a cell phone as compared to 18 to 19 year-olds in Abuja. could be valuable to tailor retention strategies. Owning a cell phone was associated with better retention in Prior studies have shown that men can be particularly diffi- our study, which could be explained by the fact that phones cult to retain in HIV care [36-39], and while most prior stud- were a primary mode of contact for reminders of upcoming ies focused on heterosexual and cisgender adults, the same and missed appointments. For participants without cell phones gender norms may play a role in retention for MSM and or working phone numbers, clinic staff relied on social net- TGW. A qualitative study in eastern Africa suggested that works to maintain contact. To reach young participants with- men perceived health clinics as women’s spaces and engaging out phones, a peer-mentoring programme could be employed in care was more of a concern and activity of women [40]. In to facilitate social network-based communication [25]. Social our study, living with HIV was independently associated with media may be another avenue, as even those without cell improved retention among cisgender men and participants phones may be able to access the internet and social media with other/unknown gender, but HIV status did not predict accounts through shared devices. retention among TGW. Additional research examining percep- Individuals who had not been tested for HIV prior to study tions of health clinics among TGW in sub-Saharan Africa could enrolment were at higher risk of LTFU and less likely to lend insight to this non-effect. adhere to the visit schedule. For many, prior avoidance of Healthcare engagement can also be facilitated by concern healthcare engagement for HIV testing may have been driven about specific signs, symptoms, or diagnoses rather than pre- by anticipated stigma [26-30]. Other qualitative studies have vention of ailments that are yet to occur. Our findings sug- shown that sexual and gender minorities avoid testing because gested higher retention of participants with a diagnosis of HIV a diagnosis of HIV or rectal STIs could lead to unintended dis- or STIs at enrolment, likely driven by their need for treatment. closure of anal sex practices [4,31,32]. We have previously When a person feels healthy or tests negative for HIV, active reported substantial anticipated and enacted stigma engagement in care tends to dissipate [41]. In prior studies,

46 Kayode BO et al. Journal of the International AIDS Society 2020, 23(S6):e25592 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25592/full | https://doi.org/10.1002/jia2.25592

Table 3. Visit adherence among participants enrolled in TRUST/RV368 by demographic, behavioural and clinical characteristics

Characteristic Visit adherence rate (95% CI) Crude RR (95% CI) p value Adjusted RR (95% CI) p value

Overall 0.62 (0.61 to 0.64) –– HIV status At risk 0.51 (0.49 to 0.54) 0.72 (0.68 to 0.75) <0.01 0.80 (0.75 to 0.85) <0.01 Living with HIV 0.72 (0.69 to 0.74) ref –– – Age (years) 16 to 19 0.55 (0.51 to 0.59) 0.82 (0.75 to 0.88) <0.01 0.95 (0.87 to 1.04) 0.28 20 to 24 0.60 (0.57 to 0.62) 0.89 (0.84 to 0.94) <0.01 0.92 (0.87 to 0.98) <0.01 25+ 0.67 (0.65 to 0.70) ref –– – Study site Abuja 0.59 (0.57 to 0.60) ref – Lagos 0.70 (0.67 to 0.73) 1.19 (1.13 to 1.25) <0.01 Education ≤High school 0.59 (0.57 to 0.61) 0.87 (0.83 to 0.92) <0.01 >High school 0.68 (0.65 to 0.70) ref – Employment status Unemployed 0.65 (0.62 to 0.78) ref – Employed/student 0.62 (0.60 to 0.64) 0.96 (0.91 to 1.02) 0.18 Sexual orientation Homosexual 0.65 (0.62 to 0.68) ref – Bisexual 0.61 (0.59 to 0.63) 0.94 (0.89 to 0.99) 0.02 Gender Cisgender man 0.61 (0.59 to 0.63) ref – Transgender woman 0.66 (0.62 to 0.71) 1.09 (1.00 to 1.17) 0.04 Other/unknown 0.66 (0.61 to 0.72) 1.09 (0.99 to 1.19) 0.06 Owns a cell phone No 0.42 (0.36 to 0.48) 0.66 (0.57 to 0.77) <0.01 0.81 (0.70 to 0.94) <0.01 Yes 0.63 (0.62 to 0.65) ref –– – Number of male sexual partners in the past year 0 to 4 0.63 (0.60 to 0.65) ref – 5 to 9 0.62 (0.59 to 0.75) 0.99 (0.94 to 1.06) 0.87 10+ 0.64 (0.60 to 0.67) 1.01 (0.95 to 1.08) 0.66 Receptive and/or insertive anal sexual practices in the past year RAI only 0.67 (0.64 to 0.71) ref –– – IAI only 0.58 (0.55 to 0.61) 0.86 (0.80 to 0.93) <0.01 0.89 (0.82 to 0.97) <0.01 IAI and RAI 0.63 (0.61 to 0.66) 0.95 (0.89 to 1.01) 0.08 0.92 (0.87 to 0.98) 0.02 Condom used at last anal sex with male partner No 0.61 (0.58 to 0.63) 0.96 (0.91 to 1.01) 0.10 Yes 0.63 (0.61 to 0.65) ref – Sold sex in the past year No 0.66 (0.64 to 0.69) ref –– – Yes 0.57 (0.55 to 0.60) 0.86 (0.82 to 0.91) <0.01 0.90 (0.85 to 0.95) <0.01 Bought sex in the past year No 0.63 (0.61 to 0.65) ref – Yes 0.61 (0.58 to 0.64) 0.97 (0.91 to 1.03) 0.27 Ever tested for HIV No 0.45 (0.42 to 0.48) 0.67 (0.63 to 0.73) <0.01 0.74 (0.69 to 0.80) <0.01 Yes 0.67 (0.65 to 0.68) ref –– – At least somewhat worried about HIV in the past year No 0.61 (0.59 to 0.63) 0.96 (0.91 to 1.01) 0.11 Yes 0.64 (0.62 to 0.67) ref – Ever been verbally harassed for being MSM No 0.62 (0.60 to 0.64) ref – (Continued)

47 Kayode BO et al. Journal of the International AIDS Society 2020, 23(S6):e25592 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25592/full | https://doi.org/10.1002/jia2.25592

Table 3. (Continued)

Characteristic Visit adherence rate (95% CI) Crude RR (95% CI) p value Adjusted RR (95% CI) p value

Yes 0.64 (0.61 to 0.67) 1.03 (0.98 to 1.09) 0.25 Ever disclosed MSM status to healthcare worker No 0.60 (0.58 to 0.62) 0.90 (0.85 to 0.97) <0.01 Yes 0.67 (0.64 to 0.69) ref – Ever been afraid to access health services No 0.60 (0.58 to 0.62) 0.92 (0.87 to 0.97) <0.01 Yes 0.66 (0.63 to 0.68) ref – Friends with whom socialize are MSM and heterosexual No 0.53 (0.47 to 0.59) ref –– – Yes 0.63 (0.61 to 0.65) 1.20 (1.07 to 1.34) <0.01 1.15 (1.02 to 1.28) 0.02 Trusts the majority of other MSM they know No 0.68 (0.65 to 0.70) ref –– – Yes 0.57 (0.55 to 0.60) 0.85 (0.81 to 0.89) <0.01 0.90 (0.85 to 0.94) <0.01 Chlamydia trachomatis Negative 0.62 (0.61 to 0.64) 0.97 (0.90 to 1.04) 0.34 Positive 0.64 (0.61 to 0.68) ref – Neisseria gonorrhoeae Negative 0.61 (0.59 to 0.62) 0.87 (0.82 to 0.92) <0.01 0.89 (0.83 to 0.94) <0.01 Positive 0.70 (0.66 to 0.74) ref –– –

Bolding indicates p < 0.05. CI, confidence interval; IAI, insertive anal intercourse; MSM, men who have sex with men; RAI, receptive anal intercourse; RR, risk ratio. higher attrition has been observed among PLWH with indica- HIV may have altered their engagement with the clinic over tors of asymptomatic disease, such as high CD4 counts [42] time, likely leading to overestimated retention for participants or no AIDS-defining diagnosis [43]. Education about preven- at risk for HIV. The unavailability of pre-exposure prophylaxis tive medical needs may be needed to shift perceptions on medication in Nigeria at the time of this study should also be accessing healthcare to improve retention for individuals at noted in context of our retention estimates. Enrolment charac- risk for HIV and other STIs. teristics were used in analyses rather than time-varying ones Transactional sex involves a culmination of factors that con- to allow for comparability with future prospective studies, tribute to a lack of engagement such as stigmas from commer- which must determine eligibility and anticipate retention cial sex work and same-sex sexual practices as well as power based on initial evaluations. However, retention may have imbalances and socio-economic vulnerability [44-46]. These been influenced by characteristics that changed over time. may co-occur with mental health challenges, low self-esteem Finally, this study originated in two cities in Nigeria and may and withdrawal [35,47]. Sexual and gender minorities with not be generalizable to other areas, or on a national scale. many of these vulnerabilities [48,49] may be less inclined to attend visits regularly. Our participants who engaged in transactional sex were also younger [45], reinforcing further disen- 5 | CONCLUSIONS gagement because adolescents have a lower perceived risk of HIV infection [45,50]. Incorporating peer-led social groups at Overall, retention in this study was suboptimal but within the the clinics to promote trust and agency may increase reten- range of other regional and continental estimates. Participants tion, similar to what has been done for female sex workers in living with HIV demonstrated better retention as compared to Zimbabwe [51]. those at risk for HIV, although HIV status did not impact This study has some limitations. First, understanding the retention specifically for TGW in this study. Marketing the interplay between HIV status, sexual behaviours, stigma and benefits of testing, improving access to cell phones and nur- retention required test results and completion of both enrol- turing trust with clients may further improve retention among ment questionnaires, which were unavailable for 491 and 323 marginalized Nigerian MSM and TGW. Clinic-level interven- participants, respectively. This may have resulted in an under- tions to improve retention must be accompanied by rights-af- estimate of overall LTFU, including a substantial minority of firming structural interventions to maximize benefits to key participants who were lost between visits 0 and 1 but populations in Nigeria. unevaluable because of incomplete data. Second, this study could not assess whether lost participants re-engaged in care AUTHORS’ AFFILIATIONS elsewhere [52]. If re-engagement occurred, then we overesti- 1Institute of Human Virology Nigeria, Abuja, Nigeria; 2Institute of Human Virol- mated the true loss and underestimated overall retention. ogy, University of Maryland School of Medicine, Baltimore, MD, USA; 3HJF Third, a proportion of men seroconverted and the diagnosis of Medical Research International, Abuja, Nigeria; 4U.S. Military HIV Research

48 Kayode BO et al. Journal of the International AIDS Society 2020, 23(S6):e25592 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25592/full | https://doi.org/10.1002/jia2.25592

Program, Walter Reed Army Institute of Research, Silver Spring, MD, USA; therapy for HIV patients at sites providing pre-ART care in Nigeria, 2004–2012. 5Maryland Global Initiatives Corporation- A University of Maryland Baltimore PLoS One. 2017;12:e0183823. Affiliate, Abuja, Nigeria; 6Henry M. Jackson Foundation for the Advancement of 3. Santos G-M, Makofane K, Arreola S, Do T, Ayala G. Reductions in access Military Medicine, Bethesda, MD, USA; 7Department of Epidemiology, Johns to HIV prevention and care services are associated with arrest and convictions Hopkins School of Public Health, Baltimore, MD, USA in a global survey of men who have sex with men. Sex Transm Infect. 2017;93:62–4. 4. Stannah J, Dale E, Elmes J, Staunton R, Beyrer C, Mitchell KM, et al. HIV COMPETING INTEREST testing and engagement with the HIV treatment cascade among men who have The authors have no conflicts of interest. sex with men in Africa: a systematic review and meta-analysis. Lancet HIV. 2019;6:e769–87. 5. Geng EH, Nash D, Kambugu A, Zhang Y, Braitstein P, Christopoulos KA, AUTHORS’ CONTRIBUTIONS et al. Retention in care among HIV-infected patients in resource-limited set- BK, AM and RN conceived the analysis for the manuscript. Data collection and tings: emerging insights and new directions. Curr HIV/AIDS Rep. 2010;7:234– management was facilitated by BK, AK, NN and HR. AM conducted the data 44. analysis with input from RN and TC. BK, AM and RN drafted the manuscript 6. Cornell M, McIntyre J, Myer L. Men and antiretroviral therapy in Africa: our and AK, NN, HR, SA, MR, SB, JA, MC and TC provided critical review and edit- blind spot. Trop Med Int Health. 2011;16:828–9. ing. All authors have seen and approved the paper. 7. Rodriguez-Hart C, Bradley C, German D,Musci R, Orazulike I, Baral S, et al. The synergistic impact of sexual stigma and psychosocial well-being on HIV testing: a mixed-methods study among Nigerian men who have sex with men. AIDS ACKNOWLEDGEMENTS Behav. 2018;22:3905–15. The authors thank the study participants and staff at the TRUST/RV368 clinics 8. Augustinavicius JL, Baral SD, Murray SM, Jackman K, Xue Q-L, Sanchez TH, for their contributions and commitment to this research. The research reported et al. Characterizing cross-culturally relevant metrics of stigma among men who in this publication was supported by funding from the U.S National Institutes of have sex with men across eight sub-Saharan African countries and the United Health [R01 MH099001, R01 AI120913, R01 MH110358, 1K07CA225403]; States. Am J Epidemiol. 2020;189(7):690–7. the Henry M. Jackson Foundation for the Advancement of Military Medicine, 9. Sharma M, Ying R, Tarr G, Barnabas R. Systematic review and meta-analysis Inc., and the U.S. Department of Defense [W81XWH-11-2-0174, W81XWH-18- of community and facility-based HIV testing to address linkage to care gaps in 2-0040]; Fogarty Epidemiology Research Training for Public Health Impact in sub-Saharan Africa. Nature. 2015;528:S77–85. Nigeria programme [D43TW010051]; and the President’s Emergency Plan for 10. Wirtz AL, Trapence G, Jumbe V, Umar E, Ketende S, Kamba D, et al. Feasi- AIDS Relief through a cooperative agreement between the Department of bility of a combination HIV prevention program for men who have sex with men Health and Human Services/Centers for Disease Control and Prevention, Global in Blantyre, Malawi. J Acquir Immune Defic Syndr. 2015;70:155–62. AIDS Program, and the Institute for Human Virology-Nigeria 11. Lyons CE, Ketende S, Diouf D, et al. Potential impact of integrated stigma [NU2GGH002099]. mitigation interventions in improving HIV/AIDS service delivery and uptake for The TRUST/RV368 Study Group is constituted as follows The TRUST/ key populations in Senegal. J Acquir Immune Defic Syndr. 2017;74 Suppl 1:S52– RV368 Study Group includes Principal Investigators: Manhattan E Charurat 9. (IHV, University of Maryland, Baltimore, MD, USA), Julie A Ake (MHRP, Walter 12. Rees K, Radebe O, Arendse C, et al. Utilization of sexually transmitted Reed Army Institute of Research, Silver Spring, MD, USA); Co-Investigators: Aka infection services at 2 health facilities targeting men who have sex with men in Abayomi, Sylvia Adebajo, Stefan D Baral, Trevor A Crowell, Charlotte Gaydos, South Africa: a retrospective analysis of operational data. Sex Transm Dis. Afoke Kokogho, Jennifer Malia, Olumide Makanjuola, Nelson Michael, Nicaise 2017;44:768–73. Ndembi, Rebecca G Nowak, Oluwasolape Olawore, Zahra Parker, Sheila Peel, 13. Trapence G, Collins C, Avrett S, Carr R, Sanchez H, Ayala G, et al. From Habib O Ramadhani, Merlin L Robb, Cristina Rodriguez-Hart, Eric Sanders-Buell, personal survival to public health: community leadership by men who have sex Elizabeth Shoyemi, Sodsai Tovanabutra, Sandhya Vasan; Institutions: Institute of with men in the response to HIV. Lancet. 2012;380:400–10. Human Virology at the University of Maryland School of Medicine (IHV-UMB), 14. Jones MU, Ramadhani HO, Adebajo S, Gaydos CA, Kokogho A, Baral SD, Johns Hopkins Bloomberg School of Public Health (JHSPH), Johns Hopkins Uni- et al. Seizing opportunities for intervention: changing HIV-related knowledge versity School of Medicine (JHUSOM), U.S. Military HIV Research Program among men who have sex with men and transgender women attending trusted (MHRP), Walter Reed Army Institute of Research (WRAIR), Henry M. Jackson community centers in Nigeria. PLoS One. 2020;15:e0229533. Foundation for the Advancement of Military Medicine (HJF), Henry M. Jackson 15. Crowell TA, Baral SD, Schwartz S, Nowak RG, Kokogho A, Adebajo S, et al. Foundation Medical Research International (HJFMRI), Institute of Human Virol- Time to change the paradigm: limited condom and lubricant use among Nigerian ogy Nigeria (IHVN), International Centre for Advocacy for the Right to Health men who have sex with men and transgender women despite availability and (ICARH), The Initiative for Equal Rights (TIERS), Population Council Nigeria. counseling. Ann Epidemiol. 2019;31:11–19.e3. 16. Keshinro B, Crowell TA, Nowak RG, Adebajo S, Peel S, Gaydos CA, et al. High prevalence of HIV, chlamydia and gonorrhoea among men who have sex DISCLAIMER with men and transgender women attending trusted community centres in The content is solely the responsibility of the authors and should not be con- Abuja and Lagos, Nigeria. J Int AIDS Soc. 2016;19:21270. strued to represent the positions of the National Institutes of Health, the U.S. 17. Crowell TA, Hardick J, Lombardi K, Parker Z, Kokogho A, Amusu S, et al. Army or the Department of Defense or the Department of Health and Human Asymptomatic lymphogranuloma venereum among Nigerian men who have sex Services. The investigators have adhered to the policies for protection of human with men. Sex Transmit Infect. 2018;94(8):578–81. subjects as prescribed in AR-70. 18. Crowell TA, Lawlor J, Lombardi K, Nowak RG, Hardick J, Odeyemi S, et al. Anorectal and urogenital mycoplasma genitalium in Nigerian men who have sex with men and transgender women. Sex Transmit Dis. 2020;47(3):202–6 DATA AVAILABILITY STATEMENT 19. Nowak RG, Ndembi N, Dauda W,Jibrin P, Bentzen SM, Nnaji CH, et al. Implementation of and early outcomes from anal cancer screening at a commu- The data that support the findings of this study are available upon request from nity-engaged health care facility providing care to Nigerian men who have sex the TRUST/RV368 Study Team. As a precautionary measure due to the criminal- with men. J Glob Oncol. 2019;5:1–11. ization of same-sex behaviour in Nigeria, the research data are kept confidential. 20. Nowak RG, Mitchell A, Crowell TA, Liu H, Ketende S, Ramadhani HO, et al. Individual and sexual network predictors of HIV incidence among men who have sex with men in Nigeria. J Acquir Immune Defic Syndr. 2019;80:444–53. 21. Ramadhani HO, Ndembi N, Nowak RG, Ononaku U, Gwamna J, Orazulike I, REFERENCES et al. Individual and network factors associated with HIV care continuum out- 1. Chi BH, Yiannoutsos CT, Westfall AO, Newman JE, Zhou Ji, Cesar C, et al. comes among Nigerian MSM accessing health care services. J Acquir Immune Universal definition of loss to follow-up in HIV treatment programs: a statistical Defic Syndr. 2018;79:e7–16. analysis of 111 facilities in Africa, Asia, and Latin America. PLoS Med. 2011;8: 22. Charurat ME, Emmanuel B, Akolo C, Keshinro B, Nowak RG, Kennedy S, e1001111. et al. Uptake of treatment as prevention for HIV and continuum of care among 2. Agolory SG, Auld AF, Odafe S, Shiraishi RW, Dokubo EK, Swaminathan M, HIV-positive men who have sex with men in Nigeria. J Acquir Immune Defic et al. High rates of loss to follow-up during the first year of pre-antiretroviral Syndr. 2015;68 Suppl 2:S114–23.

49 Kayode BO et al. Journal of the International AIDS Society 2020, 23(S6):e25592 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25592/full | https://doi.org/10.1002/jia2.25592

23. Baral SD, Ketende S, Schwartz S,Orazulike I, Ugoh K, Peel SA, et al. Evalu- 43. Israelski D, Gore-Felton C, Power R, Wood MJ, Koopman C. Sociodemo- ating respondent-driven sampling as an implementation tool for universal cover- graphic characteristics associated with medical appointment adherence among age of antiretroviral studies among men who have sex with men living with HIV. HIV-seropositive patients seeking treatment in a county outpatient facility. Prev J Acquir Immune Defic Syndr. 2015;68 Suppl 2:S107–13. Med. 2001;33:470–5. 24. National Agency for the Control of AIDS (NACA). 2015 Global AIDS 44. Tsang EY-H, Qiao S, Wilkinson JS, Lai-Chu Fung A, Lipeleke F, Li X, et al. response country report. Nigeria: Abuja; 2015. Multilayered stigma and vulnerabilities for HIV infection and transmission: a 25. Simoni JM, Nelson KM, Franks JC, Yard SS, Lehavot K. Are peer interven- qualitative study on male sex workers in Zimbabwe. Am J Mens Health. tions for HIV efficacious? A systematic review. AIDS and Behav. 2011;15:1589– 2019;13:1557988318823883. 95. 45. Crowell TA, Keshinro B, Baral SD, Schwartz SR, Stahlman S, Nowak RG, 26. Risher K, Adams D, Sithole B, Ketende S, Kennedy C, Mnisi Z, et al. Sexual et al. Stigma, access to healthcare, and HIV risks among men who sell sex to stigma and discrimination as barriers to seeking appropriate healthcare among men in Nigeria. J Int AIDS Soc. 2017;20:21489. men who have sex with men in Swaziland. J Int AIDS Soc. 2013;16:18715. 46. Baral SD, Friedman MR, Geibel S, Rebe K, Bozhinov B, Diouf D, et al. Male 27. Stahlman S, Bechtold K, Sweitzer S, Mothopeng T, Taruberekera N, Nko- sex workers: practices, contexts, and vulnerabilities for HIV acquisition and nyana J, et al. Sexual identity stigma and social support among men who have transmission. Lancet. 2015;385:260–73. sex with men in Lesotho: a qualitative analysis. Reprod Health Matters. 47. Li H, Tucker J, Holroyd E, Zhang J, Jiang B. Suicidal ideation, resilience, and 2015;23:127–35. healthcare implications for newly diagnosed HIV-positive men who have sex 28. Schwartz SR, Nowak RG, Orazulike I, Keshinro B, Ake J, Kennedy S, et al. with men in China: a qualitative study. Arch Sex Behav. 2017;46:1025–34. The immediate eff ECT of the same-sex marriage prohibition act on stigma, dis- 48. Oldenburg CE, Perez-Brumer AG, Reisner SL, Mimiaga MJ. Transac- crimination, and engagement on HIV prevention and treatment services in men tional sex and the hiv epidemic among men who have sex with men (MSM): who have sex with men in Nigeria: analysis of prospective data from the TRUST results from a systematic review and meta-analysis. AIDS Behav. cohort. Lancet HIV. 2015;2:e299–306. 2015;19:2177–83. 29. Kushwaha S, Lalani Y, Maina G, Ogunbajo A, Wilton L, Agyarko-Poku T, 49. Beyrer C, Crago A-L, Bekker L-G, Butler J, Shannon K, Kerrigan D, et al. An et al. ‘But the moment they find out that you are MSM..’: a qualitative investiga- action agenda for HIV and sex workers. Lancet. 2015;385:287–301. tion of HIV prevention experiences among men who have sex with men (MSM) 50. MacKellar DA, Valleroy LA, Secura GM, Behel S, Bingham T, Celentano DD, in Ghana’s health care system. BMC Public Health. 2017;17:770. et al. Perceptions of lifetime risk and actual risk for acquiring HIV among young 30. Lorenc T, Marrero-Guillamon I, Llewellyn A, Aggleton P, Cooper C, Leh- men who have sex with men. AIDS Behav. 2007;11:263–70. mann A, et al. HIV testing among men who have sex with men (MSM): system- 51. Cowan FM, Chabata ST, Musemburi S, Fearon E, Davey C, Ndori-Mharadze atic review of qualitative evidence. Health Educ Res. 2011;26:834–46. T, et al. Strengthening the scale-up and uptake of effective interventions for sex 31. Lane T, Mogale T, Struthers H, McIntyre J, Kegeles SM. ‘They see you as a workers for population impact in Zimbabwe. J Int AIDS Soc. 2019;22(S4): different thing’: the experiences of men who have sex with men with healthcare e25320. workers in South African township communities. Sex Transm Infect. 52. MacKellar DA, Williams D, Storer N, Okello V, Azih C, Drummond J, et al. 2008;84:430–3. Enrollment in HIV care two years after hiv diagnosis in the kingdom of swazi- 32. Sharma A, Bukusi E, Gorbach P, Cohen CR, Muga C, Kwena Z, et al. Sexual land: an evaluation of a national program of new linkage Procedures. PLoS One. identity and risk of HIV/STI among men who have sex with men in Nairobi. Sex 2016;11:e0150086. Transm Dis. 2008;35:352–4. 33. Kokogho A, Amusu S, Baral SD, Charurat ME, Adebajo S, Makanjuola O, et al. Disclosure of same-sex sexual practices to family and healthcare providers by men who have sex with men and transgender women in Nigeria. Arch Sex Behav. 2020. https://doi.org/10.1007/s10508-020-01644-8 SUPPORTING INFORMATION 34. Rodriguez-Hart C, Musci R, Nowak RG,German D, Orazulike I, Ononaku U, et al. Sexual stigma patterns among Nigerian men who have sex with men and Additional Supporting Information may be found in the online their link to HIV and sexually transmitted infection prevalence. AIDS Behav. 2018;22:1662–70. version of this article: 35. Rodriguez-Hart C, Nowak RG, Musci R, German D, Orazulike I, Kayode B, Table S1. Distribution of demographic, behavioral and clinical et al. Pathways from sexual stigma to incident HIV and sexually transmitted characteristics of young participants enrolled in TRUST/ infections among Nigerian MSM. AIDS. 2017;31:2415–20. RV368* by age group 36. Hawkins C, Chalamilla G, Okuma J, Spiegelman D, Hertzmark E, Aris E, Table S2. Loss-to-follow-up among participants enrolled in et al. Sex differences in antiretroviral treatment outcomes among HIV-infected adults in an urban Tanzanian setting. AIDS. 2011;25:1189–97. TRUST/RV368 by demographic, behavioral, and clinical charac- 37. Kipp W, Alibhai A, Saunders LD, Senthilselvan A, Kaler A, Konde-Lule J, teristics: cisgender men et al. Gender differences in antiretroviral treatment outcomes of HIV patients Table S3. Visit adherence among participants enrolled in in rural Uganda. AIDS Care. 2010;22:271–8. TRUST/RV368 by demographic, behavioral, and clinical charac- 38. Ochieng-Ooko V, Ochieng D, Sidle JE, Holdsworth M, Wools-Kaloustian K, Siika AM, et al. Influence of gender on loss to follow-up in a large HIV teristics: cisgender men treatment programme in western Kenya. Bull World Health Organ. 2010;88: Table S4. Loss-to-follow-up among participants enrolled in 681–8. TRUST/RV368 by demographic, behavioral, and clinical charac- 39. Sharma M, Barnabas RV, Celum C. Community-based strategies to teristics: transgender women ’ strengthen men s engagement in the HIV care cascade in sub-Saharan Africa. Table S5. Visit adherence among participants enrolled in PLoS Med. 2017;14:e1002262. 40. Camlin CS, Ssemmondo E, Chamie G, El Ayadi AM, Kwarisiima D, Sang N, TRUST/RV368 by demographic, behavioral, and clinical charac- et al. Men ‘missing’ from population-based HIV testing: insights from qualitative teristics: transgender women research. AIDS Care. 2016;28 Suppl 3:67–73. Table S6. Loss-to-follow-up among participants enrolled in 41. Horstmann E, Brown J, Islam F, Buck J, Agins BD, et al. Retaining HIV-in- TRUST/RV368 by demographic, behavioral, and clinical charac- fected patients in care: where are we? Where do we go from here? Clin Infect Dis. 2010;50:752–61. teristics: other or unknown gender 42. Arici C, Ripamonti D, Maggiolo F, Rizzi M, Finazzi M, Pezzotti P, et al. Fac- Table S7. Visit adherence among participants enrolled in tors associated with the failure of HIV-positive persons to return for scheduled TRUST/RV368 by demographic, behavioral, and clinical charac- medical visits. HIV Clin Trials. 2002;3:52–7. teristics: other or unknown gender

50 Kunzweiler C et al. Journal of the International AIDS Society 2020, 23(S6):e25598 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25598/full | https://doi.org/10.1002/jia2.25598

SHORT REPORT Enrolment characteristics associated with retention among HIV negative Kenyan gay, bisexual and other men who have sex with men enrolled in the Anza Mapema cohort study Colin Kunzweiler1, Robert C Bailey1,2, Duncan O Okall2, Susan M Graham3 , Supriya D Mehta1 , Boaz Otieno-Nyunya4, Gaston Djomand5 and Fredrick O Otieno2,§ §Corresponding author: Fredrick Odhiambo Otieno, Nyanza Reproductive Health Society, UNIM Research and Training Centre, Inside Lumumba Sub County Hospital, Ondiek Highway, P. O. Box 1764 Kisumu 40100, Kenya. Tel: +254 57 2024065. ([email protected])

Abstract Introduction: Most gay, bisexual and other men who have sex with men (GBMSM) live in rights-constrained environments making retaining them in research to be as hard as recruiting them. To evaluate Anza Mapema, an HIV risk-reduction programme in Kisumu, Kenya, we examined the enrolment sociodemographic, behavioural, psychosocial and clinical factors associated with missing two or more follow-up visits for GBMSM participating in Anza Mapema. Methods: Between August 2015 and November 2017, GBMSM were enrolled and followed in a prospective cohort study with quarterly visits over 12 months. At enrolment, men were tested for HIV and sexually transmitted infections and completed questionnaires via audio computer-assisted self-interview. Because the Kenya Ministry of Health recommends HIV testing every three to six months for GBMSM, the retention outcome in this cross sectional analysis was defined as missing two consecutive follow-up visits (vs. not missing two or more consecutive visits). Multivariable logistic regression estimated the adjusted odds ratios (aOR) and 95% confidence intervals (CI) for the associations of the enrolment characteristics with the binary outcome of retention. Results and discussion: Among 609 enrolled HIV-negative GBMSM, the median age was 23 years (interquartile range, 21 to 28 years), 19.0% had completed ≤8 years of education and 4.1% had resided in the study area <1 year at enrolment. After enrolment, 19.7% missed two consecutive follow-up visits. In the final multivariable model, the odds of missing two consecutive follow-up visits were higher for men who: resided in the study area <1 year at enrolment (aOR, 4.14; 95% CI: 1.77 to 9.68), were not living with a male sexual partner (aOR, 1.59; 95% CI: 1.01 to 2.50), and engaged in transactional sex during the last three months (aOR, 1.70; 95% CI: 1.08 to 2.67). Conclusions: One in five men missed two consecutive follow-up visits during this HIV prevention study despite intensive retention efforts and compensation for travel and participation. Participants with recent community arrival may require special support to optimize their retention in HIV prevention activities. Live-in partners of participants may be enlisted to support greater engagement in prevention programmes, and men who engage in transactional sex will need enhanced counselling and support to stay in longitudinal studies. Keywords: gay and bisexual men who have sex with men; GBMSM; care and treatment; cohort study; missed follow-up visits; retention; HIV; HIV negative; Kenya

Received 25 March 2020; Accepted 20 July 2020 Copyright © 2020 The Authors. Journal of the International AIDS Society published by John Wiley & Sons Ltd on behalf of the International AIDS Society. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

1 | INTRODUCTION comprehensive HIV prevention and treatment interventions tailored to the needs of GBMSM [4-7], and to identify the Studies conducted throughout sub-Saharan Africa demon- challenges and opportunities involved in engaging and retain- strate that gay, bisexual and other men who have sex with ing participants in studies and prevention programmes [8]. We men (GBMSM) have HIV prevalence rates two to four times conducted a longitudinal cohort study called Anza Mapema higher than the general male population [1] with substantial (Kiswahili for “Start Early”) whose purpose was to optimize stigma due to criminalization of their sexual practices [2,3]. To regular HIV testing, linkage to care and retention in HIV pre- address their increased risk -of HIV acquisition and transmis- vention and care among GBMSM in Kisumu, Kenya. In this sion, it is necessary to design, implement and test scalable analysis, we sought to identify enrolment factors associated

51 Kunzweiler C et al. Journal of the International AIDS Society 2020, 23(S6):e25598 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25598/full | https://doi.org/10.1002/jia2.25598 with non-retention of HIV-negative GBMSM in the Anza each quarterly study visit. For participants who relocated from Mapema study, which may translate to improved retention the study area within the country and wanted to continue par- practices. ticipation, transportation was arranged and paid for by the study. 2 | METHODS Personnel collected names, nicknames, physical address, primary telephone number, e-mail address and social media identity for each study participant, and a map was drawn of the 2.1 | Study population participant’s neighbourhood and directions to his home. The Recruitment into the Anza Mapema Cohort study occurred name and location of frequent hangouts were recorded, and from August 2015 to September 2016 using snowball contact information of family members and close friends were methodology [9]. Eligibility criteria were as follows: age recorded with the participant’s approval. ≥18 years, self-reported anal or oral intercourse with another Personnel obtained permission to contact participants via man in the previous six months, not participating in another multiple modalities including telephone calls, SMS text messag- HIV intervention or vaccine study, and residing in the study ing, WhatsApp, Facebook messaging and in-person contacts. area (within Kisumu County) [10]. Of the 711 men completing Visit reminders were sent at the beginning of each partici- enrolment procedures in Anza Mapema; 75 were HIV-positive pant’s visit window period, mid-way in the visit window period at baseline, 14 seroconverted, 4 died and 9 withdrew from and the day before the scheduled visit. Up to three reminders the study during follow-up, leaving 609 in this analysis. Men were sent at each time point via multiple modalities. The par- who withdrew from the study or died were excluded from this ticipant was called up to three times the day following a analysis because they may not have had sufficient time to missed visit and again during the first and second week fol- experience the primary outcome; HIV-positive GBMSM and lowing the missed visit. If necessary, study personnel initiated men who seroconverted were excluded from analysis because up to three attempts of physical tracing at the participant’s their follow-up was monthly. The Anza Mapema study was address. approved by the ethical review boards of Maseno University, the University of Illinois at Chicago and the University of 2.4 | Study retention definition Washington. A study visit was classified as “missed” if the participant did not attend his visit within 1.5 months before or 1.5 months 2.2 | Study procedures after his scheduled visit date. We considered men who had Following provision of written informed consent, all men pro- missed any two consecutive follow-up visits as not retained in vided detailed locator information, completed an audio com- the Anza Mapema study. This outcome aligns with the National puter-assisted self-interview (ACASI), underwent HIV AIDS and STI Control Programme of the Kenya Ministry of counselling and testing, completed a medical history and phys- Health, which recommends HIV testing among key populations ical examination and provided specimens for sexually transmit- every three to six months [11]. ted infections testing at enrolment. The same procedures were followed at quarterly follow-up visits for 12 months. Par- 2.5 | Predictor variables ticipants were referred for additional services including alcohol and substance abuse and psychological counselling by All sociodemographic, behavioural and psychosocial predictor study personnel as necessary or as requested. variables included in this analysis were collected via ACASI (Questionnaire Development Software version 3.0, NOVA Research Company, Silver Spring, MD, USA) at enrolment. 2.3 | Retention strategies ACASI questionnaires were available in DhoLuo, Kiswahili and As part of retention, personnel established and regularly com- English. All scales and the cut-offs applied for behavioural and municated with a community advisory board throughout psychosocial variables are presented elsewhere [10]. recruitment and follow-up. Personnel and peer outreach workers created and maintained a social media account (https:// 2.6 | Statistical analysis www.facebook.com/NRHSAnzaMapema/) promoting the delivery of HIV prevention and treatment services to GBMSM in The analysis is cross sectional with the outcome being two the study. They also hosted social activities, including movie consecutive missed visits. Differences between baseline nights, support groups and religious services, four to five days explanatory variables and outcome were assessed by chi- per week. Personnel conducted case reviews of specific partic- square test for categorical variables. Variables with p < 0.20 ipants with suboptimal visit attendance and developed strate- by likelihood ratio testing were entered in multivariable logis- gies for improving retention. tic regression, with backwards stepwise selection, retaining To minimize missed follow-up visits, participants received a variables with p < 0.05 by likelihood ratio test. reminder card with the date of their next visit at the end of each visit. The study offered flexible hours including early 3 | RESULTS AND DISCUSSION morning, evening and weekend appointments. A second clinic site was opened in an office building in the centre of Kisumu 3.1 | Study population City to enhance confidentiality for men who did not want to attend the main study site. Participants were compensated The baseline sociodemographic, behavioural and psychosocial 500 Kenyan shillings (US$5) for their time and travel costs at characteristics of the participants are shown in Table 1.

52 Kunzweiler C et al. Journal of the International AIDS Society 2020, 23(S6):e25598 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25598/full | https://doi.org/10.1002/jia2.25598

Table 1. Distribution of sociodemographic, behavioural and psychosocial characteristics collected at baseline among HIV-negative GBMSM in Kenya by retention status (N = 609)

Missed 2 consecutive visits

No Yes Total Variable N = 489 (80.3%) N = 120 (19.7%) N = 609 (100%) Sample n (%)a n (%)a n (%)b

Age, years (median [IQR]) 23 (21 to 28) 23 (21 to 26) 23 (21 to 28) Age, years 18 to 24 279 (78.8) 75 (21.2) 354 (58.1) ≥25 210 (82.3) 45 (17.7) 255 (41.9) Education, years 0 to 8 91 (78.5) 25 (21.5) 116 (19.1) 9 to 12 249 (79.3) 65 (20.7) 314 (51.6) ≥13 149 (83.2) 30 (16.8) 179 (29.4) Currently enrolled in school No 334 (79.3) 87 (20.7) 421 (69.1) Yes 155 (82.5) 33 (17.5) 188 (30.9) Employment status Less than full-time employed 224 (78.9) 60 (21.1) 284 (46.6) Full-time employed 265 (81.5) 60 (18.5) 325 (53.4) Uncertain/Very uncertain financial status No 132 (78.6) 36 (21.4) 168 (27.6) Yes 357 (81.0) 84 (19.0) 441 (72.4) Tribe/ethnicity Other 84 (76.4) 26 (23.6) 110 (18.1) Luo 405 (81.2) 94 (18.8) 499 (81.9) Resided in Kisumu for less than 1 year No 458 (81.1) 107 (18.9) 565 (92.8) Yes 13 (54.2) 11 (45.8) 24 (3.9) Missing 18 (90.0) 2 (10.0) 20 (3.3) Any religious affiliation No 35 (81.4) 8 (18.6) 43 (7.1) Yes 454 (80.2) 112 (19.8) 566 (92.9) Marital status Single 361 (79.9) 91 (20.1) 452 (74.2) Married or living with female partner 53 (81.5) 12 (18.5) 65 (10.7) Separated or divorced from female partner 75 (81.5) 17 (18.5) 92 (15.1) Gay or homosexual sexual identity No 149 (78.8) 40 (21.2) 189 (31.0) Yes 340 (80.9) 80 (19.1) 420 (69.0) Currently living with a male sexual partner No 306 (78.1) 86 (21.9) 392 (64.4) Yes 183 (84.3) 34 (15.7) 217 (35.6) Transactional sex (participant gave or received money, food, or housing) during the last three months No 188 (83.2) 38 (16.8) 226 (37.1) Yes 301 (78.6) 82 (21.4) 383 (62.9) Ever had sex with a female partner No 138 (82.1) 30 (17.9) 168 (27.6) Yes 351 (79.6) 90 (20.4) 441 (72.4) Always used condoms during AI with a male sexual partner (last three months) No 279 (79.5) 72 (20.5) 351 (57.6) Yes 198 (81.5) 45 (18.5) 243 (39.9) Did not have anal sex 12 (80.0) 3 (20.0) 15 (2.5)

53 Kunzweiler C et al. Journal of the International AIDS Society 2020, 23(S6):e25598 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25598/full | https://doi.org/10.1002/jia2.25598

Table 1. (Continued)

Missed 2 consecutive visits

No Yes Total Variable N = 489 (80.3%) N = 120 (19.7%) N = 609 (100%) Sample n (%)a n (%)a n (%)b

Usual sexual position during sex with a male partner Receptive or versatile 209 (81.0) 49 (19.0) 258 (42.4) Insertive 271 (79.5) 70 (20.5) 341 (56.0) Missing 9 (90.0) 1 (10.0) 10 (1.6) Experienced recent trauma due to same-sex behaviours (last two weeks) (USAID HPI) No 192 (80.7) 46 (19.3) 238 (39.1) Yes 250 (79.6) 64 (20.4) 314 (51.5) Missing 47 (82.5) 10 (17.5) 57 (9.4) Experienced sexual or physical abuse during childhood (CECA) No 104 (86.7) 16 (13.3) 120 (19.7) Yes 381 (78.7) 103 (21.3) 484 (79.5) Missing 4 (80.0) 1 (20.0) 5 (0.8) Harmful alcohol use (AUDIT ≥ 8) No 242 (79.6) 62 (20.4) 304 (49.9) Yes 247 (81.3) 57 (18.7) 304 (49.9) Missing 0 (0.0) 1 (100.0) 1 (0.2) Harmful substance use (DAST ≥ 3) No 375 (80.6) 90 (19.4) 465 (76.4) Yes 114 (79.2) 30 (20.8) 144 (23.6) Any injection drug use in last year No 458 (80.1) 114 (19.9) 572 (93.9) Yes 31 (83.8) 6 (16.2) 37 (6.1) Social support (MOS-SS; continuous range: 0 50 (34 to 66) 50 (36 to 61) 50 (36 to 64) to 100 scale; median [IQR]) Moderately severe or severe depressive symptoms (PHQ-9 ≥ 15) No 440 (80.7) 105 (19.3) 545 (89.5) Yes 49 (76.6) 15 (23.4) 64 (10.5) Circumcision status No 121 (83.5) 24 (16.5) 145 (23.8) Yes 368 (79.3) 96 (20.7) 464 (76.2) STI status Negative for CT and/or NG 420 (80.5) 102 (19.5) 522 (85.7) Positive for CT and/or NG 63 (78.7) 17 (21.3) 80 (13.1) Missing 6 (85.7) 1 (14.3) 7 (1.2)

AI, anal intercourse; AUDIT, Alcohol Use Disorders Identification Test; CECA, Childhood Experiences of Care and Abuse; DAST, Drug Abuse Screening Test; IQR, interquartile range; MOS-SS, Medical Outcomes Study-Social Support scale; PHQ-9, Personal Health Questionnaire-9; USAID HPI, United States Agency for International Development Health Policy Initiative. aRow percentages are presented. Row percentages may not equal 100.0% due to rounding; bColumn percentages are presented. Column percentages may not equal 100.0% due to rounding.

to 23.1%) men missed two consecutive follow-up visits 3.2 | Retention outcome (Table 1). There were missing results in some categories Of the 609 men included in the primary analysis, 20.0%, including; Resided in Kisumu for less than one year; Usual sex- 21.5%, 22.0% and 17.9% missed their months 3, 6, 9 and 12 ual position during sex with a male partner; Experienced follow-up visits respectively. Overall, 8.5% missed all four recent trauma due to same-sex behaviours; Experienced sex- study visits; 4.8% missed three study visits; 8.5% missed 2 visual or physical abuse during childhood; Harmful alcohol use its and 15.9% missed one visit. Regarding the retention out- and STI status. These missing results we generally less than come of interest, 19.7% (95% CI (confidence interval): 16.6% 5% of the reported results thus are still generalizable.

54 Kunzweiler C et al. Journal of the International AIDS Society 2020, 23(S6):e25598 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25598/full | https://doi.org/10.1002/jia2.25598

Table 2. Results of bivariate logistic regression: baseline characteristics of HIV-negative GBMSM in Kenya associated with missing two consecutive study visits

Variable OR (95% CI) p-valuea p-valueb

Age, years 18 to 24 1.25 (0.83 to 1.89) 0.28 0.28 ≥25 1.00 (ref) Education (years) 0 to 8 1.36 (0.76 to 2.46) 0.30 0.49 9 to 12 1.30 (0.80 to 2.09) 0.29 ≥13 1.00 (ref) Enrolled in school Yes 0.82 (0.52 to 1.27) 0.37 0.37 No 1.00 (ref) Employment status Less than full-time employed 1.18 (0.79 to 1.76) 0.41 0.41 Full-time employed 1.00 (ref) Uncertain/very uncertain financial status Yes 0.86 (0.56 to 1.34) 0.51 0.51 No 1.00 (ref) Tribe/ethnicity Other 1.33 (0.81 to 2.19) 0.25 0.25 Luo 1.00 (ref) Resided in Kisumu for <1 year Yes 3.62 (1.58 to 8.31) <0.01 <0.01 No 1.00 (ref) Any religious affiliation No 0.93 (0.42 to 2.05) 0.85 0.85 Yes 1.00 (ref) Marital status Married or living with female partner 0.90 (0.46 to 1.75) 0.75 0.90 Separated or divorced from female partner 0.90 (0.51 to 1.60) 0.72 Single 1.00 (ref) Gay or homosexual sexual identity No 1.14 (0.75 to 1.75) 0.54 0.54 Yes 1.00 (ref) Currently living with a male sexual partner No 1.51 (0.98 to 2.34) 0.06 0.06 Yes 1.00 (ref) Transactional sex (participant received money, food, or housing) during the last three months Yes 1.35 (0.88 to 2.06) 0.17 0.17 No 1.00 (ref) Ever had sex with a female partner Yes 1.18 (0.75 to 1.86) 0.48 0.48 No 1.00 (ref) Always uses condoms during AI with a male sexual partner (last three months) No 1.14 (0.75 to 1.72) 0.55 0.55 Yes 1.00 (ref) Usual sexual position during sex with a male partner Receptive or versatile 0.91 (0.60 to 1.36) 0.64 0.66 Insertive 1.00 (ref) Experienced recent trauma due to same-sex behaviours (last two weeks) (USAID HPI)? Yes 1.07 (0.70 to 1.63) 0.76 0.76 No 1.00 (ref)

55 Kunzweiler C et al. Journal of the International AIDS Society 2020, 23(S6):e25598 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25598/full | https://doi.org/10.1002/jia2.25598

Table 2. (Continued)

Variable OR (95% CI) p-valuea p-valueb

Experienced sexual or physical abuse during childhood (CECA) Yes 1.76 (0.99 to 3.11) 0.05 0.05 No 1.00 (ref) Harmful alcohol use (AUDIT ≥ 8) Yes 0.90 (0.60 to 1.35) 0.61 0.61 No 1.00 (ref) Harmful substance use (DAST ≥ 3) Yes 1.10 (0.69 to 1.74) 0.70 0.70 No 1.00 (ref) Any injection drug use in last year Yes 0.78 (0.32 to 1.91) 0.58 0.58 No 1.00 (ref) Social support (MOS-SS) 1.00 (0.99 to 1.01) 0.76 0.65c Moderately severe or severe depressive symptoms (PHQ-9 ≥ 15) Yes 1.28 (0.69 to 2.38) 0.43 0.43 No 1.00 (ref) Circumcision status No 0.76 (0.46 to 1.24) 0.28 0.27 Yes 1.00 (ref) STI status Positive for CT and/or NG 1.11 (0.62 to 1.98) 0.72 0.72 Negative for CT and/or NG 1.00 (ref)

AI, anal intercourse; AUDIT, Alcohol Use Disorders Identification Test; CECA, Childhood Experiences of Care and Abuse; CI, confidence interval; DAST, Drug Abuse Screening Test; MOS-SS, Medical Outcomes Study-Social Support scale; OR, odds ratio; PHQ-9, Personal Health Question- naire-9; USAID HPI, United States Agency for International Development Health Policy Initiative. ap-value is the result of the Wald chi square test from the bivariable logistic regression; bp-value is the result of the Pearson chi square test from contingency table analysis; cp-value is the result of the Wilcoxon rank sum test for continuous variables; The bold values are the values that were significant from the analysis.

In our study, residing in the study area for <1 year, which 3.3 | Crude and adjusted regression analyses was a baseline measure of length of residence, had the stron- In bivariate analyses (Table 2), missing two consecutive fol- gest association with missing two consecutive follow-up visits. low-up visits was associated (p ≤ 0.20) with shorter length of The immediate period following migration may be character- residence in the study area at enrolment (<1 year vs. ≥1 year: ized by instability and lack of social support systems, including OR, 3.62), not living with a male sexual partner (OR, 1.51), family members, friends, sexual partners and peer support transactional sex during the last three months (OR, 1.35) and groups [15,16]. The men might also be returning to their pre- history of physical or sexual abuse during childhood (OR, vious area of residence and receiving services there. HIV pre- 1.76). In the multivariable model, the odds of missing two con- vention programmes should consider identifying individuals secutive follow-up visits were increased for men who resided who are newcomers to the study area, assess family and social in the study area for <1 year (aOR, 4.14 [95% CI 1.77 to support and link them to local prevention and support ser- 9.68] p < 0.01), who were not living with a male sexual part- vices in a timely manner. A navigator system linking such men ner (aOR, 1.59 [95% CI 1.01 to 2.50] p 0.05) and who to a permanent resident peer may be helpful, as might be reported transactional sex during the last three months (aOR, proactively asking recent arrivals about upcoming travel plans 1.70 [95% CI 1.08 to 2.67] p- 0.02). to link him with prevention services outside of the study area. While the retention rates achieved in this study are not Similar to other studies [17,18] among GBMSM in suboptimal, they are somewhat better than the limited estimates Saharan Africa, men who reported transactional sex had 70% available from other studies of GBMSM in sub-Saharan Africa. increased odds of missing two consecutive visits. In Kenya, as In a study of 449 HIV-negative GBMSM followed in coastal in many other sub-Saharan African countries, male same-sex Kenya, 25.7% did not report to the clinic within six months of behaviours are criminalized and highly stigmatized [2,3]. In our the last planned study visit [12]. In a study of 441 HIV-nega- study, men who reported transactional sex more frequently tive GBMSM in Nigeria, just 48.5% of participants attended experienced verbal insults, physical abuse, sexual abuse and their 12-month visit [13], and among 327 HIV-negative verbal threats due to their perceived male-sex behaviours GBMSM recruited from Cape Town, Nairobi and Kilifi, attri- [19] compared to men who did not report transactional sex tion rates were 21.8 per 100 person-years [14]. (data not shown). Also, GBMSM who report transactional sex

56 Kunzweiler C et al. Journal of the International AIDS Society 2020, 23(S6):e25598 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25598/full | https://doi.org/10.1002/jia2.25598 may have low retention rates due to psychosocial comorbidi- emphasize available support services, promote social network- ties including harmful alcohol use and severe depressive ing and peer support and facilitate engagement in HIV pre- symptoms [19], both of which were more common among vention services if participants decide to travel or locate men who reported transactional sex (data not shown). Pro- elsewhere. grammes to address these needs are urgently needed. The odds of missing two consecutive follow-up visits were AUTHORS’ AFFILIATIONS increased for men who reported not living with a male sex 1Division of Epidemiology and Biostatistics, School of Public Health, University partner. Only a small portion of men reported openly dis- of Illinois at Chicago, Chicago, Illinois, USA; 2UNIM Research and Training Cen- cussing their male same-sex behaviours with family members tre, Nyanza Reproductive Health Society, Kisumu, Kenya; 3Departments of Med- icine, Global Health, and Epidemiology, University of Washington, Seattle, (9.3%) or friends (12.0%). Thus, support from male sex part- 4 ners may serve as an important buffer for GBMSM against Washington, USA; Division of Global HIV/AIDS, Centers for Disease Control and Prevention, Kisumu, Kenya; 5Division of Global HIV/AIDS and Tuberculosis, limited interpersonal support regarding same-sex behaviours. Centers for Disease Control and Prevention, Atlanta, Georgia, USA Support by male sex partners may be actively strengthened by programmes to help remind participants about appoint- COMPETING INTERESTS ments, provide financial assistance and facilitate greater All authors declare no conflict of interest in this manuscript and the work asso- engagement in HIV prevention programmes. ciated with it. Although our data were collected in the context of a research study between 2015 and 2017, the experiences of AUTHORS’ CONTRIBUTIONS the staff and participants in the Anza Mapema study may help future HIV prevention programmes anticipate chal- FO, RCB, SDM, SMG, BON and GD carried out conception or design of the work – DO, FO, SDM, CK, SMG, RCB involved in acquisition, analysis or inter- lenges regarding the retention of GBMSM in rights-con- pretation of data for the work. CK, RCB, SDM, FO, DO and SMG drafting the strained settings. In our study, staff implemented extensive work. CK, RCB, DO, FO, SDM, SMG BON and GD revising the work critically retention procedures, not least of which was ensuring a for important intellectual content. All Authors read and approved the manu- safe, GBMSM affirming environment with nearly daily group script. activities to encourage engagement. Also, participants were reimbursed for their travel to the study clinic and compen- ACKNOWLEDGEMENTS sated for their time at all scheduled follow-up visits. Retain- The authors thank the men who participated in the Anza Mapema study. We ing GBMSM in prevention programmes is especially also thank all research and staff members of the Anza Mapema study and the challenging, and retention may be even lower than observed Nyanza Reproductive Health Society, including Leah Osula, Beatrice Achieng, George N’gety, Caroline Oketch, Violet Apondi, Evans Kottonya, Caroline here if significant effort and resources are not allocated to Agwanda, Paula Abuor, Ted Aloo, George Oloo, Caroline Obare, Risper Oyah, support intensive retention strategies and reasonable partici- Haron Kadieda, Lilian Jumba, Violet Awuor, Eve Obondi, Lucy Atieno, Milcah Ari- pant compensation. ongo, Peter Oketch, William Oriedo, Francis Etiat, and Edmon Obat. We thank The participants in this study may not be representative of Dr. Ross Slotten and the Slotten Scholarship in Global Health at the University of Illinois at Chicago for supporting this work and Eduard Sanders for his GBMSM in Kisumu or Kenya since we used non-probability encouragement. sampling techniques to recruit participants. We collected data via ACASI, which has been shown to reduce response bias and interviewer bias [20,21]. However, misreporting is still FUNDING possible. The psychosocial scales we used have not been vali- This research has been supported by the U.S. President’s Emergency Plan for dated specifically among Kenyan GBMSM. However, many of AIDS Relief (PEPFAR) through the US Centers for Disease Control and Preven- the scales demonstrated acceptable internal reliability (Cron- tion (CDC) (grant number U01GH000762) and by Evidence for HIV Prevention in Southern Africa (grant number MM/EHPSA/NRHS/0515008). bach’s alpha ≥ 0.70) in the study [10]. We also did not collect information on procedures used to increase retention (number DISCLAIMER of reminders, supporting transportation fees for relocated individuals, physical tracing) and could not assess their associa- The findings and conclusions in this paper are those of the authors and do not necessarily represent the official position of the Centers for Disease Control tion with retention. and Prevention or other funding agencies.

4 | CONCLUSIONS REFERENCES The Anza Mapema study implemented comprehensive reten- 1. Beyrer C, Baral SD, van Griensven F, Goodreau SM, Chariyalertsak S, Wirtz AL, et al. Global epidemiology of HIV infection in men who have sex with men. tion strategies that incorporated community engagement and Lancet. 2012;380(9839):367–77. staff capacity building, that created a GBMSM welcoming 2. Kenya National Commission on Human Rights (KNCHR). The outlawed environment, and that incorporated participant tracing via amongst us: a study of the LGBT community’s search for equality and non-discrim- multiple modalities. Despite these efforts, 19.7% of men ination in Kenya. Nairobi, Kenya: Kenya National Commission on Human Rights; 2011 [cited 2020 Mar 23]. Available from: http://www.khrc.or.ke/mobile-publi missed two consecutive follow-up visits, and the proportion of cations/equality-and-antidiscrimination/70-the-outlawed-amongst-us/file.html men missing any scheduled visit during the 12-month study 3. Human Rights Watch. The issue is violence: attacks on LGBT people on Ken- ranged from 17.9% to 22.0%. Support and involvement of ya’s Coast. United States: Human Rights Watch; 2015 [cited 2020 Mar 23]. male sexual partners with whom participants live may be a Available from: http://www.hrw.org/report/2015/09/28/issue-violence/attacks- means to improve retention in HIV prevention programmes. lgbt-people-kenyas-coast 4. Beyrer C, Sullivan PS, Sanchez J, Dowdy D, Altman D, Trapence G, et al. A For participants who have resided in an area for <1 year, call to action for comprehensive HIV services for men who have sex with men. retention may be improved if clinicians and counsellors Lancet. 2012;380(9839):424–38.

57 Kunzweiler C et al. Journal of the International AIDS Society 2020, 23(S6):e25598 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25598/full | https://doi.org/10.1002/jia2.25598

5. Beyrer C, Baral SD, Collins C, Richardson ET, Sullivan PS, Sanchez J, et al. 13. Nowak RG, Mitchell A, Crowell TA, Liu H, Ketende S, Ramadhani HO, The global response to HIV in men who have sex with men. Lancet. 2016;388 et al. Individual and sexual network predictors of HIV incidence among men (10040):198–206 who have sex with men in Nigeria. J Acquir Immune Defic Syndr. 2019;80(4): 6. Joint United Nations Programme on HIV and AIDS. UNAIDS Action Frame- 444–53. work: Universal Access for Men who have Sex with Men and Transgender Peo- 14. Price M, Rida W, Mwangome M, Mutua G, Middelkoop K, Roux S, et al. ple. Geneva, Switzerland: World Health Organization; 2009 [cited 2020 Mar Identifying at-risk populations in Kenya and South Africa: HIV incidence in 23]. Available from: http://data.unaids.org/pub/report/2009/jc1720_action_fra cohorts of men who report sex with men, sex workers, and youth. J Acquir mework_msm_en.pdf Immune Defic Syndr. 2012;59(2):185–93. 7. President’s Emergency Plan for AIDS Relief-Scientific Advisory Board. PEP- 15. Salinero-Fort MA, del Otero-Sanz L, Martın-Madrazo C, de Burgos-Lunar C, FAR Scientific Advisory Board Recommendation for the Office of the U.S. Global Chico-Moraleja RM, Rodes-Soldevila B, et al. The relationship between social AIDS coordinator: intensify programmatic activity and implementation science support and self-reported health status in immigrants: an adjusted analysis in to reduce HIV burden, increase coverage and improve PEPFAR’s impact for key the Madrid Cross Sectional Study. BMC Fam Pract. 2011;12:46. populations. Washington, DC: U.S. Department of State-Office of the U.S. Global 16. Lu Y. Household migration, social support, and psychosocial health: The AIDS Coordinator and Health Diplomacy; 2011 [cited 2020 Mar 23]. Available perspective from migrant-sending areas. Soc Sci Med. 2012;74(2):135–42. from: https://www.pepfar.gov/documents/organization/188748.pdf 17. Masvawure TB, Sandfort TGM, Reddy V, Collier KL, Lane T. ‘They think 8. Muraguri N, Geibel S, Temmerman M. A decade of research involving men that gays have money’: gender identity and transactional sex among Black men who have sex with men in sub-Saharan Africa: Current knowledge and future who have sex with men in four South African townships. Cult Health Sex. directions. SAHARA-J. 2012;9(3):137–47. 2015;17(7):891–905. 9. Plenty A. Men’s health study: a cross sectional study of HIV among men who 18. Lane T, Raymond HF, Dladla S, Rasethe J, Struthers H, McFarland W, et al. have sex with men in Kisumu, Kenya. 2012 [cited 2020 Mar 23]. Available from: High HIV prevalence among men who have sex with men in Soweto, South http://indigo.uic.edu/bitstream/handle/10027/9323/Plenty_Albert.pdf?sequence=1 Africa: results from the Soweto men’s study. AIDS Behav. 2011;15(3):626–34 10. Kunzweiler CP, Bailey RC, Okall DO, Graham SM, Mehta SD, Otieno FO. 19. Kunzweiler CP, Bailey RC, Okall DO, Graham SM, Mehta SD, Factors Depressive symptoms, alcohol and drug use, and physical and sexual abuse Otieno FO. Associated with prevalent HIV infection among Kenyan MSM: the among men who have sex with men in Kisumu, Kenya: the Anza Mapema Study. Anza Mapema Study. J Acquir Immune Defic Syndr. 2017;76:241–9. Aids Behav. 2018;22(5):1517–29. 20. Metzger DS, Koblin B, Turner C, Navaline H, Valenti F, Holte S, et al. Ran- 11. National AIDS and STI Control Programme (NASCOP). The Kenya HIV domized controlled trial of audio computer-assisted self-interviewing: utility and Testing Services Guidelines. Nairobi, Kenya: Ministry of Health; 2015 [cited acceptability in longitudinal studies. HIVNET Vaccine Preparedness Study Proto- 2020 Mar 23]. Available from: https://archive.org/details/hts_policy_kenya_2015 col Team. Am J Epidemiol. 2000;152(2):99–106 12. Sanders EJ, Okuku HS, Smith AD, Mwangome M, Wahome E, Fegan G, 21. van der Elst EM, Okuku HS, Nakamya P, Muhaari A, Davies A, McClelland et al. High HIV-1 incidence, correlates of HIV-1 acquisition, and high viral loads RS, et al. Is audio computer-assisted self-interview (ACASI) useful in risk beha- following seroconversion among men who have sex with men in Coastal Kenya. viour assessment of female and male sex workers, Mombasa, Kenya? PLoS One. AIDS. 2013;27(3):437–46. 2009;4:e5340.

58 Sandfort TGM et al. Journal of the International AIDS Society 2020, 23(S6):e25600 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25600/full | https://doi.org/10.1002/jia2.25600

RESEARCH ARTICLE The feasibility of recruiting and retaining men who have sex with men and transgender women in a multinational prospective HIV prevention research cohort study in sub-Saharan Africa (HPTN 075) Theodorus GM Sandfort1,§ , Erica L Hamilton2 , Anita Marais3, Xu Guo4, Jeremy Sugarman5 , Ying Q Chen4, Vanessa Cummings6, Sufia Dadabhai7, Karen Dominguez8, Ravindre Panchia3, David Schnabel9, Fatima Zulu7, Doerieyah Reynolds8, Oscar Radebe10, Calvin Mbeda9 , Dunker Kamba11, Brian Kanyemba8, Arthur Ogendo9, Michael Stirratt12, Wairimu Chege13, Jonathan Lucas2, Maria Fawzy14, Laura A McKinstry4 and Susan H Eshleman6 §Corresponding author: Theo GM Sandfort, New York State Psychiatric Institute, 1051 Riverside Drive, Unit 15, New York, New York 10032 USA. Tel: +1 646 774 6946. ([email protected])

Abstract Introduction: Men who have sex with men (MSM) and transgender women (TGW) in sub-Saharan Africa (SSA) are profoundly affected by HIV with high HIV prevalence and incidence. This population also faces strong social stigma and legal barriers, potentially impeding participation in research. To date, few multi-country longitudinal HIV research studies with MSM/TGW have been conducted in SSA. Primary objective of the HIV Prevention Trials Network (HPTN) 075 study was to assess feasibility of recruiting and retaining a multinational prospective cohort of MSM/TGW in SSA for HIV prevention research. Methods: HPTN 075, conducted from 2015 to 2017, was designed to enroll 400 MSM/TGW at four sites in SSA (100 per site: Kisumu, Kenya; Blantyre, Malawi; Cape Town, South Africa; and Soweto, South Africa). The number of HIV-positive persons was capped at 20 per site; HIV-positive persons already in care were excluded from participation. The one-year study included five biobehavioural assessments. Community-based input and risk mitigation protocols were included in study design and conduct. Results: Of 624 persons screened, 401 were enrolled. One in five participants was classified as transgender. Main reasons for ineligibility included: (a) being HIV positive after the cap was reached (29.6%); (b) not reporting anal intercourse with a man in the preceding three months (20.6%); and (c) being HIV positive and already in care (17.5%). Five (1.2%) participants died during the study (unrelated to study participation). 92.9% of the eligible participants (368/396) completed the final study visit and 86.1% participated in all visits. The main, overlapping reasons for early termination included being (a) unable to adhere to the visit schedule, predominantly because of relocation (46.4%), and (b) unable to contact the participant (32.1%). Participants reported strong motivation to participate and few participation barriers. Four participants reported social harms (loss of confidentiality and sexual harassment by study staff) that were successfully addressed. Conclusions: HPTN 075 successfully enrolled a multinational sample of MSM/TGW in SSA in a prospective HIV prevention research study with a high retention rate and few documented social harms. This supports the feasibility of conducting large- scale research trials in this population to address its urgent, unmet HIV prevention needs. Keywords: HIV; closed cohort study; men who have sex with men; transgender women; sub-Saharan Africa

Additional Supporting information may be found online in the Supporting Information tab for this article.

Received 17 January 2020; Accepted 21 July 2020 Copyright © 2020 The Authors. Journal of the International AIDS Society published by John Wiley & Sons Ltd on behalf of the International AIDS Society. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

1 | INTRODUCTION and transgender women (TGW) and their role in the epidemic. Earlier HIV research and public health efforts in SAA have In sub-Saharan Africa (SSA), there is increasing recognition of focused on heterosexual transmission, since that is the main the HIV burden among men who have sex with men (MSM) mode of HIV transmission this region [1,2]. However, multiple

59 Sandfort TGM et al. Journal of the International AIDS Society 2020, 23(S6):e25600 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25600/full | https://doi.org/10.1002/jia2.25600 epidemiologic studies now show that gay, bisexual and other multinational prospective cohort [25]. Although the primary MSM in SSA are profoundly affected by HIV. The first HIV focus of HPTN 075 was on MSM, TGW were not excluded prevalence study among MSM, conducted in 2004 in Senegal, because some TGW in SSA socialize and identify with MSM reported a prevalence of 21.5% [3]; pioneering work with or identify as “gay.” [7,26] This report describes the prepara- MSM has also been conducted in Kenya [4]. The observed tion of the HPTN 075 study sites for study implementation, HIV prevalence among MSM in SSA ranges from 4.1% to recruitment methods and results, retention of study partici- 49.5% [5-9]. A systematic review in 2012 estimated the over- pants and occurrence of social harms. all HIV prevalence among MSM in SSA to be 18% [10]. A more recent review [11] showed that HIV testing among 2 | METHODS MSM in SSA has significantly increased over time. However, HIV status awareness is still low, ranging from 6.7% in coun- 2.1 | Study design and population tries with the most severe legislation against the lesbian, gay, bisexual and transgender communities, to 22.0% in countries Study participation included biobehavioural assessments at with the least severe legislation. screening and at five subsequent study visits over one year. Prevention trials are urgently needed to evaluate HIV pre- Four sites participated: Kisumu, Kenya; Blantyre, Malawi; and vention interventions among MSM/TGW in SSA [12]. How- Cape Town and Soweto, South Africa. HIV-positive and HIV- ever, more information is needed about the feasibility of negative persons were eligible to enrol; the number of HIV- conducting such trials in this population. Since the early positive persons was capped at 20 per site. Although TGW 2000s, several studies have been conducted among MSM in could participate, there were no specific efforts to recruit SSA, demonstrating the feasibility of recruiting this population; them. The same considerations applied to male sex workers. however, most of these studies had a cross-sectional design Screening and enrolment started in June 2015 and ended in (e.g. [13,14]). The limited number of longitudinal cohort stud- July 2016. Data collection ended in July 2017. ies conducted generally included open cohorts at single sites or in individual countries (e.g. [15,16]). Less is known about 2.2 | Eligibility criteria the feasibility of retaining MSM in a multi-national prospective cohort over time (prior multi-country longitudinal studies with Main eligibility criteria included: (a) assigned male sex at birth; MSM in SSA were limited to six-month follow-up [17]). Infor- (b) 18 to 44 years old; (c) reporting at least one act of anal mation is also needed about the feasibility of achieving optimal intercourse in the previous three months with a person adherence to study visits and preventing social harms in this reported by the participant to be biologically male; (e) three population. concordant HIV test results at screening and (d) willing to Maintaining a cohort of MSM/TGW in SSA for HIV preven- undergo HIV testing throughout the study and receive test tion research could be challenging for several reasons, includ- results. An optimal evaluation of the study aim would require ing physical, social and legal risks that are likely to interfere a sample of persons who were na€ıve to HIV research. For that with study retention. Same-sex behaviour is criminalized in reason, persons who previously participated in a biomedical most SSA countries, with sentences ranging up to the death and/or behavioural intervention or cohort study for HIV or penalty [18]. Although enforcement of these laws varies by sexually transmitted infections (STIs) were excluded; co-enrol- country, participation in MSM/TGW research could imply dis- ment in such studies was not permitted. The study protocol closure of illegal behaviour and could thus have legal reper- was amended to ensure that participants would have access cussions. Also, compared to other parts of the world, to oral pre-exposure prophylaxis (PrEP) when it became acces- countries in SSA are among the least accepting of same-sex sible at some sites (e.g. through demonstration projects); sexuality [19]. Experiences with homophobia, including vio- although several PrEP referrals were made, no participants lence and blackmail, are well-documented in this population reported initiating PrEP. To evaluate uptake of care, HIV- [20,21]. Recruiting and retaining MSM/TGW, especially for positive participants who reported already being in HIV care studies in a medical context, also require gaining trust. Prior or on antiretroviral treatment (ART) were also excluded from discriminatory experiences in medical settings may lead MSM/ study participation. TGW to fear insensitive treatment by study staff and inappro- priate disclosure of sexual practices or HIV status [22,23]. 2.3 | Procedures There are also risks for research staff and study integrity. For example, being associated with a study of MSM/TGW Given the potential risks associated with the study, each study might be interpreted as condoning or promoting same-sex site was instructed to develop a site-specific risk-mitigation sexuality. There is also the possibility of physical attacks at plan (RMP; Appendix S1), guided by international guidelines study sites and negative media coverage of the study or the for HIV prevention trials [27,28], research with MSM in study population. In other studies of MSM in SSA, offices have rights-constrained environments [29] and ethical guidance been attacked by community members and staff arrested, with from the HPTN [30]. This approach was intended to help both the allegation that same-sex sexual activities were being pro- researchers and community organizations safely conduct moted and that young people were being recruited to become meaningful research in challenging social, political and human MSM [24]. These types of social harms have not been system- rights contexts; this included use of a checklist of factors to atically studied. be considered in the design, conduct, and implementation of In this context, the HIV Prevention Trials Network (HPTN) the study. initiated the HPTN 075 study, with the objective to assess the Preparation of RMPs included: (a) establishing ongoing feasibility of recruiting and retaining 400 MSM/TGW in a engagement with the MSM community and local MSM

60 Sandfort TGM et al. Journal of the International AIDS Society 2020, 23(S6):e25600 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25600/full | https://doi.org/10.1002/jia2.25600 organizations; (b) building rapport and support with the gen- Facebook or appointment cards); arrangement of transporta- eral community, including health authorities, media, religious tion to the study site or reimbursement of transportation leaders and local police; (c) creating a Community Advisory costs; home visits based on regularly updated locator informa- Board (CAB); (d) forming a Protocol Advisory Committee tion; a welcoming study site environment (courteous treat- (PAC) for active oversight of study implementation that ment by staff; addressing participants by preferred names and included members of the MSM community and persons with pronouns; food and refreshments; magazines, video and Inter- direct expertise on MSM issues; (e) developing procedures for net access in waiting rooms); free medical services and contin- ensuring study participants’ confidentiality, that included a ued outreach and support through community events, such as requirement for staff to sign confidentiality agreements and educational events, beach days, weekend camps and pageants. interact with participants in a non-judgemental, MSM-affirming Some study participants were not interested in these events way; (f) sensitizing relevant stakeholders, including study staff, due to risk of disclosure. and informing them that disclosing any information about par- Behavioural assessments included collection of demographic, ticipants was subject to disciplinary measures, up to termina- behavioural, psychosexual and socioeconomic data, and inter- tion; (g) developing procedures for responding to problems est in potential HIV prevention strategies. Evaluation of study and establishing an emergency committee to facilitate a direct participation included barriers and facilitators to participation, response to any urgent situations, including a communication study burden and social harms and benefits from participation. plan; and (h) systematically assessing possible social harms at As much as possible, assessment tools were adopted that study visits, developing a priori responses for addressing such were successfully used in this population in SSA. Other mea- harms and training the study staff on the collection of social sures were adapted from existing assessments. harm data and reporting. The preparation of these RMPs likely After study completion, information was collected from vari- facilitated the process of obtaining in-country ethics approval ous stakeholders, including research staff, via questionnaires for the study; the study’s focus did not cause any difficulties. and in-person interviews to characterize the process at each HPTN 075 aimed to recruit a diverse sample of MSM at site for building stakeholder support and determining ideal high risk of acquiring HIV infection. In consultation with the recruitment strategies. In these evaluations, the following community, each site developed site-specific strategies to pro- topics were addressed: MSM community involvement and mote study awareness and acceptability and recruitment. This impact; recruitment and retention of participants; ongoing approach (a) allowed for optimal use of the local community’s community engagement; study site preparation and implemen- expertise and customization to local circumstances; and (b) tation; incentives and services; CABs/PACs; emergency com- made it easier to adjust strategies if recruitment outcomes mittee and future research needs. lagged at specific sites. Various recruitment strategies were implemented: (a) peer outreach: MSM, hired and trained as 2.4 | Data analysis peer-outreach workers (from one to eight per site) who approached potential study participants based on their per- Descriptive statistics were used to characterize study recruit- sonal knowledge of and connections to the MSM population; ment, enrolment and retention, as well as participant demo- (b) participant referral: participants were asked to refer graphics, motivation to take part in the study and adverse friends for participation in the study (not incentivized); (c) incidents. Univariate and multivariable logistic regression informational sessions about the study; (d) key informant was used to compare characteristics of participants who com- referral: trusted persons with access to MSM networks dis- pleted all study visits with those who did not complete the tributed study information and encouraged MSM to partici- study or missed one of more visits. A stepwise model was pate; and (e) indirect recruitment: distribution of used for multivariable analysis; the significance level for entry announcements via in person and web-based “gay” venues and and exit of variables in the model was set at 0.3 and 0.35 events. respectively. Screening for HPTN 075 included administrative procedures, collection of biological samples and HIV and STI testing. 2.5 | Ethics statement Eligible persons who consented to participate subsequently had an enrolment visit and follow-up visits at weeks 13, 26, Study sites received approval from their respective institu- 39 and 52. All study visits included structured behavioural tional review boards (IRBs) and the Division of AIDS, National assessments, HIV risk reduction counselling, assessment of Institute of Allergy and Infectious Diseases. Informed consent social impacts, collection of biological samples, HIV testing (if was obtained separately for screening and enrolment. Partici- HIV negative at the prior visit) and medical examinations. STI pants provided written consent at three sites and oral consent treatment was provided; some participants, if so desired, were at one site, as directed by the local IRB, because signing a referred to a clinic of their choice; one site offered treatment consent form could lead to unintended disclosure. to participants’ sexual partners. Condoms and lubricants were available at each visit. ART adherence assessments and counselling were provided as appropriate. Research participation 3 | RESULTS was incentivized according to local standards (ranging from $4 to $10 US). Participation of employed participants was facili- A summary of study recruitment and participation outcomes tated by offering flexible appointment times, including in the is presented in Figure 1. In total, 624 persons were screened; weekend. 223 were ineligible. The main reasons for ineligibility included: Sites implemented a variety of retention strategies. These (a) being HIV positive after the cap was reached (n = 66, included visit reminders (via telephone, text messages, email, 29.6%); (b) not reporting anal intercourse with a man in the

61 Sandfort TGM et al. Journal of the International AIDS Society 2020, 23(S6):e25600 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25600/full | https://doi.org/10.1002/jia2.25600 preceding three months (n = 46, 20.6%); (c) being HIV posi- care, resulting in many persons being excluded (eligibility crite- tive and already in care (n = 39, 17.5%); (d) past or current ria were not communicated during recruitment or screening; participation in an HIV study (n = 31, 13.9%) and (e) not one site made clear that they were looking for persons who returning for enrolment within 30 days of the screening visit were not living with HIV after a substantial number of per- (n = 29, 13.0%). The time needed to recruit 100 participants sons had to be rejected because they were living with HIV). varied by site from 18.7 to 39.1 weeks (average 31.1 weeks). Co-enrolment in another HIV-study was observed at one The average number of participants recruited per week varied study site where a PrEP demonstration study had started. An by site from 2.6 to 5.3 (overall average 3.5 per week). attempt to prevent co-enrolment by jointly introducing bio- In reviewing the recruitment process, research staff noted metrics was unsuccessful, because of challenges with obtaining the importance of collaborating with local MSM communities approval from the respective authorities. and described most of the recruitment strategies as useful and successful. Peer referral was less successful at one site 3.1 | Cohort description, motivation to participate due to stigma and fear of disclosure. Participation incentives and perceived participation barriers attracted persons at some sites who were not MSM, reinforcing the need for rigorous screening. All sites noted the impor- Table 1 presents a description of the 401 men enrolled in the tance of having multiple recruitment strategies. One site study cohort. The average age was 24.2 years (range 18 to observed that specific efforts could have led to a better rep- 44 years). Seventy-one (17.8%) participants tested positive for resentation of older and more wealthy participants. A common HIV infection at enrolment (one participant had inconclusive challenge was the stringent exclusion criterion of being in HIV HIV test results). Most participants (62.4%) identified as gay;

Screened, N = 624 Ineligible: n = 223

Visit 1/Enrollment Eligible and enrolled: n = 401

Did not return: n = 8

Deceased: n = 3

Visit 2/Week 13 Participated: Missed: n = 381 n = 7 + 2†

Did not return: n = 3

Deceased: n = 2

Visit 3/Week 26 Participated: Missed: n = 370 n = 14 + 1†

Did not return: n = 5

Visit 4/Week 39 Participated: Missed: n = 368 n = 12

Did not return: n = 12

Visit 5/Week 52 Participated: n = 368

Figure 1. Overview of screening and study participation in HPTN 075. †Visits of men who returned for at least one follow-up visit but did not complete Visit 5.

62 Sandfort TGM et al. Journal of the International AIDS Society 2020, 23(S6):e25600 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25600/full | https://doi.org/10.1002/jia2.25600 one in five (20.0%) identified as female or transgender (in line two murders) were determined by local research staff, after with recommended procedures [31,32], these persons were extensive investigation, to be unrelated to study participation. categorized as TGW). About two-thirds of the participants Of the remaining 396 participants, 368 (92.9%) completed the (65.6%) had at least completed secondary education. Week 52 Visit and 317 (86.1%) completed all visits; 28 (7.1%) At screening, most participants expressed a strong motiva- participants did not return after either the Enrolment Visit or tion to participate in the study. Only two participants any subsequent visit. The main, overlapping reasons for early responded “no” to the question whether, if enrolled, they study termination included: (a) unable to adhere to the visit intended to participate in all scheduled assessments. All partici- schedule, predominantly because of relocation (46.4%); (b) pants indicated that it was likely or very likely that they would unable to contact the participant (32.1%); (c) refusal to partici- be able to remain in the study for at least one year. Most par- pate further (17.9%) and (d) incarceration (3.6%). The propor- ticipants described themselves as very (81.1%) or moderately tion of early terminations differed by site, ranging from 0% (17.0%) committed to this study. In response to the question (Soweto) to 14.0% (Blantyre). A comparison between partici- how important or unimportant participants considered the pants who terminated early and those who participated in all study to be for their community, the majority said “very impor- visits, showed that participants in Blantyre and participants tant” (88.8%; 9.0% said “moderately important” and 2% “slightly with children were more likely to terminate early compared to important”). In response to an open question about the single participants in Kisumu and those without children respectively most important reason to participate, one third (33.5%) (Table 3). Study site (Blantyre, compared to Kisumu) was the reported their interest in receiving HIV counselling and testing, only factor that remained significant in multivariable analysis. and knowing their status. Participants’ answers frequently Early termination was not associated with any of the percep- included more than one reason (Table 2). Most participants tions of the study, including perceived barriers to participa- (96.5%) felt it was easy or very easy to set up study appoint- tion. ments, make time to come to study visits (91.8%), and travel to Thirty participants missed a total of 36 visits (including the clinic for study visits (89.3%). The distance that participants three visits by participants who terminated early). The number had to travel ranged from less than a mile up to 20 miles, with of participants who missed visits varied from two to 18 per travel times ranging from a few minutes to about 60 minutes. site. Reasons for missed visits, based on the total number of visits, included (a) unable to schedule a visit, including because of temporary relocation (55.6 %); (b) unable to contact partici- 3.2 | Retention pant (25.0%); (c) refused visit (5.6%); (d) incarcerated (2.8%); Five participants (1.2%) died during the study; the causes of (e) other reasons (11.1%). Compared to participants who death (one sports injury, one case of malaria, one suicide and completed all visits, participants who missed any visit

Table 1. Characteristics of the study cohort (N = 401)

Kisumu, Kenya Blantyre, Malawi Cape Town, South Africa Soweto, South Africa (N = 100) (N = 100) (N = 100) (N = 101) M (median) / n/Na (%) M (median) / n/Na (%) M (median) / n/Na (%) M (median) / n/Na (%)

Age, in years 25.1 (23) 25.2 (24) 23.5 (22) 23.2 (22) Education Grade 11 or lower 33/100 (33.0) 45/96 (46.9) 37/100 (37.0) 27/99 (27.3) Completed Grade 12 38/100 (38.0) 36/96 (37.5) 39/100 (39.0) 56/99 (56.6) Completed college 29/100 (29.0) 15/96 (15.6) 24/100 (24.0) 16/99 (16.2) Married/legal partnership 7/100 (7.0) 10/99 (10.1) 8/100 (8.0) 3/100 (3.0) Any child 28/100 (28.0) 29/99 (29.3) 20/100 (20.0) 10/100 (10.0) Transgenderb 20/99 (20.2) 27/97 (27.8) 18/100 (18.0) 14/99 (14.1) Sexual attraction Men and women 81/100 (81.0) 71/98 (75.4) 36/99 (36.4) 22/100 (22.0) Men only 19/100 (19.0) 27/98 (27.6) 63/99 (63.6) 78/100 (78.0) Sexual identity Bisexual and other 52/100 (52.0) 50/99 (50.5) 27/100 (27.0) 21/100 (21.0) Gay 48/100 (48.0) 49/99 (49.5) 73/100 (73.0) 79/100 (79.0) Ever sex with women 77/100 (77.0) 67/99 (67.7) 50/100 (55.0) 30/100 (30.0) In ongoing same-sex, Intimate relationship 87/98 (88.8) 90/99 (90.9) 60/95 (63.2) 78/100 (78.0) HIV positive at screeningc 15/100 (15.0) 16/99 (16.2) 20/100 (20.0) 20/101 (19.8)

M, mean; n, number with characteristic; N, total number. aDue to missing values, some n’s do not add up to sample totals; bpersons who identified their gender as female or transgender; cthe HIV status of one participant could not be determined.

63 Sandfort TGM et al. Journal of the International AIDS Society 2020, 23(S6):e25600 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25600/full | https://doi.org/10.1002/jia2.25600

Table 2. Most important reason for participating in HPTN 075 3.3 | Risk mitigation and social harms (N = 391)a Four social harms were reported. Two study participants Reason % Example reported indecent treatment by a male study nurse (inappro- priate touching and sexual propositioning). Research staff Receiving HIV counselling and 33.5 “I needed to know about my quickly contacted these participants to address the issue and testing; knowing one’s status” apologize; staff followed up to help ensure that the partici- status pants regained a sense of safety in the study. Site staff were Receiving HIV risk reduction 30.9 “I will learn how to keep retrained in appropriate behaviour with participants. After a education myself from HIV and get thorough investigation, the nurse involved resigned. Explo- ration with the site’s outreach workers indicated no negative the protective measures repercussions in the community. In a different incident, one and information” participant left his job because his employer did not allow him “ Knowing more about MSM as 20.5 To get more information to attend study visits. Finally, one participant reported loss of a community; meeting new about MSM and my lifestyle confidentiality related to being gay; a co-worker found his people, gaining support and challenges that we face informed consent form and told colleagues, which was fol- from other MSM or being as gay people” lowed by homophobic comments from his manager. This man empowered as MSM considered quitting his job in response. This event resulted in Learning more about one’s 16.4 “To pass through the medical instructing study staff at all sites to be clear to participants own health: getting tested tests that would allow me about the risks involved in having a signed consent form and ’ for other things than HIV, know my health” more clearly offering the option not to take one s copy. getting free check-ups and After data collection was completed, discussions among research staff indicated that the development of the RMPs receiving treatment sensitized the sites and prepared research staff to deal with a Improving one’s general 14.8 “To know more about HIV range of problems that might occur. One site commented that ” knowledge of health, and my health preparing the RMP had helped them to focus on dealing with beyond HIV and STIs site emergencies more generally. Learning about MSM research 10.5 “Because the study involves or contributing to MSM MSM and I am one of them research I think I should participate” 4 | DISCUSSION

MSM, men who have sex with men; STIs, sexually transmitted infections. The HPTN 075 study successfully enrolled a large multinational a ’ Based on answers to an open question. Some of the participants sample of MSM and TGW in SSA in a prospective HIV preven- answers to the open question included more than one reason. tion research study with high rates of retention and few documented social harms. This indicates that longitudinal research (but completed the Week 52 Visit) were more likely to be with MSM and TGW in SSA is feasible and can be safely con- younger, to live in Cape Town, to be exclusively attracted to ducted when there is close attention to community engagement men, to identify as gay and to conceal their sexuality (Table 4). and risk mitigation procedures. These results support future Sexual identity was the only factor that remained a significant efforts to conduct large-scale HIV prevention research studies predictor of having missed any visits in multivariable analysis. and trials with MSM and TGW in SSA to address the urgent Missed visits were not associated with any perceptions of the and unmet HIV prevention needs in this group. study, including perceived barriers to study participation. HPTN 075 represents one of the largest, longest, prospec- Research staff described the implemented retention activities tive, multi-country closed-cohort research study with MSM as effective and necessary. Continuing community involvement and TGW in SSA to date. The study followed 401 men for helped to promote trust in the study. Interaction with partici- 12 months, with one screening visit and five study visits. Most pants and recruiters further offered the opportunity to obtain prior longitudinal studies of MSM cohorts in SSA were con- feedback about the study, and to address concerns and miscon- ducted at a single site or in a single country, enrolled open ceptions (e.g. the misconception that blood draws were used cohorts, or were associated with ongoing clinical care rather for commercial purposes). Intense mobility in the study popula- than research [33,34]. tion made it hard to reach some participants, despite frequently The legal status and social marginalization experienced by updating of locator information. In addition, some participants MSM/TGW in SSA has prompted questions about the feasibil- did not have phones, and some lost their phones during the ity of engaging them in research. The HPTN 075 study had study. A few participants did not have a street address; this high rates of participant accrual across four sites through a required creation of maps to collect locator information, making mix of direct and indirect recruitment approaches, including it harder to locate these participants. Staff reported that some peer outreach, participant and key informant referral, and participants provided incorrect locator information because venue-based recruitment combined with findings from prior they had not yet disclosed their sexual orientation to their cross-sectional studies of MSM in SSA [35]. This indicates that family. Others had concerns that coming to the study site or it is possible to address recruitment and enrolment challenges being seen with other participants might disclose their sexual in this population. orientation. School and work obligations made it hard for some The HPTN 075 study had high rates of participant reten- participants to meet all appointment times. tion over one year at four sites in three countries, averaging

64 adotTMe al. et TGM Sandfort Table 3. Factors associated with loss to follow-up during study, HPTN 075 study, Kenya, Malawi, South Africaa http://onlinelibrary.wiley.com/doi/10.1002/jia2.25600/full

Univariate Multivariableb Mean (SD) / n/N (%) Completed all visits (N = 341) Did not complete the study (N = 28) OR 95% CI p value AOR 95% CI p value

Countryc Society AIDS International the of Journal Kisumu, Kenya 90/94 (95.7%) 4/94 (4.3%) REF REF Blantyre, Malawi 84/98 (85.7%) 14/98 (14.3%) 3.73 1.11, 16.17 0.030 3.77 1.12, 16.39 0.029 Cape Town, South Africa 74/84 (88.1%) 10/84 (11.9%) 3.02 0.83, 13.75 0.105 3.29 0.89, 15.09 0.079 Soweto, South Africa 93/93 (100.0%) 0/93 (0.0%) 0.19 0.00, 1.11 0.124 0.22 0.00, 1.32 0.172 HIV status at screening Negative 281/304 (92.4%) 23/304 (7.6%) REF Positive 59/64 (92.2%) 5/64 (7.8%) 1.04 0.38, 2.83 0.946 Age 24.30 (5.50) 25.26 (6.36) 1.03 0.97, 1.10 0.386

Education | https://doi.org/10.1002/jia2.25600 Low (less than grade 12) 121/132 (91.7%) 11/132 (8.3%) REF 2020, Middle (at least grade 12) 144/153 (94.1%) 9/153 (5.9%) 0.69 0.28, 1.71 0.421 23

High (beyond secondary school) 71/79 (89.9%) 8/79 (10.1%) 1.24 0.48, 3.23 0.660 (S6):e25600 Employment status Full or part time employed 102/110 (92.7%) 8/110 (7.3%) REF Self-employed 45/52 (86.5%) 7/52 (13.5%) 1.98 0.68, 5.80 0.211 Unemployed (including in-between jobs) 96/100 (96.0%) 4/100 (4.0%) 0.53 0.15, 1.82 0.314 Student 86/95 (90.5%) 9/95 (9.5%) 1.33 0.49, 3.61 0.570 Other 10/10 (100.0%) 0/10 (0.0%) N.A. Marital status Single/divorced/widowed 318/342 (93.0%) 24/342 (7.0%) REF Married/civil union/legal partnership 21/25 (84.0%) 4/25 (16.0%) 2.52 0.80, 7.95 0.113 Any children No 268/285 (94.0%) 17/285 (6.0%) REF REF Yes 71/82 (86.6%) 11/82 (13.4%) 2.44 1.10, 5.45 0.029 2.05 0.87, 5.05 0.139 Transgender No 266/289 (92.0%) 23/289 (8.0%) REF Yes 69/74 (93.2%) 5/74 (6.8%) 0.84 0.31, 2.28 0.730 Sexual attraction Men and women 183/200 (91.5%) 17/200 (8.5%) REF Men only 155/166 (93.4%) 11/166 (6.6%) 0.76 0.35, 1.68 0.503 Sexual identity Bisexual and other 133/144 (92.4%) 11/144 (7.6%) REF Gay 206/223 (92.4%) 17/223 (7.6%) 1.00 0.45, 2.20 0.996 Negative feelings of homosexuality 1.98 (0.54) 1.92 (0.58) 0.79 0.39, 1.59 0.503 MSM-related stigma in healthcare 1.84 (0.22) 1.92 (0.20) 9.65 0.84, 111.0 0.069

65 Concealing same-sex sexuality 2.16 (1.18) 1.78 (1.16) 0.74 0.52, 1.07 0.113 0.86 0.55, 1.34 0.496 Sandfort TGM et al. Journal of the International AIDS Society 2020, 23(S6):e25600 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25600/full | https://doi.org/10.1002/jia2.25600

92.9%. The high retention was likely driven by the strong

value commitment and motivation to the study reported by partici-

p pants. Study retention may also have been enhanced by the b novelty of the study and the sense of validation of one’s same-sex attraction. The relatively high level of education of the participants in HPTN 075 compared to other studies among MSM in SSA (e.g. [33]) might have facilitated participation, even though level of education was not associated with completion of study visits. Participants reported an interest in giving back to their communities through study participation, in addition to receiving key services, including HIV testing and risk reduction counselling; this likely reinforced study partici- , reference group; SD, standard devia- pation. The involvement of outreach workers is likely to have value AOR 95% CI facilitated retention. Findings from HPTN 075 further high- p light the importance of extensive community and site preparation (amongst others for the delivery of culturally appropriate treatment), active involvement of the community and intense study retention activities. Barriers to retention were limited Univariate Multivariable and were largely related to mobility of participants. A very important finding in HPTN 075 was that there were very few documented social harms that triggered significant risk mitigation procedures. Given the small number of documented social harms, preparing RMPs might seem superflu- ous. Alternatively, one could argue that preparation of the 28) OR 95% CI

= RMPs may have reduced the potential for social harms; this was suggested by retrospective discussions with research staff. Without RMPs, staff might have been caught off-guard and unprepared, which could have aggravated the social harms that occurred. The few social harms that were observed were related to consent procedures and staff training and supervi- sion; future studies should attend closely to these factors. Some limitations should be considered when evaluating the findings. Because of the study design, it is not possible to state with certainty which factors contributed most to the study’s success. The design also did not allow us to evaluate the efficiency of the various recruitment strategies. Even Three variables were selected using stepwise model with selection of variables at entry significance level of 0.3 and

b though the study samples collected at each site were diverse,

341) Did not complete the study (N they are not necessarily representative of the respective pop-

= ulations. Finally, this study was implemented from 2015 to 2017, and although the social situation for sexual minority persons in SSA is not stable, it is extremely likely that what was done to make the study successful is still relevant in the current situation. It is not clear whether COVID-19 would have an impact on the feasibility of recruitment and retention

Mean (SD) / n/N (%) specific to MSM and TGW. The results of this study open the door to further large- Completed all visits (N scale HIV prevention research with MSM and TGW in SSA. Research is needed to improve understanding of the risks and resiliencies of this key population with respect to HIV transmission, and to develop evidence-based approaches to meet their urgent HIV prevention needs. MSM and TGW in SSA exact logistic regression analysis is applied due to the zero frequency of participants in the study site Soweto who did not complete the study. c have previously indicated interest in HIV prevention strategies, including condom use and PrEP [36,37]. The results of HPTN 075 support the conduct of future trials to advance

) integrated behavioural and biomedical HIV prevention in these key populations. MSM and TGW in SSA could benefit from inclusion in the next generation of HIV prevention trials to Continued

( determine whether promising interventions are feasible and effective for this key population, and to facilitate future implementation of HIV prevention interventions in these popula- Five participants who died during the study were excluded from this table; Table 3. a Likelihood to remain in studyHow for committed a they year felt toImportance participating of study for MSMTravel community to study siteMaking time for visitSetting up appointmentAOR, 1.32 adjusted 1.22 (0.46) odds (0.46) ratio;tion. CI, 1.14 confidence (0.42) intervals; MSM, men who have sex with men; n, number with characteristic; N, total number; OR, odds ratio; REF 1.92 (0.56) 1.22 1.28 (0.42) 1.90 1.82 (0.54) (0.52) (0.48) 1.10 (0.32) 0.59 1.33 0.79 0.23, 0.63, 1.49 2.81 0.27, 2.28 1.92 (0.54) 1.82 0.264 1.78 (0.54) 0.454 (0.50) 0.658 1.03 0.76 0.89 0.52, 2.03 0.36, 0.40, 1.60 1.99 0.942 0.470 0.775 exit significance level of 0.35; tions in SSA.

66 adotTMe al. et TGM Sandfort Table 4. Factors associated with missing one or more study visits, HPTN 075 study, Kenya, Malawi, South Africaa http://onlinelibrary.wiley.com/doi/10.1002/jia2.25600/full

Mean (SD) / n/N (%) Univariate Multivariableb

Completed all visits (N = 341) Missed ≥ 1 visits (N = 27) OR 95% CI p value AOR 95% CI p value

Country Society AIDS International the of Journal Kisumu, Kenya 90/93 (96.8%) 3/93 (3.2%) REF Blantyre, Malawi 84/86 (97.7%) 2/86 (2.3%) 0.71 0.12, 4.38 0.716 Cape Town, South Africa 74/89 (83.1%) 15/89 (16.9%) 6.08 1.70, 21.81 0.006 Soweto, South Africa 93/100 (93.0%) 7/100 (7.0%) 2.26 0.57, 9.00 0.248 HIV status at screening Negative 281/302 (93.0%) 21/302 (7.0%) REF Positive 59/65 (90.8%) 6/65 (9.2%) 1.36 0.53, 3.52 0.524 Age 24.30 (5.50) 21.82 (3.36) 0.88 0.79, 0.98 0.024

Education | https://doi.org/10.1002/jia2.25600 Low (less than grade 12) 121/129 (93.8%) 8/129 (6.2%) REF 2020, Middle (at least grade 12) 144/157 (91.7%) 13/157 (8.3%) 1.37 0.55, 3.40 0.504 23

High (beyond secondary school) 71/76 (93.4%) 5/76 (6.6%) 1.07 0.34, 3.38 0.915 (S6):e25600 Employment status Full or part time employed 102/110 (92.7%) 8/110 (7.3%) REF Self-employed 45/47 (95.7%) 2/47 (4.3%) 0.57 0.12, 2.78 0.483 Unemployed (including in-between jobs) 96/104 (92.3%) 8/104 (7.7%) 1.06 0.38, 2.94 0.907 Student 86/93 (92.5%) 7/93 (7.5%) 1.04 0.36, 2.98 0.945 Other 10/12 (83.3%) 2/30 (16.7%) 2.55 0.48, 13.68 0.275 Marital status Single/divorced/widowed 318/344 (92.4%) 26/344 (7.6%) REF Married/civil union/legal partnership 21/22 (95.5%) 1/22 (4.5%) 0.58 0.08, 4.50 0.604 Any children No 268/292 (91.8%) 24/292 (8.2%) REF Yes 71/74 (95.9%) 3/74 (4.1%) 0.47 0.14, 1.61 0.231 Transgender No 266/289 (92.0%) 23/289 (8.0%) REF Yes 69/73 (94.5%) 4/73 (5.5%) 0.67 0.22, 2.00 0.474 Sexual attraction Men and women 183/191 (95.8%) 8/191 (4.2%) REF Men only 155/173 (89.6%) 18/173 (10.4%) 2.66 1.12, 6.28 0.026 Sexual identity Bisexual and other 133/137 (97.1%) 4/137 (2.9%) REF REF Gay 206/229 (90.0%) 23/229 (10.0%) 3.71 1.26, 10.97 0.018 4.65 1.29, 16.83 0.019 Negative feelings of homosexuality 1.98 (0.54) 2.04 (0.62) 1.19 0.58, 2.46 0.637 1.91 0.80, 4.53 0.144 MSM-related stigma in healthcare 1.84 (0.22) 1.78 (0.24) 0.31 0.06, 1.49 0.143 0.39 0.07, 2.23 0.288

67 Concealing same-sex sexuality 2.16 (1.18) 2.66 (1.26) 1.42 1.01, 1.99 0.045 1.29 0.87, 1.90 0.206 Sandfort TGM et al. Journal of the International AIDS Society 2020, 23(S6):e25600 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25600/full | https://doi.org/10.1002/jia2.25600

5 | CONCLUSIONS value

p Enrolling and retaining MSM and TGW in SSA in a multi-coun-

b try, longitudinal, biobehavioural cohort study is feasible and can be conducted safely and successfully. This is especially the case when the local community of MSM and TGW as well as the community more generally are involved in the preparation of the study, and when MSM and TGW play a role in the actual study implementation. Extensive study site preparation seems indispensable. The primary barrier to study participation is the mobility of participants. Retention can be promoted in a variety of ways, including by providing needed services , reference group; SD, standard devia- and validation of participants’ sexual minority status. These findings strongly suggest that needed prevention trials with

value AOR 95% CI MSM and TGW in SSA are viable. p

AUTHORS’ AFFILIATIONS 1HIV Center for Clinical and Behavioral Studies, New York State Psychiatric Institute and Columbia University, New York, NY, USA; 2Science Facilitation Department, FHI 360, Durham, NC, USA; 3Perinatal HIV Research Unit, Univer- sity of the Witwatersrand, Soweto HPTN CRS, Soweto, South Africa; 4Vaccine and Infectious Disease Division, Fred Hutchinson Cancer Research Center, Seat- tle, WA, USA; 5Berman Institute of Bioethics, Johns Hopkins University, Balti- more, MD, USA; 6Department of Pathology, Johns Hopkins University School of Medicine, Baltimore, MD, USA; 7Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Blantyre, Malawi; 8Desmond Tutu HIV Cen- tre, UCT Medical School, Cape Town, South Africa; 9Kenya Medical Research Institute (KEMRI) CDC, Kisumu, Kenya; 10Anova Health Institute, Johannesburg, 27) OR 95% CI South Africa; 11Centre for the Development of People (CEDEP), Blantyre, = Malawi; 12Division of AIDS Research, National Institute of Mental Health, Bethesda, MD, USA; 13Division of AID, National Institute of Allergy and Infec- tious Disease, National Institutes of Health, Bethesda, MD, USA; 14FHI 360, Durham, NC, USA 1 visits (N ≥ COMPETING INTERESTS Jeremy Sugarman is a member of Merck KGaA’s Bioethics Advisory Panel and Stem Cell Research Oversight Committee, IQVIA’s Ethics Advisory Panel and Aspen Neuroscience’s Scientific Advisory Board; he has consulted for Biogen and Portola Pharmaceuticals Inc. None of these relationships are related to the four variables were selected using stepwise model with selection of variables at entry significance level of 0.3 and b material discussed in this manuscript. None of the other authors has a conflict 341) Missed of interest or a potential conflict of interest to report. = Mean (SD) / n/N (%) Univariate Multivariable AUTHORS’ CONTRIBUTIONS TS, EH, YC, VC, JS, SD, KD, RP, DS, FZ, DR, OR, CM, DK, BK, AO, MS, WC, JL, MF, LM and SE involved in study design and implementation. AM, XG and YC analysed the data. TS, EH and SE wrote the manuscript. EH, AM, XG, VC, JS, SD, KD, DS, MS, WC, JL, YC, MF and LM reviewed manuscript and offered comments and revisions. All authors have read and approved the final manuscript. Completed all visits (N ACKNOWLEDGEMENTS The authors thank the local community advisory boards, study staff and study participants. The protocol team included the following members (in addition to authors of the current manuscript): Stefan D Baral, Linda-Gail Bekker, Vanessa Elharrar, Lynda Emel, Chris(tie) Heiberg, Noel Kayange, Josh Kikuvi, Tim Lane, Lebah Lugalia, Yamikani Mbilizi, Ernest Moseki, Scott Rose, Paul Semugoma, Jer- ome Singh, Patrick S Sullivan and Carlos Toledo. Mark Barnes assisted in the preparation of the risk management plans.

) FUNDING Research reported in this publication was supported by the National Institute of Allergy and Infectious Diseases and the National Institute of Mental Health of

Continued the National Institutes of Health under award number UM1AI068619 (HPTN ( Leadership and Operations Center), UM1AI068617 (HPTN Statistical and Data Management Center) and UM1AI068613 (HPTN Laboratory Center). The content is solely the responsibility of the authors and does not necessarily repre- Five participants who died during the study were excluded from this table; Table 4. a Likelihood to remain in studyHow for committed a they year felt toImportance participating of study for MSMTravel community to study siteMaking time for visitSetting up appointmentAOR, adjusted odds 1.32 ratio;tion. 1.22 (0.46) CI, (0.46) confidence intervals; MSM, 1.14 men (0.42) who have sex with men; n, number with characteristic; N, total number; OR, odds ratio; REF 1.30 1.08 (0.46) (0.26) 1.12 1.92 (0.42) (0.56) 1.90 1.82 (0.52) (0.48) 0.91 0.34 0.82 0.39, 0.08, 2.14 1.38 0.29, 2.32 1.92 0.826 (0.48) 0.131 1.92 1.82 (0.54) (0.48) 0.703 1.02 1.13 1.01 0.50, 2.05 0.52, 0.44, 2.44 2.32 0.962 0.756 0.979 exit significance level of 0.35. sent the official views of the National Institute of Allergy and Infectious

68 Sandfort TGM et al. Journal of the International AIDS Society 2020, 23(S6):e25600 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25600/full | https://doi.org/10.1002/jia2.25600

Diseases or the National Institute of Mental Health. Dr. Sandfort received addi- 21. Baral S, Trapence G, Motimedi F, Umar E, Iipinge S, Dausab F, et al. HIV tional support from the National Institute of Mental Health, P30 MH43520. prevalence, risks for HIV infection, and human rights among men who have sex The funders had no role in study design, data collection and analysis, decision to with men (MSM) in Malawi, Namibia, and Botswana. PLoS One. 2009;4:e4997. publish or preparation of the manuscript. 22. Crowell TA, Keshinro B, Baral SD, Schwartz SR, Stahlman S, Nowak RG, et al. Stigma, access to healthcare, and HIV risks among men who sell sex to men in Nigeria. J Int AIDS Soc. 2017;20:21489. 23. Duby Z, Nkosi B, Scheibe A, Brown B, Bekker LG. ’Scared of going to the REFERENCES clinic’: Contextualising healthcare access for men who have sex with men, 1. Mayer KH, Beyrer C. HIV epidemiology update and transmission factors: female sex workers and people who use drugs in two South African cities. South risks and risk contexts—16th International AIDS Conference epidemiology ple- Afr. J HIV Med. 2018;19:701. nary. Clin Infect Dis. 2007;44(7):981–7. 24. Kombo B, Sariola S, Gichuru E, Molyneux S, Sanders EJ, van der Elst E. 2. Brown T, Peerapatanapokin W. Evolving HIV epidemics: the urgent need to “Facing Our Fears”: Using facilitated film viewings to engage communities in refocus on populations with risk. Curr Opin HIV Aids. 2019;14(5):337–53. HIV research involving MSM in Kenya. Cogent Med. 2017;4:1330728. 3. Wade AS, Kane CT, Diallo PA, Diop AK, Gueye K, Mboup S, et al. HIV infec- 25. The HIV Prevention Trials Network. Feasibility of HIV prevention cohort tion and sexually transmitted infections among men who have sex with men in studies among MSM in sub-Saharan Africa 2018 [cited 2018 Aug 16]. Available Senegal. AIDS. 2005;19(18):2133–40. from: https://www.hptn.org/research/studies/hptn075 4. Sanders EJ, Graham SM, Okuku HS, van der Elst EM, Muhaari A, Davies A, 26. Dominguez K, Kanyemba B, Kikuvi J, Cummings V, Guo X, Chege W, et al. et al. HIV-1 infection in high risk men who have sex with men in Mombasa, Gender nonconformity and HIV risk among sub-Saharan African men who have Kenya. AIDS. 2007;21(18):2513–20. sex with men in HPTN 075. 13th AIDS Impact Conference; 15 November 5. Rispel LC, Metcalf CA, Cloete A, Reddy V, Lombard C. HIV prevalence and 2017; Cape Town. risk practices among men who have sex with men in two South African cities. 27. UNAIDS. Good participatory practice: guidelines for biomedical HIV pre- J Acquir Immune Defic Syndr. 2011;57(1):69–76. vention trials. Geneva: UNAIDS; 2007. 6. Wirtz AL, Trapence G, Kamba D, Gama V, Chalera R, Jumbe V, et al. Geo- 28. Joint United Nations Programme on HIV/AIDS. Ethical considerations in graphical disparities in HIV prevalence and care among men who have sex with biomedical HIV prevention trials. Geneva: Joint United Nations Programme on men in Malawi: results from a multisite cross-sectional survey. Lancet HIV. HIV/AIDS (UNAIDS); 2012. 2017;4(6):e260–9. 29. amFAR. International AIDS Vaccine Initiative, Johns Hopkins University- 7. Sandfort TG, Lane T, Dolezal C, Reddy V. Gender expression and risk of HIV Center for Public Health and Human Rights, United Nations Development Pro- infection among black South African men who have sex with men. AIDS Behav. gram. Respect, protect, fulfill: best practices guidance in conducting HIV 2015;19(12):2270–9. research with gay, bisexual, and other men who have sex with men (MSM) in 8. Lane T, Raymond HF, Dladla S, Rasethe J, Struthers H, McFarland W, et al. rights-constrained environments. New York: amFAR; 2011. High HIV prevalence among men who have sex with men in Soweto, South 30. Rennie S, Sugarman J. Developing ethics guidance for HIV prevention Africa: results from the Soweto Men’s Study. AIDS Behav. 2011;15(3):626–34. research: the HIV Prevention Trials Network approach. J Med Ethics. 2010;36 9. Lahuerta M, Patnaik P, Ballo T, Telly N, Knox J, Traore B, et al. HIV preva- (12):810–5. Current version: Brown BJ, Sugarman J. Why ethics guidance needs lence and related risk factors in men who have sex with men in Bamako, Mali: to be updated for contemporary HIV prevention research. J Int AIDS Soc. findings from a bio-behavioral survey using respondent-driven sampling. AIDS 2020;23(5):e25500. Behav. 2018;22(7):2079–88. 31. Reisner SL, Biello K, Rosenberger JG, Austin SB, Haneuse S, Perez-Brumer 10. Beyrer C, Baral SD, van Griensven F, Goodreau SM, Chariyalertsak S, Wirtz A, et al. Using a two-step method to measure transgender identity in Latin AL, et al. Global epidemiology of HIV infection in men who have sex with men. America/the Caribbean, Portugal, and Spain. Arch Sex Behav. 2014;43(8):1503– Lancet. 2012;380(9839):367–77. 14. 11. Stannah J, Dale E, Elmes J, Staunton R, Beyrer C, Mitchell KM, et al. HIV 32. Reisner SL, Deutsch MB, Bhasin S, Bockting W, Brown GR, Feldman J, testing and engagement with the HIV treatment cascade among men who have et al. Advancing methods for US transgender health research. Curr Opin Endo- sex with men in Africa: a systematic review and meta-analysis. Lancet HIV. crinol Diabetes Obes. 2016;23(2):198–207. 2019;6(11):e769–87. 33. Sanders EJ, Okuku HS, Smith AD, Mwangome M, Wahome E, Fegan G, 12. Stahlman S, Lyons C, Sullivan PS, Mayer KH, Hosein S, Beyrer C, et al. HIV et al. High HIV-1 incidence, correlates of HIV-1 acquisition, and high viral loads incidence among gay men and other men who have sex with men in 2020: following seroconversion among MSM. AIDS. 2013;27(3):437–46. where is the epidemic heading? Sexual Health. 2017;14(1):5–17. 34. Keshinro B, Crowell TA, Nowak RG, Adebajo S, Peel S, Gaydos CA, et al. 13. Rao A, Stahlman S, Hargreaves J, Weir S, Edwards J, Rice B, et al. Sampling High prevalence of HIV, chlamydia and gonorrhoea among men who have sex Key populations for HIV surveillance: results from eight cross-sectional studies with men and transgender women attending trusted community centres in using respondent-driven sampling and venue-based snowball sampling. JMIR Abuja and Lagos, Nigeria. J Int AIDS Soc. 2016;19:21270. Public Health Surveill. 2017;3(4):e72. 35. Nyato D, Kuringe E, Drake M, Casalini C, Nnko S, Shao A, et al. Partici- 14. Stahlman S, Johnston LG, Yah C, Ketende S, Maziya S, Trapence G, et al. pants’ accrual and delivery of HIV prevention interventions among men who Respondent-driven sampling as a recruitment method for men who have sex have sex with men in sub-Saharan Africa: a systematic review. BMC Public with men in southern sub-Saharan Africa: a cross-sectional analysis by wave. Health. 2018;18(1):370. Sex Transm Infect. 2016;92(4):292–8. 36. Karuga RN, Njenga SN, Mulwa R, Kilonzo N, Bahati P, O’Reilley K, et al. 15. Moller LM, Stolte IG, Geskus RB, Okuku HS, Wahome E, Price MA, et al. "How I Wish This Thing Was Initiated 100 Years Ago!" willingness to take daily Changes in sexual risk behavior among MSM participating in a research cohort oral pre-exposure prophylaxis among men who have sex with men in Kenya. in coastal Kenya. AIDS. 2015;29:S211–9. PLoS One. 2016;11:e0151716. 16. Schwartz SR, Nowak RG, Orazulike I, Keshinro B, Ake J, Kennedy S, et al. 37. Shaver J, Sullivan P, Siegler A, de Voux A, Phaswana-Mafuya N, Bekker LG, The immediate effect of the Same-Sex Marriage Prohibition Act on stigma, dis- et al. Comparing provider and client preferences for HIV prevention services in crimination, and engagement on HIV prevention and treatment services in men South Africa among men who have sex with men. J Int Assoc Provid AIDS Care. who have sex with men in Nigeria: analysis of prospective data from the TRUST 2017;16(6):562–71. cohort. Lancet HIV. 2015;2(7):e299–306. 17. Couderc C, Dembele Keita B, Anoma C, Wade AS, Coulibaly A, Ehouman S, et al. Is PrEP needed for MSM in West Africa? HIV incidence in a prospective multi-country cohort. J Acquir Immune Defic Syndr. 2017;75(3):e80–e2. 18. Carroll A, Mendos LR. State-sponsored homophobia. Geneva: Ilga; 2017. SUPPORTING INFORMATION 19. Kohut A, Wike R, Bell J, Horowitz JM, Simmons K, Stokes B, et al. The global divide on homosexuality. Washington, DC: Pew Research Center; 2013. 20. Thoreson R, Cook S. Nowhere to turn: blackmail and extortion of LGBT Additional Supporting Information may be found in the online people in sub-Saharan, Africa. New York: International Gay and Lesbian Human version of this article: Rights Commission; 2011. Appendix S1. HPTN 075 site-specific risk mitigation plans.

69 Ramadhani HO et al. Journal of the International AIDS Society 2020, 23(S6):e25599 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25599/full | https://doi.org/10.1002/jia2.25599

RESEARCH ARTICLE Association of age with healthcare needs and engagement among Nigerian men who have sex with men and transgender women: cross-sectional and longitudinal analyses from an observational cohort Habib O Ramadhani1,§ , Trevor A Crowell2,3 , Rebecca G Nowak1 , Nicaise Ndembi4, Blessing O Kayode4, Afoke Kokogho3,5, Uchenna Ononaku4, Elizabeth Shoyemi6, Charles Ekeh6, Sylvia Adebajo7, Stefan D Baral8 and Manhattan E Charurat 1 for the TRUST/RV368 Study Group §Corresponding author: Habib O Ramadhani, Division of Epidemiology and Prevention, Institute of Human Virology, University of Maryland School of Medicine, 725 W Lombard St, Baltimore, Maryland 21201, USA. Tel: 4107061283. ([email protected])

Abstract Introduction: Young men who have sex with men (MSM) and transgender women (TGW) face stigmas that hinder access to healthcare. The aim of the study was to understand age-related determinants of healthcare needs and engagement among MSM and TGW. Methods: The TRUST/RV368 cohort provides integrated prevention and treatment services for HIV and other sexually transmitted infections (STIs) tailored to the needs of sexual and gender minorities. MSM and TGW aged ≥16 years in Abuja and ≥18 years Lagos, Nigeria, completed standardized behavioural questionnaires and were tested for HIV, Neisseria gonorrhoeae (NG) and Chlamydia trachomatis (CT) every three months for up to 18 months. Logistic regression was used to estimate adjusted odds ratios (aORs) for associations of age and other factors with outcomes of interest upon enrolment, including HIV care continuum steps – HIV testing, ART initiation and viral suppression <1000 copies/mL. Cox proportional hazards models were used to calculate adjusted hazard ratios (aHRs) for associations with incident infections. Results: Between March 2013 and February 2019, 2123 participants were enrolled with median age 23 (interquartile range 21 to 27) years. Of 1745 tested, 865 (49.6%) were living with HIV. HIV incidence was 11.6/100 person-years [PY], including 23.1/100PY (95% CI 15.5 to 33.1) among participants aged 16 to 19 years and 23.8/100 PY (95% CI 13.6 to 39.1) among TGW. Compared to participants aged ≥25 years, those aged 16 to 19 years had decreased odds of prior HIV testing (aOR 0.40 [95% CI 0.11 to 0.92]), disclosing same-sex sexual practices to healthcare workers (aOR 0.53 [95% CI 0.36 to 0.77]) and receiving HIV prevention information (aOR 0.60 [95% CI 0.41 to 0.87]). They had increased odds of avoiding healthcare (aOR 1.94 [95% CI 1.3 to 2.83]) and engaging in transactional sex (aOR 2.76 [95% CI 1.92 to 3.71]). Age 16 to 19 years was independently associated with increased incidence of HIV (aHR 4.09 [95% CI 2.33 to 7.49]), NG (aHR 3.91 [95% CI 1.90 to 8.11]) and CT (aHR 2.74 [95% CI 1.48 to 5.81]). Conclusions: Young MSM and TGW demonstrated decreased healthcare engagement and higher incidence of HIV and other STIs as compared to older participants in this Nigerian cohort. Interventions to address unique obstacles to healthcare engagement by adolescents and young adults are needed to curb the spread of HIV and other STIs among MSM and TGW in Nigeria. Keywords: men who have sex with men; transgender people; HIV care continuum; STI; Delivery of Health Care; sexual and gender minorities; Africa South of the Sahara

Additional Supporting information may be found online in the Supporting Information tab for this article.

Received 18 January 2020; Accepted 20 July 2020 Copyright © 2020 The Authors. Journal of the International AIDS Society published by John Wiley & Sons Ltd on behalf of the International AIDS Society. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

1 | INTRODUCTION TGW are more likely than their older counterparts to engage in transactional sex, have multiple sexual partners and have Men who have sex with men (MSM) and transgender women condomless sex with partners whose HIV status remains (TGW) are disproportionately impacted by HIV and other sex- unknown [1]. They tend to have lower levels of education, ually transmitted infections (STIs) [1,2]. Young adult MSM and decreased ability to negotiate condom use and increased

70 Ramadhani HO et al. Journal of the International AIDS Society 2020, 23(S6):e25599 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25599/full | https://doi.org/10.1002/jia2.25599 susceptibility to financial coercion to engage in condomless The study was approved by the University of Maryland Bal- sex [3,4]. They also tend to have more sexual partners and timore IRB, Baltimore, MD, USA; the Federal Capital Territory membership in higher risk sexual networks [5]. Health Research Ethics Committee, Abuja, Nigeria; Ministry of Same-sex sexual practices are discouraged by cultural norms Defense Health Research Ethics Committee, Abuja, Nigeria; in many parts of the world and, as a result, MSM and TGW and Walter Reed Army Institute of Research IRB, Silver are frequently subject to stigma in their communities and Spring, MD, USA. from healthcare providers [6-8]. The synergistic interaction of stigma-related challenges and sexual practices contributes to a 2.2 | Data collection decrease in healthcare seeking behaviour and engagement and an increase in the burden of HIV and other STIs among Face-to-face interviews using structured questionnaires were young adult MSM and TGW [8,9]. In the United States where administered to study participants at enrolment (spread same-sex practices are not criminalized, HIV incidence among across two visits spaced two weeks apart) and every three MSM aged <25 years was 2.5 times that of MSM aged months thereafter for up to 18 months. Among other topics, ≥25 years [10]. This phenomenon of higher HIV incidence the questionnaires captured data on demographic characteris- among young adult MSM has consistently been replicated in tics, sexual practices, healthcare-seeking behaviours, disclosure other parts of the world [11,12]. of HIV status and receipt of HIV prevention education. Partici- In Nigeria, Africa’s most populous country, same-sex sexual pants were also asked whether they were currently experienc- practices are not only stigmatized, but criminalized [10,13]. ing potential STI symptoms including penile or rectal Experienced and anticipated stigma, such as expected rejec- discharge, painful urination, genital and ulcer/sore/lesion/rash. tion by their communities, lead many Nigerian MSM and Clinical data and specimens such as blood, rectal swabs and TGW to avoid seeking healthcare [8,14]. HIV screening is sub- urine samples were collected at enrolment and each follow-up stantially less common among Nigerian MSM and TGW aged visit. Blood samples were tested in real-time for HIV using rapid ≤19 as compared to those 30 years or older despite prior HIV antibody tests following the parallel testing algorithm for studies showing that awareness of HIV status lowers the risk at-risk participants according to national guidelines in Nigeria of acquisition and transmission [15]. Young adult Nigerian with determine, Uni-gold and HIV1/2 Stat Pak as a tie-breaker MSM and TGW demonstrate a higher prevalence of sexual for discrepant results [27]. For participants who were at risk for risk practices increasing their vulnerability to HIV and STIs as HIV, HIV testing continued every three months. Among people compared to their older counterparts [16]. We have previ- living with HIV (PLWH), plasma HIV RNA was quantified every ously demonstrated exceptionally high HIV incidence among three months (Abuja) or six months (Lagos) using COBAS Taq- Nigerian MSM and TGW, particularly those under 19 years of Man HIV-1 Test (Roche Molecular Diagnostics, Pleasanton, CA, age [17]. USA). At every visit, urine and rectal swabs were tested for NG The present analyses build upon this prior work to charac- and CT using the Aptima Combo 2 assay (Gen-Probe, San terize age-related determinants of healthcare engagement Diego, CA, USA). All participants who tested positive for any that may intersect with HIV acquisition risk. We explored infection were offered treatment according to national guideli- healthcare-seeking behaviour, HIV prevention practices, HIV nes [28]. Partner tracing was not conducted as part of the care continuum outcomes and incidence of HIV, Neisseria study, but participants were encouraged to bring sexual part- gonorrhoeae (NG) and Chlamydia trachomatis (CT) infections in ners to the clinic for STI diagnostics and treatment. an updated data set with a focus on associations between age and each of these outcomes among MSM and TGW in 2.3 | Definition of variables Nigeria. HIV care continuum outcome variables included lifetime history 2 | METHODS of ever having been tested for HIV prior to enrolment, ART initiation at enrolment and viral suppression at enrolment. Partici- < 2.1 | Study design and population pants who initiated ART and had HIV RNA 1000 copies/mL at six-months after ART initiation were considered virally sup- This was a secondary data analysis of data collected in the pressed, consistent with WHO Health [29]. Other outcomes TRUST/RV368 study of MSM and TGW in Nigeria. Partici- included disclosure of HIV status and same-sex sexual practices pants were prospectively recruited into a combination HIV to health care workers, avoidance of seeking healthcare, partici- prevention and treatment study using respondent driven sam- pation in HIV prevention meetings, receipt of HIV prevention pling (RDS) at two clinics in Abuja and Lagos as previously education and sex in exchange for money. described [18-20]. Incidence of HIV, rectal NG and CT infections were evalu- The study was conducted in collaboration with local non- ated among participants who did not have each of these infec- governmental organizations at facilities tailored to the needs tions at enrolment. We chose to focus on rectal NG and CT of sexual and gender minorities. Providers were sensitized to for these analyses because we have previously reported that the social, legal and sexual health needs of MSM and TGW. urogenital STIs are uncommon among Nigerian MSM and Staff included members of the MSM and TGW communities. TGW [23,25]. Only the first occurrence of NG and CT were Services provided at the clinics included education about safer included in these analyses. The main exposure of interest for sex practices [21], distribution of condoms and condom com- all analyses was participant age at enrolment categorized as patible lubricants [22], as well as diagnosis and treatment of 16 to 19, 20 to 24 and ≥25 years. Age categories were HIV and other STIs [23-26]. chosen with a focus on the youngest participants in the

71 Ramadhani HO et al. Journal of the International AIDS Society 2020, 23(S6):e25599 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25599/full | https://doi.org/10.1002/jia2.25599 cohort – those in their teenage years – with the remaining outcomes explored. Statistical analyses were conducted using older participants divided into two groups of roughly equal SAS version 9.3 (SAS Institute Inc., Cary, NC, USA). size. Categorizing ≤19 as younger is consistent with our previous work [15]. All participants in the cohort were birth males 3 | RESULTS and reported sex with men. We categorized participants as cisgender MSM, TGW, or other based on their self-reported 3.1 | Study population gender identity of “man,”“woman” or “other.” Network size was defined as the number of MSM and TGW the participant Between March 2013 and February 2019, a total of 2123 knew and/or had seen or communicated with in the past six MSM and TGW were enrolled in TRUST/RV368, including months. Network density was calculated as the total number 336 (15.8%) aged 16 to 19 years, 905 (42.6%) 20 to 24 years of actual connections divided by the total number of potential and 882 (41.5%) 25 years or older. Their median age was 23 connections, with higher proportions indicating larger network (interquartile range 21 to 27) years. Among 2108 participants densities [30]. Each study participant was asked to report up with self-reported gender, 1697 (80.5%) were cisgender to 5 of his most recent MSM/TGW has interacted with within MSM, 234 (11.1%) were TGW and 177 (8.4%) reported other the past six months. Potential connections were then com- gender identities, such as non-binary, gender neutral or gen- puted based on the number of MSM/TGW participant has der fluid. Among 1745 participants who were tested for HIV interacted with. Network size was roughly categorized into upon enrolment, 865 (49.6%) were living with HIV. The preva- four quartiles consistent to our previous work [15]. Social sup- lence of HIV was significantly higher among TGW compared port was defined as the extent of assistance an individual was to cisgender MSM (59.5% vs. 47.6%, p = 0.003). Proportion likely to receive from friends in times of need, based on of missing data was in the range 0.7% to 26% for different responses to five different social support questions, each mea- outcome variables. Table 1 presents the distribution of covari- sured using a Likert scale ranging from (0) strongly disagree ates stratified by age groups. Table 2 presents outcome vari- to (3) strongly agree. Scores were summed (score range: 0 to ables including HIV care continuum, markers of healthcare 15) and higher scores indicated stronger social support. For engagement and incident outcomes stratified by age group. analyses, social support scores were dichotomized at the med- Figure 1 presents HIV care continuum outcomes. ian in our study population. 3.2 | Prevalence of HIV and anorectal STIs 2.4 | Statistical analyses At enrolment, 865 participants were PLWH, of whom 380 Frequencies of categorical variables were calculated as the (43.9%) were newly diagnosed during study enrolment. Among proportion of participants. Chi-square tests were used to com- 1539 participants tested for NG, 311 (20.2%) had prevalent pare participants in relation to lifetime history of HIV testing, infection at enrolment and among 1537 tested for CT, 208 ART initiation and viral suppression. Logistic regression mod- (13.5%) had prevalent infection. els were used to calculate odds ratios (ORs) and 95% confidence intervals (CIs) for age and other factors such as prior 3.3 | HIV care continuum and other markers of HIV testing, ART initiation, viral suppression, exposure to HIV healthcare engagement education, disclosure of same-sex sexual practices, healthcare- seeking behaviours and transactional sex. Poisson regression Analyses of HIV care continuum and other markers of health- models were used to compute incidence rates and 95% CIs of care engagement (Table 3) showed that, compared to partici- HIV, NG and CT infections. Cox proportional hazards models pants aged ≥ 25 years, those 16 to 19 years had decreased were used to calculate hazard ratios (HRs) associated with odds of prior HIV testing (adjusted odds ratios (aORs) = 0.40; disease incidence. Person time was computed from the time 95% CI, 0.11 to 0.92), disclosing that they have sex with men of enrolment up to incident infection or study completion. to healthcare workers (aOR = 0.53; 95% CI, 0.36 to 0.77), Individuals with no incident infection were censored at their and receiving information about HIV prevention (aOR = 0.60; last study visit. In the primary analyses, cisgender MSM and 95% CI, 0.41 to 0.87). In comparison to older participants, TGW were pooled. Then, to explore whether the effect of age those who were between 20 and 24 years old also had on the incidence of STIs was modified by HIV status or gen- decreased odds of disclosing that they have sex with men to der identity, analyses were stratified. Complete-case analyses healthcare workers at enrolment (aOR = 0.80; 95% CI, 0.64 were conducted, excluding participants with missing data, to 1.00) and for those living with HIV, had decreased odds of except for analyses of HIV testing and linkage to care in which initiating ART (aOR = 0.60; 95% CI, 0.44 to 0.85). Young par- missing was recorded as not tested or not linked respectively. ticipants, had increased odds of having sex in exchange for Enrolment data were used in cross-sectional analyses to money (16 to 19 vs. ≥ 25 aOR = 2.67; 95% CI, 1.92 to 3.71 assess factors associated with prior HIV testing, ART initiation, and 20 to 24 vs. ≥ 25 aOR = 1.68; 95% CI, 1.34 to 2.10). exposure to HIV education, disclosure of same-sex sexual practices, healthcare-seeking behaviours and transactional sex. 3.4 | Incident HIV and other STIs Longitudinal data were used to assess factors associated with incidence of HIV, NG and CT infections in participants who The incidence of HIV per 100 person years among individuals did not have each infection already at enrolment. Multivariable aged 16 to 19 years was 23.1 (95% CI, 15.5 to 33.1). The models were adjusted for gender identity, network size, net- incidence of HIV for those aged 20 to 24 and ≥25 years was work density and social support. These variables were 13.6 (95% CI, 9.9 to 18.1) and 5.8 (95% CI, 3.6 to 8.8) adjusted because of their plausibility in relation to the respectively. The incidence of anorectal NG per 100 person

72 Ramadhani HO et al. Journal of the International AIDS Society 2020, 23(S6):e25599 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25599/full | https://doi.org/10.1002/jia2.25599

Table 1. Distribution of covariates and other characteristics: comparison between younger and older MSM in Abuja and Lagos, Nigeria, 2013 to 2019

16 to 19 years old 20 to 24 years old ≥25 years old N = 336 N = 905 N = 882 Characteristics n (%) n (%) n (%) p-value

Cis gender Cisgender MSM 266 (79.2) 711 (78.6) 720 (81.6) TGW 44 (13.1) 108 (11.9) 82 (9.3) 0.210 Other 26 (7.7) 80 (8.8) 72 (8.2) Missing 0 (0.0) 6 (0.7) 8 (0.9) Network density <50% 69 (20.5) 244 (27.0) 289 (32.8) ≥50% 132 (39.3) 355 (39.2) 342 (38.8) <.001 Missing 135 (40.2) 306 (33.8) 241 (28.5) Network size 1 to 10 220 (65.5) 611 (67.5) 592 (67.1) 11 to 20 47 (14.0) 131 (14.5) 132 (15.0) 21 to 30 26 (7.7) 57 (6.3) 45 (5.1) 0.669 31+ 43 (12.8) 102 (11.3) 110 (12.5) Missing 0 (0.0) 4 (0.4) 3 (0.3) Social support

MSM, Men who have sex with men; n, number of study participants with a given characteristic; N, total number of study participants in that age category; TGW, transgender women. years among individuals aged 16 to 19, 20 to 24 and participants living with HIV. There were no statistically signifi- ≥25 years was 37.2 (95% CI, 25.2 to 53.1), 28.6 (95% CI, cant association between network density, network size, social 23.4 to 34.6) and 15.8 (95% CI 12.8 to 19.3) respectively. The support and incidence of HIV or other STIs. incidence of CT per 100 person years among individuals aged The effect of age on incident HIV, NG and CT infections 16 to 19, 20 to 24 and ≥25 years was 41.1 (95% CI, 28.8 to was modified by whether a participant identified as cisgender 57.0), 26.9 (95% CI, 22.2 to 32.3) and 18.5 (95% CI, 15.2 to MSM or TGW. Cisgender MSM who were 16 to 19 vs. 22.3) respectively. Of 248 cumulative incident NG and CT ≥25 years (aHR = 4.70; 95% CI: 2.28 to 9.51) and 20 to 24 infections, 201 (81.0%) were asymptomatic. vs. ≥25 years (aHR = 2.83; 95% CI: 1.48 to 5.32) had Compared to older participants, young participants had increased hazards of incident HIV (Table S1). Cisgender MSM increased hazards of incident HIV, (adjusted hazard ratios who were 16 to 19 vs. ≥25 years (aHR = 2.70; 95% CI: 1.65 (aHR) = 4.09; 95% CI, 2.33 to 7.47), NG (aHR = 3.91; 95% to 4.41) and 20 to 24 vs. ≥25 years (aHR = 1.81; 95% CI: CI, 1.90 to 8.11) and CT (aHR = 2.74; 95% CI, 1.48 to 5.81), 1.30 to 2.51) had increased hazards of incident NG infections after adjusting for gender, network size, network (Table S2). Cisgender MSM who were 16 to 19 vs. ≥25 years density and social support (Table 4). As compared to cisgen- (aHR = 1.63; 95% CI: 1.66 to 4.16) and 20 to 24 vs. der MSM, TGW had a higher hazards of incident HIV, ≥25 years (aHR = 1.60; 95% CI: 1.17 to 2.18) had increased (aHR = 2.00; 95% CI, 1.09 to 3.61), NG (aHR = 1.71; 95% CI, hazards of incident CT (Table S3). There were no statistically 1.00 to 2.79) and CT (aHR = 1.42; 95% CI, 1.00 to 2.01). significant associations between age and incident HIV, NG or Analyses stratified by HIV (Table 5) showed that age, 16 to CT among TGW. 19 vs.> 25 years (aHR = 2.00; 95% CI: 1.19 to 3.01) and 20 to 24 vs. ≥25 years (aHR = 3.88; 95% CI: 1.91 to 3.61), was significantly associated with an increased hazards of incident 4 | DISCUSSION NG among participants living with HIV. Similarly, those who were 16 to 19 vs. ≥25 years (aHR = 1.73; 95% CI: 1.22 to The incidence of HIV and other STIs was higher among cis- 2.59) and 20 to 24 vs. ≥25 years (aHR = 2.68; 95% CI: 1.30 gender MSM and TGW youth aged 24 and younger as com- to 5.72) had an increased hazard of incident CT among pared to older cisgender MSM and TGW. The incidence of

73 Ramadhani HO et al. Journal of the International AIDS Society 2020, 23(S6):e25599 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25599/full | https://doi.org/10.1002/jia2.25599

Table 2. HIV care continuum outcomes, sti and health seeking behavior characteristics: comparison between younger and older MSM in Abuja and Lagos, Nigeria, 2013 to 2019

16 to 19 years old 20 to 24 years old ≥25 years old

Characteristics N n (%) N n (%) N n (%)

Tested for HIV 336 241 (71.7) 905 744 (82.2) 882 760 (86.2) Initiated ART 70 51 (72.9) 349 239 (68.5) 446 348 (78.0) Virally suppressed 47 31 (66.0) 178 140 (78.7) 263 206 (78.3) Avoided seeking healthcare 336 82 (24.4) 902 196 (21.7) 874 179 (20.5) Disclosed sex with men to healthcare worker 333 60 (18.0) 901 275 (30.5) 875 330 (37.7) Disclosed HIV status to healthcare worker 36 29 (80.6) 187 152 (81.3) 262 218 (83.2) Participated in HIV prevention meetings 197 45 (22.8) 609 186 (30.5) 660 191 (28.9) Received information about HIV 197 146 (74.1) 610 483 (79.2) 663 548 (82.7) Having anal sex in exchange for money 327 199 (60.9) 878 433 (49.3) 848 288 (34.0) Cumulative HIV incidence 214 27 (12.6) 491 43 (8.7) 390 19 (4.9) Cumulative Gonorrhoea incidence 122 28 (23.0) 453 101(22.3) 538 89 (16.5) Cumulative Chlamydia incidence 120 33 (27.5) 473 110 (23.3) 548 105 (19.2)

ART, antiretroviral therapy; HIV, human immunodeficiency viruses; MSM, men who have sex with men; n, number of study participants with a given characteristic; N, total number of study participants in that age category; STI, sexually transmitted infections.

800

700 86% 82% 760

600 744

500 Age(in years) 400 16 –19

78% 20 –24 ≥ 25 300 348 72% 69%

200 78% 241 239 79% 206

Number of clients at each point cascade 100 66% 140 73% 31 0 51 HIV testing ART initiation Viral supression

Figure 1. HIV care continuum outcomes. Comparison between younger and older MSM in Abuja and Lagos, Nigeria, 2013 to 2019. ART, antiretroviral therapy; MSM, men who have sex with men

HIV and other STIs was also higher among Nigerian TGW other STIs were significantly higher among cisgender MSM compared to cisgender MSM. As previously reported, TGW and TGW youth compared to older cisgender MSM and TGW who have sex with men predominantly practice receptive anal in Nigeria. Globally, young populations may face many develop- sex that carried an increased risk of HIV and other STI acqui- mental, psychological and social factors that predispose them sition as compared to other sex acts [31]. High burdens of to an increased risk of STIs [35]. These may include lack of bacterial STIs have been reported among TGW, which also knowledge about HIV risk, negative or complacent attitudes increase risk of incident HIV infection [32,33]. High incidence towards safer sex, disinformation provided by peers on safer of HIV has similarly been reported among TGW in Kenya sexual behaviours, not having experienced the severity of dis- [34]. We have also shown that the incidence rates of HIV and ease at the onset of the illness and perception of HIV as a

74 Ramadhani HO et al. Journal of the International AIDS Society 2020, 23(S6):e25599 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25599/full | https://doi.org/10.1002/jia2.25599

Table 3. Association between health seeking behaviour, HIV prevention characteristics and age: comparison between younger and older MSM/TGW in Abuja and Lagos, Nigeria, 2013 to 2019

16 to 19 years old 20 to 24 years old ≥25 years old Characteristics aOR (95% CI) aOR (95% CI) aOR

Tested for HIV 0.40 (0.11 to 0.92) 0.50 (0.28 to 1.02) Ref Initiated ART 0.75 (0.43 to 1.38) 0.60 (0.44 to 0.85) Ref Viral suppression 0.82 (0.75 to 1.05) 1.10 (0.91 to 1.10) Ref Avoided seeking healthcare 1.94 (1.43 to 2.83) 1.11 (0.88 to 1.48) Ref Disclosed MSM status to healthcare 0.53 (0.36 to 0.77) 0.80 (0.64 to 1.00) Ref Disclosed HIV status to healthcare 0.84 (0.35 to 2.02) 0.90 (0.55 to 1.48) Ref Participated in HIV prevention meetings 0.92 (0.42 to 2.03) 1.36 (0.90 to 2.04) Ref Received information about HIV prevention 0.60 (0.41 to 0.87) 0.82 (0.46 to 1.48) Ref Having sex in exchange for money 2.67 (1.92 to 3.71) 1.68 (1.34 to 2.10) Ref aOR, adjusted odds ratio; ART, antiretroviral therapy; CI, confidence Intervals; HIV, human immunodeficiency viruses; MSM, men who have sex with men; n, number of study participants with a given characteristic; N, total number of study participants in that age category; TGW, transgender women. chronic disease, which may lead to underestimation of per- engage in sex-selling. It is also important to note that selling sonal risk [18,36,37]. Our findings also showed cisgender sex is associated with high-risk behaviours such as condomless MSM and TGW youth were less likely to participate in HIV sex [46,47]. Higher incidence of HIV and other STIs among prevention meetings or receive information about HIV preven- cisgender MSM and TGW youth in this cohort could also be tion and therefore less likely to be knowledgeable of HIV pre- due to the high proportion of engagement in transactional sex vention strategies. High HIV incidence among cisgender MSM and mediated by condomless sex. and TGW youth underscores the need for biomedical inter- These analyses included a large sample from a well- ventions to prevent HIV, such as pre-exposure prophylaxis characterized cohort of cisgender MSM and TGW in combina- (PrEP), and the need for innovative strategies that improve tion with evaluation of HIV care continuum, markers of uptake and adherence to these interventions. Potential barri- healthcare engagement and incident outcomes, improving ers to PrEP use by young MSM include aversion to daily pills upon earlier cross-sectional and cohort studies with smaller and frequent clinic visits [38]. sample sizes evaluating age-related determinants of healthcare Regular HIV testing is critical for ensuring that cisgender needs and engagement [48,49]. A limitation of this study was MSM and TGW youth are linked to appropriate HIV preven- loss to follow-up which contributed to a high amount of miss- tion and treatment services. Frequent testing and treatment ing data, ranging from 0.7% to 26% for different outcome are also beneficial in diagnosing other STIs and reducing undi- variables. Also, our use of a complete-case strategy for analy- agnosed HIV infection [39-41]. In these analyses, we noted ses resulted in different sample sizes and denominators for young cisgender MSM and TGW reported avoidance of most analyses and may have introduced some bias, but healthcare. Prior studies have shown that individuals at risk enabled inclusion of diverse parameters without making any for HIV fear rejection and additional stigma if they acquire assumptions for imputation of missing values. Stigma, mental HIV, hindering them from accessing HIV testing [42]. Qualita- health and substance use were not used in the secondary tive literature has shown that young MSM experience fear of analysis and this limited our ability to describe their associa- HIV testing results, rejection and unfriendly testing environ- tion with incidence of HIV and other STIs. However, our previ- ments [43]. Prior studies also indicate that PLWH may delay ous work showed that higher stigma scores were associated or not access care in order to avoid anticipated rejection by with increased prevalence of HIV and STIs [6]. In addition, the providers, families and the general public [15,16]. Low levels small TGW sample size diminished power to detect significant of testing, particularly among MSM and TGW youth, could differences in HIV or STI incidence by age group, though the hinder implementation of HIV prevention strategies such as observed rates were indeed higher among younger age groups PrEP. among TGW. RDS weights were not included in our models Over half of the cisgender MSM and TGW youth in this based on our prior work showing that equilibrium for our HIV cohort reported having sex in exchange for money. A prior incidence analysis was reached after recruitment of 100 to study from Nigeria also showed that transactional sex was 300 individuals [17]. HIV testing was assessed using an ever most common among MSM aged between 15 and 19 years measure, which limited our ability to understand with cer- [16]. We have previously reported that sex-selling was inde- tainty if young individuals were less likely than older ones to pendently associated with decreased uptake of healthcare ser- test for HIV. Furthermore, differences in the lower age limit vices in our cohort [44]. Researchers have noted that (16 for Abuja and 18 for Lagos) for recruitment may have transactional sex is more prevalent for those who are socio- introduced bias in our estimates of age-related associations. economically strained and increases the rates of HIV and To distinguish new incident infections from unresolved infec- other STIs [45]. Compared to older MSM, young MSM are tions, we excluded prevalent anorectal STI and considered more likely to be economically disadvantaged and hence only the first incident infection in all incidence analyses.

75 aahn Oe al. et HO Ramadhani http://onlinelibrary.wiley.com/doi/10.1002/jia2.25599/full ora fteItrainlAD Society AIDS International the of Journal

Table 4. Factors associated with incident HIV/STI among MSM and TGW in Abuja and Lagos, Nigeria, 2013 to 2019

HIV (N = 880) Gonorrhoea (N = 1141) Chlamydia (N = 1113)

Cases/ Cases/ Cases/ person Incidence /100 person Incidence /100 person Incidence /100 Characteristic years person years aHR (95% CI) years person years aHR (95% CI) years person years HR (95% CI)

Age | https://doi.org/10.1002/jia2.25599

≥25 19/330.0 5.8 Ref 89/563.7 15.8 Ref 105/567.6 18.5 Ref 2020, 20 to 24 43/317.0 13.6 2.41 (1.39 to 4.11) 101/353.7 28.6 1.86 (1.36 to 2.43) 110/408.5 26.9 1.67 (1.18 to 2.60) 16 to 19 27/117.0 23.1 4.09 (2.33 to 7.47) 28/75.2 37.2 3.91 (1.90 to 8.11) 33/80.3 41.1 2.74 (1.48 to 5.81) 23 (S6):e25599 Cisgender Cisgender 65/631.2 10.3 Ref 166/813.0 20.4 Ref 187/854.2 21.9 Ref MSM TGW 14/58.7 23.8 2.00 (1.09 to 3.61) 31/92.7 33.4 1.72 (1.00 to 2.79) 36/106.1 33.9 1.42 (1.00 to 2.01) Other 8/49.9 16.0 1.39 (0.64 to 2.88) 18/62.3 28.9 1.69 (0.92 to 3.18) 19/70.8 26.8 1.28 (0.78 to 1.99) Network density <50% 45/337.0 13.4 Ref 123/537.5 22.9 Ref 111/482.6 23.0 Ref ≥50% 40/352.7 11.3 0.82 (0.49 to 1.17) 95/454.0 20.9 0.71 (0.49 to 1.13) 132/546.6 24.0 1.00 (0.71 to 1.53) Network size 1 to 10 61/478.0 12.8 Ref 132/577.2 22.9 Ref 136/621.2 21.9 Ref 11 to 20 7/115.1 6.1 0.52 (0.21 to 1.14) 33/176.0 18.8 0.78 (0.54 to 1.22) 50/171.8 29.1 1.32 (1.01 to 1.78) 21 to 30 6/68.4 8.8 0.58 (0.24 to 1.28) 14/68.4 20.5 0.84 (0.50 to 1.51) 21/80.7 26.0 1.11 (0.72 to 1.84) 31+ 14/85.4 16.4 1.32 (0.68 to 2.32) 39/167.3 23.3 0.93 (0.71 to 1.46) 41/179.7 22.8 1.00 (0.68 to 1.41) Social support

CI, confidence intervals; HIV, human immunodeficiency viruses; HR, hazard ratio; MSM, men who have sex with men; STI, sexually transmitted infections; TGW, transgender women. 76 aahn Oe al. et HO Ramadhani http://onlinelibrary.wiley.com/doi/10.1002/jia2.25599/full ora fteItrainlAD Society AIDS International the of Journal

Table 5. Factors associated with incident GC and CT among MSM and TGW in Abuja and Lagos, Nigeria, 2013 to 2019

Gonorrhoea Chlamydia

HIV-infected HIV-uninfected HIV-infected HIV-uninfected | https://doi.org/10.1002/jia2.25599

Characteristic Cases/person years aHR (95% CI) Cases/person years aHR (95% CI) Cases/person years aHR (95% CI) Cases/person years aHR (95% CI) 2020,

Age 23

≥25 70/369.5 Ref 19/193.9 Ref 79/382.5 Ref 26/184.9 Ref (S6):e25599 20 to 24 81/224.4 2.00 (1.29 to 3.01) 20/129.0 1.51 (0.78 to 3.02) 87/270.6 1.73 (1.22 to 2.59) 23/137.5 1.18 (0.72 to 2.12) 16 to 19 22/40.4 3.88 (1.91 to 7.99) 6/34.9 2.01 (0.83 to 4.99) 23/50.0 2.68 (1.30 to 5.72) 10/30.3 2.89 (1.52 to 6.14) Network density <50% 71/228.5 Ref 30/213.5 Ref 71/266.0 Ref 40/216.6 Ref ≥50% 85/322.1 0.99 (0.71 to 1.43) 32/213.9 1.22 (0.72 to 1.89) 95/348.3 0.99 (0.72 to 1.53) 37/197.9 1.14 (0.67 to 1.71) Network size 1 to 10 88/309.2 Ref 44/268.4 Ref 93/344.8 Ref 43/276.2 Ref 11 to 20 24/106.9 0.92 (0.63 to 1.49) 9/69.0 0.74 (0.29 to 1.52) 33/105.8 1.33 (0.89 to 2.01) 17/65.9 1.42 (0.78 to 2.62) 21 to 30 11/40.5 1.01 (0.48 to 1.87) 3/27.8 0.63 (0.21 to 2.24) 13/48.4 0.91 (0.48 to 1.70) 8/32.1 1.51 (0.72 to 3.32) 31+ 33/105.9 1.12 (0.67 to 1.73) 6/61.3 0.49 (0.24 to 1.29) 30/127.2 0.82 (0.58 to 1.31) 11/52.4 1.11 (0.63 to 2.28) Social support

aHR, adjusted Hazard ratio; CI, confidence intervals; CT, Chlamydia trachomatis; GC, Neisseria gonorrhoeae; HIV, human immunodeficiency viruses; TGW, transgender women. 77 Ramadhani HO et al. Journal of the International AIDS Society 2020, 23(S6):e25599 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25599/full | https://doi.org/10.1002/jia2.25599

5 | CONCLUSIONS with men (MSM): a prospective cohort in Abuja and Lagos, Nigeria. Sex Transm Infect. 2017;93(5):348–55. 2. Harney BL, Agius PA, El-Hayek C, Fairley CK, Chow EPF, Roth N, et al. Risk of The incidence of HIV and other STIs among Nigerian cisgen- subsequent HIV infection following sexually transmissible infections among men der MSM and TGW youth participants was remarkably high in who have sex with men. Open Forum Infectious Diseases. 2019;6(10):ofz376. this cohort. Developmental and social transitions that occur 3. Musinguzi G, Bastiaens H, Matovu JK, Nuwaha F, Mujisha G, Kiguli J, et al. during this period of life limit the exposure of youth to HIV Barriers to condom use among high risk men who have sex with men in and STI prevention strategies. This calls for focused and com- Uganda: a qualitative study. PLoS One. 2015;10:e0132297. 4. Chemnasiri T, Varangrat A, Chaikummao S, Chitwarakorn A, Holtz TH. Why prehensive prevention messaging as well as care facilities that young MSM do not use condoms consistently: a qualitative exploration. AIDS Res specialize in adolescent care to better engage cisgender MSM Hum Retroviruses. 2014;30(S1):A199. and TGW youth. In many settings, exposure to sexual health 5. Pines HA, Karris MY, Little SJ. Sexual partner concurrency among partners and HIV prevention education doesn’t begin until after individ- reported by MSM with recent HIV infection. AIDS Behav. 2017;21(10):3026–34. 6. Rodriguez-Hart C, Musci R, Nowak RG, German D, Orazulike I, Ononaku U, uals reach 18 years of age, however, sexual debut can be et al. Sexual stigma patterns among Nigerian men who have sex with men and much earlier. Early engagement to address the healthcare their link to HIV and sexually transmitted infection prevalence. AIDS Behav. needs of youth at risk for HIV and other STIs is critical in 2018;22(5):1662–70. order to achieve an AIDS free generation. 7. Rodriguez MM, Madera SR, Diaz NV. Stigma and homophobia: persistent challenges for HIV prevention among young MSM in Puerto Rico. Rev Cienc Soc. 2013;26:50–9. AUTHORS’ AFFILIATIONS 8. Risher K, Adams D, Sithole B, Ketende S, Kennedy C, Mnisi Z, et al. Sexual 1Institute of Human Virology, University of Maryland School of Medicine, Balti- stigma and discrimination as barriers to seeking appropriate healthcare among more, MD, USA; 2Henry M. Jackson Foundation for the Advancement of Mili- men who have sex with men in Swaziland. J Int AIDS Soc. 2013;16:18715. tary Medicine Inc, Bethesda, MD, USA; 3U.S. Military HIV Research Program, 9. Shangani S, Naanyu V, Operario D, Genberg B. Stigma and healthcare-seeking Walter Reed Army Institute of Research, Silver Spring, MD, USA; 4Institute of practices of men who have sex with men in Western Kenya: a mixed-methods – Human Virology Nigeria, Abuja, Nigeria; 5HJF Medical Research International, approach for scale validation. AIDS Patient Care STDS. 2018;32(11):477 86. Abuja, Federal Capital Territory, Nigeria; 6Population Council, Abuja, Federal 10. Sullivan PS, Rosenberg ES, Sanchez TH, Kelley CF, Luisi N, Cooper HL, Capital Territory, Nigeria; 7Maryland Global Initiatives Corporation- A University et al. Explaining racial disparities in HIV incidence in black and white men who of Maryland Baltimore Affiliate, Abuja, Nigeria; 8Johns Hopkins School of Public have sex with men in Atlanta, GA: a prospective observational cohort study. Ann – Health, Baltimore, MD, USA Epidemiol. 2015;25(6):445 54. 11. Wahome E, Thiong’o AN, Mwashigadi G, Chirro O, Mohamed K, Gichuru E, et al. An empiric risk score to guide PrEP targeting among MSM in coastal COMPETING INTERESTS Kenya. AIDS Behav. 2018;22 Suppl 1:35–44. The authors declare no conflict of interest. 12. Piyaraj P, van Griensven F, Holtz TH, Mock PA, Varangrat A, Wimonsate W, et al. The finding of casual sex partners on the internet, methamphetamine use for sexual pleasure, and incidence of HIV infection among men who have sex AUTHORS’ CONTRIBUTIONS with men in Bangkok, Thailand: an observational cohort study. Lancet HIV. – HOR designed this analysis and authored the first draft of the manuscript. 2018;5(7):e379 89. BOK, UO, ES, CE and AK oversaw the collection of clinical data. TAC, RGN, NN, 13. Liu A, Coleman K, Bojan K, Serrano PA, Oyedele T, Garcia A, et al. Develop- SA, SDB and MEC, contributed to the design of the analysis and assisted in the ing a mobile App (LYNX) to support linkage to HIV/sexually transmitted infec- interpretation of results. All authors reviewed this manuscript, provided feed- tion testing and pre-exposure prophylaxis for young men who have sex with back and approved of the manuscript in its final form. men: protocol for a randomized controlled trial. JMIR Res Protoc. 2019;8: e10659. 14. Wirtz AL, Jumbe V, Trapence G, Kamba D, Umar E, Ketende S, et al. HIV ACKNOWLEDGEMENT among men who have sex with men in Malawi: elucidating HIV prevalence and The authors thank the study participants and staff at the TRUST/RV368 clinics correlates of infection to inform HIV prevention. J Int AIDS Soc. for their contributions and commitment to this research. 2013;16:18742. 15. Ramadhani HO, Ndembi N, Nowak RG, Ononaku U, Gwamna J, Orazulike I, et al. Individual and network factors associated with HIV care continuum out- FUNDING comes among Nigerian MSM accessing healthcare services. J Acquir Immune – This work was supported by a cooperative agreement between the Henry M. Defic Syndr. 2018;79(1):e7 16. Jackson Foundation for the Advancement of Military Medicine, Inc., and the U.S. 16. Bamgboye EA, Badru T, Bamgboye A. Transactional sex between men and Department of Defense (W81XWH-11-2-0174); the National Institutes of its implications on HIV and sexually transmitted infections in Nigeria. J Sex – Health (R01 MH099001, R01 AI120913, R01 MH110358); Fogarty Epidemiol- Transm Dis. 2017;2017:1 7. ogy Research Training for Public Health Impact in Nigeria program 17. Nowak RG, Mitchell A, Crowell TA, Liu H, Ketende S, Ramadhani HO, et al. (D43TW010051) and the President’s Emergency Plan for AIDS Relief through a Individual and sexual network predictors of HIV incidence among men who have – cooperative agreement between the Department of Health and Human Ser- sex with men in Nigeria. J Acquir Immune Defic Syndr. 2019;80(4):444 53. vices/Centers for Disease Control and Prevention, Global AIDS Program and 18. Charurat ME, Emmanuel B, Akolo C, Keshinro B, Nowak RG, Kennedy S, the Institute for Human Virology-Nigeria (NU2GGH002099). et al. Uptake of treatment as prevention for HIV and continuum of care among HIV-positive men who have sex with men in Nigeria. J Acquir Immune Defic Syndr. 2015;68 Suppl 2:S114–23. DISCLAIMER 19. Baral SD, Ketende S, Schwartz S, Orazulike I, Ugoh K, Peel SA, et al. Evalu- The views expressed are those of the authors and should not be construed to ating respondent-driven sampling as an implementation tool for universal cover- represent the positions of the U.S. Army, the Department of Defense, or the age of antiretroviral studies among men who have sex with men living with HIV. – Department of Health and Human Services. The investigators have adhered to J Acquir Immune Defic Syndr. 2015;68 Suppl 2:S107 13. the policies for protection of human subjects as prescribed in AR-7. 20. Kayode BO, Mitchell A, Ndembi N, Kokogho A, Ramadhani HO, Adebajo S, et al. Retention of a cohort of men who have sex with men and transgender women at risk for and living with HIV in Abuja and Lagos, Nigeria: a longitudinal analysis. J Int AIDS Soc. 2020;23(S6):e25592 REFERENCES 21. Jones MU, Ramadhani HO, Adebajo S, Gaydos CA, Kokogho A, Baral SD, 1. Ramadhani HO, Liu H, Nowak RG, Crowell TA, Ndomb T, Gaydos C, et al. et al. Seizing opportunities for intervention: changing HIV-related knowledge Sexual partner characteristics and incident rectal Neisseria gonorrhoeae and among men who have sex with men and transgender women attending trusted Chlamydia trachomatis infections among gay men and other men who have sex community centers in Nigeria. PLoS One. 2020;15:e0229533.

78 Ramadhani HO et al. Journal of the International AIDS Society 2020, 23(S6):e25599 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25599/full | https://doi.org/10.1002/jia2.25599

22. Crowell TA, Baral SD, Schwartz S, Nowak RG, Kokogho A, Adebajo S, et al. oral and injectable PrEP clinical trials among adolescent men who have sex with Time to change the paradigm: limited condom and lubricant use among Nigerian men. PLoS One. 2018;13:e0200560. men who have sex with men and transgender women despite availability and 39. Fairley CK, Law M, Chen MY. Eradicating syphilis, hepatitis C and HIV in counseling. Ann Epidemiol. 2019;31:11–19.e3. MSM through frequent testing strategies. Curr Opin Infect Dis. 2014;27(1): 23. Keshinro B, Crowell TA, Nowak RG, Adebajo S, Peel S, Gaydos CA, et al. 56–61. High prevalence of HIV, chlamydia and gonorrhoea among men who have sex 40. Burns DN, DeGruttola V, Pilcher CD, Kretzschmar M, Gordon CM, Flana- with men and transgender women attending trusted community centres in gan EH, et al. Toward an endgame: finding and engaging people unaware of Abuja and Lagos, Nigeria. J Int AIDS Soc. 2016;19:21270. their HIV-1 infection in treatment and prevention. AIDS Res Hum Retroviruses. 24. Crowell TA, Hardick J, Lombardi K, Parker Z, Kokogho A, Amusu S, et al. 2014;30(3):217–24. Asymptomatic lymphogranuloma venereum among Nigerian men who have sex 41. Raymond HF, Chen YH, Ick T, Scheer S, Bernstein K, Liska S, et al. A new with men. Sex Transm Infect. 2018;94(8):578–81. trend in the HIV epidemic among men who have sex with men, San Francisco, 25. Crowell TA, Lawlor J, Lombardi K, Nowak RG, Hardick J, Odeyemi S, et al. 2004–2011. J Acquir Immune Defic Syndr. 2013;62(5):584–9. Anorectal and urogenital mycoplasma genitalium in Nigerian men who have sex 42. Gamarel KE, Nelson KM, Stephenson R, Santiago Rivera OJ, Chiaramonte with men and transgender women: prevalence, incidence, and association with D, Miller RL. Anticipated HIV stigma and delays in regular HIV testing behaviors HIV. Sex Transm Dis. 2020;47(3):202–6. among sexually-active young gay, bisexual, and other men who have sex with 26. Nowak RG, Ndembi N, Dauda W, Jibrin P, Bentzen SM, Nnaji CH, et al. men and transgender women. AIDS Behav. 2018;22(2):522–30. Implementation of and early outcomes from anal cancer screening at a commu- 43. Pharr JR, Lough NL, Ezeanolue EE. Barriers to HIV testing among young nity-engaged health care facility providing care to nigerian men who have sex men who have sex with men (MSM): experiences from Clark count, Nevada. with men. J Glob Oncol. 2019;5:1–11. Glob J Health Sci. 2015;8(7):9–17. 27. Bassey O, Bond K, Adedeji A, Oke O, Abubakar A, Yakubu K, et al. Evalua- 44. Crowell TA, Keshinro B, Baral SD, Schwartz SR, Stahlman S, Nowak RG, tion of nine HIV rapid test kits to develop a national HIV testing algorithm in et al. Stigma, access to healthcare, and HIV risks among men who sell sex to Nigeria. Afr. J Lab Med. 2015;4(1):1–17. men in Nigeria. J Int AIDS Soc. 2017;20:21489. 28. National guidelines for HIV prevention treatment and care. National AIDS 45. Bauermeister JA, Eaton L, Meanley S, Pingel ES. Transactional sex with reg- and STI’s Control Progamme, Federal Ministry of Health. [CIted 2020 Apr 14]. ular and casual partners among young men who have sex with men in the Available from: https://aidsfree.usaid.gov/sites/default/files/2016_nigeria_natl_ Detroit metro area. Am J Mens Health. 2017;11(3):498–507. guidelines_hiv_treat_prev.pdf 46. Beyrer C, Baral SD, Walker D, Wirtz AL, Johns B, Sifakis F. The expanding 29. Chinaeke EE, Fan-Osuala C, Bathnna M, Ozigbu CE, Olakunde B, Ramad- epidemics of HIV type 1 among men who have sex with men in low- and mid- hani HO, et al. Correlates of reported modern contraceptive use among post- dle-income countries: diversity and consistency. Epidemiol Rev. 2010;32:137– partum HIV-positive women in rural Nigeria: an analysis from the MoMent 51. prospective cohort study. Reprod Health. 2019;16(1):2. 47. Beyrer C, Trapence G, Motimedi F, Umar E, Iipinge S, Dausab F, et al. Bisex- 30. WHO. Consolidated guidelines on the use of antiretroviral drugs for treat- ual concurrency, bisexual partnerships, and HIV among Southern African men ing and preventing HIV infection. Recommendations for a public health who have sex with men. Sex Transm Infect. 2010;86(4):323–7. approach. [CIted 2020 Apr 14]. Available from: http://apps.who.int/iris/bitstrea 48. Sekoni AO, Ayoola OO, Somefun EO. Experiences of social oppression m/10665/208825/1/9789241549684_eng.pdf?ua=1 among men who have sex with men in a cosmopolitan city in Nigeria. HIV AIDS. 31. Guida J, Hu L, Liu H. The impact of occupational stigma on the social net- 2015;7:21–7. works of older female sex workers: results from a three-site egocentric network 49. Tun W, Vu L, Dirisu O, Sekoni A, Shoyemi E, Njab J, et al. Uptake of HIV study in China. AIDS Patient Care STDS. 2016;30(1):1–3. self-testing and linkage to treatment among men who have sex with men 32. Silva-Santisteban A, Raymond HF, Salazar X, Villayzan J, Leon S, McFarland (MSM) in Nigeria: A pilot programme using key opinion leaders to reach MSM. W, et al. Understanding the HIV/AIDS epidemic in transgender women of Lima, J Int AIDS Soc. 2018;21(S5):e25124. Peru: results from a sero-epidemiologic study using respondent driven sampling. AIDS Behav. 2012;16(4):872–81. 33. Guadamuz TE, Wimonsate W, Varangrat A, Phanuphak P, Jommaroeng R, McNicholl JM, et al. HIV prevalence, risk behavior, hormone use and surgical SUPPORTING INFORMATION history among transgender persons in Thailand. AIDS Behav. 2011;15(3):650–8. 34. Kimani M, van der Elst EM, Chiro O, Oduor C, Wahome E, Kazungu W, et al. PrEP interest and HIV-1 incidence among MSM and transgender women Additional Supporting Information may be found in the online in coastal Kenya. J Int AIDS Soc. 2019;22(6):e25323. version of this article: 35. Mofenson LM, Cotton MF. The challenges of success: adolescents with peri- Table S1. Association between incident HIV and age: compar- natal HIV infection. J Int AIDS Soc. 2013;16:18650. ison between MSM and TGW in Abuja and Lagos, Nigeria, 36. Mutchler MG, McDavitt B. ’Gay boy talk’ meets ’girl talk’: HIV risk assessment assumptions in young gay men’s sexual health communication with best 2013 to 2019 friends. Health Educ Res. 2011;26(3):489–505. Table S2. Association between incident gonorrhea and age: 37. Baral SD, Ketende S, Schwartz S, Orazulike I, Ugoh K, Peel SA et al. Evalu- comparison between MSM and TGW in Abuja and Lagos, ating respondent-driven sampling as an implementation tool for universal cover- Nigeria, 2013 to 2019 age of antiretroviral studies among men who have sex with men living with HIV. Table S3. Association between incident chlamydia and age: J Acquir Immune Defic Syndr. 2015;68 Suppl 2:S107–13 38. Fisher CB, Fried AL, Ibrahim Puri L, Macapagal K, Mustanski B. "Free Test- comparison between MSM and TGW in Abuja and Lagos, ing and PrEP without Outing Myself to Parents:" motivation to participate in Nigeria, 2013 to 2019

79 Twahirwa Rwema JO, et al. Journal of the International AIDS Society 2020, 23(S6):e25604 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25604/full | https://doi.org/10.1002/jia2.25604

RESEARCH ARTICLE HIV infection and engagement in HIV care cascade among men who have sex with men and transgender women in Kigali, Rwanda: a cross-sectional study Jean Olivier Twahirwa Rwema1,§ , Carrie E Lyons1 , Sara Herbst5, Benjamin Liestman1, Julien Nyombayire2, Sosthenes Ketende1, Amelia Mazzei2, Oluwasolape Olawore1 , Sabin Nsanzimana3, Placidie Mugwaneza3, Aflodis Kagaba4, Patrick S Sullivan5 , Susan Allen5, Etienne Karita2 and Stefan D Baral1 §Corresponding author: Jean Olivier Twahirwa Rwema, Department of Epidemiology, Key Populations Program, Center for Public Health and Human Rights, Johns Hopkins Bloomberg School of Public Health, 615 N Wolfe Street E 7133 Baltimore, Maryland 21205, USA. Tel: +1 443 972 0243. ([email protected])

Abstract Introduction: Given intersecting biological, network and structural risks, men who have sex with men (MSM) and transgender women (TGW) consistently have a high burden of HIV. Although MSM are a key population in Rwanda, there are limited epidemiologic data to guide programming. This study aimed to characterize HIV prevalence and care cascade among MSM and TGW in Kigali. Methods: MSM and TGW ≥ 18 years were recruited using respondent-driven sampling (RDS) from March–August 2018 in Kigali. Participants underwent a structured interview including measures of individual, network and structural determinants. HIV and sexually transmitted infections (STI) including syphilis, Neisseria gonorrhoea (NG) and Chlamydia trachomatis (CT) were tested. Viral load was measured for MSM living with HIV. Robust Poisson regression was used to characterize the determinants of HIV infection and engagement in the HIV treatment cascade. Results: A total of 736 participants were enrolled. The mean age was 27 years (range:18 to 68) and 14% (106) were TGW. HIV prevalence was 10% (RDS-adjusted: 9.2% (95% CI: 6.4 to 12.1)). Unadjusted prevalence of any STI was 20% (147); syphilis: 5.7% (42); CT: 9.1% (67) and NG: 8.8% (65). Anticipated (41%), perceived (36%) and enacted stigmas (45%) were common and higher among TGW (p < 0.001). In multivariable RDS adjusted analysis, higher age (aPR: 1.08 (95% CI: 1.05 to 1.12)) and ever having sex with women (aPR: 3.39 (95% CI: 1.31 to 8.72)) were positively associated with prevalent HIV. Being circumcised (aPR: 0.52 (95% CI: 0.28 to 0.9)) was negatively associated with prevalent HIV infection. Overall, 61% (45/74) of respondents reported knowing their HIV-positive status. Among these, 98% (44/45) reported antiretroviral therapy use (ART); 75% (33/44) were virally suppressed using a cut-off of <200 copies/mL. Of the 29 participants who did not report any previous HIV diagnosis or ART use, 38% (11/29) were virally suppressed. Cumulatively, 59% (44/74) of all participants living with HIV were virally suppressed. Conclusions: These data show a high burden of HIV among MSM/TGW in Kigali, Rwanda. Bisexual concurrency was common and associated with prevalent HIV infection, demonstrating the need of comprehensive screening for all sexual practices and preferences in the provision of comprehensive HIV prevention services in Rwanda. Viral suppression was below the UNAIDS target suggesting poor adherence and potential ART resistance. There is a need for adherence support, screening for primary and secondary ART resistance and stigma mitigation interventions to optimize HIV-related outcomes for MSM in Rwanda. Keywords: HIV care continuum; structural determinants; men who have sex with men; Kigali; Rwanda

Additional Supporting information may be found online in the Supporting Information tab for this article.

Received 5 February 2020; Accepted 21 July 2020 Copyright © 2020 The Authors. Journal of the International AIDS Society published by John Wiley & Sons Ltd on behalf of the International AIDS Society. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

1 | INTRODUCTION homogenous among people living in those settings (1). More recent studies across Africa have consistently shown that key Traditionally, HIV epidemics in eastern and southern African populations including men who have sex with men (MSM) and countries have been described as generalized epidemics. This transgender women (TGW) have higher HIV prevalence and definition is predicated on an assumption that HIV risks are incidence compared with other adults of reproductive age

80 Twahirwa Rwema JO, et al. Journal of the International AIDS Society 2020, 23(S6):e25604 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25604/full | https://doi.org/10.1002/jia2.25604

(2-6). Data from eight countries across sub-Saharan Africa Recruitment occurred from March-August 2018 using (SSA) found a pooled HIV prevalence of 14% among cisgender respondent-driven sampling (RDS), a method used to recruit MSM and 25% among TGW (7). The higher risk of HIV infection marginalized populations (19). Two seeds, MSM who were among MSM and TGW is due in part to individual-level risks well-connected in the MSM community were recruited to initi- including condomless anal sex among serodiscordant and virae- ate recruitment. A third seed was recruited during the study mic partners, and higher prevalence of sexually transmitted to enhance recruitment of older participants. Efforts were infections (STIs) (3,8). Individual HIV risks are contextualized by made to select individuals with different characteristics to community-level and structural factors including stigma, sexual maximize sample heterogeneity. and physical violence and exposure to human rights violations, Study procedures were performed at the study site and which influence individual HIV risk factors (2,3). Furthermore, included a structured interview and biological testing. Trained structural determinants increase HIV risk by limiting the uptake nurse counsellors conducted the eligibility screening using a and provision of HIV prevention and treatment services for questionnaire and obtained signed informed consent MSM (9). Consequently, MSM generally have lower engagement (Table S1). The interview guide was based on a modified social in HIV prevention and treatment services and poorer outcomes ecological model to assess individual-, community-, network- compared to other adults in SSA (10,11). and structural-level risks of HIV risk among MSM (20). To In Rwanda, the overall HIV prevalence is 3% among repro- protect participant’s privacy, no personal identifying informa- ductive-aged adults and 2.2% among men (12). According to tion was collected. the 2019 global AIDS update, Rwanda can achieve the HIV status was determined by serial rapid testing. HIV UNAIDS 90-90-90 targets (13). Despite a successful national screening used Alere HIV Combo – Determine (Alere, Inc, HIV programme, there are limited data on the epidemiology of Waltham, MA) and confirmatory testing used HIV ½ STAT- HIV and STIs among MSM and TGW in Rwanda. In 2011, PAK (Medford, NY, USA) as per Rwandan national guidelines. Chapman et al. discussed the “lack of clear HIV policy for Viral load testing was performed for all participants biologi- MSM due in part to the denial of existence of MSM in Rwanda cally confirmed to be living with HIV at the National Refer- by some HIV experts” (14). This study did not estimate HIV ence Laboratory of Rwanda (NRL). Syphilis antibodies were prevalence, but reported sexual practices that would predis- screened using rapid plasma reagin (RPR – Santa Coloma, pose MSM to HIV and STI risks including condomless sex and Spain) and screened positive specimens were confirmed using frequent sex work (14). A 2015 study estimated an HIV preva- the Treponema Pallidum haemagglutination assay (TPHA – lence among MSM (4.8%) higher than the national average Langdorp, Belgium). Chlamydia trachomatis and Neisseria gon- (15). Furthermore, these studies described existence of bisex- orrhoea were tested using Cepheid GeneXpert platform Xpert ual partnerships among Rwandan MSM similar to settings CT/NG (Solna, Sweden) on self-collected urine and rectal around the world where same-sex practices are stigmatized swabs. Newly HIV diagnosed participants were referred to a and sustained relationships are challenged (16-18). However, health facility of their choice for ART initiation. Those who whether MSM who have sex with both men and women in tested positive for any STI were treated without charge at the Rwanda have differential HIV and STIs risks remains unknown. study site according to the Rwanda national guidelines. Additionally, the burden of STIs has mostly relied on self-re- Upon completion of study procedures, each participant was ported data and engagement in HIV care cascade of MSM and given three study coupons to recruit peers into the study. Par- TGW is not well characterized. Finally, though same-sex prac- ticipants received 3000Frw and the same amount for each eli- tices are not explicitly criminalized in Rwanda, stigma, sexual gible participant they referred to the study. This study was and physical violence against MSM have been reported but approved by the Institutional Review Board of Emory Univer- have not been comprehensively investigated as determinants sity (IRB00089599) and the Rwanda National Research Ethics of HIV infection among Rwandan MSM and TGW (14,18). committee. This study was conducted to address these knowledge gaps and inform the content and implementation of HIV prevention 2.2 | Outcomes and treatment services for MSM and TGW in Rwanda. The primary outcome was HIV prevalence. HIV-positive status 2 | METHODS was defined as a positive result on both the screening and confirmatory tests. In case of discordant results, samples were sent to NRL for ELISA and adjudication. 2.1 | Study population and procedures Engagement in HIV care and description of the progress to This was a cross-sectional behavioural and biological assess- achieve the 90-90-90 goals among MSM in Kigali were ment for MSM in Kigali, Rwanda. Eligible participants were indi- defined as the proportion of MSM who knew their HIV-posi- viduals assigned the male sex at birth, aged 18 years or older, tive status (first 90) before the study test and who were on who had lived in Kigali for at least three months before the ART (second 90). Viral suppression (third 90) was based on study, and who reported anal sex with a man in the preceding laboratory testing and was defined as a viral load of 12 months. Community engagement leveraged existing connec- <200 copies/mL as per the national guidelines (21). tions of Projet San Francisco (PSF) and Health Development Initiative (HDI) with 10 MSM and transgender associations in 2.3 | Covariates of interest Kigali. These community members were hired as study staff, were involved in the translation of the survey instrument in A list of individual and structural determinants of HIV and Kinyarwanda to ensure the use of appropriate local language, their hypothesized association with HIV infection among community mobilization and selection of the study site. MSM/TGW is in the Figure S1.

81 Twahirwa Rwema JO, et al. Journal of the International AIDS Society 2020, 23(S6):e25604 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25604/full | https://doi.org/10.1002/jia2.25604

Table 1. Demographic characteristics of MSM and TGW 2.4 | Individual determinants enrolled in Kigali, Rwanda 2018 (N = 736) Individual determinants were demographic, biological, behavioural and health-related factors (Tables 1,2,3). Gender iden- Crude RDS-Adjusted % tity was self-reported using a two-step instrument and Characteristic N % (95% CI) dichotomized into cisgender-MSM or TGW based on gender assignment at birth and current gender identity (22). Sexual Age in years preference was self-reported and categorized as homosexual, 18 to 24 335 45.6 48.9 (43.8 to 54.1) bisexual and heterosexual. Biological and health-related factors 25 to 34 295 40.0 35.9 (30.9 to 41.1) included STI infection, circumcision and mental health status. >35 106 14.4 15.2 (11.5 to 18.5) STI diagnosis was determined by biological testing and circum- Education cision was self-reported. Depression was assessed using the Never attended school 24 3.3 4.5 (2.0 to 6.9) Patient Health Questionnaire-9 (PHQ-9) (23,24) and categorized using established cut-points: none (0 to 4), mild (5 to 9), Primary 185 25.2 30.3 (25.6 to 35.1) moderate (10 to 14), moderately severe (15 to 19) and severe Some secondary 245 33.2 31.5 (26.7 to 36.4) depression (≥20). Secondary or above 282 38.3 33.7 (28.7 to 38.6) Condom use with male partners was assessed by partner Marital status type (regular vs. casual male partners), time scale (last sex and Single/Never married 653 88.7 87.4 (83.8 - 90.9) last six months) and anal sex type (receptive or insertive). Cohabitating with male 27 3.7 3.8 (1.7 - 5.8) Consistent condom use (CCU) was defined as having always partner used condoms during anal sex with male partners in the last Cohabitating/married with 24 3.3 3.3 (1.4 - 5.2) six months. For participants who also reported sex with female partner women, frequency and condom use during anal sex with their Divorced/Separated/Widow 32 4.3 5.5 (2.9 - 8.1) female partners was assessed. Hazardous alcohol use was Occupation screened using the AUDIT-C (25,26), with a score of four indicating hazardous alcohol use. Unemployed 130 17.6 15.1 (11.4 - 18.8) Student 89 12.1 12.1 (8.4 - 15.8) Employed/Self employed 517 70.3 72.8 (68.1 - 77.6) 2.5 | Structural determinants Monthly Income (Frw) Structural-level risks measured stigma in healthcare settings, Less than 50,000 482 65.5 70.3 (65.5 - 75.1) from family and friends, uniformed officers and in the commu- 50,000 to 100,000 178 24.2 23.6 (18.8 - 27.9) nity. These scales have been validated to assess stigma among Over 100,000 76 10.3 6.3 (4.3 - 8.3) MSM in other settings (27). Scores for anticipated stigma, Self-reported gender identity enacted stigma and perceived stigma were dichotomized at 1, Cisgender Men who Sex with 630 85.6 89.4 (86.4 to 92.2) where a score of ≥1 for a given stigma scale was considered Men as experiencing that type of stigma (Table 3). The complete Transgender 106 14.4 10.6 (7.8 to 13.5) list of the stigma-related questions and the corresponding scales is in the Table S2. Other structural factors included Self-reported sexual preferences community level access to condom and condom-compatible Gay or Homosexual 475 64.6 58.1 (52.8 - 63.4) lubricants. Bisexual 227 30.8 35.6 (30.4 - 40.8) Heterosexual 34 4.6 6.3 (3.7 - 8.8) 2.6 | Statistical analyses Sex with women Never 187 25.4 25.6 (20.7 - 30.4) Crude numbers and proportions were calculated for all covari- Yes, but not in the last 279 37.9 36.7 (31.7 - 41.6) ates of interest. RDS-adjusted proportions with 95% confi- 12 months dence intervals (CIs) were calculated for sociodemographic Yes, in the last 12 months 270 36.7 37.7 (32.9 - 42.6) and outcome variables (Tables 1,3) using Stata’s RDS II esti- Engagement in sex work mator package (28). Pearson Chi-squared tests were used to Never 453 61.7 68.9 (64.3 - 73.7) compare participants who have sex with men only (MSMO) and those who report sex with both men and women Ever provided sex acts in 171 23.2 22.0 (17.8 - 26.2) (MSMW) (Table 2); and Cisgender MSM and TGW (Table S3). exchange for money Given the high HIV prevalence (>10%), Poisson regression Main source of income in the 112 15.1 9.0 (6.5 - 11.5) with robust variance estimation was used to model the associ- previous year ation between the outcome and covariates as log-binomial models failed to converge. Bivariable analyses estimated RDS, respondent-driven sampling. prevalence ratios (PR) and 95%CIs. The final multivariable model used variables associated with the outcome in the 3 | RESULTS bivariable analyses at (p < 0.2). Finally, RDS adjustment of the final model was performed to estimate these associations with In total, 738 participants including the three seeds were inference to the Kigali MSM population. All analyses were per- recruited in 12 recruitment waves (Figure 1). Analyses were formed in Stata Version.14.2 (StataCorp, College Station, TX). restricted to 736 participants with complete data. The mean

82 Twahirwa Rwema JO, et al. Journal of the International AIDS Society 2020, 23(S6):e25604 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25604/full | https://doi.org/10.1002/jia2.25604

Table 2. Comparison of men who report sex with men only Table 2. (Continued) and men who report sex with both men and women on key sociodemographic, biological, behavioural and HIV/STI out- MSMO MSMW comes Kigali, Rwanda 2018 % (N) % (N) p value

MSMO MSMW Receptive anal sex % (N) % (N) p value Never 34.6 (53) 37.4 (150) 0.546 Ever 65.4 (100) 62.6 (251) Sociodemographic characteristics Sexual practices with casual male partners Age in years Insertive anal sex 18 to 24 49.7 (93) 44.3 (243) 0.033 Never 16.3 (24) 5.6 (27) 0.0001 25 to 34 41.7 (78) 39.5 (217) Ever 83.7 (123) 94.4 (455) Over 35 8.6 (16) 16.2 (89) Receptive anal sex Education Never 32.6 (48) 39.2 (189) 0.151 Primary level or less 28.3 (53) 28.6 (157) 0.991 Ever 67.4 (99) 60.8 (293) Some secondary 33.7 (63) 33.2 (182) Alcohol use Secondary or above 38 (71) 38.2 (210) Nonhazardous alcohol use 27.8 (52) 22.6 (124) 0.148 Marital status Hazardous alcohol use 72.2 (135) 77.4 (425) Single/Never married 93.6 (175) 87.1 (478) 0.000 HIV and sexually transmitted infections-related outcomes Cohabitating with male 6.4 (12) 2.7 (15) HIV infection partner Negative 93.5 (174) 88.7 (487) 0.058 Cohabitating/married with 0 (0) 4.4 (24) Positive 6.5 (12) 11.3 (62) female partner Virally suppressed (based on < 200 copies/mL) Divorced/Separated/Widow 0 (0) 5.8 (32) No 50.0 (6) 37.7 (23) 0.426 Monthly income (Frw) Yes 50.0 (6) 62.3 (38) Less than 50,000 66.8 (125) 65 (356) 0.640 STI diagnosis Over 50,000 33.2 (62) 35.0 (192) Negative 80.7 (151) 79.8 (438) 0.775 Gender identity Positive 19.3 (36) 20.2 (111) Cisgender Men who have 74.3 (139) 89.4 (491) 0.0001 Gonorrhoea Sex with Men Negative 91.4 (170) 91.1 (499) 0.88 Transgender women 25.7 (48) 10.6 (58) Positive 8.6 (16) 8.9 (49) Self-reported sexual preference Chlamydia Gay or Homosexual 86.6 (162) 57.0 (313) 0.0001 Negative 92.5 (172) 90.3 (495) 0.380 Bisexual 12.8 (24) 37.0 (203) Positive 7.5 (14) 9.7 (53) Heterosexual 0.5 (1) 6.0 (33) Syphilis Biological Negative 94.1 (175) 94.3 (518) 0.892 Circumcision Positive 5.9 (11) 5.6 (549) No 24.6 (46) 20.7 (114) 0.272 The bold values are variables that were statistically significantly different Yes 75.4 (141) 79.3 (435) among participants who report sex with women and those who did not Behavioural based on the chi‐square p values. The cutoff of significance was p <0.05. Age of first sex with male MSMO, men who have sex with men only; MSMW, men who have sex Before 19 years 67.4 (126) 49.1 (269) 0.0001 with men and women. 19 to 22 years 21.9 (41) 22.8 (125) age was 27 years [range: 18 to 68], 89% were single and Over 22 years 10.7 (20) 28.1 (154) 14% were TGW and 31% were bisexual (Table 1). Number of regular sexual partners in the last month None 31.2 (58) 45.9 (252) 0.001 3.1 | HIV prevalence 0ne to two 48.4 (90) 32.6 (179) Two to three 9.7 (18) 10.2 (56) The overall HIV prevalence was 10% (n = 74) RDS-adjusted Over three 10.7 (20) 11.3 (62) 9.2% (95% CI: 6.3 to 12.1). There was no difference in HIV Number of casual sexual partners in the last month prevalence among cis-MSM (10.2%) and TGW (9.4%) (p = 0.818). HIV prevalence was significantly higher with age: None 60.4 (113) 48.8 (267) 0.018 2% among 18- to 24-year olds, 12% among 25- to 34-year One to three 27.8 (52) 38.2 (209) olds and 30% at ≥35 years (p < 0.0001). Over three 11.8 (22) 12.9 (71) Sexual practices with regular male partners 3.2 | Individual determinants Insertive anal sex One in five participants 20% (n = 147) was diagnosed with at Never 17.7 (27) 4.2 (17) 0.0001 least one STI, 9% (n = 65) had chlamydia, 9% (65) had Ever 82.3 (126) 95.8 (384)

83 Twahirwa Rwema JO, et al. Journal of the International AIDS Society 2020, 23(S6):e25604 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25604/full | https://doi.org/10.1002/jia2.25604

Table 3. Individual and structural determinants of HIV infection among MSM and TGW in Kigali, Rwanda

Unadjusted PR with p Adjusted PR with p RDS adjusted PR p N% 95%CI value [95%CI] value [95%CI] value

Sociodemographic characteristics Age in years Age (from 18 years of age) 736 1.06 (1.05 to 1.07) 0.0001 1.08 (1.06 to 1.12) 0.0001 1.08 (1.05 to 1.12) 0.0001 District of residence Gasabo 89 12.2 Ref Ref Ref Ref Ref Kicukiro 162 21.9 0.59 (0.27 to 1.30) 0.196 0.57 (0.26 to 1.25) 0.163 0.96 (0.32 to 2.89) 0.950 Nyarugenge 485 65.9 0.86 (0.46 to 1.57) 0.605 0.72 (0.41 to 1.25) 0.240 1.15 (0.52 to 2.58) 0.725 Education Primary level or less 210 28.5 Ref Ref Ref Ref Ref Some secondary 244 33.2 0.41 (0.23 to 0.70) 0.001 0.93 (0.52 to 1.65) 0.806 0.76 (0.33 to 1.75) 0.528 Secondary or above 282 38.3 0.43 (0.26 to 0.72) 0.001 1.28 (0.65 to 2.51) 0.479 2.01 (0.82 to 4.90) 0.124 Marital status Single/Never married 653 88.7 Ref Ref Ref Ref Ref Cohabitating with male 27 3.7 0.46 (0.06 to 3.18) 0.429 0.41 (0.05 to 3.01) 0.380 0.41 (0.04 to 3.86) 0.440 partner Cohabitating/married with 24 3.3 3.46 (1.76 to 6.79) 0.0001 0.54 (0.21 to 1.35) 0.188 0.35 (0.11 to 1.23) 0.104 female partner Divorced/Separated/Widow 32 4.3 4.08 (2.37 to 7.02) 0.0001 0.85 (0.41 to 1.76) 0.661 0.97 (0.38 to 2.47) 0.950 Monthly Income (Frw) Less than 50,000 481 65.4 Ref Ref Ref Ref Ref Ref Over 50,000 254 34.6 0.61 (0.36 to 1.01) 0.056 0.56 (0.33 to 0.96) 0.036 0.36 (0.16 to 0.82) 0.015 Self-reported gender identity Cis MSM 630 85.6 Ref Ref Transgender woman 106 14.4 0.92 (0.49 to 1.75) 0.819 Self-reported sexual preference Gay or Homosexual 475 64.6 Ref Ref Bisexual 227 30.8 1.03 (0.65 to 1.63) 0.91 Heterosexual 34 4.6 0.29 (0.04 to 2.07) 0.219 Biological and other health-related factors STI diagnosis No 589 80.1 Ref Ref Ref Ref Ref Ref Yes 147 19.9 1.81 (1.14 to 2.86) 0.011 1.45 (0.95 to 2.19) 0.081 1.56 (0.94 to 2.58) 0.08 Mental health Not depressed 475 64.6 Mild depression 194 26.3 1.17 (0.72 to 1.84) 0.514 Moderate depression 48 6.5 0.86 (0.32 to 2.29) 0.767 Moderately severe depression 12 1.6 0.86 (0.13 to 5.7) 0.878 Severe major 7 1 1.72 (0.28 to 10.5) 0.555 Circumcision No 161 21.8 Ref Ref Ref Ref Ref Ref Yes 575 78.2 0.43 (0.28 to 0.67) 0.0001 0.74 (0.46 to 1.21) 0.229 0.52 (0.28 to 0.97) 0.041 Behavioural Age of first sex with male Before 19 years 396 53.8 Ref Ref Ref Ref Ref 19 to 22 years 166 22.6 1.46 (0.81 to 2.66) 0.21 1.13 (0.65 to 1.95) 0.665 0.90 (0.39 to 2.05) 0.805 Over 22 years 174 23.6 2.79 (1.72 to 4.54) 0.0001 1.28 (0.74 to 2.21) 0.375 0.87 (0.43 to 1.75) 0.688 CCU with male partners in the last six months Non-consistent condom use 642 87.3 Ref Ref Ref Ref Ref Consistent condom use 94 12.7 0.60 (0.27 to 1.35) 0.219 0.72 (0.33 to 1.58) 0.419 1.15 (0.53 to 2.48) 0.719 Lubricant use Never 95 12.9 Ref Ref Ever used lubricant 641 87.1 0.95 (0.50 to 1.78) 0.87

84 Twahirwa Rwema JO, et al. Journal of the International AIDS Society 2020, 23(S6):e25604 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25604/full | https://doi.org/10.1002/jia2.25604

Table 3. (Continued)

Unadjusted PR with p Adjusted PR with p RDS adjusted PR p N% 95%CI value [95%CI] value [95%CI] value

Number of regular sexual partners in the last month None 310 42.2 Ref Ref One to two 269 36.6 1.28 (0.78 to 2.09) 0.327 Two to three 74 10.1 1.08 (0.49 to 2.39) 0.838 Over three 82 11.1 1.26 (0.62 to 2.57) 0.526 Number of casual sexual partners in the last month None 379 51.8 Ref Ref Ref Ref Ref Ref One to three 262 35.6 1.00 (0.62 to 1.63) 0.985 1.25 (0.76 to 2.04) 0.383 0.96 (0.49 to 1.90) 0.910 Over three 93 12.6 1.47 (0.81 to 2.66) 0.201 2.51 (1.42 to 4.41) 0.001 2.13 (0.86 to 5.30) 0.102 Sex with women Never 187 25.5 Ref Ref Ref Ref Ref Yes, but not in the last 279 37.9 1.88 (1.00 to 3.55) 0.049 1.32 (0.70 to 2.48) 0.386 3.39 (1.31 to 8.72) 0.011 12 months Yes, in the last 12 months 270 36.6 1.61 (0.89 to 3.08) 0.153 1.33 (0.66 to 2.69) 0.422 4.3 (1.46 to 12.69) 0.008 Professional sex worker in the last 12 months No 624 84.8 Ref Ref Ref Ref Ref Yes 112 15.2 0.58 (0.27 to 1.23) 0.159 0.66 (0.31 to 1.44) 0.298 0.89 (0.28 to 2.87) 0.849 Alcohol use Nonhazardous alcohol use 175 23.8 Ref Ref Hazardous alcohol use 561 76.2 0.97 (0.58 to 1.60) 0.907 Structural Perceived stigma score No 474 64.5 Ref Ref Yes 262 35.5 1.16 (0.75 to 1.81) 0.496 Anticipated stigma score No 438 59.4 Ref Ref Yes 299 40.6 0.94 (0.60 to 1.46) 0.791 Enacted stigma score No 407 55.4 Ref Ref Yes 329 44.6 1.05 (0.68 to 1.62) 0.820 Access to condom in the last six months Easy 501 68.1 Ref Ref Difficult 136 18.5 1.09 (0.62 to 1.93) 0.748 Not applicable 99 13.4 1.39 (0.79 to 2.49) 0.252 Access to lubricant in the last six months Easy 372 58.1 Ref Ref Difficult 163 25.4 0.72 (0.39 to 1.31) 0.288 Not applicablea 106 16.5 0.90 (0.54 to 1.49) 0.692

Bold values are for variables that were found to be significantly associated with prevalent HIV infection in the regression analyses. Cis-MSM, cisgender men who have sex with men; CCU, consistent condom use; PR, prevalence ratios. aParticipants who did not try to get lubricants in the last six months. gonorrhoea and 6% (42) had syphilis. Most participants self- anal sex and 63% reported receptive anal sex; and CCU in reported being circumcised (78%). Overall, 26% reported mild the previous six months was 37% and 33 %, respectively, for depression and 8% reported moderate or worse depression insertive and receptive anal sex. Among those reporting casual (Table 3). partners, insertive anal sex was more common than receptive Overall, 54% reported sexual debut with men before the anal sex (92% vs 62%); and CCU was 38% and 36%, respec- age of 19. Multiple sexual partnerships in the preceding tively, for insertive and receptive anal sex with casual part- month included 21% reporting ≥2 regular partners; and 36% ners. reporting 1 to 3 casual male sexual partners. Among partici- Three quarters of participants reported ever having sex pants reporting regular partners, 92% engaged in insertive with women, among whom 49% reported sex with women in

85 Twahirwa Rwema JO, et al. Journal of the International AIDS Society 2020, 23(S6):e25604 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25604/full | https://doi.org/10.1002/jia2.25604

Figure 1. Respondent-Driven Sampling plot demonstrating recruitment networks and HIV status among men who have sex with men and transgender women in Kigali, Rwanda. The blue nodes indicate participants living with HIV and the red nodes are for HIV-negative participants. The size of the node is proportional to the network size, participants who know more MSM and TGW have larger nodes compared to participants who are less connected in the MSM/TGW community in Kigali

the prior year. Overall, 38% of the 226 who reported ever (p < 0.001) compared to cisgender MSM. There were no dif- having a regular female partner and 30% of 482 who ferences in HIV and STI outcomes (Table S3). reported casual female partners reported anal sex with their female partners. Overall, condom use at last anal sex was 50% 3.5 | Men who have sex with men only and men and 54%, respectively, with regular and casual female part- who have sex with both men and women ners. Overall, 87% ever used lubricants during sex, including com- MSMW differed from those reporting MSMO on several char- mercial water-based lubricants (63%) and petroleum jelly/ acteristics including age, marital status, gender identity, sexual Vaseline (30%). Among respondents, 38% disclosed ever pro- preference and practices. MSMW reported a higher age of viding sexual acts in exchange for money and 15% reported sexual debut with male partners (<0.001), higher number of sex work as their main source of income in the previous year. casual male partners (p = 0.018) but lower number of regular Hazardous alcohol use was common (76%). partners (p = 0.001). MSMW reported more insertive, but similar levels of receptive anal sex with male partners. How- ever, there were no significant differences with condom use 3.3 | Structural determinants with any male partner type or STIs apart from HIV (Table 2). Overall, 13% reported ever feeling excluded from family activities and 20% reported discriminatory remarks from family 3.6 | HIV care cascade members. Additionally, 28% of participants reported feeling scared to be in public places and 31% experienced verbal Most participants (91%) reported ever being tested for HIV. harassment. Some participants reported past physical (16%) Of the 74 participants who tested positive for HIV, 61% (45) or sexual (14%) violence. Based on the stigma scales scores, reported knowing their status before the study. Of those, 44 41% reported anticipated stigma, 36% perceived stigma and (98%) reported being on ART and 33 (75%) were virally sup- 45% reported enacted stigma. pressed. Among the 29 participants living with HIV, unaware of HIV status prior to the study and reported not being on ART, 38% 3.4 | Cisgender and transgender individuals (11/29) were virally suppressed. Hence, the total proportion TGW were different from cisgender MSM in age distribution of viral suppression among all participants living with HIV was (p = 0.038), marital status (p = 0.075) and reported more 59% (44/74) (Figure 2). depressive symptoms (p < 0.0001). TGW reported younger age of sexual debut with men (p = 0.005), engaged more in 3.7 | Factors associated with prevalent HIV sex work (p = 0.0001) and reported more lubricant use com- infection pared to cisgender MSM (p = 0.007). There were no significant differences in condom or alcohol use. TGW experienced In the multivariable RDS adjusted analysis, age higher levels of anticipated, perceived and enacted stigmas (aPR = 1.08;95%:CI 1.05 to 1.12), ever having had sex with

86 Twahirwa Rwema JO, et al. Journal of the International AIDS Society 2020, 23(S6):e25604 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25604/full | https://doi.org/10.1002/jia2.25604 women (aPR = 3.39; 95% CI: 1.31 to 8.72) and sex with This study found a high level of concurrent partnerships women in the 12 months preceding the study (aPR = 4.3; with men and women and an increased risk of HIV infection 95% CI: 1.46 to 12.69) remained significantly associated with among MSMW. This highlights the need for prevention inter- prevalent HIV infection. Having a prevalent STI infection was ventions tailored to the needs of MSMW and screening for all marginally significantly associated with HIV (aPR = 1.56; 95% sexual practices in health encounters. Bisexual partnerships CI: 0.94 to 2.58). Being circumcised (aPR = 0.52; 95% CI: are common among MSM across SSA and around the world 0.28 to 0.97) and having income >50,000 Frw (PR = 0.56; (16,32,33), however, evidence on HIV risk associated with 95% CI: 0.34 to 0.93) was significantly inversely associated bisexual relationships among MSM is mixed. A study in with HIV infection. (Table 3). Mozambique found that MSMW had lower odds of HIV infection compared to MSMO (17). A systematic review focused on China found a higher prevalence of HIV among MSMW 4 | DISCUSSION compared to MSMO (34), whereas another systematic review focused on Asia found no difference in HIV prevalence These data show that MSM and TGW have a higher burden between those two groups (32). For Rwanda, it is difficult to of HIV compared to adult men in Kigali, and across Rwanda conclude the reason that HIV among MSMW was higher (29). They also highlighted factors associated with HIV infec- based on these cross-sectional data. This might be explained tion among MSM in Kigali. Older age, low income, STI infec- in part by the differences in age, age of sexual debut, gender tion and sex with women were positively associated with HIV identity, sexual preference and practices among MSMW and infection, whereas circumcision was negatively associated with MSMO. MSMW were older and reported more male casual HIV. These findings are consistent with findings from other partners in the previous month. Furthermore, HIV acquisition studies across SSA demonstrating a high burden of HIV among MSMW may be from their female sexual partners among MSM and TGW (7,30,31). Furthermore, HIV preva- through unprotected vaginal and anal sex as this study lence in this study is higher than previously estimated for observed that anal sex with female partners is common and MSM in Rwanda (15). condom use is limited. It may be that MSMW engage in sex

80 100% N=74 70

60 61% 98% 50 45/74 44/45 75% 40 33/44 30

0 Living with HIV Knew their HIV On ART before the Virally suppressed posiƟve status before study among those among those on ART the study who knew their status

80 100% N=74 70

60 61% 59% 59% 50 (45/74) (44/74) (44/74) 40

0 Living with HIV Knew their HIV On ART before the Virally suppressed posiƟve status before study among all living among all living with the study with HIV HIV

Figure 2. (A) HIV care cascade among men who have sex with men and transgender women who knew their HIV-positive status and were on ART before the study in Kigali, Rwanda. (B) HIV care cascade among all men who have sex with men and transgender women living with HIV in Kigali, Rwanda

87 Twahirwa Rwema JO, et al. Journal of the International AIDS Society 2020, 23(S6):e25604 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25604/full | https://doi.org/10.1002/jia2.25604 with female partners with higher risk of HIV; or that MSMW strategies including self-testing could improve identification of may access MSM-friendly prevention services less often than MSM living with HIV and their linkage to care. Furthermore, MSMO. Further studies, including those leveraging phyloge- given that a quarter of participants reporting ART use were netic data, are needed to help understand HIV transmission viraemic, HIV drug resistance testing and adherence pro- dynamics in different sexual networks. There is a demon- grammes could improve quality of care. Finally, longitudinal strated need to reinforce current prevention strategies for studies are needed to understand the implication of these MSM in Rwanda and implement and scale up HIV Pre-Expo- findings on HIV incidence among MSM in Kigali. sure Prophylaxis (PrEP) given the low CCU, multiple sexual This study has several limitations. The cross-sectional nature partnerships, high prevalence of STIs and the prevalence of of the study limits the ability to make any temporal relation- viraemia found among MSM and TGW in this study. ships between HIV and the determinants of interest. The small The association of circumcision with lower prevalence of HIV sample of TGW within the study was underpowered to con- infection is consistent with findings from a recent systematic duct multivariable analyses stratified by gender identity. The review and meta-analysis showing that circumcision was associ- RDS-II estimator relies on individual reported network size to ated with 42% decreased odds of HIV infection among MSM in calculate sampling weights that are used to estimate popula- low- and middle-income countries (35). This suggests that vol- tion estimates. Thus, any biases in reporting individuals’ net- untary medical male circumcision (VMMC) could be an addi- work size would have introduced bias in the RDS adjusted tional prevention strategy for MSM as well (36). In Rwanda, analyses. The sensitive nature of questions assessing HIV VMMC is currently implemented for prevention of HIV among infection and/or risk makes it subject to social-desirability bias. men and MSM can benefit from this prevention tool. This limited the ability to precisely estimate the proportion of In this study, one in six participants self-identified as trans- MSM who knew their HIV status and who were engaged in gender or a woman, showing that transgender individuals con- care. There is also potential for recall bias for questions asking stitute a large portion of the MSM/TGW population in Kigali. about longer time frames including those on condom use and This proportion of TGW is comparable to findings from other number and type of partners. However, in this case, the risk is African countries (7). Consistent with findings worldwide that the data reported are an underestimate of the actual (7,37,38), TGW in Kigali were more likely to engage in sex numbers. Finally, restricting recruitment to Kigali city makes work and experience more stigma and mental health issues the findings less generalizable to other areas of the country. compared to cis-MSM. These structural factors are known to potentiate the risk of HIV acquisition by preventing access to HIV prevention and treatment services (3,20). Structural inter- 5 | CONCLUSION ventions including stigma and violence mitigation interventions may support HIV programming for MSM/TGW in Rwanda. These data reinforce that MSM/TGW are at high risk for HIV The lack of association of gender identity with HIV and STIs and are currently underserved by HIV prevention and treat- may be explained by the small number of transgender individ- ment services in Rwanda. Unmet HIV prevention needs uals in the study resulting in limited power to detect differ- included high STI prevalence, limited consistent use of con- ences. However, the differences found on known HIV risk doms and of condom compatible lubricants. From the perspec- factors suggest they may have a higher risk of HIV acquisition tive of treatment, the level of viral load suppression was and urge to consider their needs in HIV programming in below the third target of UNAIDS 90-90-90 suggesting Rwanda. A larger study focused on TGW in Rwanda could opportunities for optimization of linkage and retention ser- help understand the epidemiology of HIV in this group. vices. Moreover, the disconnect between reported treatment Among participants living with HIV, under 60% were virally status and viral suppression may suggest poor adherence and/ suppressed suggesting suboptimal ART linkage, retention and or significant primary and secondary ART resistance though adherence; as well as potential acquired and transmitted ART this was not assessed. These data suggest the utility of novel resistance. The high proportion of viral suppression among HIV treatment and prevention services including decentralized MSM who reported not knowing their status or using ART, HIV testing strategies and pre-exposure prophylaxis for MSM/ suggests that more MSM participants knew their HIV status TGW in Rwanda. and were on ART than those who self-reported. This misreporting of HIV status and engagement in care has been AUTHORS’ AFFILIATIONS described previously (10). Although, there is a possibility that 1 some are elite controllers, this proportion is much higher than Department of Epidemiology, Key Populations Program, Center for Public Health and Human Rights, Johns Hopkins Bloomberg School of Public Health, what would be expected based on several studies of elite con- Baltimore, MD, USA; 2Projet San Francisco, Kigali, Rwanda; 3HIV and AIDS Divi- trollers in Africa (39-41). Thus, although bio-behavioural sur- sion, Rwanda Biomedical Center, Kigali, Rwanda; 4Health Development Initiative, veys are sometimes the only way to assess MSM and other Kigali, Rwanda; 5Emory University, Atlanta, GA, USA KPs engagement in care due to the difficulty of identifying them in routine HIV programme data (42,43), these data reit- COMPETING INTERESTS erate the shortcomings of relying on self-reported data to The authors declare no competing interests to disclose. estimate HIV status and ART coverage and the potential utility of testing for plasma ART levels in addition to viral load AUTHORS’ CONTRIBUTIONS (44). Despite the misclassification, the high proportion of par- SDB, SA, EK, PS, AK, SN and PM conceived and designed the study. JOTR, BL, ticipants living with HIV who were not virally suppressed is JN, SH, AM and CEL oversaw implementation and data collection. JOTR, OO evidence of gaps in the first two UNAIDS targets. Conse- and SK analysed the data. JOTR wrote the first draft of the paper. All authors quently, promotion and implementation of novel HIV testing reviewed, edited and approved the manuscript.

88 Twahirwa Rwema JO, et al. Journal of the International AIDS Society 2020, 23(S6):e25604 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25604/full | https://doi.org/10.1002/jia2.25604

ABBREVIATIONS 13. UNAIDS. Global AIDS update. UNAIDS. 2019. ART, Antiretroviral treatment; CCU, Consistent Condom Use; CT, Chlamydia 14. Chapman J, Koleros A, Delmont Y, Pegurri E, Gahire R, Binagwaho A. High Trachomatis; ELISA, Enzyme-linked Immunosorbent Assay; Frw, Francs Rwan- HIV risk behavior among men who have sex with men in Kigali, Rwanda: making – dais; HDI, Health Development Initiative; HIV, Human Immunodeficiency virus; the case for supportive prevention policy. AIDS Care. 2011;23(4):449 55. IQR, Interquartile range; MSM, men who have sex with men; MSMW, men who 15. Ntale RS, Rutayisire G, Mujyarugamba P, Shema E, Greatorex J, Frost SDW, have sex with men and women; NG, Neisseria Gonorrhoea; NRL, National refer- et al. HIV seroprevalence, self-reported STIs and associated risk factors among ence Laboratory of Rwanda; PrEP, Pre-exposure Prophylaxis; PSF, Projet San men who have sex with men: a cross-sectional study in Rwanda, 2015. Sex – Francisco; RDS, Respondent-Driven Sampling; RPR, Rapid Plasma Reagin; SSA, Transm Infect. 2019;95(1):71 4. sub-Saharan Africa; STI, Sexually Transmitted Infection; TGW, Transgender 16. Beyrer C, Trapence G, Motimedi F, Umar E, Iipinge S, Dausab F, et al. Bisex- women; TPHA, Treponema Pallidum Haemagglutination Assay; UNAIDS, Joint ual concurrency, bisexual partnerships, and HIV among Southern African men – United Nations Programme on HIV/AIDS. who have sex with men. Sex Transm Infect. 2010;86(4):323 7. 17. Isabel Sathane RH, Young Peter, Inguane Celso, Nala Rassul, Miranda Angelica E, Tim Lane H, et al. Risk factors associated with HIV among men who ACKNOWLEDGEMENTS have sex only with men and men who have sex with both men and women in – The authors thank all members of MSM and transgender associations operating three urban areas in Mozambique. AIDS Behav. 2016;20:2296 308. in Kigali city who helped during the study design and implementation. We also 18. Adedimeji A, Sinayobye JD, Asiimwe-Kateera B, Chaudhry J, Buzinge L, thank all the study participants for their time, commitment and contribution to Gitembagara A, et al. Social contexts as mediator of risk behaviors in Rwandan this study men who have sex with men (MSM): implications for HIV and STI transmission. PLoS One. 2019;14:e0211099. 19. Heckathorn DD, Wangroongsarb P, Thwing J, Eliades J, Satimai W, Delacol- FUNDING lette C, et al. Respondent-driven sampling: a new approach to the study of hid- – This study was supported by the Center for Disease Control and Prevention den populations. Soc Probl. 1997;44(2):174 99. through PEPFAR COAG NU2GGH001443 and supervised by CDC – Rwanda 20. Baral S, Logie CH, Grosso A, Wirtz AL, Beyrer C. Modified social ecological AIDS office. SB’s effort was supported in part by a grant from the National model: a tool to guide the assessment of the risks and risk contexts of HIV epi- Institutes of Mental Health (R01MH110358). PS’s effort was supported in part demics. BMC Public Health. 2013;13(1):482. by the Emory University Center for AIDS Research (P30AI050409). The con- 21. UNAIDS. Global AIDS update 2017. Ending AIDS - progress towards the – – tent is solely the responsibility of the authors and does not necessarily repre- 90 90 90 targets. Geneva, Switzerland: UNAIDS; 2017. sent the official views of the NIH. 22. Tordoff DM, Morgan J, Dombrowski JC, Golden MR, Barbee LA. Increased Ascertainment of transgender and non-binary patients using a 2-step versus 1- Step gender identity intake question in an STD Clinic setting. Sex Transm Dis. 2019;46(4):254–9. REFERENCES 23. Kroenke K, Spitzer RL, Williams JBW. The PHQ-9. J Gen Intern Med. 1. Baral S, Phaswana-Mafuya N. Rewriting the narrative of the epidemiology of 2001;16(9):606–13. HIV in sub-Saharan Africa. SAHARA-J: journal of Social Aspects of HIV/AIDS. 24. Monahan PO, Shacham E, Reece M, Kroenke K, Ong’or WO, Omollo O, SAHARA J. 2012;9(3):127–30. et al. Validity/reliability of PHQ-9 and PHQ-2 depression scales among adults 2. Baral SD, Grosso A, Holland C, Papworth E. The epidemiology of HIV among living with HIV/AIDS in Western Kenya. J Gen Intern Med. 2009;24(2):189–97. men who have sex with men in countries with generalized HIV epidemics. Cur- 25. Frank D, DeBenedetti AF, Volk RJ, Williams EC, Kivlahan DR, Bradley KA. rent Opinion in HIV and AIDS. 2014;9(2):156–67. Effectiveness of the AUDIT-C as a screening test for alcohol misuse in three 3. Beyrer C, Baral SD, van Griensven F, Goodreau SM, Chariyalertsak S, Wirtz race/ethnic groups. J Gen Intern Med. 2008;23(6):781–7. AL, et al. Global epidemiology of HIV infection in men who have sex with men. 26. Bush KKD, McDonell MB, Fihn SD, Bradley KA, for the Ambulatory Care The Lancet. 2012;380(9839):367–77. Quality Improvement Project (ACQUIP). The AUDIT alcohol consumption ques- 4. Keshinro B, Crowell TA, Nowak RG, Adebajo S, Peel S, Gaydos CA, et al. tions (AUDIT-C): an effective brief screening test for problem drinking. Arch High prevalence of HIV, chlamydia and gonorrhoea among men who have sex Intern Med. 1998;158(16):1789–95. with men and transgender women attending trusted community centres in 27. Ashley Lynn Grosso SK, Stahlman S, Ky-Zerbo O, Ouedraogo HG, Kouanda Abuja and Lagos, Nigeria. J Int AIDS Soc. 2016;19:21270. S, Samadoulougou C, et al. Development and reliability of metrics to character- 5. Mmbaga EJ, Moen K, Leyna GH, Mpembeni R, Leshabari MT. HIV Preva- ize types and sources of stigma among men who have sex with men and female lence and associated risk factors among men who have sex with men in Dar es sex workers in Togo and Burkina Faso. BMC Infect Dis. 2019;19:208. https://d Salaam, Tanzania. J Acquir Immune Defic Syndro. 2018;77(3):243–9. oi.org/10.1186/s12879-019-3693-0 6. Kunzweiler CP, Bailey RC, Okall DO, Graham SM, Mehta SD, Otieno FO. 28. Volz E, Heckathorn DD. Probability based estimation theory for respondent Factors associated with prevalent HIV infection among Kenyan MSM. J Acquir driven sampling. Journal of Offic Stat. 2008;24(1):79. Immune Defic Syndr. 2017;76(3):241–9. 29. UNAIDS. Rwanda HIV and AIDS estimates. 2018. 7. Poteat T, Ackerman B, Diouf D, Ceesay N, Mothopeng T, Odette KZ, et al. 30. Mmbaga EJ, Moen K, Leyna GH, Mpembeni R, Leshabari MT. HIV Preva- HIV prevalence and behavioral and psychosocial factors among transgender lence and associated risk factors among men who have sex with men in Dar Es women and cisgender men who have sex with men in 8 African countries: A Salaam, Tanzania. J Acquir Immune Defic Syndr. 2018;77(3):243–249. cross-sectional analysis. PLoS Med. 2017;14:e1002422. 31. Baral SD, Ketende S, Mnisi Z, Mabuza X, Grosso A, Sithole B, et al. A 8. Galvin SR, Cohen MS. The role of sexually transmitted diseases in HIV trans- cross-sectional assessment of the burden of HIV and associated individual- and mission. Nat Rev Microbiol. 2004;2(1):33–42. structural-level characteristics among men who have sex with men in Swaziland. 9. Augustinavicius JL, Baral SD, Murray SM, Jackman K, Xue Q-L, Sanchez TH, J Int AIDS Soc. 2013;16:18768 et al. Characterizing cross-culturally relevant metrics of stigma among men who 32. Bowring AL, Veronese V, Doyle JS, Stoove M, Hellard M. HIV and Sexual have sex with men across eight sub-Saharan African countries and the United risk among men who have sex with men and women in Asia: a systematic States. Am J Epidemiol. 2020;189(7):690–7. https://doi.org/10.1093/aje/kwz270 review and meta-analysis. AIDS Behav. 2016;20(10):2243–65. 10. Sandfort TGM, Dominguez K, Kayange N, Ogendo A, Panchia R, Chen YQ, 33. Lau JT, Wang M, Wong HN, Tsui HY, Jia M, Cheng F, et al. Prevalence of et al. HIV testing and the HIV care continuum among sub-Saharan African men bisexual behaviors among men who have sex with men (MSM) in China and who have sex with men and transgender women screened for participation in associations between condom use in MSM and heterosexual behaviors. Sex HPTN 075. PLoS One. 2019;14:e0217501. Transm Dis. 2008;35(4):406–13. 11. Stannah J, Dale E, Elmes J, Staunton R, Beyrer C, Mitchell KM, et al. HIV 34. Wang HY, Xu JJ, Zou HC, Reilly KH, Zhang CM, Yun K, et al. Sexual risk testing and engagement with the HIV treatment cascade among men who have behaviors and HIV infection among men who have sex with men and women in sex with men in Africa: a systematic review and meta-analysis. Lancet HIV. China: evidence from a systematic review and meta-analysis. Biomed Res Int. 2019;6(11):e769–87. 2015;2015:850132. 12. NISR. Rwanda Demographic and Health Survey 2014–15 - Final Report. 35. Yuan T, Fitzpatrick T, Ko N-Y, Cai Y, Chen Y, Zhao J, et al. Circumcision to Kigali - Rwanda. National Insitute of Statistics of Rwanda. 2016. prevent HIV and other sexually transmitted infections in men who have sex with

89 Twahirwa Rwema JO, et al. Journal of the International AIDS Society 2020, 23(S6):e25604 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25604/full | https://doi.org/10.1002/jia2.25604

men: a systematic review and meta-analysis of global data. Lancet Global Health. 44. Johnston LG, Sabin ML, Prybylski D, Sabin K, McFarland W, Baral S, et al. 2019;7(4):e436–47. The importance of assessing self-reported HIV status in bio-behavioural surveys. 36. Pintye J, Baeten JM. Benefits of male circumcision for MSM: evidence for Bull World Health Organ. 2016;94(8):605–12. action. Lancet Global Health. 2019;7(4):e388–9. 37. Stahlman S, Liestman B, Ketende S, Kouanda S, Ky-Zerbo O, Lougue M, et al. Characterizing the HIV risks and potential pathways to HIV infection among transgender women in Cote d’Ivoire, Togo and Burkina Faso. J Int AIDS Soc. 2016;19:20774. SUPPORTING INFORMATION 38. Jobson G, Tucker A, de Swardt G, Rebe K, Struthers H, McIntyre J, et al. Gender identity and HIV risk among men who have sex with men in Cape Additional Supporting Information may be found in the online Town, South Africa. AIDS Care. 2018;30(11):1421–5. 39. Alex Kayongo EG-G, Gum€ us€ Emrah, Niwaha Anxious J, Semitala Fred, version of this article: Gum€ usg€ oz€ Emrah, Kalyesubula Robert, et al. Identification of Elite and viremic controllers from a large urban hiv ambulatory center in Kampala, Uganda. J Table S1. Study inclusion criteria Acquir Immune Defic Syndr. 2018;79(3):394–8. Table S2. Items used to assess stigma experiences among 40. Sivay MV, Fogel JM, Wang J, Zhang Y, Piwowar-Manning E, Clarke W, et al. Natural control of HIV infection in young women in South Africa: HPTN 068. men who have sex with men and transgender women in Kigali, HIV Clin Trials. 2018;19(5):202–8. Rwanda 41. Kiros YK, Elinav H, Gebreyesus A, Gebremeskel H, Azar J, Chemtob D, Figure S1. Individual and structural determinants hypothe- et al. Identification and characterization of HIV positive Ethiopian elite con- sized to be associated with HIV infection among men who – trollers in both Africa and Israel. HIV Med. 2019;20(1):33 7. have sex with men and transgender women in Kigali, Rwanda. 42. Risher K, Mayer KH, Beyrer C. HIV treatment cascade in MSM, people who inject drugs, and sex workers. Curr Opin HIV AIDS. 2015;10(6):420–9. Table S3. Comparison of cis-gender MSM and transgender 43. Hladik W, Benech I, Bateganya M, Hakim AJ. The utility of population-based women on key sociodemographic, biological, behavioral, and surveys to describe the continuum of HIV services for key and general popula- HIV/STI outcomes Kigali, Rwanda, 2018. tions. Int J STD AIDS. 2016;27(1):5–12.

90 Fearon E et al. Journal of the International AIDS Society 2020, 23(S6):e25603 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25603/full | https://doi.org/10.1002/jia2.25603

RESEARCH ARTICLE Online socializing among men who have sex with men and transgender people in Nairobi and Johannesburg and implications for public health-related research and health promotion: an analysis of qualitative and respondent-driven sampling survey data Elizabeth Fearon1,§ , Adam Bourne2, Siyanda Tenza3, Thesla Palanee-Phillips3, Rhoda Kabuti4, Peter Weatherburn5, Will Nutland5, Joshua Kimani4,6 and Adrian D Smith7 §Corresponding author: Elizabeth Fearon, 15-17 Tavistock Place, London, WC1H 9SH, United Kingdom. Tel: +44 (0)20 7927 2877. ([email protected])

Abstract Introduction: There is little published literature about gay, bisexual and other men who have sex with men and transgender individuals (MSM and TG)’s use of social media in sub-Saharan Africa, despite repressive social and/or criminalizing contexts that limit access to physical HIV prevention. We sought to describe MSM and TG’s online socializing in Nairobi and Johannes- burg, identifying the characteristics of those socializing online and those not, in order to inform the development of research and health promotion in online environments. Methods: Respondent-driven sampling surveys were conducted in 2017 in Nairobi (n = 618) and Johannesburg (n = 301) with those reporting current male gender identity or male sex assigned at birth and sex with a man in the last 12 months. Online socializing patterns, sociodemographic, sexual behaviour and HIV-testing data were collected. We examined associations between social media use and sociodemographic characteristics and sexual behaviours among all, and only those HIV- uninfected, using logistic regression. Analyses were RDS-II weighted. Thirty qualitative interviews were conducted with MSM and TG in each city, which examined the broader context of and motivations for social media use. Results: Most MSM and TG had used social media to socialize with MSM in the last month (60% Johannesburg, 71% Nairobi), mostly using generic platforms (e.g. Facebook), but also gay-specific (e.g. Grindr). HIV-uninfected MSM and TG reporting risk- ier recent sexual behaviours had raised odds of social media use in Nairobi, including receptive anal intercourse (adjusted OR = 2.15, p = 0.006), buying (aOR = 2.24, p = 0.015) and selling sex with men (aOR = 2.17, p = 0.004). Evidence for these associations was weaker in Johannesburg, though socializing online was associated with condomless anal intercourse (aOR = 3.67, p = 0.003) and active syphilis (aOR = 13.50, p = 0.016). Qualitative findings indicated that while online socializing can limit risk of harm inherent in face-to-face interactions, novel challenges were introduced, including context collapse and a fear of blackmail. Conclusions: Most MSM and TG in these cities socialize online regularly. Users reported HIV acquisition risk behaviours, yet this space is not fully utilized for sexual health promotion and research engagement. Effective, safe and acceptable means of using online channels to engage with MSM/TG that account for MSM and TG’s strategies and concerns for managing online security should now be explored, as complements or alternatives to existing outreach. Keywords: men who have sex with men; transgender persons; social media; internet; HIV; sexual health

Additional Supporting information may be found online in the Supporting Information tab for this article.

Received 22 January 2020; Accepted 21 July 2020 Copyright © 2020 The Authors. Journal of the International AIDS Society published by John Wiley & Sons Ltd on behalf of the International AIDS Society. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

1 | INTRODUCTION particularly among young people, men and those with a higher education [1,2]. Gay, bisexual and other men who have sex Access to the Internet, smartphones, online platforms and with men and transgender people (MSM and TG) have been social media is rising rapidly in sub-Saharan Africa (SSA), disproportionately affected by HIV around the world, including

91 Fearon E et al. Journal of the International AIDS Society 2020, 23(S6):e25603 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25603/full | https://doi.org/10.1002/jia2.25603 in generalized epidemic settings in East and Southern Africa 2 | METHODS [3-6]. The use of social media for socializing and partner- seeking among gay men and other MSM is well-documented 2.1 | Population and settings and described in many parts of the world [7,8]. Behavioural interventions delivered on Internet and social media platforms We used mixed methods to describe and investigate qualitative have demonstrated impact on sexual health among MSM in and population-representative quantitative survey data col- the United States, East Asia and Peru [9] and are adaptable lected from MSM and transgender individuals from Nairobi and to changing technology and HIV prevention tools [10]. How- Johannesburg from 2016 to 2017. MSM and TG were eligible if ever, despite the severity of HIV epidemics among MSM and they were aged at least 18 years; reported consensual sex with TG in parts of SSA [5] and the high prevalence of stigma and a man in the previous 12 months; resided in Johannesburg or discrimination towards MSM and TG in many African coun- within 50 km of Nairobi county; and either currently identified tries [4,25], sexual health interventions rarely harness this as a man or had been assigned male sex at birth. route of HIV prevention delivery and there are few validated online interventions [11,12]. 2.2 | Data Collection There is evidence that MSM in SSA do use social media [13] and many do so to seek partners. Surveys of MSM from 2.2.1 | Phase 1 qualitative urban Lesotho (2011) and eSwatini (2014) found that 39% and 44% of MSM, respectively, reported having met a sexual We recruited 30 MSM/TG at each site (total n = 60) between partner online [14], and in Nigeria (2013 to 2015), 62% May 2016 and July 2017 for an in-depth face-to-face inter- reported having found sexual partners online [15]. Online view. They were purposively selected to provide diversity in partner-seeking varied by educational attainment, religion, age, age, socio-economic status and ethnicity. Participants were sexual and gender identity and size of social network of MSM. recruited through existing community outreach activities, com- To further consider the utility of Internet or app-based munity organizations and sexual health clinics. Interviews last- methods to complement existing approaches to surveillance, ing 60 to 90 minutes were conducted in English, Kiswahili intervention delivery and research engagement among MSM (Kenya) or Zulu (South Africa), depending on the preference and TG in SSA, we need better understanding of online use of the participant, and were audio-recorded, transcribed and within broader social and sexual networking. The Internet has translated prior to data coding. In addition to issues relating been successfully used in other regions to survey MSM to sexual health and HIV, the interviews explored how men [16,17], and proved to be a cost-effective method to engage met and engaged with other men in both physical and online large and broadly representative population samples [18]. In environments. SSA, there are examples of surveys conducted entirely online [19,20] or that utilize a combination of online surveys and 2.2.2 | Phase 2 respondent driven sampling surveys referral from community organizations [21]. However, the rep- resentativeness of MSM and TG who socialize online to wider Respondent-driven sampling (RDS) surveys [31] were con- MSM and TG populations in SSA remains unclear, and ducted in Nairobi (April -December 2017) and Johannesburg prompts concern that employing online environments for (April - November 2017) with intended sample sizes of 600 research sampling or intervention delivery may only reach a and 300 respectively. There were 10 initial “seed” participants relatively affluent and educated sub-population [22]. in Nairobi and nine in Johannesburg (not all productive). Par- Online engagement with MSM and TG populations must ticipants were given two coupons for onwards recruitment also navigate the dual role that social media can play in safety valid for two weeks. They were reimbursed for their participa- and security [23]. Studies have highlighted how online socializ- tion and for each enrolled recruit. More detail about RDS ing can be viewed as safer than meeting in physical locations recruitment in Johannesburg is given elsewhere [32]. [14-15,24]. This is unsurprising in a region where same-gender After screening, participants gave informed consent, pri- sexual relationships are usually criminalized, and where not, vately completed a self-administered questionnaire on a tablet remain highly stigmatized [25-29]. Physical venues used for computer, undertook HIV counselling and testing according to socialization in the region are often covert and short-lived as national guidelines, and visited a clinician for examination, they are targeted by police, authorities or unofficial groups blood draw and where indicated STI treatment, linkage to HIV [30]. However, while providing physical security, online social- treatment services or for pre-exposure prophylaxis. Blood izing and partner-seeking may present other risks that MSM samples were tested for active syphilis (Treponema pallidum and TG must negotiate, including blackmail from partners met haemmaglutination (TPHA) & Rapid Plasma Reagin (RPR) and online [24]. How MSM and TG in SSA weigh up and mitigate HIV viral load (GeneXpert HIV-1 VL) if HIV-seropositive. the risks and benefits of in-person versus online socializing The questionnaire included sociodemographics, sexual iden- and partner-seeking is currently understudied. tity and gender identity, online and in-person socializing pat- Here we have sought to: (1) examine the extent, nature and terns and social network size (number of adult MSM in the means of engagement among MSM and TG individuals who relevant city with whom the participant had spoken in the use social media in Johannesburg, South Africa and Nairobi, previous two weeks). For the previous three months, partici- Kenya; (2) identify their demographic and sexual behaviour- pants reported whether they had sex with a man and/or a related characteristics and (3) identify opportunities to use woman; what types of sex they had had; and frequencies of social media settings to facilitate both public health research condom use during anal and vaginal intercourse. For the previ- and health promotion interventions. ous 12 months they reported: the number of sexual partners

92 Fearon E et al. Journal of the International AIDS Society 2020, 23(S6):e25603 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25603/full | https://doi.org/10.1002/jia2.25603 they had had; whether they had received money, gifts or Research Ethics Committee of the University of the Witwater- favours in exchange for sex with men and with women; and srand for Johannesburg and the Kenya Medical Research whether they had given gifts, money or favours in exchange Institute and University of Oxford for Nairobi. for sex with men. Participants reported whether they had experienced urethral or rectal STI symptoms over that period. 3 | RESULTS Participants self-reported if they had socialized online with other MSM in the previous month; previous year; more than 3.1 | Quantitative findings one year ago, or never. If any use was reported, they were asked to indicate all the sites they had visited or apps that they 3.1.1 | Characteristics of MSM and TG had used in the previous month from a list populated with all sites/apps known to operate in the country, with space to add A total of 618 MSM and TG participants were recruited in options. Social media were further categorized as generic Nairobi and 301 MSM and TG participants in Johannesburg. (sites/apps in widespread use among the whole population, e.g. Convergence of reporting social media engagement was Facebook, Instagram), gay-specific (sites/apps targeted to MSM achieved (Figure S1). only, e.g. Grindr, Planet Romeo) and dating-specific (sites/apps The majority of MSM and TG in Nairobi and Johannesburg targeted to users seeking sex or relationships, e.g. Badoo). were young and self-identified as gay/homosexual and cisgender (Table 1). Most had completed secondary education, yet many were unemployed. Most MSM and TG had been sexu- 2.3 | Analysis ally active with another man in the previous three months. 2.3.1 | Qualitative analysis Nearly two-fifths of men in Nairobi and one-fifth of those in Johannesburg had sold sex to a man within the previous Digital recordings of the interviews were transcribed verbatim 12 months. and translated to English. Debriefing reports were written and discussed after the interview to identify emergent themes. 3.1.2 | Use of social media apps and sites for Interview transcripts were subject to a detailed thematic anal- socializing with other MSM ysis [33], supported by NVIVO 10. Transcripts were read and re-read by a panel of researchers and interviewers to identify Most participants reported having socialized online with initial codes (relevant or significant features) that comprised MSM in the previous month, significantly higher in Nairobi the coding framework. The meaning and conceptual distinction than Johannesburg (70.9%; 95% CI 66.5 to 75.0 and 60.1%; of these codes was discussed and agreed upon among the 95% CI 53.2 to 66.6 respectively; p < 0.0068). Few reported qualitative research team, following which all sections of each never having done so: 14.3% in Nairobi and 21.5% in Johan- transcript were coded using this framework. Data within each nesburg. code were then carefully reviewed and formulated into higher level themes and cross-referenced against the rest of the cod- 3.1.3 | Sites/apps used for socializing with MSM ing framework for conceptual clarity. The most popular sites/apps used for socializing with MSM in 2.3.2 | Quantitative analysis the last month were similar in both cities. Generic social media sites/apps were the most widely used, notably Face- We report findings from each city separately. All percentages book (53.7% Nairobi; 38.2% Johannesburg) and WhatsApp are weighted using RDS-II estimation [34] using the self- (46.0% and 42.9%, respectively, Figure 1). Fewer participants reported measure of social network degree. We examined used gay- or dating-specific apps, nearly all of whom also used sociodemographic associations with online socializing within generic apps/sites to socialize with MSM (Figure 1B). Grindr the previous month using logistic regression, dropping seed was the most frequently cited gay-specific service in both participants and probability weighting by inverse network size, cities, used by 8.2% (Nairobi) and 8.1% (Johannesburg) in the and used Wald tests to assess statistical evidence for associa- previous month. tions. We then examined the associations between online socializing and sexual behaviours, first among all MSM and TG 3.1.4 | Characteristics associated with online in each city, then restricted to those HIV-uninfected to assess socializing with MSM in the previous month the association between online socializing and measures of behavioural HIV acquisition risk. We first examined crude Few sociodemographic differences were associated with associations, then adjusted for sociodemographic characteris- recent online socializing with MSM, and these were inconsis- tics found to be associated with online socializing and age tent across cities (Table 2). (Model 1), and finally examined associations with sexual beha- There was no clear association between online socializing viours, partners and STIs among only HIV-uninfected partici- and age, nor sexual or gender identity, nor were there strong pants (Model 2). For multivariate models, covariates with differences in the sociodemographic profiles of MSM and TG p < 0.100 in bivariate associations were retained. by use of gay-specific or dating apps. (We show usage of sites apps separately by gender identity in Table S3 and Figures S2 and S3.) Those in the second lowest income category in Nair- 2.4 | Ethics obi were more likely to use gay-specific apps than those in Ethical approval was obtained from the London School of the lowest, but there was no consistent trend. Those born Hygiene and Tropical Medicine for both sites, the Human outside of Kenya were more likely to use dating apps

93 Fearon E et al. Journal of the International AIDS Society 2020, 23(S6):e25603 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25603/full | https://doi.org/10.1002/jia2.25603

Table 1. Characteristics of MSM and TG in Nairobi and Johannesburg

Nairobi Johannesburg

Characteristics n Unweighted % RDS % n Unweighted % RDS %

Age group in years 18 to 21 162 26.2 28.3 75 24.9 24.1 22 to 24 177 28.6 28.6 67 22.3 23.8 25 to 29 136 22.0 20.9 72 23.9 22.7 30+ 143 23.1 22.2 87 28.9 29.5 Born in Nairobi/Johannesburg Born in Nairobi/Johannesburg 179 29.0 30.8 182 60.5 62.0 Born elsewhere in Kenya/South Africa 299 48.4 47.7 98 32.6 31.6 Born outside Kenya/South Africa 123 19.9 21.5 18 6.0 6.5 Religion Christianity 536 86.7 89.5 260 86.4 83.7 Islam 53 8.6 7.6 4 1.3 1.5 Other 4 0.6 0.3 1 0.3 0.0 None 18 2.9 2.6 35 11.6 14.8 Neighbourhood (Nairobi) Dagoretti 95 15.4 16.0 Embakasi 146 23.6 23.7 Kamukunji 18 2.9 2.9 Kasarani 111 18.0 18.7 Langata 34 5.5 5.6 Makadara 18 2.9 2.2 Starehe 83 13.4 13.3 Westlands 53 8.6 8.2 Outskirts 53 8.6 8.1 Missing 7 1.1 1.2 Neighbourhood (Johannesburg) Soweto 158 52.5 55.3 Hillbrow 44 14.6 16.5 Brammfontein 28 9.3 7.4 Orange Farm 13 4.3 4.0 Othera 58 19.3 16.7 Sexual identity Gay 448 72.5 73.2 216 71.8 70.2 Bisexual 143 23.1 23.4 71 23.6 26.5 Heterosexual 2 0.3 0.4 3 1.0 0.8 Other, None, Don’t know 16 2.6 2.9 9 3.0 2.5 Gender Identityb Cisgender male 528 85.4 86.2 233 77.4 78.3 Transfeminine 70 11.3 11.3 45 15.0 13.2 Transmasculine 3 0.5 0.4 2 0.7 0.3 Non-binary 17 2.8 2.1 21 7.0 8.2 Monthly income (Nairobi) <5000 KSH 224 36.2 37.8 5000 to 9999 KSH 166 26.9 25.6 10,000 to 19,999 KSH 129 20.9 20.5 20,000 KSH + 55 8.9 8.6 Monthly income (Johannesburg) 0 to 499 ZAR 82 27.2 28.1 500 to 999 ZAR 39 13.0 17.5 1000 to 1999 ZAR 57 18.9 19.4 2000 to 4999 ZAR 74 24.6 24.4 5000 + ZAR 30 10.0 10.5

94 Fearon E et al. Journal of the International AIDS Society 2020, 23(S6):e25603 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25603/full | https://doi.org/10.1002/jia2.25603

Table 1. (Continued)

Nairobi Johannesburg

Characteristics n Unweighted % RDS % n Unweighted % RDS %

Employment status Employed full-time 57 9.2 8.8 32 10.6 8.8 Employed part-time 122 19.7 19.2 42 14.0 15.0 Self-employed 159 25.7 27.4 33 11.0 9.7 Unemployed 247 40.0 41.7 168 55.8 59.3 Student 12 1.9 1.6 19 6.3 5.3 Other 11 1.8 1.3 6 2.0 1.9 Missing 44 7.1 7.6 Completed educational attainment Primary 111 18.0 18.1 25 8.3 8.2 Secondary 329 53.2 55.0 200 66.4 68.8 Higher education 171 27.7 26.9 75 24.9 23.1 Marital status Not married 496 80.3 81.8 267 88.7 88.3 Married to a man or transgender individual 65 10.5 10.8 30 10.0 11.1 Married to a woman 50 8.1 7.4 2 0.7 0.6

Online socializing with MSM Last time socialized online with MSM using social media, website or mobile app. In the last month 461 74.6 70.9 201 66.8 60.1 In the last year but not the last month 40 6.5 7.6 31 10.3 11.2 More than one year ago 43 7.0 7.2 21 7.0 7.2 Never 74 12.0 14.3 48 15.9 21.5

Sexual behaviours Sex with a man (three months) 543 87.9 87.2 234 77.7 75.7 Sex with a woman (three months) 174 28.2 28.3 82 27.2 31.7 Condomless anal intercourse (3 months) 265 42.9 41.8 113 37.5 35.9 Receptive anal sex (three months) 321 51.9 49.0 143 47.5 43.0 Number of sexual partners (three months) 0 75 12.1 12.8 67 22.3 24.3 1 148 23.9 28.3 94 31.2 35.8 2 171 27.7 30.6 70 23.3 22.2 3 to 5 154 24.9 20.9 52 17.3 13.9 6+ 70 11.3 7.3 18 6.0 3.9 Sold sex to a man (12 months) 297 48.1 43.8 69 22.9 21.9 Bought sex from a man (12 months) 177 28.6 28.2 31 10.3 10.0 STI symptoms (12 months) 225 36.4 35.2 301 100.0 28.6 Syphilis (active, RPR and TPHA positive) 5 0.8 1.1 28 9.3 9.7 CT (urethral) 39 6.3 7.3 18 6.0 6.2 NG (urethral) 27 4.4 4.4 4 1.3 1.6 CT (rectal) 53 8.6 8.1 –– – NG (rectal) 76 12.3 13.2 –– –

RDS-II weighted percentage given (inverse network size weighting, seed participants dropped). Seed participants are included in frequency counts. n’s do not add to full sample size where responses were missing. CT, Chlamydia trachomatis; KSH, Kenyan shillings (currency); MSM/ and TG, men who have sex with men and transgender people; NG, Neisseria gonorrhoeae; RPR and TPHA positive, Rapid Plasma Reagin and Treponema pallidum haemmaglutination positive, indicating active syphilis infection; ZAR, South African Rand (currency). aOther Johannesburg neighbourhoods include all those with fewer than 10 participants each bGender identity was assessed using what at the time was considered best practice via a two-step approach [35], comprising assessment of sex assignment at birth (male, female or prefer not to say) and current gender identity (male, female, transgender or none of these). Alongside recommendations [36], we described participants as transmasculine where they had been assigned female sex at birth, but now identified as male or transgender, and transfeminine where they had been assigned male sex at birth, but now identified as female or transgender. Participants who did not currently identify as male, female or transgender were described as ‘Non-binary’.

95 Fearon E et al. Journal of the International AIDS Society 2020, 23(S6):e25603 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25603/full | https://doi.org/10.1002/jia2.25603

(a) Sites/apps used to socialise with MSM in the last month: Facebook WhatsApp Instagram TwiƩer Grindr Skype Badoo Gay.com Men2Men 2go Planet Romeo GayXchange WeChat Gaydar Hornet Total by app/site type, of MSM/TG socialising online Snapchat Nairobi (n, RDS%) Johannesburg (n, RDS%) Gay Radar Generic 421, 65.6% 177, 52.7% GuySpy Get Male Gay specific 171, 22.1% 71, 18.1% Adam4Adam DaƟng speciﬁc 69, 8.4% 37, 11.5% Mamba Online DaƟng Buzz ManToManPlus Sex Trader South Africa

0102030405060

RDS-II weighted % reporƟng use in the last month, of all MSM/TG in each city

Upper bars are Nairobi; lower bars are Johannesburg.

(b) Overlaps in types of sites/apps used: Nairobi Johannesburg

Unweighted counts used to show overlaps in use. MSM/TG = men who have sex with men and transgender people.

Figure 1. Sites/apps used to socialize with MSM among MSM and TG in Nairobi (n = 618) and Johannesburg (n = 301) during the month prior to interview. (A) Sites/apps used to socialize with MSM in the last month. Upper bars are Nairobi; lower bars are Johannesburg. (B) Overlaps in types of sites/apps used. Unweighted counts used to show overlaps in use. MSM/TG, men who have sex with men and transgender people.

compared to MSM and TG born in Nairobi. Students in Johan- 3.1.5 | Associations between sexual behaviour, STIs nesburg were more likely than others to use gay apps, and and online socializing there was variation by neighbourhood, but there was no evidence for a difference by educational attainment or income, Evidence for associations between online socializing and sexual (Tables S4 and S5). behaviour differed between cities (Table 3). In Nairobi,

96 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25603/full ernEe al. et E Fearon Table 2. Associations between socializing online with MSM in the previous month and sociodemographic characteristics of MSM and TG in Nairobi and Johannesburg

Nairobi, n = 608, without seeds Johannesburg, n = 292, without seeds

Crude Adjusted Crude Adjusted Society AIDS International the of Journal

p p p p Characteristic n RDS % OR 95% CI value aOR 95% CI value n RDS % OR 95% CI value aOR 95% CI value

Overall socialized 461/618 70.9 201/301 60.1 online with other MSM in the last month Age group in years 18 to 21 112/162 64.0 1.00 0.133 1.00 0.114 55/75 67.0 1.00 0.319 1.00 0.236 2020,

22 to 24 134/177 73.2 1.54 0.89 2.65 1.55 0.89 2.73 47/67 58.9 0.71 0.31 1.63 0.66 0.28 1.53 | https://doi.org/10.1002/jia2.25603

25 to 29 111/136 78.5 2.05 1.10 3.84 2.23 1.15 4.34 48/72 65.6 0.94 0.41 2.17 0.94 0.39 2.24 23 30+ 104/143 69.5 1.28 0.73 2.26 1.49 0.82 2.71 51/87 51.2 0.52 0.24 1.13 0.46 0.20 1.08 (S6):e25603 Born in Nairobi/Johannesburg Born in Nairobi/ 124/179 66.6 1.00 0.44 113/182 54.8 1.00 0.052 1.00 0.073 Johannesburg Born elsewhere 226/299 71.9 1.28 0.80 2.06 75/98 73.2 2.25 1.16 4.36 2.29 1.10 4.78 in Kenya/ South Africa Born outside 98/123 74.0 1.43 0.78 2.61 12/18 53.9 0.87 0.29 3.18 1.12 0.30 4.17 Kenya/South Africa Religion Christianity 404/536 72.0 1.00 0.002 1.00 0.004 175/260 60.9 1.00 0.192 Islam 42/53 75.8 1.22 0.54 2.75 1.25 0.58 2.72 2/4 17.8 0.14 0.02 1.17 Other 1/4 8.5 0.04 0.00 0.54 0.04 0.00 0.49 0/1 0.0 None 10/18 32.7 0.19 0.06 0.56 0.20 0.06 0.65 23/35 59.0 0.92 0.40 2.13 Neighbourhood (Nairobi) Dagoretti 73/95 72.7 1.00 0.905 Embakasi 102/146 66.1 0.73 0.37 1.43 Kamukunji 12/18 64.3 0.68 0.20 2.32 Kasarani 82/111 67.3 0.77 0.38 1.58 Langata 23/34 73.3 1.03 0.37 2.87 Makadara 15/18 79.9 1.49 0.30 7.48 Starehe 64/83 74.3 1.08 0.49 2.40 Westlands 41/53 76.5 1.22 0.48 3.08 Outskirts 44/53 77.3 1.28 0.48 3.38 97 Table 2. (Continued) http://onlinelibrary.wiley.com/doi/10.1002/jia2.25603/full al. et E Fearon

Nairobi, n = 608, without seeds Johannesburg, n = 292, without seeds

Crude Adjusted Crude Adjusted Society AIDS International the of Journal

p p p p Characteristic n RDS % OR 95% CI value aOR 95% CI value n RDS % OR 95% CI value aOR 95% CI value

Neighbourhood (Johannesburg) Soweto 97/158 56.4 1.00 0.100 1.00 0.127 Hillbrow 30/44 60.7 1.19 0.53 2.68 0.94 0.34 2.55 Brammfontein 23/28 77.6 2.67 0.72 9.89 1.90 0.50 7.20 Orange Farm 12/13 94.7 13.74 1.63 115.50 13.60 1.39 132.44 Othera 39/58 55.9 0.98 0.47 2.07 0.74 0.34 1.62 2020,

Sexual identity | https://doi.org/10.1002/jia2.25603

Gay 338/448 71.3 1.00 0.848 152/216 63.3 1.00 0.439 23

Bisexual 108/143 73.4 1.11 0.67 1.83 41/71 53.2 0.66 0.35 1.26 (S6):e25603 Heterosexual 0/2 0.0 – 0/3 0.0 – Other 5/9 65.0 0.75 0.16 3.46 6/9 55.4 0.72 0.13 3.98 Gender Identity Cisgender 394/528 71.5 1.00 0.84 159/233 61.3 1.00 0.819 male Transfeminine 52/70 65.8 0.77 0.41 1.44 29/45 59.2 0.92 0.42 2.00 Transmasculine 2/3 80.0 1.59 0.14 17.88 1/2 40.0 0.42 0.03 6.98 Non–binary 13/17 71.0 0.98 0.26 3.65 12/21 51.1 0.66 0.23 1.92 Monthly income (Nairobi) <5000 KSH 165/224 72.3 1.00 0.572 5000 to 9999 124/166 66.7 0.77 0.46 1.30 KSH 10,000 to 103/129 75.5 1.18 0.65 2.15 19,999 KSH 20,000 KSH + 40/55 69.7 0.88 0.41 1.92 Monthly income (Johannesburg) 0 to 499 ZAR 56/82 63.1 1.00 0.930 500 to 999 25/39 57.6 0.80 0.32 1.97 ZAR 1000 to 1999 40/57 61.1 0.92 0.39 2.19 ZAR 2000 to 4999 48/74 58.2 0.82 0.37 1.81 ZAR 5000 + ZAR 21/30 68.3 1.26 0.43 3.69 98 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25603/full Table 2. (Continued) al. et E Fearon

Nairobi, n = 608, without seeds Johannesburg, n = 292, without seeds

Crude Adjusted Crude Adjusted Society AIDS International the of Journal

p p p p Characteristic n RDS % OR 95% CI value aOR 95% CI value n RDS % OR 95% CI value aOR 95% CI value

Employment status Employed 42/57 67.9 0.84 0.40 1.79 0.255 20/32 45.4 0.58 0.23 1.48 0.486 full-time Employed 95/122 76.2 1.27 0.70 2.32 28/42 65.5 1.33 0.58 3.07 part-time Self-employed 113/159 67.1 0.81 0.49 1.35 20/33 56.1 0.90 0.35 2.32 2020,

Unemployed 187/247 71.5 1.00 110/168 58.7 1.00 | https://doi.org/10.1002/jia2.25603

Student 11/12 95.4 8.34 1.00 69.59 16/19 75.9 2.22 0.54 9.11 23

Other 7/11 59.1 0.58 0.12 2.78 6/6 100.0 – (S6):e25603 Completed educational attainment Primary 69/111 58.8 0.47 0.28 0.80 0.022 0.40 0.23 0.70 0.006 1/3 66.7 1.38 0.12 15.81 0.892 Secondary 256/329 75.1 1.00 1.00 146/222 59.1 1.00 Higher Education 132/171 71.0 0.81 0.49 1.34 0.70 0.42 1.19 53/75 62.5 1.15 0.59 2.27 Marital status Not married 369/496 70.9 1.00 0.971 179/267 61.0 1.00 0.598 Married to 51/65 72.5 1.08 0.54 2.18 20/30 55.0 0.78 0.31 1.96 a man or transgender person Married to a 37/50 70.2 0.97 0.50 2.09 0/2 0.0 – woman

Models weighted using RDS-II weights (inverse network size) with seed participants dropped. Adjusted models include age a priori and those variables in bivariate analyses that showed an association of p < 0.1. aOR, adjusted odds ratio; KSH, Kenyan shillings (currency); MSM/TG, men who have sex with men and transgender people; OR, odds ratio; RDS, respondent-driven sampling; ZAR, South African Rand (currency). aOther Johannesburg neighbourhoods include all those with fewer than 10 participants each. 99 Fearon E et al. Journal of the International AIDS Society 2020, 23(S6):e25603 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25603/full | https://doi.org/10.1002/jia2.25603

Table 3. Associations between socializing with MSM online in the previous month, recent sexual behaviours and sexually transmitted infection

Adjusted, all participants Adjusted, HIV uninfected Socialized online Crude, n = 608, (Model 1), n = 608, participants (Model 2), previous month without seeds without seeds n = 424, without seeds

Nairobi n RDS % OR 95% CI p value aOR 95% CI p value aOR 95% CI p value

Overall socialized 461/618 70.9 online with MSM in the last month Sex with a man (past three months) Yes 417/543 72.9 2.02 1.13 3.61 0.018 1.97 1.08 3.59 0.028 1.74 0.91 3.34 0.095 No 44/75 57.1 1.00 1.00 1.00 Sex with a woman (past three months) Yes 116/174 63.5 0.62 0.40 0.96 0.032 0.61 0.38 0.96 0.032 0.55 0.33 0.94 0.029 No 345/444 73.8 1.00 1.00 1.00 Condomless anal intercourse (past three months) Yes 212/265 75.5 1.48 0.97 2.28 0.072 1.37 0.88 2.14 0.161 1.67 0.97 2.87 0.063 No 249/353 67.6 1.00 1.00 1.00 Receptive anal intercourse (past three months) Yes 202/321 76.6 1.72 1.13 2.62 0.011 1.69 1.10 2.60 0.017 2.15 1.25 3.69 0.006 No 259/321 65.5 1.00 1.00 1.00 No. of male sexual partners (past three months) 0 44/75 57.1 1.00 0.003 1.00 0.012 1.00 0.008 1 104/148 66.0 1.46 0.75 2.81 1.48 0.75 2.93 1.14 0.55 2.39 2 121/171 69.9 1.75 0.91 3.34 1.72 0.88 3.36 1.65 0.78 3.50 3 to 5 132/154 84.4 4.05 1.91 8.60 3.71 1.72 7.96 4.88 1.92 12.39 6+ 60/70 79.4 2.89 1.10 7.58 2.60 0.95 7.10 2.23 0.74 6.73 Sold sex to a man (past 12 months) Yes 240/297 77.8 1.82 1.19 2.79 0.006 1.78 1.15 2.77 0.010 2.17 1.29 3.65 0.004 No 218/316 65.7 1.00 1.00 1.00 Bought sex from a man (past 12 months) Yes 145/177 80.2 1.98 1.19 3.28 0.009 1.77 1.05 2.99 0.033 2.24 1.17 4.27 0.015 No 313/437 67.1 1.00 1.00 1.00 STI symptoms (past 12 months) Yes 175/225 74.5 1.30 0.83 2.02 0.251 1.26 0.79 2.02 0.234 1.48 0.81 2.71 0.206 No 282/387 69.3 1.00 1.00 1.00 Syphilis (active) Positive 5/5 100.0 –––– – ––– – –– – Negative 454/609 70.7 HIV Positive 146/186 76.2 1.43 0.89 2.32 0.134 1.37 0.81 2.31 0.243 –––– Negative 314/431 69.0 1.00 1.00

Crude, n = 292, without seeds Model 1, n = 292, without seeds Model 2, n = 179, without seeds

Johannesburg n RDS % OR 95% CI p value aOR 95% CI p value aOR 95% CI p value

Overall socialized 201/301 60.1 online with MSM in the last month Sex with a man (past three months) Yes 163/234 62.1 1.42 0.74 2.74 0.292 1.32 0.66 2.64 0.431 1.72 0.76 3.91 0.194 No 38/67 53.6 1.00 1.00 1.00

100 Fearon E et al. Journal of the International AIDS Society 2020, 23(S6):e25603 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25603/full | https://doi.org/10.1002/jia2.25603

Table 3. (Continued)

Crude, n = 292, without seeds Model 1, n = 292, without seeds Model 2, n = 179, without seeds

Johannesburg n RDS % OR 95% CI p value aOR 95% CI p value aOR 95% CI p value

Sex with a woman (past three months) Yes 47/82 51.5 0.59 0.33 1.09 0.093 0.61 0.32 1.15 0.126 0.59 0.27 1.26 0.170 No 154/219 64.1 1.00 1.00 1.00 Condomless anal intercourse (past three months) Yes 89/113 73.6 2.51 1.34 4.68 0.004 1.46 0.80 2.67 0.216 3.67 1.57 8.58 0.003 No 112/188 52.6 1.00 1.00 1.00 Receptive anal intercourse (past three months) Yes 105/140 67.5 1.70 0.94 3.06 0.077 1.48 0.81 2.72 0.205 1.70 0.94 3.06 0.077 No 91/154 54.9 1.00 1.00 1.00 No. of male sexual partners (past three months) 0 38/67 53.6 1.00 0.192 1.00 0.301 1.00 0.323 1 61/94 55.8 1.09 0.52 2.30 1.03 0.47 2.26 1.37 0.55 3.46 2 50/70 68.4 1.87 0.81 4.31 1.90 0.80 4.55 2.56 0.88 7.43 3 to 5 36/52 61.9 1.41 0.56 3.56 1.16 0.44 3.05 1.42 0.40 5.06 6+ 16/18 86.5 5.54 1.07 28.63 5.39 0.71 41.02 11.49 0.50 266.40 Sold sex to a man (past 12 months) Yes 47/69 60.1 0.97 0.50 1.91 0.941 0.92 0.45 1.88 0.809 0.87 0.34 2.25 0.777 No 152/228 60.7 1.00 1.00 1.00 Bought sex from a man (past 12 months) Yes 17/31 39.4 0.38 0.16 0.93 0.036 0.35 0.14 0.85 0.022 0.31 0.09 1.03 0.056 No 182/266 63.0 1.00 1.00 1.00 STI symptoms (past 12 months) Yes 61/89 60.7 1.07 0.57 2.00 0.840 1.11 0.57 2.16 0.766 0.83 0.35 1.97 0.672 No 137/209 59.2 1.00 1.00 1.00 Syphilis (active) Positive 23/28 83.6 3.77 1.06 13.38 0.041 4.40 1.14 17.08 0.033 13.50 1.63 111.96 0.016 Negative 178/273 57.6 1.00 1.00 1.00 HIV Positive 82/118 63.9 1.29 0.72 2.33 0.387 1.75 0.86 3.55 0.122 –––– Negative 118/182 57.8 1.00 1.00

Models weighted using RDS-II weights (inverse network size) with seed participants dropped. Model 1: adjusted for age and sociodemographic characteristics found to be associated with online socializing in the previous month. Model 2: among only HIV uninfected participants, adjusted for age and sociodemographic characteristics found to be associated with online socializing in the previous month. aOR, adjusted odds ratio; KSH, Kenyan shillings (currency); MSM and TG, men who have sex with men and transgender people; OR, odds ratio; RDS, respondent-driven sampling; ZAR, South African Rand (currency).

socializing with MSM online was more common among those point estimates for associations between online socializing who had: sex with a man in the last three months; receptive anal with men and having had sex with a man, sex with a woman intercourse in the last three months; more sexual partners; and and receptive anal intercourse in the previous three months those who had either bought or sold sex to a man in the last among MSM and TG in Johannesburg were similar to those 12 months. Socializing online was less common among those observed in Nairobi, the statistical evidence was weaker. who had sex with a woman in the previous three months. Associ- There was no significant difference in online socializing ations persisted after adjustment (Model 1) and when restricted between the proportion of participants reporting STI symp- to HIV-uninfected participants (Model 2), though sex with a man toms or living with HIV by city. However, in Johannesburg, in the last three months was not significant in the latter. online socializing was associated with active syphilis. In Johannesburg, there was evidence for higher odds of Among MSM and TG who did report socializing online, online socializing among those reporting condomless anal those reporting some sexual behaviours associated with intercourse in the last three months (crude model and among higher HIV transmission risks did have raised odds of using those HIV-uninfected), but online socializing was significantly gay-specific and dating-specific apps, though not uniformly less often reported by participants who purchased sex from a (Tables S6 and S7). In Nairobi, MSM and TG with more part- man in the last 12 months, in both adjusted models. While the ners and who sold sex were more likely to report using gay-

101 Fearon E et al. Journal of the International AIDS Society 2020, 23(S6):e25603 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25603/full | https://doi.org/10.1002/jia2.25603 specific apps, while in Johannesburg the association with sell- 3.2.2 | Perceiving and mitigating harm ing sex was reversed (though the statistical evidence was somewhat weak). Use of dating apps was associated with both For many participants across both cities, interacting with men buying and selling sex in Nairobi, but not in Johannesburg. In in social media environments was considered safer than both settings, socializing with MSM using dating apps was less attending physical locations where MSM congregate. This was common among MSM and TG who reported having had sex especially true for participants in Kenya who reported that with a woman. police raids or general hostility towards gay bars or hotspots were commonplace. However, while facilitating a broad range 3.1.6 | Associations between online socializing and of social and sexual connectivity, social media apps and sites engagement with HIV prevention and care technologies were not without their own risks. Concerns were raised by participants in both cities regard- No significant associations were detected between online ing the possibility for “context collapse” [25]. This notion refers socializing in the last month and use of antiretroviral therapy to when different aspects of one’s life and experience, which (ART), virological suppression, recency of HIV testing, PrEP- are usually kept separate in the physical world, come to over- related knowledge (among those HIV uninfected or untested) lap in online environments; for example, circumstances where or use or access to condoms or lubricants in either city a man’s family is unaware or disapproving of his attraction to (Table S1). men, this could be threatening. While having multiple profiles was a common strategy to avoid this occurrence, the risk remained. Particular concern was raised for pictures shared 3.2 | Qualitative findings (especially sex-related images) and how these might come to The sociodemographic characteristics of interview participants the attention of non-MSM friends and family or be used for were consistent with those of the RDS samples in both cities. the purposes of blackmail. “In social media the majority of them The majority (n = 54) of those interviewed reported regular are looking for money. The others are blackmailers.” [Aged 22, use of social media to engage with other MSM and TG. Two Nairobi]. Indeed, a concern for blackmail was pervasive among clear themes were identified relevant to the delivery of inter- many, such that some were hesitant to share any personal ventions or research in online spaces. information or images of themselves until they felt “safe” with the person(s) they were interacting with online. Ultimately 3.2.1 | Navigating online environments choosing to meet face-to-face was challenged by these concerns. Risk mitigation strategies included “screening” the per- Participants commonly reported having multiple profiles on son online with questions about sex between men to ensure generic social media sites: one used for family or work friends they were “legitimate” MSM, requesting that they send multi- and a second for engaging with other MSM. Much use was ple photographs of themselves to ensure they had not simply made of gay or MSM-specific forums, groups and Facebook adopted another person’s photograph), and actively discussing pages where participants reported how posting within such physical safety concerns before meeting. groups and searching their members for mutual friends was typically a reliable way of identifying other MSM. Posting a “Is it safe? Will people see me? Is there parking? Will photo of oneself and seeing which men “liked” was considered there be people around?” [Aged 26, Nairobi] a good initial strategy to online interaction.

“There are those gay groups where you post your pic- ture and say, ‘Please like me’ and you get like 50 mes- 4 | DISCUSSION sages in your inbox and like 20 friend requests in a day.” [Aged 20, Johannesburg] Use of social media to socialize with MSM was common among MSM and TG in Nairobi and Johannesburg, with a The use of gay-specific apps was less common, though they majority among population-representative samples reporting enabled a more direct engagement with others known to be having done so recently. Those socializing online showed a MSM. Positioning oneself in such an environment did, how- wide diversity of characteristics and clear HIV prevention and ever, present its own risks which were often cited as the rea- sexual health needs. MSM and TG used different platforms, son for using generic sites/apps. A number of interviewees but generic socializing apps/sites were most common, while also expressed the belief that gay-specific apps remain the use of gay-specific and dating apps/sites were less prevalent, purview of MSM described as “higher class.” However, in both a pattern also seen amongst MSM seeking partners online in contexts, participants described using generic social media Nigeria [15]. All types of social media were used for partner- sites to find sexual partners, or to buy/sell sex. seeking as well as socializing, but the role that social media Those with less sexual experience or confidence negotiating plays in providing anonymity and security in the context of with men reported this setting as more amenable to flirtation criminalizing and socially stigmatizing settings is complex. and sexual-planning. More common in both cities, however, Online socializing with MSM was not restricted to particular was the use of social media to establish friendships, and to subgroups of socio-economic status, age or sexual and gender alleviate feelings of loneliness or isolation. A small number of identity. Nor was there good evidence that this varied very men in both cities described how their first forays into MSM much by type of app, and the distribution of their use was bars, clubs, hotels or other hot-spots were facilitated by very similar across cities. This is important because it suggests friendships originally made in online environments. that a representative diversity of MSM and TG in each city

102 Fearon E et al. Journal of the International AIDS Society 2020, 23(S6):e25603 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25603/full | https://doi.org/10.1002/jia2.25603 might be reached via social media channels. There was some Africa should be further explored to inform the targeting and evidence that among MSM and TG in Nairobi online socializ- design of intervention packages to address them. ing was more common among those with at least a secondary We found some strong similarities in online socializing with education, compared to none or primary, but the use was high MSM across the two cities, with similar types of site/apps across all education groups. Although not directly comparable, used and concordance on the motivations for their use. Some our findings are unlike earlier surveys from Swaziland, Lesotho differences in the associations with sexual behaviours may and Nigeria, in which online partner-seeking was strongly arise from differences in sample power, or different population associated with higher education and younger age [14,15]. prevalence of STIs (e.g. active syphilis is more prevalent in There was some evidence that social media use was higher Johannesburg than Nairobi). There were different patterns amongst those who were born outside of Johannesburg, com- associated with sexual exchange in each city, but this also var- pared to those born in the city. This may reflect lower accessi- ied substantially in prevalence; buying and selling of sex was bility or tolerance of offline MSM communities outside of twice as common among MSM and TG in Nairobi as in Johan- large cities, as reported in rural and semi-rural parts of Mpu- nesburg. The relative usefulness and strategies for using social malanga, South Africa [22]. media for engaging MSM and TG engaged in transactional sex Our findings confirm that MSM and TG who are active might differ across cities. online have unmet needs for sexual health information, ser- A strength of our approach was the use of a representative vices and referral. In both cities, MSM and TG socializing survey method that permitted the comparison of those who online reported high levels of HIV and STI transmission beha- are, and are not, socializing online (which cannot be accom- viours, yet we found no evidence they were more engaged in plished in an online-only survey). The qualitative data aid inter- HIV prevention or care. The latter observation differs from pretation and make clear that, while not without its own studies in the United States suggesting that whilst Grindr challenges, online socializing helps facilitate in contexts of users may engage in higher risk behaviours than non-users, social censure. However, as a cross-sectional survey, we are they also report higher uptake of prevention (PrEP use [37]) unable to determine causality relating to the experiences that and HIV testing [7]). The delivery of HIV-related education as influence social media use, nor infer direction of association well as signposting to HIV testing venues via engagement with between social media use and HIV transmission risk beha- MSM in social media (often termed “Netreach”) has been viours. Both study sites were urban and findings are not nec- shown effective in several countries [38,39], whereas social essarily generalizable to rural areas. While widely used to media facilitated HIV testing (e.g. for home-based self-testing obtain population-representative estimates among MSM in or self-sampling) has shown early promise as acceptable and SSA [40], assumptions underlying RDS estimation are difficult feasible [40]. to meet in practice [51]. This survey lacked the power to Effective and convenient sexual health promotion and ser- explore issues specific to transgender participants and we vice models facilitated by social media have been developed in acknowledge the need for such work. While there were lower high-resource settings, and such models typically expand user percentages of transfeminine persons reporting use of gay- choice beyond traditional, facility-based services rather than specific and dating sites/apps in the last month, we did not replacing existing models completely [12]. Online interventions find statistical evidence for differences in use of sites/apps to might help in mitigating the effects of healthcare associated socialize with MSM in the last month by gender identity. stigma via facilitating better choice among services by Finally, the use of specific sites/apps changes quickly over enabling peer service reviews, for example [30,41]. Site/app- time, but there is little more recent literature documenting based channels of communication provide an additional option and investigating online socializing among MSM and TG across for peer support interventions, for instance, those aiming to SSA. In trying to understand motivations behind usage we improve ART adherence [42]. Importantly, our findings suggest hope that the data will be more enduringly informative. that a sizeable minority of MSM and TG are not regularly active online (40% in Johannesburg, 30% in Nairobi). Some may not have access to the necessary technology, however, 5 | CONCLUSIONS our qualitative findings indicate that for some, the risks of establishing a presence online are perceived to be greater Online environments are widely used by African MSM and TG than the benefits [24,43]. Providers planning online services to socialize and partner-seek, and offer routes to deliver sex- and interventions must be aware of and mitigate what may be ual health promotion, services and research opportunities that unfamiliar risks, and should anticipate that such services will are currently under-utilized. The lack of demographic variation not be acceptable to all. Furthermore, the design of online in social media use suggests that access is not limited to those interventions or intervention components need to consider with higher socio-economic status only. While evidence of an the role of syndemics of mental health disorders, substance elevated HIV risk profile among those using social media may use and experience of harassment and abuse, which are part complicate the use of Internet-facilitated samples to estimate of the context in which sexual behaviours and risks among HIV prevalence, MSM and TG recruited online may provide MSM and TG in Kenya and South Africa occur [44-47], but valuable insight into the factors that influence risk behaviour which vary across settings [48]. A systematic review of sub- as well as uptake/engagement with HIV testing, care and pre- stance use among African MSM found that use of recreational vention interventions. In addition, these data strengthen the drugs and alcohol were frequently used as part of sexual evidence base for delivery of health education, social market- experiences [49]. The ways in which these factors interact ing and peer support programmes in online spaces. Since with the motivations for and experiences of online socializing 2017, overall access to online sites/apps has likely increased and partner seeking among MSM and TG in Kenya and South and it is unlikely that the diversity of MSM and TG using

103 Fearon E et al. Journal of the International AIDS Society 2020, 23(S6):e25603 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25603/full | https://doi.org/10.1002/jia2.25603 them has narrowed. It is also possible that disruptions both to 5. Baral SD, Grosso A, Holland C, Papworth E. The epidemiology of HIV among in-person HIV and sexual health services, as well as to the men who have sex with men in countries with generalized HIV epidemics. Curr Opin HIV AIDS. 2014;9(2):156–67. options available for socializing amongst MSM and TG caused 6. Poteat T, Ackerman B, Diouf D, Ceesay N, Mothopeng T, Odette KZ, et al. by the COVID-19 pandemic and associated physical distancing HIV prevalence and behavioral and psychosocial factors among transgender policies further strengthen the need for online site/app medi- women and cisgender men who have sex with men in 8 African countries: a ated interventions. While needing to be attentive to concerns cross-sectional analysis. PLoS Medicine. 2017;14:e1002422. relating to context collapse and blackmail or digital security, 7. Paz-Bailey G, Hoots BE, Xia M, Finlayson T, Prejean J, Purcell DW, et al. Trends in internet use among men who have sex with men in the United States. community-based organizations in particular are well-placed to J Acquir Immune Defic Syndr. 2017;75 Suppl 3:S288–95. deliver online interventions at scale, potentially side-stepping 8. Cao B, Liu C, Durvasula M, Tang W, Pan S, Saffer AJ, et al. Social Media some of the traditional challenges inherent in reaching MSM engagement and HIV testing among men who have sex with men in China: A and TG in hostile social environments that often present phys- Nationwide Cross-Sectional Survey. J Med Internet Res. 2017;19:e251. 9. Knight R, Karamouzian M, Salway T, Gilbert M, Shoveller J. Online interven- ical safety concerns. tions to address HIV and other sexually transmitted and blood-borne infections among young gay, bisexual and other men who have sex with men: a systematic AUTHORS’ AFFILIATIONS review. J Int AIDS Soc. 2017;20:e25017. 10. Grov C, Westmoreland D, Rendina HJ, Nash D. Seeing Is believing? Unique 1Department of Global Health & Development, London School of Hygiene & capabilities of internet-only studies as a tool for implementation research on HIV Tropical Medicine, London, United Kingdom; 2Australian Research Centre in Sex, prevention for men who have sex with men: a review of studies and methodologi- Health & Society, La Trobe University, Melbourne, Australia; 3Wits Reproductive cal considerations. J Acquir Immune Defic Syndr. 2019;82 Suppl 3:S253–60. Health and HIV Institute, School of Clinical Medicine, University of the Witwa- 11. Nguyen LH, Tran BX, Rocha LEC, Nguyen HLT, Yang C, Latkin CA, et al. A tersrand, Johannesburg, South Africa; 4Partners for Health and Development, systematic review of eHealth interventions addressing HIV/STI prevention Nairobi, Kenya; 5Sigma Research, Department of Public Health, Environments among men who have sex with men. AIDS Behav. 2019;23(9):2253–72. and Society, London School of Hygiene & Tropical Medicine, London, United 12. Cao B, Gupta S, Wang J, Hightow-Weidman LB, Muessig KE, Tang W, et al. Kingdom; 6Department of Community Health Sciences, University of Manitoba Social media interventions to promote hiv testing, linkage, adherence, and reten- (UoM), Winnipeg, Canada; 7Nuffield Department of Population Health, Univer- tion: systematic review and meta-analysis. J Med Internet Res. 2017;19:e394. sity of Oxford, Oxford, United Kingdom 13. Baral S, Turner RM, Lyons CE, Howell S, Honermann B, Garner A, et al. Population size estimation of gay and bisexual men and other men who have COMPETING INTEREST sex with men using social media-based platforms. JMIR Public Health Surveill. 2018;4(1):e15. No author has conflicts of interest to declare. 14. Stahlman S, Grosso A, Ketende S, Mothopeng T, Taruberekera N, Nkonyana J, et al. Characteristics of men who have sex with men in southern Africa who AUTHORS’ CONTRIBUTIONS seek sex online: a cross-sectional study. J Med Internet Res. 2015;17(5):e129. 15. Stahlman S, Nowak RG, Liu H, Crowell TA, Ketende S, Blattner WA, et al. EF contributed to conceiving and designing the study and data collection instru- Online Sex-seeking among men who have sex with men in nigeria: implications ments, carried out quantitative analyses and drafting of the manuscript; AB con- for online intervention. AIDS Behav. 2017;21(11):3068–77. tributed to conceiving and designing the study and data collection instruments, 16. Weatherburn P, Schmidt AJ, Hickson FCI, Reid D, Berg RC, Hospers HJ, carried out qualitative analyses and drafting of the manuscript; ST and RK con- et al. The European men-who-have-sex-with-men internet survey (EMIS): design tributed to managing data collection, conducted interviews and approved the and methods. Sex Res Soc Policy. 2013;10(4):243–57. final draft; ADS contributed to designing the study, data collection instruments, 17. Holt M, Lea T, Mao L, Zablotska I, Lee E, de Wit JBF, et al. Adapting beha- carried out quantitative analyses, commented on and approved the final draft; vioural surveillance to antiretroviral-based HIV prevention: reviewing and antici- WN helped design the data collection instruments, analysed qualitative data and pating trends in the Australian Gay Community Periodic Surveys. Sex Health. approved the final draft; JK, PW and TPP contributed to conceiving and design- 2017;14(1):72–9. ing the study and data collection instruments, commented on the manuscript 18. Prah P, Hickson F, Bonell C, McDaid LM, Johnson AM, Wayal S, et al. Men and approved the final draft. All authors have approved the final manuscript. who have sex with men in Great Britain: comparing methods and estimates from probability and convenience sample surveys. Sex Transm Infect. 2016;92 ACKNOWLEDGEMENTS (6):455–63. 19. Chard AN, Metheny NS, Sullivan PS, Stephenson R. Social stressors and We thank and acknowledge the participants of the TRANSFORM study, and are intoxicated sex among an online sample of men who have sex with men (MSM) grateful to our community-based partners, ANOVA Health Institute, Ten81 and drawn from seven countries. Subst Use Misuse. 2018;53(1):42–50. SOHACA in Johannesburg and GALCK, HOYMAS and ISHTAR in Nairobi. 20. Wagenaar BH, Sullivan PS, Stephenson R. HIV knowledge and associated factors among internet-using men who have sex with men (MSM) in South FUNDING Africa and the United States. PLoS One. 2012;7:e32915. 21. Emmanuel G, Folayan M, Undelikwe G, Ochonye B, Jayeoba T, Yusuf A, This study was funded through the Evidence for HIV prevention in Southern et al. Community perspectives on barriers and challenges to HIV pre-exposure Africa programme, funded by UK aid from the Department for International prophylaxis access by men who have sex with men and female sex workers Development, and Sweden, through the Swedish International Development access in Nigeria. BMC Public Health. 2020;20(1):69. Agency (SIDA), mandated to represent the Norwegian Agency for Development 22. Rebe K, Hoosen N, McIntyre JA. Strategies to improve access for MSM in Cooperation (Norad), and managed by Mott Macdonald. Further analyses were low-income and middle-income countries. Curr Opin HIV AIDS. 2019;14 funded via the Medical Research Council grant number MR/S020462/1. The (5):387–92. funders played no role in this reporting of findings. 23. Shio J, Moyer E. Navigating norms of masculinity: tactical gender perfor- mances among gay men in Dar Es Salaam, Tanzania. Gender, Place Culture. 2020;1–17. https://doi.org/10.1080/0966369X.2020.1759513 REFERENCES 24. Thomann M, Grosso A, Wilson PA, Chiasson MA. ‘The only safe way to find a partner’: rethinking sex and risk online in Abidjan, Cote^ d’Ivoire. Critical Public 1. Pew Research Center. Social media use continues to rise in developing coun- Health. 2020;30(1):53–67. tries but plateaus across developed ones. 2018. 25. ILGA World: Mendos LR. State-sponsored homophobia 2019: global legisla- 2. Measuring the information society report. Volume 1. Geneva: International tion overview update. Geneva: ILGA; 2019. Telecommunications Union; 2018. 26. Zahn R, Grosso A, Scheibe A, Bekker LG, Ketende S, Dausab F, et al. 3. Beyrer C, Baral SD, Collins C, Richardson ET, Sullivan PS, Sanchez J, et al. Human rights violations among men who have sex with men in Southern Africa: The global response to HIV in men who have sex with men. Lancet. 2016;388 comparisons between legal contexts. PLoS One. 2016;11:e0147156. (10040):198–206. 27. Stahlman S, Sanchez TH, Sullivan PS, Ketende S, Lyons C, Charurat ME, 4. Smith AD, Tapsoba P, Peshu N, Sanders EJ, Jaffe HW. Men who have sex et al. The prevalence of sexual behavior stigma affecting gay men and other with men and HIV/AIDS in sub-Saharan Africa. Lancet. 2009;374(9687):416–22.

104 Fearon E et al. Journal of the International AIDS Society 2020, 23(S6):e25603 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25603/full | https://doi.org/10.1002/jia2.25603

men who have sex with men across sub-Saharan Africa and in the United States. 45. Daniels J, Struthers H, Lane T, Maleke K, McIntyre J, Coates T. "Booze is JMIR Public Health Surveill. 2016;2(2):e35. the main factor that got me where I am today": alcohol use and HIV risk for 28. Rispel LC, Metcalf CA, Cloete A, Moorman J, Reddy V. You become afraid MSM in rural South Africa. AIDS Care. 2018;30(11):1452–8. to tell them that you are gay: health service utilization by men who have sex 46. Korhonen C, Kimani M, Wahome E, Otieno F, Okall D, Bailey RC, et al. with men in South African cities. J Public Health Policy. 2011;32 Suppl 1: Depressive symptoms and problematic alcohol and other substance use in 1476 S137–51. gay, bisexual, and other MSM at three research sites in Kenya. AIDS. 2018;32 29. Jobson G, de Swardt G, Rebe K, Struthers H, McIntyre J. HIV risk and (11):1507–15. prevention among men who have sex with men (MSM) in peri-urban 47. Kunzweiler CP, Bailey RC, Okall DO, Graham SM, Mehta SD, Otieno FO. townships in Cape Town, South Africa. AIDS Behav. 2013;17 Suppl 1: Factors associated with prevalent HIV infection among kenyan MSM: the Anza S12–22. Mapema Study. J Acquir Immune Defic Syndr. 2017;76(3):241–9. 30. Bourne A, Fearon E, Nutland W.Mapping and appraisal of HIV prevention 48. Sandfort TGM, Bos H, Reddy V. Gender expression and mental health in and care interventions among men who have sex with men in Kenya, Uganda, black South African men who have sex with men: further explorations of unex- Tanzania and Zimbabwe: a report of the SHARP programme. International HIV/ pected findings. Arch Sex Behav. 2018;47(8):2481–90. AIDS Alliance Sigma Research. 2016. 49. Sandfort TGM, Knox JR, Alcala C, El-Bassel N, Kuo I, Smith LR. Substance 31. Heckathorn DD. Respondent-driven sampling ii: deriving valid population use and HIV risk among men who have sex with men in Africa: a systematic estimates from chain-referral samples of hidden populations. Soc Probl. 2002;49 review. J Acquir Immune Defic Syndr. 2017;76(2):e34–46. (1):11–34. 50. Sabin KM, Johnston LG. Epidemiological challenges to the assessment of HIV 32. Fearon E, Tenza S, Mokoena C, Moodley K, Smith AD, Bourne A, Weather- burdens among key populations: respondent-driven sampling, time-location sam- burn P, Palanee-Phillips T. HIV testing, care and viral suppression among men pling and demographic and health surveys. Curr Opin HIV AIDS. 2014;9(2):101–6 who have sex with men and transgender individuals in Johannesburg, South 51. Gile KJ, Beaudry IS, Handcock MS, Ott MQ. Methods for inference from Africa. PLoS One. 2020;15(6):e0234384. respondent-driven sampling data. Ann Rev Stat Appl. 2018;5:65–93. 33. Braun V, Clarke V. Using thematic analysis in psychology. Qualit Res Psy- chol. 2006;3:77–101. 34. Volz E, Heckathorn DD. Probability based estimation theory for respondent driven sampling. J Offic Stat. 2008;24(1):79–97. 35. Bauer GR, Braimoh J, Scheim AI, Dharma C. Transgender-inclusive mea- SUPPORTING INFORMATION sures of sex/gender for population surveys: mixed-methods evaluation and recommendations. PLoS One. 2017;12:e0178043. Additional Supporting Information may be found in the online 36. Reisner SL, Poteat T, Keatley J, Cabral M, Mothopeng T, Dunham E, et al. Global health burden and needs of transgender populations: a review. Lancet. version of this article: 2016;388(10042):412–36. Figure S1. Convergence of the estimate of the proportion of 37. Hoenigl M, Little SJ, Grelotti D, Skaathun B, Wagner GA, Weibel N, et al. MSM/TG in Nairobi and Johannesburg who report having GrindrTM users take more risks, but are more open to HIV pre-exposure pro- socialised online in the previous month. phylaxis: could this dating app provide platform for HIV prevention outreach? Figure S2. Sites and App usage in Nairobi in the previous Clin Infect Dis. 2019. 38. Lampkin D, Crawley A, Lopez TP, Mejia CM, Yuen W, Levy V. Reaching sub- month by gender identity. urban men who have sex with men for STD and HIV services through online Figure S3. Sites and App usage in Johannesburg in the pre- social networking outreach: a public health approach. J Acquir Immune Defic vious month by gender identity. Syndr. 2016;72(1):73–8. Table S1. Associations between frequency of online socialising 39. Mowlabocus S, Haslop C, Dasgupta RK. From Scene to Screen: The challenges and opportunities of commercial digital platforms for HIV community and engagement with HIV prevention and care technologies outreach. Social Media Soc. 2018;2(4):1–27. Table S2. Demographic characteristics of qualitative interview 40. Witzel TC, Rodger AJ, Burns FM, Rhodes T, Weatherburn P. HIV Self-Test- participants ing among Men Who Have Sex with Men (MSM) in the UK: A Qualitative Study Table S3. Online socialising with MSM in the last month by of Barriers and Facilitators, Intervention Preferences and Perceived Impacts. gender identity PLoS One. 2016;11(9):e0162713. 41. Lyons CE, Ketende S, Diouf D, Drame FM, Liestman B, Coly K, et al. Poten- Table S4. Sociodemographic characteristics associated with tial impact of integrated stigma mitigation interventions in improving HIV/AIDS socialising with MSM in the last month using gay-specific apps, service delivery and uptake for key populations in senegal. J Acquir Immune among MSM/TG who had socialised online in the last month Defic Syndr. 2017;74 Suppl 1:S52–9. Table S5. Sociodemographic characteristics associated with 42. Graham SM, Micheni M, Chirro O, Nzioka J, Secor AM, Mugo PM, et al. A randomized controlled trial of the Shikamana intervention to promote antiretro- socialising with MSM in the last month using dating-specific viral therapy adherence among gay, bisexual, and other men who have sex with apps, among MSM/TG who had socialised online in the last men in Kenya: feasibility, acceptability, safety and initial effect size. AIDS Behav. month 2020;24(7):2206–19. Table S6. Sexual behaviours and STIs associated with socialis- 43. Witzel TC, Guise A, Nutland W, Bourne A. It starts with me: privacy con- ing with MSM in the last month using gay-specific apps, cerns and stigma in the evaluation of a Facebook health promotion intervention. Sex Health. 2016;13(3):228–33. among MSM/TG who had socialised online in the last month 44. Secor AM, Wahome E, Micheni M, Rao D, Simoni JM, Sanders EJ, et al. Table S7. Sexual behaviours and STIs associated with socialis- Depression, substance abuse and stigma among men who have sex with men in ing with MSM in the last month using dating-specific apps, coastal Kenya. AIDS. 2015;29 Suppl 3:S251–9. among MSM/TG who had socialised online in the last month

105 Palmer S et al. Journal of the International AIDS Society 2020, 23(S6):e25590 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25590/full | https://doi.org/10.1002/jia2.25590

REVIEW Acute and early HIV infection screening among men who have sex with men, a systematic review and meta-analysis Shaun Palmer1,2,*, Maartje Dijkstra3,4,§,* , Johannes CF Ket5, Elizabeth W Wahome1 , Jeffrey Walimbwa6, Evanson Gichuru1, Elise M van der Elst1 , Maarten F Schim van der Loeff3,4, Godelieve J de Bree4 and Eduard J Sanders1,7,8 §Corresponding author: Maartje Dijkstra, Nieuwe Achtergracht 100, 1018WT, Amsterdam, the Netherlands. Tel: +31 20 5555792. ([email protected]) PROSPERO Number: CRD42019124963. *These authors contributed equally to the work

Abstract Introduction: Screening for acute and early HIV infections (AEHI) among men who have sex with men (MSM) remains uncommon in sub-Saharan Africa (SSA). Yet, undiagnosed AEHI among MSM and subsequent failure to link to care are important dri- vers of the HIV epidemic. We conducted a systematic review and meta-analysis of AEHI yield among MSM mobilized for AEHI testing; and assessed which risk factors and/or symptoms could increase AEHI yield in MSM. Methods: We systematically searched four databases from their inception through May 2020 for studies reporting strategies of mobilizing MSM for testing and their AEHI yield, or risk and/or symptom scores targeting AEHI screening. AEHI yield was defined as the proportion of AEHI cases among the total number of visits. Study estimates for AEHI yield were pooled using random effects models. Predictive ability of risk and/or symptom scores was expressed as the area under the receiver operator curve (AUC). Results: Twenty-two studies were identified and included a variety of mobilization strategies (eight studies) and risk and/or symptom scores (fourteen studies). The overall pooled AEHI yield was 6.3% (95% CI, 2.1 to 12.4; I2 = 94.9%; five studies); yield varied between studies using targeted strategies (11.1%; 95% CI, 5.9 to 17.6; I2 = 83.8%; three studies) versus universal testing (1.6%; 95% CI, 0.8 to 2.4; two studies). The AUC of risk and/or symptom scores ranged from 0.69 to 0.89 in development study samples, and from 0.51 to 0.88 in validation study samples. AUC was the highest for scores including symptoms, such as diarrhoea, fever and fatigue. Key risk score variables were age, number of sexual partners, condomless receptive anal intercourse, sexual intercourse with a person living with HIV, a sexually transmitted infection, and illicit drug use. No studies were identified that assessed AEHI yield among MSM in SSA and risk and/or symptom scores developed among MSM in SSA lacked validation. Conclusions: Strategies mobilizing MSM for targeted AEHI testing resulted in substantially higher AEHI yields than universal AEHI testing. Targeted AEHI testing may be optimized using risk and/or symptom scores, especially if scores include symptoms. Studies assessing AEHI yield and validation of risk and/or symptom scores among MSM in SSA are urgently needed. Keywords: acute HIV infection; early HIV infection; men who have sex with men; targeted screening; risk score; mobilization; systematic review

Additional Supporting information may be found under the Supporting Information tab for this article.

Received 14 January 2020; Accepted 14 July 2020 Copyright © 2020 The Authors. Journal of the International AIDS Society published by John Wiley & Sons Ltd on behalf of the International AIDS Society. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

1 | INTRODUCTION Africa: ranging from 5.1/100 person years (PY) (95% confidence interval [CI], 2.6 to 9.8) in Kenya to 12.5/100 PY (95% In 2018, sub-Saharan Africa (SSA) faced approximately one CI, 8.1 to 19.2) in South Africa and 15.4/100 PY (95% CI, million new HIV infections [1]. Although HIV disproportionally 12.3 to 19.0) in Nigeria [5-7]. An important driver in the affects men who have sex with men (MSM) globally [2,3], HIV ongoing HIV epidemic among MSM in SSA could be acute and testing and treatment cascade estimates among African MSM early HIV infections (AEHI), as high viral loads during AEHI are well below target goals set by UNAIDS [4]. lead to a high probability of transmission [8,9]. Therefore, HIV incidence estimates among MSM in sub-Saharan Africa AEHI is important to diagnose and treat to mitigate onward (SSA) are 10 to 15 fold higher than in general populations in transmission risk in MSM [10]. Furthermore, immediate

106 Palmer S et al. Journal of the International AIDS Society 2020, 23(S6):e25590 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25590/full | https://doi.org/10.1002/jia2.25590 treatment after identification of AEHI restores the immune and “mobilization” from database inception to the search date function of people with AEHI [11-14]. mentioned earlier, without geographical or language restric- Acute HIV infection (AHI) is typically defined as the first tions. The keywords represented three domains: domains one weeks after HIV acquisition, during which HIV antibodies are and two identified studies pertaining to MSM and AEHI undetectable [15]. AHI can be diagnosed with HIV-RNA test- respectively. The third domain sought to capture studies that ing using nucleic acid amplification testing (NAAT) and/or HIV focused on mobilization strategies, which included methods of p24-antigen testing [16,17]. Early HIV infection (EHI) is usu- communication with MSM. The full search strategy is ally defined as the first months after HIV acquisition [18,19]. described in Table S1. Experts in the field and secondary ref- In this period, HIV antibody tests are often indeterminate. erence searching on included studies identified additional Therefore, diagnosis of EHI requires a combination of HIV studies. antibody, HIV-RNA, and/or p24 assays [8,18-20]. While AEHI testing, here defined as testing with a combination of HIV 2.2 | Inclusion criteria and screening antibody, HIV-RNA and p24 assays, was not available in most of SSA until recently, the emergence of point-of-care HIV- Studies were included when the following inclusion criteria RNA testing in SSA enables AEHI testing among a range of were met: (1) the study described a strategy of mobilizing populations [21]. In some well-resourced countries, national MSM for AEHI testing; or (2) the study described the devel- guidelines recommend AEHI testing for people who report opment or validation of a risk and/or symptom score which risk behaviour and symptoms associated with AEHI [22,23], could increase the yield of AEHI in MSM. Studies were and facility-based AEHI testing with HIV-RNA can successfully excluded if they merely assessed knowledge of AEHI among identify AEHI among MSM [16,24-29]. Unfortunately, global MSM, assessed AEHI laboratory testing techniques, described policies do not recommend AEHI testing for MSM [30]. AEHI testing among MSM who had already presented for HIV Modelling and phylogenetic transmission studies suggest testing, did not include the number of AEHI cases, or that 10% to 50% of HIV transmission events occur during described AEHI testing among MSM who had already pre- AEHI [8,31-35]. In order to reduce HIV incidence among sented for HIV testing (e.g. laboratory evaluations of pooled MSM, screening strategies should target MSM with the high- samples obtained from MSM who had tested for HIV). Peer- est risk behaviour, as AEHI yield will be the highest [36]. Ide- reviewed articles and conference abstracts were included. For ally, all people at risk of HIV acquisition should be tested for each conference abstract meeting the inclusion criteria, a AEHI. However, this may not be feasible in less-resourced set- specific search was set out to identify the subsequent peer- tings due to the high costs of AEHI testing. Focussing on yield reviewed article of the study, as such, no conference abstracts would therefore limit the number of people that require AEHI were included in the final review. Two independent reviewers testing, while increasing the number of people diagnosed with (SP and MD) used rayyan.qcri.org to screen titles and AEHI [36]. Behaviour risk scores can identify MSM with high- abstracts of records identified through the search to remove risk behaviour [37,38]. Thus, risk and/or symptom scores may non-relevant records. Full-text records were then assessed for assist in defining which subpopulations should be targeted for eligibility. Discrepancies were resolved by discussion with a AEHI testing [39,40]. third and fourth reviewer (EJS and GJB). We assessed study Recently, a systematic review assessed strategies to quality using the Appraisal tool for Cross-sectional Studies increase HIV testing among MSM [41]. Authors concluded (AXIS; Table S2)[43]. that social network-based strategies, community-based testing, HIV self-testing and modifications to the traditional facility- 2.3 | Data extraction based model can effectively reach urban MSM. However, AEHI testing strategies were not reviewed. The aim of this Data were extracted by two independent reviewers (SP and study was to conduct a systematic review and meta-analysis MD) using a standardized form. If studies reported on both of (1) AEHI yield among MSM mobilized for AEHI testing; and MSM and other populations, we extracted data for MSM only if (2) assess which risk factors and/or symptoms could increase disaggregated data were available, otherwise we included esti- AEHI yield in MSM. mates of the whole sample. We contacted study authors when additional information was needed. A modified framework from Campbell et al. was applied [41]. Studies were categorized 2 | METHODS according to two principal testing categories: (1) mobilization for AEHI testing, and (2) risk and/or symptom score screening. The Preferred Reporting Items for Systematic Reviews and Mobilization for AEHI testing included three subcategories: Meta-Analysis (PRISMA)-statement was followed, which pro- media campaigns, partner notification services (PNS) and vides items for reporting in systematic reviews and meta- community-based testing. The data extracted included the fol- analyses [42]. lowing: AEHI cases identified, the total number of visits during which AEHI was assessed, year of publication, year of conduct, country, study population and study design. For the papers con- 2.1 | Search strategy cerning mobilization strategy, we extracted the mobilization On 25 May 2020, we searched PubMed, Embase.com, Clari- strategy, eligibility criteria for AEHI testing, and AHI and EHI vate Analytics/Web of Science Core Collection and Ebsco/ definitions. For risk and/or symptoms scores a list of risk factors ERIC using search terms, including synonyms and related and/or symptoms included in the score, the recall period, cut-off terms, and keywords such as “men who have sex with men,” value of the score, the area under the receiver operator curve “homosexuality,”“acute HIV infection,”“early HIV infection,” (AUC), sensitivity and specificity of the score.

107 Palmer S et al. Journal of the International AIDS Society 2020, 23(S6):e25590 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25590/full | https://doi.org/10.1002/jia2.25590

records were identified from other sources: five from sec- 2.4 | Mobilization for acute and early HIV infection ondary reference searching [38,49-52] and two from expert testing recommendation [53,54]. Taken together, 22 records met the In literature, different definitions are being used for AEHI inclusion criteria: eight studies concerned strategies mobilizing based on the interval between infection and evolution of HIV MSM for AEHI testing [51,55-61] and another 14 studies tests as well as dynamics in antibodies over time. We used dealt with risk and/or symptom score screening [17,37- AEHI definitions as proposed by authors of the included stud- 40,49,50,52-54,62-65]. Critical appraisal showed that none of ies. These varying definitions may have biased the cumulative the included studies justified their sample size and most stud- results of this systematic review, however, we were unable to ies did not address, categorize or describe information about standardize AEHI definitions across the included studies as non-responders (Table S3). study authors reported results based on the above-described definitions. We defined AEHI yield as the proportion of identi- 3.2 | Characteristics of mobilization studies fied AEHI cases among the number of visits during which AEHI was assessed. Targeted AEHI testing was defined as Of the eight studies that assessed strategies mobilizing MSM testing among a selected subgroup of MSM based on high-risk for AEHI testing, seven studies originated from well-resourced behaviour and/or AEHI symptoms. This was opposed to uni- settings [51,55-59,61]. One study originated from a less- versal AEHI testing, defined as testing all MSM. Outcomes resourced setting and was conducted in Thailand [60] included type of mobilization strategy, and AHI and AEHI (Table 1). All eight studies were cross-sectional studies and yield. were conducted between 1996 and 2017 [51,55-61]. Seven studies exclusively targeted MSM [55-61]. One study included sexual or injection drug equipment partners of people living 2.5 | Data analysis with HIV (PLWH) [51]. Although this study did not specify the We pooled independent study estimates for AEHI yield using number of MSM included, they predominantly targeted MSM the Freeman-Tukey double arcsine transformation in random during recruitment. effects models based on the method of DerSimonian and Laird [44,45]. Exact binomial procedures were used to calcu- 3.3 | Strategies for mobilization for acute and early late 95% CIs [46]. Pooled estimates were back-transformed HIV infection testing on their original scale. Heterogeneity across estimates was assessed using the I2 statistic [47]. After observing large The eight studies that assessed strategies mobilizing MSM for heterogeneity across the estimates, we performed sub-group AEHI testing included four studies assessing the impact of analyses of studies assessing targeted AEHI and AHI testing media campaigns [51,56,57,61], one describing PNS for people and studies assessing universal AEHI and AHI testing. Analy- with AEHI [58], and three describing community-based testing ses were performed using the Metaprop package [48] in Stata for AEHI [55,59,60]. Three studies reported on targeted AEHI (version 15.1; StataCorp). testing [51,58,61] and five studies on universal AEHI testing [55-57,59,60]. Media campaigns aimed to target MSM to increase knowl- 2.6 | Risk and/or symptom score screening edge and awareness of AEHI, the increased transmission risk, Outcomes included AUC, sensitivity and specificity for risk AEHI symptoms, AEHI tests and early treatment. Further- and/or symptom scores. We extracted (or calculated, if not more, they aimed to increase motivation to test for AEHI and provided by authors) sensitivity and specificity at the score included referral for facility-based AEHI testing. The cam- cut-off as proposed by the authors of included studies. We paigns were developed and promoted in conjunction with defined internal validation as assessment of predictive ability MSM community-based organizations [51,56,57,61]. Resources (AUC, sensitivity and specificity) of a risk and/or symptom included print advertisements, condom packs, billboards, pos- score in a different study sample from the same location as ters, web-based advertisements (e.g. on dating websites and the study sample in which the score was developed (i.e. the applications) and campaign websites. These were promoted at dataset was randomly split in a development and validation MSM community-based events and MSM venues such as bars dataset or split based on calendar year). We defined external and bathhouses, MSM-targeted magazines and HIV testing validation as assessment of predictive ability of a risk and/or facilities. symptom score in a study sample from a different location as One study offered PNS to people with AEHI (index clients) the study sample in which the score was developed. [58]. The target population included MSM sexual or injection drug equipment partners of index clients with AEHI. Referral 3 | RESULTS was done by index clients, with or without assistance of a healthcare provider, or by a healthcare provider without disclosing the identity of the index client. 3.1 | Study selection Three studies assessed community-based AEHI testing at We identified 1632 records through the database search MSM venues [55,59,60]. The target population consisted of (Figure 1). Following the removal of 685 duplicates, 947 MSM visiting the venues. Venues included bathhouses, saunas, records were screened for title and abstract. Of these, 873 spas, bars, clubs and local non-governmental organizations. non-relevant records were excluded and 74 full-text records Collection of samples, conduction of rapid antibody tests and were assessed for eligibility, of which 15 records met the eligi- delivery of rapid antibody test results took place on-site at bility criteria and were included in this study. Seven additional the venues. AEHI testing was laboratory based.

108 Palmer S et al. Journal of the International AIDS Society 2020, 23(S6):e25590 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25590/full | https://doi.org/10.1002/jia2.25590

Figure 1. Study selection. AEHI, acute and early HIV infection; ERIC, Education Resources Information Center; MSM, men who have sex with men.

the five studies assessing universal testing this was 0.2% 3.4 | Definitions of acute and early HIV infection (95% CI, 0.1 to 0.3; I2 = 49.3%) [55-57,59,60]. The highest AHI was defined as a positive HIV-RNA test and a negative AHI yield was recorded in a study among MSM partners of antibody test in six included studies [55-57,59-61], as a pos- people with AEHI: 4.9% (95% CI, 1.6 to 11.0) [58]. Three itive HIV-RNA test and an indeterminate antibody test in studies assessed whether implementation of the media cam- one study [58], or as a positive HIV-RNA test and a posi- paign led to increased AHI yield compared with pre- tive antibody test and a documented negative antibody test implementation: AHI yield increased in Vancouver and Ams- in the previous 30 days in one study [51]. Five included terdam post-implementation, but not in Seattle [56,57,61]. studies defined and reported on EHI, varying from a nega- This assessment was quantified by two studies, therefore, tive or indeterminate Western blot test to a documented or we included post-implementation estimates in the pooled self-reported negative antibody test in the previous six analysis [57,61]. months [51,55,58,59]. HIV tests included (pooled) HIV plasma viral load, point-of-care HIV-RNA tests, fourth gener- 3.7 | Characteristics of risk and/or symptom score ation antigen/antibody tests, rapid antibody tests and Wes- studies tern blot. Of the 14 studies that assessed risk and/or symptom score screening, 11 studies originated from well-resourced settings 3.5 | Acute and early HIV infection yield [37,39,40,49,50,52,54,62-65] (Table 2). The three studies The above-described mobilization strategies resulted in a from less-resourced settings originated from Kenya pooled AEHI yield of 6.3% (95% CI, 2.1 to 12.4; I2 = 94.9%; [17,38,53]. There were four cross-sectional studies [39,40, five studies [51,55,58,59,61]); this was 11.1% (95% CI, 5.9 to 62,64], seven prospective cohort studies [17,37,38,53, 17.6; I2 = 83.8%) among the three studies assessing targeted 54,63,65], one retrospective cohort study [52], one study testing [51,58,61], and 1.6% (95% CI, 0.8 to 2.4) among the analysed both cross-sectional data and data from a random- two studies assessing universal testing [55,59] (Figure 2). ized controlled trial (RCT) [49], and one study analysed data solely originating from RCTs [50]. These studies used data- sets collected between 1984 and 2018. Twelve studies 3.6 | Acute HIV infection yield exclusively included MSM [17,37-39,49,50,52-54,63-65] and The overall pooled AHI yield was 0.7% (95% CI, 0.4 to 1.2; two studies focused on people who had presented for HIV I2 = 90.9%; eight studies) [51,55-61]. Among the three stud- testing (e.g. clients of sexually transmitted infection [STI] ies assessing targeted testing, the pooled AHI yield was clinics) [40,62], of which MSM were the vast majority 3.3% (95% CI, 2.2 to 4.6; I2 = 0%) [51,58,61], and among (>70%) of participants.

109 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25590/full amrSe al. et S Palmer

Table 1. Studies assessing strategies to mobilize men who have sex with men for testing for acute and early HIV infection Society AIDS International the of Journal

Partner Media campaigns notification Community-based testing services First author Silvera Stekler Gilbert Dijkstra Green Daskalakis Liang Pankam

Site New York City [51] Seattle [56] Vancouver [57] Amsterdam [61] San Diego [58] New York City [55] Hong Kong [59] Bangkok [60] Country USA USA Canada The Netherlands USA USA Hong Kong Thailand Years study 2004 to 2008 2004 to 2009 2006 to 2012 2008 to 2017 1996 to 2014 2007 2010 to 2011 2011 to 2012 conducted 2020,

Year of publication 2010 2013 2013 2020 2017 2009 2015 2018 | https://doi.org/10.1002/jia2.25590

Study design Cross-sectional Cross-sectional Cross-sectional Cross-sectional Cross-sectional Cross-sectional Cross-sectional Cross-sectional 23

Target population Heterosexual men MSM MSM MSM MSM MSM MSM MSM (S6):e25590 and women, MSM Mobilization Media campaign Media campaign Media campaign Media campaign PNS MSM venue-based MSM venue-based MSM venue-based strategy testinga testingb testingc Eligibility criteria for Sex or sharing injection MSM presenting Men, TGP presenting ARS and CAIe MSMg partners MSM venue visitors MSM venue MSM venue AEHI testing drug equipment with for HIV testing for HIV testingf of people visitors visitors and reporting aPLWHd,e with AEHI sex with menh Targeted AEHI testing Yes No No Yes Yes No No No AHI definition RNA +i +j +k +l +m +n +o +p Ag/Ab NP NP À or + q À or Ær NP NP NP À or + r Ab À or + s Àt Àt Àu À or Æu Àu Àu Àt,u WB À or Æ or + v ÀÀ À NP À NP NP EHI definition Infection NS NS WBÀ or WBÆ Infection Seroconversion Seroconversion NS <129 daysw <170 daysw <170 daysw <6 monthsx

Ab, second or third generation rapid antibody test; AEHI, acute and early HIV infection; Ag/Ab, fourth generation antigen/antibody test; AHI, acute HIV infection; ARS, acute retroviral syndrome; CAI, condomless anal intercourse; EHI, early HIV infection; MSM, men who have sex with men; NP, not performed; NS, not specified; PLWH, person living with HIV; PNS, partner notification services; RNA, ribonucleic acid; TGP, transgender people; USA, United States of America; WB, western blot. aMSM bathhouses; bMSM bars, saunas, clubs and a local non-governmental organization; cMSM saunas and spa venues; dor of unknown HIV status; ein the previous three months; for if sex was missing; gthe original study did not report solely on MSM, disaggregated data on MSM partners (as reported here) were provided by the authors; hat least once in their lifetime; iHIV plasma viral load; jpooled HIV nucleic acid; k<2009: HIV nucleic acid; ≥2009: Pooled HIV nucleic acid; lpoint-of-care HIV-RNA; m<2007: Quantitative HIV-RNA; ≥2007: HIV nucleic acid; npooled HIV viral load; opoint-of-care real-time dried blood spot-based quantitative polymerase chain reaction; pHIV nucleic acid, HIV viral load; q<2009: p24 antigen, discontinued from 2009; rfourth generation antigen/antibody; sEzyme-linked immunosorbent assay; tEnzyme immunoassay; urapid antibody; vor a documented negative antibody test in the previous 30 days; westimated by recency assays or a serologic testing algorithm for recent seroconversion[18,19]; xpositive rapid antibody test with self-reported negative antibody test in the previous six months. 110 Palmer S et al. Journal of the International AIDS Society 2020, 23(S6):e25590 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25590/full | https://doi.org/10.1002/jia2.25590

4 | DISCUSSION 3.8 | Risk and/or symptom score screening The 14 studies assessed predictive ability of 13 independent In this systematic review and meta-analysis, we showed sub- risk and/or symptom scores to target AEHI testing among stantial AHI and AEHI yields when MSM were mobilized for MSM [17,37-40,49,50,52-54,62-65]. In total, the 14 studies AEHI testing in studies predominantly conducted in well- included 26 score outcomes (including AUC, sensitivity and resourced settings. With the severe ongoing HIV epidemic specificity from nine development and 17 validation out- among MSM in SSA [5-7], infrequent HIV testing and poor comes), as most scores were assessed multiple times linkage to care and viral suppression outcomes [4], there is an (Table 3). Four scores were not validated [17,38,52,53]. urgent need to better identify AEHI in MSM. As such, targeted AEHI testing will likely result in high AEHI yields among MSM in SSA. Unfortunately, the World Health Organization 3.9 | Variables included in risk and/or symptom (WHO) has no targeted AEHI testing recommendation for key scores populations, including MSM who have among the highest inci- The recall period for risk factors and symptoms included in the dences [5-7]. Thus, AEHI testing should be offered to MSM, scores varied from two weeks to two years. The 13 scores com- be supported by specific policy recommendations for MSM, prised eight scores only including demographic or behavioural and AEHI testing guidelines tailored to SSA need to be devel- risk factors for HIV acquisition [17,39,49,50,52,54,62], four oped and endorsed by WHO. scores including risk factors and AEHI symptoms [38,53,65] Strategies mobilizing MSM for targeted AEHI testing and one score including only AEHI symptoms [40] (Table 3). resulted in higher AEHI yields than strategies mobilizing MSM Most frequently included risk factors were age, number of sex- for universal AEHI testing. Targeted AEHI testing may be opti- ual partners, condomless receptive anal intercourse (CRAI), sex- mized by screening with risk and/or symptom scores. The ual intercourse with a PLWH, self-reported diagnosis of an STI pooled AEHI yield was the highest when testing was targeted and illicit drug use. Most frequently included symptoms were to MSM partners of people with AEHI, to partners of PLWH, self-reported diarrhoea, fever and fatigue [17,38,40,53,65]. or to MSM with AEHI symptoms who reported CRAI (11.1%). Three scores were incorporated in MSM-targeted websites, to Although our review identified one study with a high AEHI allow for self-assessment of HIV risk (www.hebikhiv.nl/en; www. yield resulting from PNS [58], two other studies did not IsPrEPforMe.org; http://sdet.ucsd.edu [39,52,61]). assess and report on AEHI yield resulting from PNS for index clients with AEHI, and were therefore not included in this review [66,67]. When focussing only on AHI, the pooled AHI 3.10 | Predictive ability of the risk and/or symptom yield among studies assessing targeted testing was 3.3%. scores Collaboration with MSM community-based organizations The AUC ranged from 0.69 to 0.89 in development study was key in successfully mobilizing MSM for AEHI testing, samples, and from 0.51 to 0.88 in validation study samples either through the design and promotion of AEHI media cam- (Table 4 and Figure 3). Sensitivity at the cut-off proposed by paigns, or through the delivery of community-based testing the authors ranged from 74% to 98% in development study [51,55-57,59-61]. In the studies included in this review, on- samples, and from 25% to 94% in validation samples. Speci- site AEHI diagnosis was not possible in community-based test- ficity was between 17% and 90% in development study sam- ing settings, but required laboratory-based tests and skilled ples, and between 15% and 96% in validation study samples. laboratory personnel. The emergence of point-of-care HIV- Internal and external validation resulted in lower predictive RNA tests may enable on-site community-based AEHI testing ability for most scores. For example the San Diego Early Test in SSA [21]. However, no study approached AEHI testing in a (SDET) score yielded an AUC of 0.74 (95% CI, 0.70 to 0.79) comprehensive, culturally sensitive and integrated fashion in in the development study sample, and between 0.55 (95% CI, SSA. As such, these strategies need to be urgently developed 0.44 to 0.66) to 0.70 (95% CI, 0.63 to 0.78) in external valida- in close collaboration with local community-based organization samples [37,39,63]. A study in Atlanta validated three tions, including the need to include learning about point-of- scores (SDET, HIRI-MSM and the Menza score) in a cohort care HIV-RNA testing when locally available. While WHO rec- with a high proportion of HIV seroconversions among Black ommends regular HIV testing for MSM, we suggest that MSM MSM, whereas the scores had been developed and previously with unknown or HIV-negative status who experience AEHI validated in study samples consisting of predominantly white symptoms or meet risk criteria be evaluated for AEHI, espe- MSM [63]. The three scores performed poorly in this valida- cially when PrEP initiation is considered [68]. tion study sample among Black MSM and had markedly lower Opportunities to diagnose AEHI are often missed, due to AUC values than in other validation study samples. This was the non-specificity of symptoms and the costly diagnostic also the case for a validation study in Chicago among young assays required for AEHI diagnosis [69-72]. The studies Black MSM [54]. Two scores showed high predictive ability in included in this review used several testing strategies to iden- both the development and validation study samples: the Ams- tify AEHI, including point-of-care HIV-RNA testing and terdam score yielded AUC values of 0.78 (95% CI, 0.74 to (pooled) HIV viral load testing. A study in San Diego showed 0.82) and 0.88 (95% CI, 0.84 to 0.91) in external validation that AEHI testing with HIV-RNA testing was cost-effective in study samples [64,65], the San Diego Symptom Score (SDSS) populations of MSM with an HIV prevalence above 0.4% [73]. yielded an AUC of 0.85 (95% CI, 0.78 to 0.92) in internal vali- Since HIV prevalence in MSM in SSA is estimated to be well dation [40]. Both scores included symptoms. Other scores, all above this threshold [2], AEHI testing among SSA MSM may from Kenya, with high AUC values in development study sam- also be cost-effective, although evidence hereof is lacking. Fur- ples (0.76 to 0.89) have not been validated [17,38,53]. thermore, targeting resources to specific subpopulations of

111 Palmer S et al. Journal of the International AIDS Society 2020, 23(S6):e25590 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25590/full | https://doi.org/10.1002/jia2.25590

AEHI yield (A) % AEHI Total

Study ES (95% CI) Weight cases visits

Targeted screening

Silvera (2010) - USA* 9.5 (7.0, 12.4) 20.65 47 497

Green (2017) - USA 20.4 (13.1, 29.5) 18.16 21 103

Dijkstra (2020) - NL 6.8 (4.0, 10.7) 19.94 17 249

Subtotal 11.1 (5.9, 17.6) 58.74

Universal screening

Daskalakis (2009) - USA 1.6 (0.7, 3.2) 20.64 8 493

Liang (2015) - Hong Kong 1.5 (0.6, 3.0) 20.61 7 474

Subtotal 1.6 (0.8, 2.4) 41.26

Overall 6.3 (2.1, 12.4) 100.00

0 5 10 15 20 25 30 Proportion (%)

AHI yield (B) % AHI Total Study ES (95% CI) Weight cases visits

Targeted screening Silvera (2010) - USA* 2.8 (1.5, 4.7) 12.14 14 497 Green (2017) - USA 4.9 (1.6, 11.0) 4.53 5 103 Dijkstra (2020) - NL 4.0 (1.9, 7.3) 8.44 10 249 Subtotal 3.3 (2.2, 4.6) 25.12

Universal screening Daskalakis (2009) - USA 0.2 (0.0, 1.1) 12.10 1 493 Stekler (2013) - USA 0.2 (0.1, 0.2) 21.44 52 27661 Gilbert (2013) - Canada 0.2 (0.2, 0.3) 21.36 54 21967 Liang (2015) - Hong Kong 0.6 (0.1, 1.8) 11.89 3 474 Pankam (2018) - Thailand 0.9 (0.1, 3.1) 8.10 2 233 Subtotal 0.2 (0.1, 0.3) 74.88

Overall 0.7 (0.4, 1.2) 100.00

0 5 10 15 20 25 30 Proportion (%)

Figure 2. Forest plots of acute HIV infection yield and acute and early HIV infection yield among men who have sex with men. Study estimates and their 95% CIs, and pooled estimates and their 95% CIs for AEHI yield, overall and stratified by testing strategy: targeted testing and universal testing. (A) Displays AEHI yield, (B) displays AHI yield. Yield was defined as the proportion of AEHI cases among the number of visits during which AEHI was assessed. The size of the grey boxes represents a study’s weight in the meta-analysis. *The study population was men who have sex with men in all studies, with the exception of Silvera et al. In this study, heterosexual men, women and men who have sex with men were included, however, they predominantly targeted MSM during recruitment. AHI, acute HIV infection; AEHI, acute and early HIV infection; CI, confidence interval; ES, effect size; N/A, not accessible; NL, the Netherlands; USA, United States of America.

112 Palmer S et al. Journal of the International AIDS Society 2020, 23(S6):e25590 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25590/full | https://doi.org/10.1002/jia2.25590

Table 2. Overview of published risk and/or symptom scores to assist screening for acute and early HIV infection among men who have sex with men

Development Years study Year of Study Score (D) and/or First author conducted publication Site Country Study design population namea validation (V)

Menza [49] 1999 to 2008 2009 Boston, Chicago, USA Cross-sectional/ RCT MSM Menza D and V Denver, New York, San Francisco, Seattle Facente [62] 2004 to 2007 2011 San Francisco USA Cross-sectional STI clinic Facente D and V clients Smith [50] 1998 to 2001 2012 57 cities USA RCT MSMb HIRI-MSM D and V Wahome [38] 2005 to 2012 2013 Kilifi Kenya Prospective cohort MSM CDRSS D UMRSSc V Hoenigl [39] 2008 to 2014 2015 San Diego USA Cross-sectional MSM SDET D and V HIRI-MSM V Menza V Sanders [53] 1993 to 2012 2015 Kilifi Kenya Prospective cohort MSM Sanders D Beymer [52] 2009 to 2014 2017 Los Angeles USA Retrospective cohort MSM Beymer D Jones [63] 2010 to 2014 2017 Atlanta USA Prospective cohort MSM SDET V HIRI-MSM V Menza V Dijkstra [65] 1984 to 2009 2017 Amsterdam, The Netherlands, Prospective cohort MSM Amsterdam D and V Baltimore, USA score Chicago, Pittsburg, Los Angeles Lancki [54] 2013 to 2016 2018 Chicago USA Prospective cohort MSM CDC V HIRI-MSM V Gilead V Wahome [17] 2005 to 2016 2018 Kilifi Kenya Prospective cohort MSM Wahome D Lin [40] 2007 to 2017 2018 San Diego USA Cross-sectional STI clinic SDSS D and V clientsd Lin [64] 2007 to 2017 2018 San Diego USA Cross-sectional MSM Amsterdam V score Dijkstra [37] 2003 to 2018 2019 Amsterdam The Netherlands Prospective cohort MSM SDET V

CDC, Centers For Disease Control and Prevention; CDRSS, Cohort Derived Risk Screening Score; D, development; HIRI-MSM, HIV Incidence Risk Index for MSM; MSM, men who have sex with men; RCT, randomized controlled trial; SDET, San Diego Early Test; SDSS, San Diego Symptom Score; STI, sexually transmitted infection; UMRSS, University of North Carolina Malawi Risk Screening Score; USA, United States of America; V, validation. a14 studies assessed predictive ability of 13 independent risk and/or symptom scores, five scores were assessed multiple times; b75.0% (9472/ 12622) of participants were MSM; cThe development study of the UMRSS was not included in this review, as it did not include MSM; d73.8% (737/998) of participants were MSM.

MSM (e.g. those reporting high-risk behaviour and/or symp- AEHI. Several risk and/or symptom scores have been included toms) can substantially reduce costs compared with universal in MSM-targeted websites, facilitating self-assessment of HIV AEHI testing [36]. acquisition risk [39,52,61], although outcomes of these self- We identified 13 risk and/or symptom scores that may assessment tools need to be evaluated. increase AEHI yield in MSM. Key risk factors included in these Predictive ability of the 13 risk and/or symptom scores var- scores were age, number of sexual partners, CRAI, sexual ied greatly and was highest for scores that included symptoms intercourse with a PLWH, self-reported diagnosis of an STI [40,53,64,65]. Validation showed lower discriminate ability of and illicit drug use. Key symptoms were self-reported diar- most risk and/or symptom scores in the validation study sample rhoea, fever and fatigue. As knowledge of symptoms of AEHI than in the development study sample [52,54,63]. This was among MSM is low [74,75], these risk factors and symptoms specifically the case for validation of risk and/or symptom scores may be used to educate MSM and help them self-recognize among Black MSM in the USA, as the risk and/or symptom

113 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25590/full amrSe al. et S Palmer Table 3. Point values of risk factors and symptoms included in published risk and/or symptom scores to assist screening for acute and early HIV infection among men who have sex with men

HIRI- Society AIDS International the of Journal Menza Facente MSM CDRSS UMRSS Sanders Beymer Amsterdam Gilead Wahome SDSS Score name [49] [62] [50] [38] [38]SDET[39] [53] [52] score [65] CDC [54] [54] [17] [40]

Recall period 6 to 12 2 yearsa 6 4to12 4to12 12 4to12 1to12 6 6 NS 1 to 12 2 months months weeks weeks months weeks months months months weeks weeks Cutoff ≥1 ≥2 ≥10 ≥2 ≥2 ≥5 ≥2 ≥5b ≥1.5 ≥1c ≥1 ≥1 ≥11 Risk or symptom Risk Risk Risk Risk/ Risk/ Risk Risk/symptom Risk Risk/ Risk Risk Risk Symptom score symptom symptom symptom Point values Risk factors 2020, d e e f g

Age 2 to 8 1 1 0.27 to 0.48 1 | https://doi.org/10.1002/jia2.25590 h

Ethnicity 0.27 to 0.68 23

MSM 1 (S6):e25590 Sex with only men 1 IDU 1 Incarceration 1 No. of partners 3i 4to7j 1k 2i 0.01l 0.9m Partner 0.005 to 0.45n 1o characteristics IPV 0.31 RAI 0.35p 1 CI 11 CRAI 1 10 3q 0.61 1.1 1r HIV + partner 1 4 to 8s 1 CAI with 16t 3u HIV + partner Group sex 1 Transactional sex 1 Self-reported STI 4 1v 1 2 0.19 to 0.75w 1.6x 11 Methamphetamine 11y 5 0.49 1z use Inhaled nitrites 3 0.45 Ecstasy use 0.21 Discordant HIV rapid 44 antibody tests 114 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25590/full Table 3. (Continued) al. et S Palmer

HIRI- Menza Facente MSM CDRSS UMRSS Sanders Beymer Amsterdam Gilead Wahome SDSS ora fteItrainlAD Society AIDS International the of Journal Score name [49] [62] [50] [38] [38]SDET[39] [53] [52] score [65] CDC [54] [54] [17] [40]

Symptoms Body pains/ myalgia 1 8 Diarrhoea 1 2 1 Fever 1 1 1 1.6 11 Fatigue 1 2 1 Genital ulcers 3 Lymphadenopathy 1.5 Oral thrush 1.7

Sore throat 1 2020, | Weight loss 0.9 4aa https://doi.org/10.1002/jia2.25590 23

Number of validations (S6):e25590 Internalbb 0100010000001 Externalcc 3040120021100

CAI, condomless anal intercourse; CDC, Center for Disease Control and Prevention; CDRSS, Cohort Derived Risk Screening Score; CI, condomless intercourse; CRAI, condomless receptive anal intercourse; HIRI-MSM, HIV Incidence Risk Index for MSM; HIV+, HIV-infected; IDU, injection drug use; IPV, intimate partner violence; MSM, men who have sex with men; NS, not specified; RAI, receptive anal intercourse; SDET, San Diego Early Test; SDSS, San Diego Symptom Score; STI, sexually transmitted infection; UMRSS, UNC Malawi Risk Screening Score. aOr since last HIV test; bfor all risk and/or symptom scores, the point values of the variables in the score were summed to obtain an individual’s score, except for Beymer’s score: the point values were added and then exponentiated; can individual’s score was only assessed if he reported any male sex partners in previous six months, was not in a monogamous partnership with a recently tested or HIV-uninfected man; d18 to 28 years = 2 points, 29 to 40 years = 5 points, 41 to 48 years = 2 points; e18 to 29 years; f<25 years = 0.48 points, 25 to 29 years = 0.36 points, 30 to 39 years = 0.27 points; g18 to 24 years; hblack = 0.68 points, Hispanic = 0.52 points, other = 0.27 points; i>9 partners; j6 to 10 partners = 4 points, >10 partners = 7 points; k>1 partners; l≤3or> 3 partners; m>5 partners; n>age of last sex partner five years older; within five years of age; or >5 years younger = 0.005 points, same ethnicity as last partner = 0.45 points; opartners of unknown HIV status with any of the following factors: inconsistent or no condom use, STI, transactional sex, use of illicit drugs or alcohol dependence, incarceration; pRAI with a condom; qCRAI and >4 partners; rany condomless anal sex (insertive or receptive); s1 HIV-infected partner = 4 points; >1 HIV-infected partners = 8 points; tcondomless insertive anal intercourse with >5 HIV-infected partners; ucondomless receptive anal intercourse with an HIV-infected partner; va simplified model without STI had similar performance but was not included in this review; wdiagnosed with an STI > 1 year ago = 0.19 points, <1 year ago = 0.75 points; xself-reported gonorrhoea; yor use of inhaled nitrites; zuse of illicit drugs or alcohol dependence (excluding marijuana); aa≥2.5 kg; bb assessment of predictive ability of the score in a different study sample from the same location as the study sample in which the risk and/or symptom score was developed; ccassessment of predictive ability of the risk and/or symptom score in a study sample from a location different to the study sample in which the score was developed. 115 Palmer S et al. Journal of the International AIDS Society 2020, 23(S6):e25590 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25590/full | https://doi.org/10.1002/jia2.25590

Table 4. Predictive ability of published risk and/or symptom scores to assist screening for acute and early HIV infection among men who have sex with men

Total visits (n) AEHI cases (n) AUC (95% CI) Sensitivity (%)b Specificity (%)b

First author Score namea DVDVD V DVDV

Menza [49] Menza NS NS 101 104 0.69 (0.60 to 0.74) 0.66 (0.61 to 0.71) 83% 86% 30% 29% Facente [62] Facente 12,350 12,249c 137 36 0.67 (NS) 83% 50% Smith [50] HIRI-MSM 24,391 15,582 320 171 0.74 (NS) NS 84% 81% 45% 38% Wahome [38] CDRSS 6531 73 0.85 (0.80 to 0.90) 81% 76% UMRSS 6531 73 0.79 (0.72 to 0.85) 75% 76% Hoenigl [39] SDET 5568 2758 137 63 0.74 (0.70 to 0.79) 0.70 (0.63 to 0.78) NS 60% NS 77% HIRI-MSM 8326 200 0.70 (0.67 to 0.74) 69% 60% Menza 8326 200 0.63 (0.59 to 0.68) 67% 54% Sanders [53] Sanders 7054 20 0.89 (0.79 to 0.99) 74% 90% Beymer [52] Beymer NS 370 0.6 (NS) 75% 50% Jones [63] SDET 3372 32 0.55 (0.44 to 0.66) 25% 84% HIRI-MSM 372 32 0.62 (0.52 to 0.72) 63% 57% Menza 3372 32 0.51 (0.41 to 0.60) 63% 41% Dijkstra [65] Amsterdam score 17,446 63,618 175 491 0.82 (0.79 to 0.86) 0.78 (0.74 to 0.82) 76% 56% 76% 89% Lancki [54] CDC 866 33 0.51 (NS) 52% 52% HIRI-MSM 866 33 0.580.49 to 0.68 85% 30% Gilead 866 33 0.57 (NS) 94% 15% Wahome [17] Wahome 9143 97 0.76 (0.71 to 0.80) 98% 17% Lin [40] SDSS 673 325 70 43 0.82 (0.76 to 0.88) 0.85 (0.78 to 0.92) NS 72% NS 96% Lin [64] Amsterdam score 757 110 0.88 (0.84 to 0.91) 78% 81% Dijkstra SDET 14,695 0.70 (0.64 to 0.76) 54% 78%

AEHI, acute and early HIV infection; AUC, area under receiver operator curve; CDC, Center for Disease Control and Prevention; CDRSS, Cohort Derived Risk Screening Score; CI, confidence interval; D, Development study sample; HIRI-MSM, HIV Incidence Risk Index for MSM; MSM, men who have sex with men; NS, not specified; SDET, San Diego Early Test; SDSS, San Diego Symptom Score; UMRSS, University of North Carolina Malawi Risk Screening Score; V, Validation study sample. a13 studies assessed predictive ability of 13 independent risk and/or symptom scores, five scores were assessed multiple times; bat the cutoff specified by the authors; cthe HIV negative visits were used in both the development and validation dataset. scores poorly predicted HIV acquisition [54,63]. This underlines However, risk and/or symptom scores are imperfect, and using the importance of external validation of risk and/or symptom a risk and/or symptom score to define who will be tested for scores [76]. Importantly, none of the MSM risk and/or symptom AEHI will inevitably exclude people with AEHI [85,86]. Thus scores developed in SSA were validated [17,38,53]. Further- far, no AEHI yield has been reported resulting from screening more, no risk and/or symptom scores developed in well-re- MSM with published AEHI risk and/or symptom scores. sourced settings have been validated in less-resourced settings. This study has some limitations. First, the database search Scores including symptoms may be particularly useful in strategy did not identify seven out of 22 included studies. SSA, where stigma and discrimination towards MSM behaviour Some of the included studies not identified by the search is high, and social desirability bias may prevent MSM from dis- strategy focused on PrEP screening scores rather than AEHI- closing high-risk behaviour to healthcare providers [77-79]. screening scores. Because these scores may also assist AEHI However, symptoms may vary by HIV-1 subtype [80], limiting screening, we included these studies in this review. Second, the generalizability of symptom-based scores across SSA. heterogeneity across study estimates was large. This was Risk-based scores may assist targeted AEHI screening, but partly explained by different testing strategies; heterogeneity may also be of use in identifying and prioritizing candidates was smaller when we stratified for testing strategy. Another for pre-exposure prophylaxis (PrEP) [81]. Recent studies using possible explanation is the variable definitions for AEHI as machine learning of routine health care data from electronic proposed by study authors. This has possibly overestimated patient records to identify potential PrEP candidates among the AHI yield in studies that included indeterminate or posi- the general population showed high predictive ability of gener- tive antibody tests in their AHI definition [51,58]. Additionally, ated risk-based scores, but included more than 20 variables the variable study designs may have increased heterogeneity. [82-84], which may limit practical use. Simpler risk and/or For risk and/or symptom scores, the high variability in recall symptom scores, consisting of a smaller number of variables, periods (two weeks to two years) will have likely resulted in which requires simple summation to calculate an individual’s variable outcomes. Likewise, the risk and/or symptoms recorded score, could be implemented in resource-limited settings. varied considerably between studies depending on the local

116 Palmer S et al. Journal of the International AIDS Society 2020, 23(S6):e25590 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25590/full | https://doi.org/10.1002/jia2.25590

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2 Area Under Receiver Operator Curve

0.1

0 Menza Facente HIRI-MSM CDRSS UMRSS SDET Sanders Beymer Amsterdam CDC Gilead Wahome SDSS [39,49,63] [62] [39,50,54,63] [38] [38] [37,39,63] [53] [52] score [64,65] [54] [54] [17] [40] 2009-2017 2011 2012-2018 2013 2013 2015-2019 2015 2017 2017-2018 2018 2018 2018 2018 D: USA D: USA D: Kenya D: USA D: Kenya D: USA D: NL D: Kenya D: USA V: USA V: USA V: USA V: Kenya V: USA-NL V: USA V: USA V: USA V: USA

Risk and/or symptom score

Figure 3. Area under receiver operator curves of published risk and/or symptom scores to assist screening for acute and early HIV infection among men who have sex with men. The black dots represent point estimates, the coloured lines 95% confidence intervals. If no coloured lines are displayed, the study did not report 95% confidence intervals. For each risk and/or symptom score, the first point estimate represents the area under receiver operator curve of the development study sample, the latter point estimate(s) represents the area under receiver operator curve of the validation study sample(s). The development outcomes of scores Facente, UMRSS, CDC and Gilead have not been included in this review, therefore, only validation outcomes are represented. CDC, Center for Disease Control and Prevention; CDRSS, Cohort Derived Risk Screening Score; D, Development study sample; HIRI-MSM, HIV Incidence Risk Index for MSM; MSM, men who have sex with men; NL, the Netherlands; NS, not specified; SDET, San Diego Early Test; SDSS, San Diego Symptom Score; UMRSS, University of North Carolina Malawi Risk Screening Score; USA, United States of America; V, Validation study sample. context and how their data collection was set up, thus impacting AUTHORS’ AFFILIATIONS the comparability of different scores. Furthermore, studies origi- 1Centre for Geographic Medicine Research – Coast, Kenya Medical Research nated from various locations with different HIV epidemics, which Institute, Kilifi, Kenya; 2International AIDS Vaccine Initiative, Amsterdam, the Netherlands; 3Department of Infectious Diseases, Public Health Service Amster- has likely increased heterogeneity. Third, we did not standardize 4 the cutoff at which sensitivity and specificity were assessed for dam, Amsterdam, the Netherlands; Department of Internal Medicine, Division of Infectious Diseases, and Amsterdam Institute for Infection and Immunity (AI&II), the risk and/or symptom scores, and as a result, these values Amsterdam UMC, University of Amsterdam, Amsterdam, the Netherlands; 5Medi- varied across studies. This has limited the comparison of sensitiv- cal Library, Vrije Universiteit Amsterdam, Amsterdam, the Netherlands; 6ISHTAR ities and specificities for the risk and/or symptom scores. MSM Health and Social Wellbeing, Nairobi, Kenya; 7Department of Global Health, and Amsterdam Institute for Global Health and Development, Amster- dam UMC, University of Amsterdam, Amsterdam, the Netherlands; 8Nuffield Department of Medicine, University of Oxford, Headington, United Kingdom 5 | CONCLUSIONS COMPETING INTERESTS In conclusion, strategies mobilizing MSM for targeted AEHI testing resulted in higher AEHI yields than universal AEHI test- GJB has received grants through her institution from Bristol-Meyer Squibbs and Mac Aids Fund; honoraria to her Institution for scientific advisory board ing. Targeted AEHI testing may be optimized using risk and/or participations for Gilead Sciences and speaker fees from Gilead Sciences and symptom scores, in particular scores including symptoms. How- Takeda. The remaining authors declared no competing interests. ever, yield of AEHI testing has not been assessed among MSM in SSA and validation of risk and/or symptom scores among AUTHORS’ CONTRIBUTIONS MSM in SSA is urgently needed. With the emergence of point- SP, MD, EW, JW, EG, EME and EJS designed the study. JK designed the search of-care HIV-RNA testing platforms in SSA, MSM with unknown strategy. SP and MD independently assessed records for eligibility, and conducted or HIV-negative status who have AEHI symptoms or meet AEHI data extraction, supported by EW. GJB and EJS had oversight in study selection risk behaviour criteria should be evaluated for AEHI. These pro- and data extraction. MD conducted the statistical analysis and drafted the manu- grammes should be developed in a culturally sensitive fashion, script. MFSVL had oversight in the statistical analysis. All authors critically reviewed and revised the manuscript and approved the final version for publication. for example through collaborating with local community-based organizations to promote learning about AEHI symptoms, and ACKNOWLEDGEMENTS or risk behaviour, particularly in SSA. Further studies should focus on AEHI yield and cost-effectiveness resulting from risk This work was supported by the International AIDS Vaccine Initiative (IAVI) and the KEMRI Wellcome Trust Research Programme at the Centre for Geographi- and/or symptom score screening, and the development and vali- cal Medicine Research–Kilifi, supported by core funding from the Wellcome dation of culturally sensitive approaches to target MSM for Trust [203077]. This study was made possible by the generous support of the AEHI screening in SSA. American people through the United States Agency for International

117 Palmer S et al. Journal of the International AIDS Society 2020, 23(S6):e25590 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25590/full | https://doi.org/10.1002/jia2.25590

Development (USAID). This work was also supported in part through the sub- 19. Janssen RS, Satten GA, Stramer SL, Rawal BD, O’Brien TR, Weiblen BJ, Saharan African Network for TB/HIV Research Excellence (SANTHE), a DELTAS et al. New testing strategy to detect early HIV-1 infection for use in incidence Africa Initiative [DEL-15-006]. The DELTAS Africa Initiative is an independent estimates and for clinical and prevention purposes. JAMA. 1998;280(1):42–8. funding scheme of the African Academy of Sciences (AAS) Alliance for Acceler- 20. Leyre L, Kroon E, Vandergeeten C, Sacdalan C, Colby DJ, Buranapraditkun S, ating Excellence in Science in Africa (AESA) and is supported by the New Part- et al. Abundant HIV-infected cells in blood and tissues are rapidly cleared upon nership for Africa’s Development Planning and Coordinating Agency (NEPAD ART initiation during acute HIV infection. Sci Transl Med. 2020;12:eaav3491. Agency) with funding from the Wellcome Trust [107752] and the UK govern- 21. Agutu CA, Ngetsa CJ, Price MA, Rinke de Wit TF, Omosa-Manyonyi G, San- ment. EJS receives research funding from IAVI, NIH (grant R01AI124968) and ders EJ, et al. Systematic review of the performance and clinical utility of point of the Wellcome Trust. MD receives funding through a PhD Scholarship from the care HIV-1 RNA testing for diagnosis and care. PLoS One. 2019;14:e0218369. Graduate School of Amsterdam UMC. The contents are the responsibility of the 22. The Netherlands Association of HIV Physicians. HIV guidelines [cited 2019 study authors and do not necessarily reflect the views of USAID, NIH, the US Sep 4]. Available from: http://www.richtlijn.nvhb.nl or UK government, AAS, NEPAD Agency, or the Wellcome Trust. This report 23. Panel on Antiretroviral Guidelines for Adults and Adolescents. Guidelines was published with permission from the director of KEMRI. for the use of antiretroviral agents in adults and adolescents with HIV. Depart- ment of Health and Human Services [cited 2019 Sep 4]. Available from: http:// www.aidsinfo.nih.gov/ContentFiles/AdultandAdolescentGL.pdf 24. Lama JR, Brezak A, Dobbins JG, Sanchez H, Cabello R, Rios J, et al. Design REFERENCES strategy of the sabes study: diagnosis and treatment of early HIV infection 1. UNAIDS. Trend of new HIV infections. [cited 2019 Nov 1]. Available from: among men who have sex with men and transgender women in Lima, Peru, http://aidsinfo.unaids.org/ 2013–2017. Am J Epidemiol. 2018;187(8):1577–85. 2. Beyrer C, Baral SD, van Griensven F, Goodreau SM, Chariyalertsak S, Wirtz 25. Robb ML, Eller LA, Rolland M. Acute HIV-1 Infection in Adults in East AL, et al. Global epidemiology of HIV infection in men who have sex with men. Africa and Thailand. N Engl J Med. 2016;375:1195. Lancet. 2012;380(9839):367–77. 26. Han X, Xu J, Chu Z, Dai D, Lu C, Wang X, et al. Screening acute HIV infec- 3. Smith AD, Tapsoba P, Peshu N, Sanders EJ, Jaffe HW. Men who have sex tions among Chinese men who have sex with men from voluntary counseling & with men and HIV/AIDS in sub-Saharan Africa. Lancet. 2009;374(9687):416–22. testing centers. PLoS One. 2011;6:e28792. 4. Stannah J, Dale E, Elmes J, Staunton S, Beyrer C, Mitchell KM, et al. HIV 27. Kuruc JD, Cope AB, Sampson LA, Gay CL, Ashby RM, Foust EM, et al. Ten testing and engagement with the HIV treatment cascade among men who have Years of screening and testing for acute HIV infection in North Carolina. sex with men in Africa: a systematic review and meta-analysis. Lancet HIV. J Acquir Immune Defic Syndr. 2016;71(1):111–9. 2019;6(11):e769–e787. 28. Kerndt PR, Dubrow R, Aynalem G, Mayer KH, Beckwith C, Remien RH, 5. Kimani M, van der Elst EM, Chiro O, Oduor C, Wahome E, Kazungu W, et al. et al. Strategies used in the detection of acute/early HIV infections. The NIMH PrEP interest and HIV-1 incidence among MSM and transgender women in Multisite Acute HIV Infection Study: I. AIDS Behav. 2009;13(6):1037–45. coastal Kenya. J Int AIDS Soc. 2019;22:e25323. 29. Ananworanich J, Fletcher JL, Pinyakorn S, van Griensven F, Vandergeeten 6. Lane T, Osmand T, Marr A, Struthers H, McIntyre JA, Shade SB. Brief C, Schuetz A, et al. A novel acute HIV infection staging system based on 4th Report: High HIV incidence in a South African community of men who have sex generation immunoassay. Retrovirology. 2013;10:56. with men: results from the Mpumalanga men’s study, 2012–2015. J Acquir 30. World Health Organization. Consolidated guidelines on HIV testing services Immune Defic Syndr. 2016;73(5):609–11. for a changing epidemic. Geneva, 2019 [cited 2020 Jan 9]. Available from: 7. Nowak RG, Mitchell A, Crowell TA, Liu H, Ketende S, Ramadhani HO, et al. https://www.who.int/publications-detail/consolidated-guidelines-on-hiv-testing-ser Individual and sexual network predictors of HIV incidence among men who have vices-for-a-changing-epidemic sex with men in Nigeria. J Acquir Immune Defic Syndr. 2019;80(4):444–53. 31. Miller WC, Rosenberg NE, Rutstein SE, Powers KA. Role of acute and early HIV 8. Cohen MS, Shaw GM, McMichael AJ, Haynes BF. Acute HIV-1 Infection. infection in the sexual transmission of HIV. Curr Opin HIV AIDS. 2010;5(4):277–82. N Engl J Med. 2011;364(20):1943–54. 32. Hollingsworth TD, Anderson RM, Fraser C. HIV-1 transmission, by stage of 9. Brenner BG, Roger M, Routy JP, Moisi D, Ntemgwa M, Matte C, et al. High infection. J Infect Dis. 2008;198(5):687–93. rates of forward transmission events after acute/early HIV-1 infection. J Infect 33. Abu-Raddad LJ, Longini IM Jr. No HIV stage is dominant in driving the HIV Dis. 2007;195(7):951–9. epidemic in sub-Saharan Africa. AIDS. 2008;22(9):1055–61. 10. Kroon E, Phanuphak N, Shattock AJ, Fletcher JLK, Pinyakorn S, Chomchey 34. Ratmann O, van Sighem A, Bezemer D, Gavryushkina A, Jurriaans S, Wens- N, et al. Acute HIV infection detection and immediate treatment estimated to ing A, et al. Sources of HIV infection among men having sex with men and impli- reduce transmission by 89% among men who have sex with men in Bangkok. cations for prevention. Sci Transl Med. 2016;8:320ra2. J Int AIDS Soc. 2017;20(1):21708. 35. Powers KA, Ghani AC, Miller WC, Hoffman IF, Pettifor AE, Kamanga G, 11. Group ISS, Lundgren JD, Babiker AG, Gordin F, Emery S, Grund B, Sharma et al. The role of acute and early HIV infection in the spread of HIV and implica- S, et al. Initiation of antiretroviral therapy in early asymptomatic HIV infection. tions for transmission prevention strategies in Lilongwe, Malawi: a modelling N Engl J Med. 2015;373(9):795–807. study. Lancet. 2011;378(9787):256–68. 12. D’Antoni ML, Byron MM, Chan P, Sailasuta N, Sacdalan C, Sithinamsuwan P, 36. Borges CM, Pathela P, Pirillo R, Blank S. Targeting the use of pooled HIV et al. Normalization of soluble CD163 levels after institution of antiretroviral RNA screening to reduce cost in health department STD clinics: New York City, therapy during acute hiv infection tracks with fewer neurological abnormalities. 2009–2011. Public Health Rep. 2015;130(1):81–6. J Infect Dis. 2018;218(9):1453–63. 37. Dijkstra M, Lin TC, de Bree GJ, Hoenigl M, Schim van der Loeff MF. Valida- 13. Thornhill J, Inshaw J, Kaleebu P, Cooper D, Ramjee G, Schechter M, et al. Brief tion of the San Diego Early Test Score for early HIV infection among Amster- report: enhanced normalization of CD4/CD8 ratio with earlier antiretroviral ther- dam men who have sex with men. Clin Infect Dis. 2020;70(10):2228–2230. apy at primary HIV infection. J Acquir Immune Defic Syndr. 2016;73(1):69–73. 38. Wahome E, Fegan G, Okuku HS, Mugo P, Price MA, Mwashigadi G, et al. 14. Schuetz A, Deleage C, Sereti I, Rerknimitr R, Phanuphak N, Phuang-Ngern Y, Evaluation of an empiric risk screening score to identify acute and early HIV-1 et al. Initiation of ART during early acute HIV infection preserves mucosal Th17 func- infection among MSM in Coastal Kenya. AIDS. 2013;27(13):2163–6. tion and reverses HIV-related immune activation. PLoS Pathog. 2014;10:e1004543. 39. Hoenigl M, Weibel N, Mehta SR, Anderson CM, Jenks J, Green N, et al. 15. Rutstein SE, Ananworanich J, Fidler S, Johnson C, Sanders EJ, Sued O, Development and validation of the San Diego Early Test Score to predict acute et al. Clinical and public health implications of acute and early HIV detection and early HIV infection risk in men who have sex with men. Clin Infect Dis. and treatment: a scoping review. J Int AIDS Soc. 2017;20:21579. 2015;61(3):468–75. 16. Peters PJ, Westheimer E, Cohen S, Hightow-Weidman LB, Moss N, Tsoi B, 40. Lin TC, Gianella S, Tenenbaum T, Little SJ, Hoenigl M. A simple symptom et al. Screening yield of HIV Antigen/Antibody Combination And Pooled HIV score for acute human immunodeficiency virus infection in a San Diego commu- RNA testing for acute HIV infection in a high-prevalence population. JAMA. nity-based screening program. Clin Infect Dis. 2018;67(1):105–11. 2016;315(7):682–90. 41. Campbell CK, Lippman SA, Moss N, Lightfoot M. Strategies to increase HIV 17. Wahome E, Thiong’o AN, Mwashigadi G, Chirro O, Mohamed K, Gichuru E, testing among MSM: a synthesis of the literature. AIDS Behav. 2018;22 et al. An empiric risk score to guide PrEP targeting among MSM in Coastal (8):2387–412. Kenya. AIDS Behav. 2018;22 Suppl 1:35–44. 42. PRISMA. Transparent reporting of systematic reviews and meta-analyses 18. Keating SM, Hanson D, Lebedeva M, Laeyendecker O, Ali-Napo NL, Owen [cited 2020 Jan 9]. Available from: http://www.prisma-statement.org SM, et al. Lower-sensitivity and avidity modifications of the vitros anti-HIV 1+2 43. Downes MJ, Brennan ML, Williams HC, Dean RS. Development of a critical assay for detection of recent HIV infections and incidence estimation. J Clin appraisal tool to assess the quality of cross-sectional studies (AXIS). BMJ Open. Microbiol. 2012;50(12):3968–76. 2016;6:e011458.

118 Palmer S et al. Journal of the International AIDS Society 2020, 23(S6):e25590 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25590/full | https://doi.org/10.1002/jia2.25590

44. Freeman MF, Tukey JW. Transformations related to the angular and the 68. Elliott T, Sanders EJ, Doherty M, Ndung’u T, Cohen M, Patel P, et al. Chal- square root. Ann Math Statist. 1950;21(4):607–11. lenges of HIV diagnosis and management in the context of pre-exposure pro- 45. DerSimonian R, Kacker R. Random-effects model for meta-analysis of clini- phylaxis (PrEP), post-exposure prophylaxis (PEP), test and start and acute HIV cal trials: an update. Contemp Clin Trials. 2007;28(2):105–14. infection: a scoping review. J Int AIDS Soc. 2019;22:e25419. 46. Newcombe RG. Two-sided confidence intervals for the single proportion: 69. Sudarshi D, Pao D, Murphy G, Parry J, Dean G, Fisher M. Missed opportu- comparison of seven methods. Stat Med. 1998;17(8):857–72. nities for diagnosing primary HIV infection. Sex Transm Infect. 2008;84(1):14–6. 47. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency 70. Sanders EJ, Mugo P, Prins HA, Wahome E, Thiong’o AN, Mwashigadi G, in meta-analyses. BMJ. 2003;327(7414):557–60. et al. Acute HIV-1 infection is as common as malaria in young febrile adults 48. Nyaga VN, Arbyn M, Aerts M. Metaprop: a Stata command to perform seeking care in coastal Kenya. AIDS. 2014;28(9):1357–63. meta-analysis of binomial data. Arch Public Health. 2014;72(1):39. 71. Powers KA, Cohen MS. Acute HIV-1 infection in sub-Saharan Africa: a com- 49. Menza TW, Hughes JP, Celum CL, Golden MR. Prediction of HIV acquisi- mon occurrence overlooked. AIDS. 2014;28(9):1365–7. tion among men who have sex with men. Sex Transm Dis. 2009;36(9):547–55. 72. Rafferty H, Chirro O, Oduor C, Wahome E, Ngoi C, van der Elst E, et al. 50. Smith DK, Pals SL, Herbst JH, Shinde S, Carey JW. Development of a clinical Pilot testing of an online training module about screening for acute HIV infec- screening index predictive of incident HIV infection among men who have sex with tion in adult patients seeking urgent healthcare. Int Health. 2019;11(2):93–100. men in the United States. J Acquir Immune Defic Syndr. 2012;60(4):421–7. 73. Hoenigl M, Graff-Zivin J, Little SJ. Costs per diagnosis of acute HIV infec- 51. Silvera R, Stein D, Hutt R, Hagerty R, Daskalakis D, Valentine F, et al. The tion in community-based screening strategies: a comparative analysis of four development and implementation of an outreach program to identify acute and screening algorithms. Clin Infect Dis. 2016;62(4):501–11. recent HIV infections in New York City. Open AIDS J. 2010;4:76–83. 74. Grin B, Chan PA, Operario D. Knowledge of acute human immunodeficiency 52. Beymer MR, Weiss RE, Sugar CA, Bourque LB, Gee GC, Morisky DE, et al. virus infection among gay and bisexual male college students. J Am Coll Health. Are centers for disease control and prevention guidelines for preexposure pro- 2013;61(4):232–41. phylaxis specific enough? formulation of a personalized HIV risk score for pre- 75. Wandell GM, Molina Y, Sanchez H, Greer AC, Rios J, Bain C, et al. Knowl- exposure prophylaxis initiation. Sex Transm Dis. 2017;44(1):48–56. edge and preferences concerning acute HIV testing programs among both Peru- 53. Sanders EJ, Wahome E, Powers KA, Werner L, Fegan G, Lavreys L, et al. vian men who have sex with men and transgender women. Int J STD AIDS. Targeted screening of at-risk adults for acute HIV-1 infection in sub-Saharan 2017;28(10):1010–7. Africa. AIDS. 2015;29 Suppl 3:S221–30. 76. Steyerberg EW, Harrell FE Jr. Prediction models need appropriate internal, 54. Lancki N, Almirol E, Alon L, McNulty M, Schneider JA. Preexposure prophy- internal-external, and external validation. J Clin Epidemiol. 2016;69:245–7. laxis guidelines have low sensitivity for identifying seroconverters in a sample of 77. Schwartz SR, Nowak RG, Orazulike I, Keshinro B, Ake J, Kennedy S, et al. young Black MSM in Chicago. AIDS. 2018;32(3):383–92. The immediate eff ect of the Same-Sex Marriage Prohibition Act on stigma, dis- 55. Daskalakis D, Silvera R, Bernstein K, Stein D, Hagerty R, Hutt R, et al. crimination, and engagement on HIV prevention and treatment services in men Implementation of HIV testing at 2 New York City bathhouses: from pilot to who have sex with men in Nigeria: analysis of prospective data from the TRUST clinical service. Clin Infect Dis. 2009;48(11):1609–16. cohort. Lancet HIV. 2015;2(7):e299–306. 56. Stekler JD, Baldwin HD, Louella MW, Katz DA, Golden MR. ru2hot?: a public 78. Baral S, Trapence G, Motimedi F, Umar E, Iipinge S, Dausab F, et al. HIV health education campaign for men who have sex with men to increase aware- prevalence, risks for HIV infection, and human rights among men who have sex ness of symptoms of acute HIV infection. Sex Transm Infect. 2013;89(5):409–14. with men (MSM) in Malawi, Namibia, and Botswana. PLoS One. 2009;4:e4997. 57. Gilbert M, Cook D, Steinberg M, Kwag M, Robert W, Doupe G, et al. Tar- 79. Stahlman S, Grosso A, Ketende S, Sweitzer S, Mothopeng T, Taruberekera geting screening and social marketing to increase detection of acute HIV infec- N, et al. Depression and social stigma among MSM in lesotho: implications for tion in men who have sex with men in Vancouver, British Columbia. AIDS. HIV and sexually transmitted infection prevention. AIDS Behav. 2015;19 2013;27(16):2649–54. (8):1460–9. 58. Green N, Hoenigl M, Chaillon A, Anderson CM, Kosakovsky Pond SL, Smith 80. Sanders EJ, Price MA, Karita E, Kamali A, Kilembe W, Bekker LG, et al. Dif- DM, et al. Partner services in adults with acute and early HIV infection. AIDS. ferences in acute retroviral syndrome by HIV-1 subtype in a multicentre cohort 2017;31(2):287–93. study in Africa. AIDS. 2017;31(18):2541–6. 59. Liang J, Liu L, Cheung M, Lee MP, Wang H, Li CH, et al. Community-based 81. Ortblad KF, Baeten JM. Electronic health record tools to catalyse PrEP HIV-1 early diagnosis and risk behavior analysis of men having sex with men in conversations. Lancet HIV. 2019;6(10):e644–5. Hong Kong. PLoS One. 2015;10:e0125715. 82. Marcus JL, Hurley LB, Krakower DS, Alexeeff S, Silverberg MJ, Volk JE. Use of 60. Pankam T, Saensiriphan S, Areeyolwattana S, Barisri N, Pengnonyang S, Siri- electronic health record data and machine learning to identify candidates for HIV vichayakul S, et al. The validation and evaluation of anti-HIV testing algorithm pre-exposure prophylaxis: a modelling study. Lancet HIV. 2019;6(10):e688–95. used in mobile clinic setting for men who have sex with men in metropolitan 83. Krakower DS, Gruber S, Hsu K, Menchaca JT, Maro JC, Kruskal BA, et al. Bangkok, Thailand. Asian Pac J Allergy Immunol. 2018;36(1):42–50. Development and validation of an automated HIV prediction algorithm to iden- 61. Dijkstra M, van Rooijen MS, Hillebregt MM, van Sighem A, Smit C, Hoge- tify candidates for pre-exposure prophylaxis: a modelling study. Lancet HIV. woning A, et al. Decreased time to viral suppression after implementation of 2019;6(10):e696–704. targeted testing and immediate initiation of treatment of acute HIV infection 84. Balzer LB, Havlir DV, Kamya MR, Chamie G, Charlebois ED, Clark TD, among men who have sex with men in Amsterdam. Clin Infect Dis. 2020. et al. Machine learning to identify persons at high-risk of HIV acquisition in rural ciaa505. Online ahead of print. Kenya and Uganda. Clin Infect Dis. 2019. ciz1096. Online ahead of print. 62. Facente SN, Pilcher CD, Hartogensis WE, Klausner JD, Philip SS, Louie B, 85. Calabrese SK. Implementation guidance needed for PrEP risk-prediction et al. Performance of risk-based criteria for targeting acute HIV screening in tools. Lancet HIV. 2019;6(10):e649. San Francisco. PLoS One. 2011;6:e21813. 86. Sanders EJ, Kigoro A, Thiong’o A, Nduati E, Graham SM. Symptom based 63. Jones J, Hoenigl M, Siegler AJ, Sullivan PS, Little S, Rosenberg E. Assessing scoring. Clin Infect Dis. 2019;69(4):736–7. the performance of 3 human immunodeficiency virus incidence risk scores in a cohort of black and white men who have sex with men in the South. Sex Transm Dis. 2017;44(5):297–302. 64. Lin TC, Dijkstra M, De Bree GJ, Schim van der Loeff MF, Hoenigl M. Brief report: the Amsterdam symptom and risk-based score predicts for acute hiv SUPPORTING INFORMATION infection in men who have sex with men in San Diego. J Acquir Immune Defic Syndr. 2018;79(2):e52–e5. Additional information may be found under the Supporting 65. Dijkstra M, de Bree GJ, Stolte IG, Davidovich U, Sanders EJ, Prins M, et al. Development and validation of a risk score to assist screening for acute HIV-1 Information tab for this article. infection among men who have sex with men. BMC Infect Dis. 2017;17(1):425. 66. Morgan E, Skaathun B, Nikolopoulos GK, Paraskevis D, Williams LD, Smyr- nov P, et al. A Network Intervention to Locate Newly HIV Infected Persons Table S1. Database search strategies – Within MSM Networks in Chicago. AIDS Behav. 2019;23(1):15 20. Table S2. The appraisal tool for cross-sectional studies 67. Ahrens K, Kent CK, Kohn RP, Nieri G, Reynolds A, Philip S, et al. HIV partner notification outcomes for HIV-infected patients by duration of infection, San Table S3. Critical appraisal of included studies using the Francisco, 2004 to 2006. J Acquir Immune Defic Syndr. 2007;46(4):479–84. Appraisal Tool For Cross-Sectional Studies

119 Makofane K et al. Journal of the International AIDS Society 2020, 23(S6):e25605 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25605/full | https://doi.org/10.1002/jia2.25605

COMMENTARY From general to specific: moving past the general population in the HIV response across sub-Saharan Africa Keletso Makofane1,§ , Elise M van der Elst2 , Jeffrey Walimbwa3, Steave Nemande4 and Stefan D Baral5 §Corresponding author: Keletso Makofane, 677 Huntington Avenue, Boston, Massachusetts, 02115, USA. Tel: 1 (617) 4951000. ([email protected])

Abstract Introduction: As the HIV field evolves to better serve populations which are diverse in risk and access to services, it is crucial to understand and adapt the conceptual tools used to make sense of the HIV pandemic. In this commentary, we discuss the concept of general population. Using a synthetic and historical review, we reflect on the genesis and usage of the general population in HIV research and programme literature, pointing to its moral connotations and its impact on epidemiologic reasoning. Discussion: From the early days of the HIV pandemic, the category of general population has carried implicit normative meanings. General population represented those people considered to be undeserving of HIV acquisition, and therefore deserving of a response. Framing the HIV epidemic in sub-Saharan Africa as a generalized epidemic primarily affecting the general population has contributed to the exclusion of men who have sex with men from epidemic responses. The usage of this category has also masked heterogeneity among those it includes; the increasing focus on the use of interventions such as circumcision and HIV treatment as general population HIV prevention approaches has been marked by a lack of attention to heterogeneity among beneficiaries. Conclusions: We recommend that the term general population be retired from the field’s lexicon. HIV programmes should strengthen their capacity to describe the heterogeneity of those they serve and plan their interventions accordingly. To increase the efficiency and impact of the HIV response, it is urgent to stratify the category of general population by risk. Sex- ual networks are a promising basis for this stratification. Keywords: HIV; key and vulnerable populations; men who have sex with men; Africa; LGBT persons; general population

Additional Supporting information may be found online in the Supporting Information tab for this article.

Received 10 January 2020; Accepted 21 July 2020 Copyright © 2020 The Authors. Journal of the International AIDS Society published by John Wiley & Sons Ltd on behalf of the International AIDS Society. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

1 | INTRODUCTION genesis and usage of the term in HIV literature, pointing to its moral connotations and its effects on epidemiologic thinking. The category of general population appears in policy briefs, As the HIV field evolves to better serve populations which research reports, guidance notes and policies. It punctuates are diverse in risk and access to services, it is important to the day-to-day speech of colleagues in global institutions and understand and adapt the conceptual tools used to make sense local programmes alike. It is as broad as it is ubiquitous, pur- of the HIV pandemic. To help contextualize the articles in this porting to include a large swathe of society, if not the entire special issue, we offer historical context for a concept that has society, while helping funders, programmers, and researchers fundamentally shaped the global HIV response yet is scarcely to make sense of the epidemic and choose among alternative explicitly examined. Our hope is that grappling with this context courses of action. But despite its wide usage, the term is will advance the development of conceptual tools better-suited rarely explicitly defined. In the 2019 data report by UNAIDS for understanding the contours of the current HIV pandemic. [1], for instance, it garners several mentions without ever We argue that framing HIV epidemics in sub-Saharan Africa being linked to a glossary, neither in that document nor in the as generalized epidemics led to the wide-spread understanding latest UNAIDS guideline on terminology [2]. Capacious as gen- that they are almost exclusively heterosexual and homoge- eral population appears to be, in common usage it is clear what neous in nature. This is one of the mechanisms through which it is not: general population is an antonym for specific popula- gay men and other men who have sex with men were tions who require specific responses (or sometimes, as history excluded from the HIV responses mounted in the region by proves, non-responses). In this commentary, we reflect on the global, regional and national institutions. Crucially, this framing

120 Makofane K et al. Journal of the International AIDS Society 2020, 23(S6):e25605 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25605/full | https://doi.org/10.1002/jia2.25605 masked heterogeneity among those included in the general hedonists upon hard-working innocents” [21]. The AIDS epi- population category, weakening HIV responses for them. As an demic was understood to also be an “epidemic of signification,” alternative to reporting and programming using the category [22] rapidly producing concepts that implicitly ascribe blame of general population, we suggest using granular descriptions of and innocence [23]. In these analyses, it was understood that distributions of risk. categories used in scientific work are always entangled with already-circulating cultural meanings [24]. In the context of 2 | DISCUSSION wide-spread homophobia, it followed that the categories used to understand the epidemic reflected widely held, negative attitudes towards gay men. 2.1 | The moral history of “General Population” The metaphors and meanings through which population dis- The idea of the general population made its appearance in early tributions of disease are understood shape public health HIV research whose goal was to uncover the aetiology of a responses [25]. For the first two decades of the HIV response new and alarming syndrome. In a mid 1980s case series, for in sub-Saharan Africa, men who have sex with men were instance, Ioachim et al.[3] used the term to contextualize the ignored despite the emergence of the HIV epidemic among incidence of lymphoma among gay men in New York, contrast- gay men in higher-income settings and despite early evidence ing it with population-wide, or general population, incidence. In of its impact among gay men in other settings [11,26-28]. the same period, Acheson et al.[4] used it to compare sero- There was early precedent. Citing mainly European studies of prevalence between two mutually exclusive groups: “high-risk African immigrants and small-sample studies in central Africa, groups” and “general population,” distinguishing “special pro- an influential group of authors reported in 1986 that “African grammes” – interventions designed to “meet the needs of AIDS patients rarely report a history of homosexual activity or declared male homosexuals and persons who abuse drugs by intravenous drug abuse” [29]. In the following years, this con- injection” – from “material directed to the population as a jecture would reverberate through the World Health Organi- whole,” [4] calling this latter group the general population.As zation’s communication about the global epidemic so that by these two examples illustrate, general population sometimes the close of the 80s, it was the basis of a fundamental classifi- denoted a complement-set and others a super-set in relation cation. Countries were grouped into three epidemic patterns, to some group of interest. It sometimes was employed to each pattern demanding a different kind of response (See make statistical comparisons and sometimes to distinguish and Figure 1)[30-32]: Pattern I – those in which transmission hap- characterize groups of people. It was not defined explicitly pens predominantly among men who have sex with men and and held unstable meaning, even within the same document. people who inject drugs; Pattern II – those in which “intra- These ambiguities have proven persistent [5-10] (See venous drug use and homosexual transmission [sic] are either Appendix S1). non-existent or occur at a very low level;” [31] and Pattern III Among African people who had acquired the virus through – those in which transmission was thought to have started heterosexual sex, the term was imbued with moral signifi- later than in countries classified under the first two patterns. cance. Unlike the risk-groups among whom the ravages of AIDS By the end of the 90s, the category of Pattern II had given were first registered, general population represented those way to generalized epidemic as the dominant way of under- considered to be at undeserved risk [11]. According to Eliza- standing and describing HIV epidemics in sub-Saharan Africa beth Pisani, an architect of early UNAIDS guidance on HIV [12,13]. The UNAIDS Guidelines for Second-Generation HIV surveillance [12-14], it was an explicit goal to foster this Surveillance, published in 2000, categorized HIV epidemics moralistic understanding of risk for the purpose of igniting a into three phases: low-level, concentrated, and generalized. global response: “... governments don’t like spending money Generalized epidemics were initially defined using an HIV on sex workers, gay men, or drug addicts... We had to find a prevalence threshold of 1% in ante-natal clinics [13]. Since way to translate the truth into something that governments then, the threshold was removed from surveillance guidelines, might care about... Politicians are always happy to do nice though the term generalized epidemic continues to be used. things for innocent women and babies. Perhaps if we could Now, four decades and many lost lives since the start of show that doing nice things for injectors would protect inno- the pandemic, the assumption that men who have sex with cent women and babies...” [15]. By the end of the 1990s, men do not feature in generalized epidemics has been roundly most of the bilateral investment in HIV programming was con- disproven [33]. All over sub-Saharan Africa, LGBT-led organi- ducted under the assumption that HIV in Africa was transmit- zations provide services to men who have sex with men and ted nearly exclusively through heterosexual transmission and advocate for programmatic inclusion in national, regional and that “its primary impact [was] on the ‘general’ population” global fora [34,35]. While there has been considerable success [11,16]. Only in the 2000s did resources begin to be targeted in advocating for the establishment of targeted funding at HIV programmes for men who have sex with men, though streams and inclusive normative guidance [36], governments funding levels were grossly insufficient [17-20]. and large agencies have struggled to shrug off their moralism. In the context of the epidemic in the United States, social As men who have sex with men have been increasingly scientists had begun in the 1980s to examine the normative included in national HIV responses under the banners of Key content of the distinction between general and high-risk. Jan Affected Populations, Most at Risk Populations, Key Popula- Grover, for instance, wrote in 1988 that according to the tions and other risk groupings, it has sometimes been for media, public health officials, and politicians “the general popu- explicitly instrumental ends. This is laid bare in AIDS national lation is virtuously going about its business, which is not plea- strategic plans in which it was the fear of contagion from high- sure-seeking (as drugs and gay life are uniformly imagined to risk to general populations that gives impetus for interventions be), so AIDS hits its members as an assault from diseased among the former [37]. This approach bears the imprint of

121 Makofane K et al. Journal of the International AIDS Society 2020, 23(S6):e25605 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25605/full | https://doi.org/10.1002/jia2.25605

Figure 1. Global patterns of HIV and AIDS – 1989. The map shows Global patterns of HIV and AIDS according to the World Health Organi- zation Global Programme on AIDS in 1989. Pattern I countries were those in which transmission was believed to occur predominantly among men who have sex with men and people who inject drugs; Pattern II countries were those in which “intravenous drug use and homosexual transmission [sic] are either non-existent or occur at a very low level;” and Pattern III countries were those in which transmission was thought to have started later than in countries classified under Pattern I and Pattern II. This map appeared in a conference report published by the International Commission of Jurists [31]. global guidance, which cautions governments to pay attention sex (condomless or not) with his partner, then they would to high-risk groups in order to guard against spread “into the both be considered members of a high-risk group since their general population” [12,14] rather than, for instance, dissemi- behaviour, sex with men, is considered risky in and of itself. nation across the population. The notion of risk that organizes this scheme focuses on behaviour, is shaped by sexual morals, and pays little attention to the most important factor in HIV transmission: the likeli- 2.2 | General population and epidemiologic hood of sexual contact between someone who has acquired reasoning the virus and someone who has not. General population not only reflects the moral standing of This inattention has shaped HIV programming and research those historically excluded from it, it shapes epidemiologic rea- in sub-Saharan Africa. As Baral et al. have previously argued soning in relation to those it includes. Whereas in epidemiol- [38], the focus of HIV treatment as general population preven- ogy the risk of an event – say acquiring HIV – is usually tion has not accounted for heterogeneity of risk among bene- defined as the probability of the event’s occurrence, in the ficiaries. The effectiveness of universal treatment at a logic of the categorization into general and high-risk popula- population level is predicated on there being existing risk for tions, risk is implicitly defined through behaviour. In some HIV transmission between the recipient and their sexual con- instances, this creates contradictions. For example a person tacts, yet little effort is made to understand the characteristics who has only had one sexual partner in her life – to whom of these networks. Thus, transmission dynamics within net- she is married and with whom she has condomless sex – works might be the key to understanding why powerful, indi- would be categorized as low-risk or general population. On the vidual-level HIV treatment effects have not translated into other hand, if her partner were living with HIV and unsup- similarly powerful, population-level incidence reductions. It pressed or if he were likely to acquire HIV from condomless might be that case that it is the size, composition and treat- sex with multiple other partners, then she would have an ele- ment coverage of personal sexual networks that determine vated risk (in the epidemiologic sense) of acquiring HIV. Under population HIV prevention benefits. the scheme that divides people into high-risk and general, the The task of stratifying this category is urgent and of public woman in this example would not be thought of as a member health significance. According to UNAIDS, about 800,000 peo- of some high-risk group. By contrast, if a man who has only ple in eastern and southern Africa acquired HIV in 2018 [1]. had one sexual partner in his life – another man – and had In one breakdown, 25% were attributed to men who have sex

122 Makofane K et al. Journal of the International AIDS Society 2020, 23(S6):e25605 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25605/full | https://doi.org/10.1002/jia2.25605 with men, transgender people, sex workers, and the sexual health, it has long been known that there are specific popula- partners of these groups and 75% were unattributed. In tions that bear a higher burden of risk and illness than others. another, young women between the ages of 15 and 24 were Also because of social determinants, particularly stigma and said to account for 26%, and the remainder were unattribu- discrimination operating at the interpersonal as well as at the ted. In both examples, the unattributed portion of new HIV structural level, the coverage of existing HIV prevention and infections, a portion containing much heterogeneity in risk, treatment services and commodities follows the inverse care constitutes the majority. law: Those with greatest need have the lowest access to nec- Even within the categories of young women or men who essary services [55]. have sex with men, however, there is considerable variation in Learning from epidemics other than HIV would suggest the risk. Among young women in Tanzania, for instance, having an need to find an optimal balance between what Geoffery Rose older partner and engaging in transactional sex are each asso- termed high-risk strategies – strategies that identify groups at ciated with double the HIV prevalence of not having an older higher risk and targets interventions appropriately – as partner and not having transactional sex respectively [39,40]. opposed to a near exclusive focus on population strategies,as In addition, there is substantial contact between groups defined earlier. We must shift away from programming for the [41,42]. Past epidemiological studies among men who have category of the general population to specific populations with sex with men in Botswana, Namibia, Malawi, South Africa, specific needs. These may be defined by age, gender, labour Kenya, Senegal and Nigeria suggest that sex with women is migration, or geography at different scales. By tracing trans- relatively common among men who have sex with men. The mission clusters, phylodynamic modelling promises to aid our reported proportion who recently had sex with women has understanding of HIV transmission risk both within and across ranged from 20% to 75% [43-46]. these populations [41,42,56]. In the absence of detailed sex- Epidemiologic and phylogenetic studies have consistently ual- or genetic-network data, the determinants of sexual net- demonstrated the interconnectedness of the sexual networks works (e.g. micro-geography, gender, occupation, mobility) of men who have sex with men with the networks of the should be used to stratify what are now termed general and remainder of the population [41,47]. There remains, however, key populations by risk. Interventions should be planned based limited standard reporting of the attributable fraction of HIV on this stratification. The idea that the general population is a epidemics across sub-Saharan Africa secondary to the unmet useful target for HIV surveillance and programmes should be needs of men who have sex with men. In part, the limited replaced with a granular mapping of risk by the cross-classifi- study and reporting of the population attributable fraction for cation of multiple demographic variables. In conducting this HIV among men who have sex with men across sub-Saharan mapping, the human rights of all individuals should be pro- Africa has emerged from the tacit assumption that they do tected [57]. not exist. Where they do exist, government consensus estimates often suggest such low population sizes so as not to be relevant for a comprehensive HIV response [48]. 3 | CONCLUSIONS The urgency of obtaining a granular understanding of risk in the HIV epidemic is heighted by the fact that Africa is home A shift away from the concept of general population suggests to the largest ever generation of young people moving into the need for understanding people’s individual needs, how adulthood [49,50]. It is crucial to understand the diversity these translate to a continuum of risks for HIV acquisition and among youth, including sexual and gender identities and sexual transmission, and how these dynamically change over time in practices, to design specific responses to address their actual an epidemic. These considerations should be grounded in local unmet needs. Characterizing and celebrating that diversity context, even as valuable lessons are transmitted across coun- increases the likelihood of differentiated programmes to adap- tries and regions. For people who have thus far been classified tively scale to address the diverse needs of hundreds of mil- as general population and those who have been classified as lions of youth across sub-Saharan Africa. key populations alike, this means continually monitoring spatial Geoffrey Rose is credited with the insight that, in a popula- and temporal patterns and identifying structural and beha- tion, the highest burden of a disease in absolute numbers is vioural causes of HIV transmission and HIV-related ill health. to be found not among the few at highest risk, but the many It further means using this knowledge to produce tailored with medium or low risk [51]. Following this line of reasoning, community-led interventions including those that attend to when mounting a public health response, it is the many that structural determinants of health such as stigma and violence. should be targeted to maximize impact, not the few. Dubbed Though it will take investment in research, there are pro- the “population strategy” and manifested in the ideas of the gramme strategies from which to draw inspiration. The generalized epidemic and the general population, this approach approach of micro-planning in sex worker programming, for has animated HIV programming in sub-Saharan Africa for the example acknowledges that not all programme beneficiaries first two decades of the response. share the same risks or need the same programmatic But the population approach rests on assumptions that may responses [58]. Abandoning the category of general population not be met in reality: population interventions, if they depend affords a great opportunity to learn about the diversity of on resources that are differentially distributed by risk, can needs and adaptive strategies developed to respond to HIV in widen disparities [52]. In addition, the largest number of cases key populations [58-61]. While likely necessary to advance the might come from a small proportion of the population at HIV response, this shift alone is not sufficient to overcome extremely high risk [53,54], rendering population-wide inter- the structural determinants of the HIV epidemic. Perhaps, the ventions inefficient. Given the biology of HIV transmission only utility of the distinction between general and key popula- intertwined with network-level and social determinants of tions is that it reinforces an understanding that intersecting

123 Makofane K et al. Journal of the International AIDS Society 2020, 23(S6):e25605 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25605/full | https://doi.org/10.1002/jia2.25605 stigmas, violence and criminalization play different roles for 10. Dvali N, Chkhartishvili N, Sharvadze L, Karchava M, Tsertsvadze T. Molecu- the HIV epidemic among these two categories. In moving lar tracing of heterosexual HIV-1 transmission in Georgia. Georgian Medical News. 2015;246:54–9. away from the use of general population and towards studying 11. Parker R, Khan S, Aggleton P. Conspicuous by their absence? Men who why people are at risk for the acquisition and transmission of have sex with men (MSM) in developing countries: Implications for HIV preven- HIV in countries across sub-Saharan Africa, the HIV response tion. Critical Public Health. 1998;8(4):329–46. should follow with specific interventions, including structural 12. Pisani E, Brown T, Saidel T, Rehle T, Carael M. Meeting the behavioural interventions, to address those needs. data collection needs of national HIV/AIDS and STD programmes. A joint IMPACT/FHI/UNAIDS workshop: report and conclusions. UNAIDS, USAID, and IMPACT; 1998. AUTHORS’ AFFILIATIONS 13. World Health Organization, Joint United Nations Program on HIV/AIDS. Second generation surveillance for HIV: the next decade [Internet]. Geneva: 1FXB Center for Health and Human Rights, Harvard University, Boston, MA, World Health Organization and Joint United Nations Program on HIV/AIDS; USA; 2Kenya Medical Research Institute, Kilifi, Kenya; 3ISHTAR MSM, Nairobi, 2000. Available from: https://www.who.int/reproductivehealth/publications/rtis/ Kenya; 4Evolve, Yaounde, Cameroon; 5Center for Public Health and Human CDS_CSR_EDC_2000_5/en/ Rights, Department of Epidemiology, Johns Hopkins School of Public Health, 14. UNAIDS/WHO Working Group on Global HIV/AIDS and STI Surveillance, Baltimore, USA World Health Organization. Guidelines for effective use of data from HIV surveillance systems. Geneva: World Health Organization; 2004. COMPETING INTERESTS 15. Pisani E. The wisdom of whores: bureaucrats, brothels and the business of AIDS. Granta Books. 2010. None. 16. Treichler PA.AIDS and HIV infection in the Third World: a First World chronicle. In: How to have theory in an epidemic: cultural chronicles of AIDS. AUTHORS’ CONTRIBUTIONS 1999;99–126. 17. amfAR, Johns Hopkins Bloomberg School of Public Health. Achieving an AIDS- KM led the development of the manuscript with specific sections led by SB. free generation for gay men and other MSM in Southern Africa. amfAR; 2013. EvdE supported conceptualization of the manuscript and completed background 18. amfAR, Johns Hopkins Bloomberg School of Public Health. Achieving an research with JW. SN provided background materials for review. All authors AIDS-free generation for gay men and other MSM: financing and implementa- provided overall direction of manuscript and inputs and text to specific sections tion of HIV programs targeting MSM. amfAR; 2012. of the manuscript. 19. Grosso A, Ryan O, Tram KH, Baral S. Financing the response to HIV among gay men and other men who have sex with men: case studies from eight diverse ACKNOWLEDGEMENTS countries. Global Public Health. 2015;10(10):1172–84. 20. Ayala G, Hebert P, Keatley J, Sundararaj M. An analysis of major HIV donor We acknowledge the individuals and communities around the world who advo- investments targeting men who have sex with men and transgender people. cate and serve their peers often at great risk and limited reward. We thank Oakland: The Global Forum on MSM & HIV,(MSMGF). 2011. Nicholas Diamond for copy editing the manuscript, and the International Com- 21. Grover JZ. AIDS: Keywords. In: Crimp D, editor. AIDS: cultural analysis, cul- mission of Jurists for allowing us to reproduce the map shown in Figure 1.SB’s tural activism. Cambridge, MA: MIT Press; 1988. contribution to this publication was supported by a grant focused on character- 22. Treichler PA. AIDS, homophobia and biomedical discourse: An epidemic of izing stigma among men who have sex with men around the world by US signification. Cultural Studies. 1987;1(3):263–305. National Institutes of Mental Health and the Office of AIDS Research of the US 23. Herrell RK. HIV/AIDS research and the social sciences. Curr Anthropol. National Institutes of Health under award number R01MH110358. KM’s contri- 1991;32(2):199–203. bution was supported by classmates and friends in the 2020 cohort of doctoral 24. Waldby C. AIDS and the body politic: biomedicine and sexual difference. affiliates at the Harvard FXB Center for Health and Human Rights. Routledge; 2003. 25. Krieger N. Epidemiology and the people’s health: theory and context. Oxford University Press; 2001. REFERENCES 26. Gevisser MA. A different fight for freedom: a history of South African les- 1. Joint United Nations Program on HIV/AIDS. UNAIDS data 2019 [Internet]. bian and gay organisation from the 1950s to the 1990s. Defiant desire: Gay – Joint United Nations Program on HIV/AIDS; 2019 [cited 2019 Nov 17]. Available and lesbian lives in South Africa. 1995;14 86. from: https://www.unaids.org/en/resources/documents/2019/2019-UNAIDS-data 27. Pegge JV. Living with loss in the best way we know how: AIDS and gay men in Cape Town. Defiant desire: Gay and lesbian lives in South Africa. 2. Joint United Nations Program on HIV/AIDS. UNAIDS terminology guidelines 2015 [Internet]. Joint United Nations Program on HIV/AIDS; 2015. Available 1995;301A310. from: https://www.unaids.org/en/resources/documents/2015/2015_terminology_ 28. McKay T. From marginal to marginalised: The inclusion of men who have guidelines sex with men in global and national AIDS programmes and policy. Global Public – – 3. Ioachim HL, Cooper MC, Hellman GC. Lymphomas in men at high risk for Health. 2016;11(7 8):902 22. acquired immune deficiency syndrome (AIDS) a study of 21 cases. Cancer. 29. Quinn TC, Mann JM, Curran JW, Piot P. AIDS in Africa: an epidemiologic – 1985;56(12):2831–42. paradigm. Science. 1986;234(4779):955 63. ’ 4. Acheson E. AIDS: a challenge for the public health. Lancet. 1986;1 30. Mann J. Worldwide strategies for HIV control: WHO s special programme – – (8482):662–6. on AIDS. Law Med Health Care. 1986;14(5 6):290 7. 5. Laar A, DeBruin D. Key populations and human rights in the context of HIV 31. Tomasevski K. HIV/AIDS and human rights. In: Conference of European services rendition in Ghana. BMC Int Health Human Rights [Internet]. 2017; national sections - AIDS and human rights [Internet]. Geneva; Leiden: Interna- Dec [cited 2020 Jun 13];17(1). Available from: http://bmcinthealthhumrights.b tional Commission of Jurists (ICJ); 1989. Available from: https://www.icj.org/con iomedcentral.com/articles/10.1186/s12914-017-0129-z ference-of-european-national-sections-aids-and-human-rights-european-social- 6. Hakim AJ, MacDonald V, Hladik W, Zhao J, Burnett J, Sabin K, et al. Gaps charter-the-hague-peace-palace-20-to-22-september-1989/ and opportunities: measuring the key population cascade through surveys and 32. Chin J, Mann JM. The global patterns and prevalence of AIDS and HIV services to guide the HIV response. J Int AIDS Soc. 2018;21:e25119. infection. AIDS. 1988;2:S247. 7. Ingabire R, Parker R, Nyombayire J, Ko JE, Mukamuyango J, Bizimana J, 33. Beyrer C, Baral SD, Van Griensven F, Goodreau SM, Chariyalertsak S, Wirtz et al. Female sex workers in Kigali, Rwanda: a key population at risk of HIV, sex- AL, et al. Global epidemiology of HIV infection in men who have sex with men. – ually transmitted infections, and unplanned pregnancy. Int J STD AIDS. 2019;30 Lancet. 2012;380(9839):367 77. (6):557–68. 34. Makofane K, Beck J, Ayala G. MSM in sub-Saharan Africa: health, access, ’ 8. Cheuk E, Isac S, Musyoki H, Pickles M, Bhattacharjee P, Gichangi P, et al. and HIV. Findings from the Global Men s Health and Rights Study. 2012; – Informing HIV prevention programs for adolescent girls and young women: a 1 25. modified approach to programmatic mapping and key population size estimation. 35. Making the SDGs work for key populations in Africa [Internet]. United JMIR Public Health and Surveillance. 2019;5:e11196. Nations Development Program. 2017 [cited 2019 Jan 2]. Available from: 9. Stuikyte R, Barbosa I, Kazatchkine M. Getting to grips with the HIV epidemic https://www.africa.undp.org/content/rba/en/home/presscenter/articles/2017/08/ in Russia. Current Opinion in HIV and AIDS. 2019;14(5):381–6. making-the-sdgs-work-for-key-populations-in-africa.html

124 Makofane K et al. Journal of the International AIDS Society 2020, 23(S6):e25605 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25605/full | https://doi.org/10.1002/jia2.25605

36. World Health Organization. Consolidated guidelines on HIV prevention, 52. Frohlich KL, Potvin L. Transcending the known in public health practice: the diagnosis, treatment and care for key populations. Geneva: World Health Orga- inequality paradox: the population approach and vulnerable populations. Am J nization; 2014. Public Health. 2008;98(2):216–21. 37. Makofane K, Gueboguo C, Lyons D, Sandfort T. Men who have sex with 53. Ortblad KF, Baeten JM, Cherutich P, Wamicwe JN, Wasserheit JN. The arc men inadequately addressed in African AIDS National Strategic Plans. Global of HIV epidemics in sub-Saharan Africa: new challenges with concentrating epi- Public Health. 2013;8(2):129–43. demics in the era of 90–90–90. Curr Opin HIV AIDS. 2019;14(5):354–65. 38. Baral S, Rao A, Sullivan P, Phaswana-Mafuya N, Diouf D, Millett G, et al. 54. Wasserheit JN, Aral SO. The dynamic topology of sexually transmitted dis- The disconnect between individual-level and population-level HIV prevention ease epidemics: implications for prevention strategies. J Infect Dis. 1996;174 benefits of antiretroviral treatment. Lancet HIV. 2019;6(9):e632–8. Supplement 2:S201–13. 39. Rucinski K, Mbita G, Curran K, Komba A, Casalini C, Majani E, et al. Trans- 55. Hart JT. The inverse care law. Lancet. 1971;297(7696):405–12. actional sex with older partners heightens HIV risks among AGYW in Tanzania. 56. Li Y, Liu H, Ramadhani HO, Ndembi N, Crowell TA, Kijak G, et al. Genetic Boston: CROI; 2020. clustering analysis for HIV infection among MSM in Nigeria: implications for 40. Rucinski K, Stoner M, Kilburn K, DeLong S, Gomez-Oliv e X, Kahn K, et al. intervention. AIDS. 2020;34(2):227–36. Developmental trajectories of transactional sex and age-disparate relationships 57. amfAR, International AIDS Vaccine Initiative, Johns Hopkins University – during adolescence: an HPTN 068 analysis. Mexico City: IAS; 2019. Center for Public Health and Human Rights, United Nations Development Pro- 41. Volz EM, Ndembi N, Nowak R, Kijak GH, Idoko J, Dakum P, et al. Phylody- gram. Respect, protect, fulfill: best practices guidance in conducting HIV namic analysis to inform prevention efforts in mixed HIV epidemics. Virus. Evo- research with gay, bisexual, and other men who have sex with men (MSM) in lution. 2017;3:vex014. rights-constrained environments. amfAR; 2011. 42. Nduva GM, Hassan AS, Nazziwa J, Graham SM, Esbjornsson€ J, Sanders EJ. 58. World Health Organization, United Nations Population Fund, Joint United HIV-1 transmission patterns within and between risk groups in coastal Kenya. Nations Programme on HIV/AIDS, Global Network of Sex Work Projects, World Sci Rep. 2020;10(1):1–10. Bank. Implementing comprehensive HIV/STI programmes with sex workers: 43. Beyrer C, Trapence G, Motimedi F, Umar E, Iipinge S, Dausab F, et al. Bisex- practical approaches from collaborative interventions. Geneva: World Health ual concurrency, bisexual partnerships, and HIV among Southern African men Organization; 2013. who have sex with men. Sex Transmit Infect. 2010;86(4):323–7. 59. United Nations Population Fund, Global Forum on MSM & HIV, United 44. Sandfort TG, Dominguez K, Kayange N, Ogendo A, Panchia R, Chen YQ, Nations Development Programme, World Health Organization, United States et al. HIV testing and the HIV care continuum among sub-Saharan African men Agency for International Development, World Bank. Implementing comprehensive who have sex with men and transgender women screened for participation in HIV and STI programmes with men who have sex with men: practical guidance for HPTN 075. PLoS One. 2019;14:e0217501. collaborative interventions. New York, NY: United Nations Population Fund; 2015. 45. Wade AS, Kane CT, Diallo PAN, Diop AK, Gueye K, Mboup S, et al. HIV 60. United Nations Office on Drugs and Crime, International Network of Peo- infection and sexually transmitted infections among men who have sex with ple Who Use Drugs, Joint United Nations Programme on HIV/AIDS, United men in Senegal. AIDS. 2005;19(18):2133–40. Nations Development Programme, United Nations Population Fund, World 46. Kokogho A, Amusu S, Baral SD, Charurat ME, Adebajo S, Makanjuola O, Health Organization, et al. Implementing comprehensive HIV and HCV pro- et al. Disclosure of same-sex sexual practices to family and healthcare providers grammes with people who inject drugs: practical guidance for collaborative by men who have sex with men and transgender women in Nigeria. Arch Sex interventions. Vienna: United Nations Office on Drugs and Crime; 2017. Behav. 2020;1–12. 61. United Nations Development Programme, IRGT, United Nations Population 47. Mukandavire C, Walker J, Schwartz S, Boily M-C, Danon L, Lyons C, et al. Fund, UCSF Center of Excellence for Transgender Health, Johns Hopkins Estimating the contribution of key populations towards the spread of HIV in Bloomberg School of Public Health, World Health Organization, et al. Imple- Dakar, Senegal. J Int AIDS Soc. 2018;21:e25126. menting comprehensive HIV and STI programmes with transgender people: 48. Baral S, Turner RM, Lyons CE, Howell S, Honermann B, Garner A, et al. practical guidance for collaborative interventions. New York, NY: United Nations Population size estimation of gay and bisexual men and other men who have Development Programme; 2016. sex with men using social media-based platforms. JMIR Public Health Surveill. 2018;4(1):e15. 49. Bekker L-G, Alleyne G, Baral S, Cepeda J, Daskalakis D, Dowdy D, et al. Advancing global health and strengthening the HIV response in the era of the Sustainable Development Goals: the International AIDS Society—Lancet Com- SUPPORTING INFORMATION mission. Lancet. 2018;392(10144):312–58. 50. Piot P, Karim SSA, Hecht R, Legido-Quigley H, Buse K, Stover J, et al. Additional Supporting Information may be found in the online Defeating AIDS—advancing global health. Lancet. 2015;386(9989):171–218. 51. Rose G. Sick individuals and sick populations. Int J Epidemiol. 2001;30 version of this article: (3):427–32. Appendix S1. Key Population/General Population Definitions.

125 van der Elst EM et al. Journal of the International AIDS Society 2020, 23(S6):e25597 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25597/full | https://doi.org/10.1002/jia2.25597

RESEARCH ARTICLE A more responsive, multi-pronged strategy is needed to strengthen HIV healthcare for men who have sex with men in a decentralized health system: qualitative insights of a case study in the Kenyan coast Elise M van der Elst1,2,§,* , Rita Mudza3,*, Justus M Onguso4, Leonard Kiirika5, Bernadette Kombo1, Nassim Jahangir6, Susan M Graham7 , Don Operario8 and Eduard J Sanders1,2,9 §Corresponding author: Elise M van der Elst, KEMRI CGMRC, Hospital Road, P.O. Box 230, 80108 Kilifi, Kenya. Tel: +254 710 866576. (evanderelst@kemri- wellcome.org) *These authors contributed equally to the work

Abstract Introduction: HIV healthcare services for men who have sex with men (MSM) in Kenya have not been openly provided because of persistent stigma and lack of healthcare capacity within Kenya’s decentralized health sector. Building on an evaluation of a developed online MSM sensitivity training programme offered to East and South African healthcare providers, this study assessed views and responses to strengthen HIV healthcare services for MSM in Kenya. Methods: The study was conducted between January and July 2017 in Kilifi County, coastal Kenya. Seventeen policymakers participated in an in-depth interview and 59 stakeholders, who were purposively selected from three key groups (i.e. healthcare providers, implementing partners and members of MSM-led community-based organizations) took part in eight focus group discussions. Discussions aimed to understand gaps in service provision to MSM from different perspectives, to identify potential misconceptions, and to explore opportunities to improve MSM HIV healthcare services. Interviews and focus group discussions were recorded, transcribed verbatim and analysed using Braun and Clarke’s thematic analysis. Results: Participants’ responses revealed that all key groups navigated diverse challenges related to MSM HIV health services. Specific challenges included priority-setting by county government staff; preparedness of leadership and management on MSM HIV issues at the facility level; data reporting at the implementation level and advocacy for MSM health equity. Strong power inequities were observed between policy leadership, healthcare providers and MSM, with MSM feeling blamed for their sexual orientation. MSM agency, as expressed in their actions to access HIV services, was significantly constrained by county context, but can potentially be improved by political will, professional support and a human rights approach. Conclusions: To strengthen HIV healthcare for MSM within a decentralized Kenyan health system, a more responsive, multi- pronged strategy adaptable and relevant to MSM’s healthcare needs is required. Continued engagement with policy leadership, collaboration with health facilities, and partnerships with different community stakeholders are critical to improve HIV healthcare services for MSM. Keywords: HIV healthcare services; MSM; Kenya; decentralization; health equity

Received 18 February 2020; Accepted 20 July 2020 Copyright © 2020 The Authors. Journal of the International AIDS Society published by John Wiley & Sons Ltd on behalf of the International AIDS Society. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

1 | INTRODUCTION programmatic pre-exposure prophylaxis (PrEP) offered to MSM since 2017 [4]. While HIV transmission patterns within Kenya has among the largest HIV epidemics in the world with and between risk groups in Kenya are not well understood, an estimated 1.6 million people living with HIV, 46,000 new recent HIV phylogeny research showed that 85% of transmis- infections, and 25,000 AIDS-related deaths in 2018 [1], affect- sion clusters was within risk groups, whereas 15% was shared ing both general and key populations (KP), including men who between risk groups [5], suggesting that the HIV epidemic have sex with men (MSM). MSM have been engaged for longi- among MSM require targeted interventions. tudinal research in coastal Kenya since 2005 [2], and Sanders Efforts by Kenya’s National AIDS Control Council (NACC) et al.[3] showed that MSM had an HIV-incidence of 5.1 (95% [6] to strengthen HIV healthcare services for MSM in Kenya CI: 2.6 to 9.8) per 100 person years in 2013. HIV incidence, [7,8] have been obstructed by historical practices, policies however, has remained high in coastal Kenya despite criminalizing consensual same-sex sexual practices [9], and

126 van der Elst EM et al. Journal of the International AIDS Society 2020, 23(S6):e25597 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25597/full | https://doi.org/10.1002/jia2.25597 deep-rooted prejudice and patterns of stigmatization against (Kilifi County Hospital in Kilifi, Malindi sub-County Hospital in male-same-sex practices [10,11]. The use of a human rights Malindi, and Mtwapa Health Center in Mtwapa) that serve based-approach as a strategy to facilitate access to HIV pre- approximately 1.4 million people over an area of 12,246 km2. vention and care services by MSM vis-a-vis illegality of same- Since 2005, the Kenya Medical Research Institute-Wellcome sex sexual practices has caused profound confusion among Trust Research (KWTRP) Programme has offered HIV preven- policymakers, healthcare providers, implementing partners and tion and care services for MSM at a stand-alone clinic in MSM themselves [12]. This is a challenge for many African Mtwapa, and at the sub-County Hospital in Malindi and the countries that simultaneously criminalize same-sex practices programme campus in Kilifi since 2008. The Kilifi County while including sexual and gender minorities in national HIV health offices are located adjacent to the campus of the health policy plans. KWTRP. Through in-depth interviews (IDI) and focus group In 2013, responsibilities for overseeing the implementation discussions (FGD), we collected narrative data of participants’ of healthcare services were officially transferred from the experiences and perceptions regarding the provision and qual- national Government of Kenya and Ministry of Health (MoH) ity of HIV healthcare services for MSM in Kilifi County. to local county governments [13,14]. At the national level, the MoH, and the National AIDS and sexually transmitted infec- 2.1 | Participants and procedures tions (STI) Control Programme (NASCOP) retained their function formulating policies, guidelines, regulations and standards, A total of 17 key informants (KI) were purposively selected to while provision of technical guidance and support to healthcare participate in IDI, including four staff from the MoH, two staff was transferred to the counties [15]. Kenya comprises members from the County Health Management Team, two 47 counties, and it has been challenging to design, implement members from sub-County Health Management Team, three and assess technical changes in HIV prevention and care ser- facility in-chargers, and six implementing partners of Kilifi vices [16], without a deeper understanding of the country’s County’s main Non-Governmental Organizations (NGOs). The complex political context [17]. Based on an analysis of the liter- implementing partners were either commissioned by Kilifi ature on priority setting, Barasa et al.[18] suggested that County’s MoH, NACC, or NASCOP to coordinate training and county governments should develop legislation that gives hos- education, and implemented complementary comprehensive pitals greater control over resources and key management HIV services. KI thoughts and hypotheses on HIV and access functions. Similarly, in a systematic review of research on expe- to HIV healthcare for MSM were used to finalize semi-struc- riences of decentralization in sub-Saharan Africa, Zon and tured focus group topic guides. others [19] underscored the complexity of implementing Participants for FGDs were selectively sampled and orga- decentralization schemes, and the need to transfer and nized by two main shared characteristics: residential location increase general administrative capacities and resources in Kilifi County and knowledge or familiarity with MSM before introducing more complex functions. Bossert et al.[20] through professional, community, or personal experience, yield- also discussed issues of decision space in the context of financing discrete focus group samples (i.e. HIV healthcare profes- ing and resources allocation in Ghana, Zambia, and Uganda, sionals, implementing stakeholders and members of a and showed how coordination and monitoring mechanisms community-based organization in support of MSM health and among the stakeholders might be a challenge with decentral- human rights). Three FGDs were conducted with HIV health- ization implementation bodies [19,21-23]. While these studies care providers (N = 23) from healthcare facilities, two FGDs did not assess services for vulnerable populations at decentral- were held with people working or volunteering for NGO ized levels, Makofane argued that MSM are inadequately (N = 14) (each FGD including seven to eight participants), and addressed in Africa’s AIDS National Strategic Programming three FGDs were held with members of Kilifi County’s main [23]. The objective of this paper is to gain a better understand- MSM-led Community-Based Organizations (CBO) (N = 22) ing of Kenya’s evolving awareness of MSM inclusiveness in across Malindi, Kilifi and Mtwapa (each FGD including six to health services planning, using Kilifi County as a case study. nine participants). Participants were informed that the pur- Kilifi is a compelling context for analysis because of its rela- pose of FGDs was to discuss barriers and facilitators to tively large MSM population and previous work by Kilifi-based accessing HIV services, and to share recommendations to researchers who have engaged MSM as early as 2005 [24,25]. improving MSM HIV prevention and care services in Kilifi One particularly relevant Kilifi-based research study involved a County. In a final feedback FGD, preliminary study findings NASCOP-endorsed MSM online training designed to were presented for validation to eight representatives of the strengthen healthcare providers’ skills to support healthcare above-mentioned groups of FGD participants. services for MSM patients in East and South Africa, that was Semi-structured IDI topic guides were used to collect evaluated in Kilifi County [10]. This study served as a starting detailed data on current interventions for MSM, MSM guideli- point in the discussions with the different stakeholders on nes, county support, and reporting tools. Semi-structured FGD what was needed to improve HIV service delivery for MSM to topic guides focused on perceptions and experiences of local establish synergies, and to create a collaborative platform to healthcare delivery for MSM including healthcare availability, further strengthen HIV healthcare services for MSM. quality and accessibility. IDIs lasted approximately 60 minutes while FGDs took up to 90 minutes. All IDIs and FGDs were audio-recorded, and socio-demographic characteristics of each 2 | METHODS participant were obtained in a brief questionnaire. A member of the socio-behavioural research team, fluent in English, Swa- The study was conducted between January and July 2017 in hili and the local native language “Mijikenda,” conducted the and around the three largest government health facilities IDIs and moderated the FGDs. Consents and topic guides

127 van der Elst EM et al. Journal of the International AIDS Society 2020, 23(S6):e25597 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25597/full | https://doi.org/10.1002/jia2.25597 were translated by the HIV project’s studies coordinator at included priority-setting by county government staff; pre- KEMRI, who has extensive experience with HIV studies. paredness of leadership and management on MSM HIV issues at the facility level; data reporting at the implementation level; and advocacy for MSM health. Strong power inequities were 2.2 | Ethical considerations observed between policy leadership, healthcare providers and Participants were informed about the study aims and provided MSM, with MSM feeling blamed for their sexual orientation. signed informed consent. The study procedures were MSM agency, as expressed in their ability to access HIV ser- approved by the ethical review board at KEMRI and the Kilifi vices, was significantly constrained by county context. County Department of Health Research Committee (KEMRI/ Scientific and Ethics Review Unit Ref: KEMRI/SERU/CGMR- 3.1 | Priority-setting by county government C/061/3372). Participants received 500 Kenyan shillings (approximately US$6) for travel and out of pocket expenses. Six years into decentralization, Kilifi County has not given as Reimbursement amounts were determined based on previous much priority to HIV programming as compared to other studies with these groups and were deemed appropriate and major health concerns such as malaria, tuberculosis, diarrhoea non-coercive according to local standards [26]. and other non-communicable diseases. Respondents recognized that HIV, and particularly same-sex sexuality, are still extremely sensitive topics. However, all strongly advocated for 2.3 | Analysis MSM inclusion and felt that a breakthrough was imminent – The audio files for the IDIs and FGDs were transcribed verba- that is having MSM patients openly to be attended to in pub- tim, and (if applicable) translated by a socio-science qualitative lic facilities. KIs associated the Kenyan president’s statement researcher with linguistic competency in Swahili, English and on same-sex sexuality with the local county’s priority-setting Mijikenda. The transcripts were uploaded in NVivo 11 soft- and decision making: ware for data management. Analyses followed Braun and Clarke’s thematic approach for qualitative data [27], which “The county government has not yet taken it [MSM HIV involved systematic coding, identifying and defining concepts healthcare] as a responsibility. I think it is just revolving emerging from the data across the data set, mapping the con- the fact that it is not accepted and holistically we saw cepts, creating typologies, finding associations between con- the president not accepting it, so they [county staff] are cepts and seeking explanations from the data. Findings were like who are we to do what other people are rejecting, triangulated across sites and between respondents. why should we pretend to...”

The same respondent continued: 3 | RESULTS “... HIV healthcare is in NASCOP’s guidelines and is Table 1 provides summary characteristics of the study partici- the reason why non-governmental organizations and pants. A total of 76 participants took part in the study: 17 funders are taking it up. So, it is not like it is not possi- participants in the IDIs and 59 participants in one of eight ble, it is very possible, it’s just that structures have to FGDs. Of the 17 IDI participants, 14 interviewees derived be re-established again to have it fit the health commu- from the MoH, including 11 county representatives of NAS- nity in a way that healthcare for MSM can be provided COP and three supervisors. Women and men were equally to and received by MSM without fear at the facility represented in the IDIs, their median age was 42 years level. We [policy makers] are the liaison, and it is our (range: 34 to 49), they had an average of nine years HIV goal to eradicate HIV, so we need to work with part- experience, and 29% was Muslim. In the three healthcare pro- ners, organizations, the community, and most impor- viders’ FGDs, a total of 23 participated, eight from Malindi, tantly we need to involve the MSM on all the reasons eight from Kilifi, and seven from Mtwapa. Irrespective of their why we still have high HIV infections...”. (IDI/policy clinical job role (clinicians, nurses, or HIV counselors), all had maker/F) received previous training on how to counsel MSM clients, and they had an average of six years of experience in the HIV This quote also suggests that much of the effort to include field. The median age was 36 years (range 27 to 46) and the MSM into the HIV prevention and care continuum is done by majority (70%) were women. Twenty-two healthcare providers implementing partners through bilateral donors, rather than were Christian, and one healthcare provider was Muslim. In by the county itself. Financial prioritization was highlighted as the two NGO FGDs, 14 implementing partners took part, the precursor of all quality care within the Comprehensive their median age was 33 years (range: 26 to 40), men were Care Centre (CCC) and was especially crucial for directing 57% and they had an average of eight years of experience in services to MSM who are living with HIV who otherwise the HIV field. For the three FGDs held with 22 MSM, partici- would lack access to healthcare. Due to the influences of pants were purposefully selected for their knowledge of NASCOP and implementing partners, a more positive trend healthcare provided to MSM in Kilifi County. The median age on budgeting, planning and prioritization of county govern- of the MSM participants was 27 years (range: 18 to 35), and ment’s HIV prevention and care programme, catering for 50% was Muslim. MSM-specific needs for tailored services, was noted: Irrespective of participants’ characteristics, responses revealed that all key groups navigated diverse challenges “If you look at our budget, HIV has always received zero related to MSM HIV health services. Specific challenges support from the county government not only for key

128 van der Elst EM et al. Journal of the International AIDS Society 2020, 23(S6):e25597 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25597/full | https://doi.org/10.1002/jia2.25597

Table 1. Socio-demographic characteristics of participants in Kilifi case study, 2017

Characteristic In-depth interviews (N) Focus group discussions (N) Focus group discussions (N)N Focus group discussions (N)

Number of participants 17 23 14 22 Specialty area Health policymakers HIV healthcare providers Leaders of NGO Leaders of MSM CBO Women 9 16 6 0 Men 8 7 8 22 Median age (range) 41.5 (34 to 49) 36 (27 to 45) 33 (26 to 40) 26.5 (18 to 35) Religion Christian 12 22 7 9 Muslim 5 1 7 11 Other 2 Education attainment Primary education 1 11 Secondary education 6 5 4 6 Diploma 11 15 8 2 Degree 2 2 3 Years working in HIV field 1to5 6 15 –– >5118––

CBO, Community Based Organization; NGO, Non-Governmental Organization.

populations, but for the whole HIV management pro- about specific sexual health issues, and recommended skills gramme...yeah, zero finances..., all along we have been training to improve capacities among healthcare providers: relying on NASCOP and depending on partners. HIV is still looked upon at as a programme belonging to NAS- “At XX, the doctors were shocked to learn about my COP.., we need to have a clear memorandum of under- [same-sex] sexuality. They [healthcare providers] dis- standing between the partners and the county cussed it with colleagues in tones that could be heard government to make sure that the finances are there to by other patients. You know, it will take time for them continue HIV services especially for MSM living with [healthcare providers] to accept the sexual aspects of HIV, to reduce stigma and discrimination...”. (IDI/policy MSM, they need sensitization, as we need education maker/M) and to know our rights.” (FGD/MSM)

When the strategic planning was done at the county level, policymakers were already sceptical about the implementation 3.2 | Preparedness of leadership and management of MSM HIV healthcare services. Reasons for their scepticism at the facility level ranged from the county’s overstretched healthcare system The rush towards decentralization combined with limited tech- and related organizational and staffing issues including: the nical capacity and guidance at the local level further de-priori- structure of leadership and management [including the tized the already stigmatized HIV health services in Kilifi. Due county-wide strikes of clinical officers and nurses at the time]; to the dynamics of stigma, manifested in healthcare managers’ lack of ownership; and experiencing “teething” problems by disapproval of same-sex-sexuality practices and general societal new healthcare staff. Across sites, and between respondents, disregard for MSM community members, healthcare providers limited resources were identified as reasons for underper- commonly refused MSM healthcare services to MSM clients: forming country-level responses, ranging from staff shortages, lack of STI diagnostic testing capacity and shortage of medica- “... healthcare workers don’t feel that they have sup- tions for treatment. NGO leaders stressed the importance of port from let’s say their managers or their in-charges endorsing MSM policy reforms and the integration of MSM when it comes to attending to MSM. For example, peo- guidelines to facilitate appropriate services for MSM. Drawing ple working in the CCC are assumed to be positive, on respondents’ different reflections on what needs to be those people working with MSM are assumed to also done to improve MSM inclusiveness in health services, one of be MSM... Even the manager or in-charge, when he the policymakers elucidated: reaches “the camp” [CCC], he peeps and says: “Now this place, ‘ai’, me I can’tgoin”. “Does that really support?” “I think what needs to be done is to create a very (FGD/HCP/F) strong network to be adapted for MSM health so that whatever improvements in one area should be actually In contrast, MSM described the need for HIV services known to the other party, there should... has to be a where they could safely disclose their identity and freely talk lot of transparency in all these activities, there is

129 van der Elst EM et al. Journal of the International AIDS Society 2020, 23(S6):e25597 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25597/full | https://doi.org/10.1002/jia2.25597

supposed to be a lot of cooperation and coordination, 4 | DISCUSSION because we are the custodians of health...”. (IDI/policy maker/M) This case study focused on the necessary conditions required to bring about collectively organized responses to strengthen HIV healthcare services delivery for MSM in a relatively young decentralized health system. Ideally, accurate data on 3.3 | Data reporting, impacting services MSM size estimates would have been instrumental for appro- implementation priate programming. The absence of it, however, made the Implementing partners described inaccurate reports as a case study (conducted in 2017) more valuable, that is the harmful negligence, leading to misinformation and altering stakeholders’ perspectives are relevant as to MSM’s current decision making. They felt that improving accurate data would situation. In accordance with the latest Kenya AIDS Response be crucial to strengthening MSM HIV healthcare services. For Progress Report [28,29], the study showed challenges with example the recently revised reporting tool (MoH form 731), quality and coverage of public HIV healthcare with regard to used to prepare monthly reports on CCC’s patients and feed- infrastructure and workforce. Concerns about MSM-specific ing into Kenya’s District Health Information System, does not needs, such as confidentiality, MSM-specific services, public capture the different KP categories, such as MSM. MSM sta- stigmatization and poor or no linkage to national reporting tus is only documented at the HIV testing and counselling reg- systems, restricted integration of MSM into the existing ister, different from the MoH 731 tool- and is subsequently county health system. Efforts to improve the attitudes of rele- lost at the national level. The following statement represents vant health policy agents (e.g. staff, managers and other policy- responses from various participants: makers) with regard to the basic dignity and rights to inclusive health services can contribute to MSM’s trust in the “We have data, but it is not specifically saying who the health system at the local level [30]. Here, Rondinelli’s early HIV patients are. Only, when we go back to the HTC appraisal of East African decentralization programmes [31], [HIV Testing and Counselling] register there is a place including characterization of limited resources, infrastructure, specifically indicating whether so and so is MSM, a sex workforce and commodities management in general, were still worker, or an IDU. But basically, it ends here, at the applicable. Also, in line with previous findings [28], the man- HTC register”. (FGD/NGO/F) agers or “in-chargers,” lacked trust in the health system which outwardly contributed to fear of engaging MSM into preven- Implementing partners stressed that with proper data col- tion and care programmes across the local units. Although the lection tools would allow them to contribute more efficiently number of relevant publications on decentralization and its towards appropriate planning and implementation of HIV pre- impact on MSM health services in sub-Saharan Africa is lim- vention and care programmes that target MSM. ited, our findings correspond with the critical need to complement institutional capacity and political and economic support with specific MSM programming [18]. For example as much as 3.4 | Advocacy of MSM health test-and-treat programmes in sub-Saharan Africa have evinced All participants pointed out that in order to give momentum general success [32], mobilizing MSM to engage with HIV to and endorse policy reforms that support MSM HIV ser- testing has been challenging at the local level [33]. Given the vices, members from the MSM communities need to be empowering effects of decentralization on county-level policy- acknowledged. As for now MSM issues are silent issues in the makers, prioritization of HIV services for MSM can only result communities and in healthcare facilities. in the presence of clear leadership commitment, managerial MSM respondents expressed the dire need for internal capacities, and provision of funding. Implementation of local coherence and consistency. They urged the different policy- HIV prevention and the care continuum for MSM must be makers and partners to address the structural inequities MSM introduced incrementally and accompanied by comprehensive face both at county and national levels. Following an exchange guidelines on how to provide effective and sensitive MSM about their experiences at public health facilities, one MSM HIV services. In order to achieve improved MSM HIV health- participant summarized their discussion, representing the care services on a larger scale, reporting tools should aggre- point of view of his fellow MSM participants: gate data into transmission risk groups, including MSM, men who have sex with both men and women, men who receive “Many of us [MSM] experience trauma due to prejudice payment for sex with cash, living expenses or goods and men and discrimination. We are not welcome at health facili- who inject drugs, explicitly indicating the services needed at ties, and many of us discontinue care, yet we are in the local level as well as specifying performance criteria that need of critical HIV services ...”. (FGD/MSM). should be met. It is of note that the focus of this study was on HIV healthcare services for MSM in decentralized public They strongly recommended improving the clinical training health facilities, however, other models of differentiated HIV of HIV healthcare providers on matters specific to MSM services delivery for MSM could be considered and may be an issues, as well as strengthening collaboration between MSM alternative option to the current public HIV-related clinics and health advocacy groups together with national entities such as patient–provider relations. NACC and NASCOP. The highest priority expressed by MSM If decentralization structures would not have been created respondents was the need to raise awareness of MSM health in Kenya, the HIV response would have remained highly cen- equity in Kenya’s HIV prevention and care continuum. tralized, local decision-making powers would not have been

130 van der Elst EM et al. Journal of the International AIDS Society 2020, 23(S6):e25597 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25597/full | https://doi.org/10.1002/jia2.25597 consolidated, community involvement would have remained importantly the voices of the MSM communities being served weak, capacity gaps continued, and the policy focus would are critical to improve MSM HIV healthcare services. have exclusively emphasized outcome (e.g. metrics) over process (e.g. engagement). AUTHORS’ AFFILIATIONS The impacts of decentralization with regard to provision of 1Kenya Medical Research Institute-Wellcome Trust Research Programme, Kilifi, healthcare are experienced by all counties and by all popula- Kenya; 2Department of Global Health, Academic Medical Centre, University of tions. However, as MSM are uniquely vulnerable to substan- Amsterdam, the Netherlands; 3Jomo Kenyatta University of Agriculture and Technology, Nairobi, Kenya; 4Institute for Biotechnology Research, Jomo Keny- dard services, there is a need for stronger and focused 5 commitment to the provision of prevention and care including atta University of Agriculture and Technology, Nairobi, Kenya; Department of Horticulture and Food Security, School of Agriculture and Environmental antiretroviral drugs for treating and preventing HIV infection Sciences, Jomo Kenyatta University of Agriculture and Technology, Nairobi, for their unique risks. Improvements in healthcare services for Kenya; 6Ministry of Health, Kilifi County, Kenya; 7Departments of Medicine, MSM cannot happen without endorsing policy reforms that Global Health, and Epidemiology, University of Washington, Seattle, WA, USA; 8Department of Behavioral and Social Sciences, School of Public Health, Brown support integration of MSM services unambiguously at the 9 ’ University, Providence, RI, USA; Nuffield Department of Medicine, University county level. Referring to Kenya s guidelines [7,34,35], barriers of Oxford, Oxford, United Kingdom to the right to necessary and appropriate health services should immediately be identified and addressed [36], and con- COMPETING INTERESTS crete recommendations for healthcare providers, policymakers and stakeholders should be invested in as strategies to assure The authors declare that they have no competing interests. the right to health for MSM populations. First, MSM rights ’ should be incorporated into the country’s legal and cultural AUTHORS CONTRIBUTIONS code, with a focus on laws and decriminalization of same-sex EMvdE, RM, NJ, EG, LK, JO, SG, DO and EJS contributed significantly to the behaviour in order to ensure that access to health services study design. EMvdE and EJS conceived the study. RM conducted the FGDs and interviews. EMvdE, RM and BK analysed the data. EMvdE, RM and EJS dis- are not impacted. Second, the County Management should cussed full texts. EMvdE and RM drafted the manuscript. DO, SG and EJS criti- take a lead in norm diffusion, by implementing policies coun- cally edited the manuscript. All authors read and approved the manuscript. tering all forms of homophobia. Third, HIV-related service delivery for MSM should be prioritized in order to sufficiently ACKNOWLEDGEMENTS impact the HIV epidemic in the county. Fourth, substantial ’ ’ The authors acknowledge NASCOP s team at the Ministry of Health for their improvements are needed to translate NASCOP and NACC s ongoing support, in particular at Kilifi County level; the authors thank the key centralized guidelines to the local levels. Finally, decentralized informants for sharing their insights into local policy, social and political settings governance, including leadership and training should assist as well as general community norms and processes. The authors thank Kilifi with implementation of policies that exist at the central level sub-Counties (i.e. North Kilifi, South Kilifi, Ganze, Kaloleni/Rabai, Malindi and Margarini), the County Health Management Team and Sub-County Health Man- for HIV prevention and care for MSM. agement Teams, as well as all healthcare providers who participated in the Limitations to this research must be acknowledged. This FGDs for their willingness to share “lived” experiences, building meaning into study took place in Kilifi County, where MSM research has discussions. The authors acknowledge the implementing partners across the occurred since 2005; therefore, findings might not be general- three sites, and they thank the LGBTQI + organizations for their valuable con- izable to other counties in Kenya. This paper focuses exclu- tributions, input and enthusiasm in discussions. sively on MSM and does not take into account other KP or competing needs for resources. We recognize that participants FUNDING in this sample may not have represented a broader MSM pop- This study is made possible by the generous support of the American people ulation in Kenya, as many MSM may remain hidden. Further- through the US Agency for International Development (USAID), and the sub- Saharan African Network for TB/HIV-1 Research Excellence (SANTHE), a DEL- more, participants might have been prone to socially desirable TAS Africa Initiative [grant # DEL-15-006]. The DELTAS Africa Initiative is an reporting about their attitudes and experiences related to independent funding scheme of the African Academy of Sciences (AAS)’s Alli- health services for MSM. Finally, it should be acknowledged ance for Accelerating Excellence in Science in Africa (AESA) and the Initiative that we based the above conclusions on limited empirical evi- To Develop African Research Leaders (IDEAL), supported by the New Partner- ’ dence, recognizing that the use of qualitative methods pro- ship for Africa s Development Planning and Coordinating Agency (NEPAD Agency) with funding from the Wellcome Trust [grant # 107752/Z/15/Z] and duces inherently subjective data which may impose limits on the UK government. The KWTRP at the Centre for Geographical Medicine the transferability of the knowledge uncovered. Research-Kilifi is supported by core funding from the Wellcome Trust (grant #077092). Dr. Operario’s involvement in this project was supported in part by NIH grant NIAID P30AI042853. Dr. Graham was supported by NIH grant NIMH R34MH099946 and the Robert W. Anderson Endowed Professorship in 5 | CONCLUSIONS Medicine. The contents are the responsibility of the study authors and do not necessarily reflect the views of USAID, the NIH, the US government or the To strengthen HIV healthcare for MSM within a decentralized Wellcome Trust. This report was published with permission from KEMRI. Kenyan health system, a more responsive, multi-pronged strategy adaptable and relevant to local county settings is needed. High-level government decision makers need to acknowledge REFERENCES that HIV transmission occurs within marginalized groups and 1. UNAIDS. Eastern and Southern Africa UNAIDS data UNAIDS data. 2019 consolidate expert guidance to support implementation of dif- [cited 2020 Aug 10]. Available from: https://www.unaids.org/sites/default/files/ media_asset/2019-UNAIDS-data_en.pdf ferentiated HIV services delivery for MSM. Continued engage- 2. Sanders EJ, Graham SM, Okuku HS, van der Elst EM, Muhaari A, Davies A, ment with policy leadership, collaboration with health facilities, et al. HIV-1 infection in high risk men who have sex with men in Mombasa, partnerships with different community stakeholders and most Kenya. AIDS. 2007;21(18):2513–20.

131 van der Elst EM et al. Journal of the International AIDS Society 2020, 23(S6):e25597 http://onlinelibrary.wiley.com/doi/10.1002/jia2.25597/full | https://doi.org/10.1002/jia2.25597

3. Sanders EJ, Okuku HS, Smith AD, Mwangome M, Wahome E, Fegan G, et al. 19. Zon H, Pavlova M, Drabo KM, Groot W. Municipal health services provision High HIV-1 incidence, correlates of HIV-1 acquisition, and high viral loads fol- by local governments: a systematic review of experiences in decentralized sub- lowing seroconversion among MSM. AIDS. 2013;27(3):437–46 Saharan African countries. Health Policy Plan. 2017;32(9):1327–36 4. Wahome E, Graham SM, Thiong’o AN, Khamisi M, Oduor TH, Gichuru E, 20. Bossert T, Health Beauvais J. Decentralization of health system in Ghana, et al. HIV-1 incidence, PrEP uptake, and adherence among Kenyan MSM. Con- Zambia, Uganda and Philippines: a comparative analysis of decision space. ference on Retroviruses and Opportunistic Infections. Boston, MA; 2020. Health Policy Plan. 2002;17:14–31. 5. Nduva GM, Hassan AS, Nazziwa J, Graham SM, Esbjornsson J, Sanders EJ. 21. Munga MA, Songstad NG, Blystad A, Mæstad O. The decentralisation- HIV-1 transmission patterns within and between risk groups in coastal Kenya. centralisation dilemma: recruitment and distribution of health workers in remote Sci Rep. 2020;10(1):6775. districts of Tanzania. BMC Int Health Human Rights. 2009;9:9. 6. Sanders EJ, Jaffe H, Musyoki H, Muraguri N, Graham SM. Kenyan MSM: no 22. Kilifi County Integrated Development Plan 2018–2022. Volume I; 1.24.8. longer a hidden population. AIDS. 2015;29 Suppl 3:S195–9. 2020 [cited 2020 Aug 10]. Available from: https://www.kilifi.go.ke/lib.php?com= 7. Gruskin S, Tarantola D. Universal Access to HIV prevention, treatment and 5&res_id=1202 care: assessing the inclusion of human rights in international and national strate- 23. Makofane K, Gueboguo C, Lyons D, Sandfort T. Men who have sex with gic plans. AIDS. 2008;22 Suppl 2:S123–32. men inadequately addressed in African AIDS National Strategic Plans. Global 8. KASF, 2018/19. 2020 [cited 2020 Aug 10]. Available from: https://nacc.or. Public Health. 2013;8(2):129–43. ke/kenya-aids-strategic-framework-kasf/NACC 24. Barasa DW. Indigeous knowledge systems and sustainable development in 9. Cohen MS, Smith MK, Muessig KE, Hallett TB, Powers KA, Kashuba AD. Africa: case study on Kenya. Tribes Tribals. 2007;2007:141–56. Antiretroviral treatment of HIV-1 prevents transmission of HIV-1: where do we 25. Shepherd G. Rank, gender and homosexuality: Mombasa as a key to undergo from here? Lancet. 2013;382(9903):1515–24. standing sexual options. In: Caplan P, editor. The Cultural Construction of Sexu- 10. van der Elst EM, Kombo B, Gichuru E, Omar A, Musyoki H, Graham SM, ality Volume 353 of Social science paperbacks. London, Tavistock: Tavistock et al. The green shoots of a novel training programme: progress and identified Publications; 1987. key actions to providing services to MSM at Kenyan health facilities. J Int AIDS 26. van der Elst EM, Kombe B, Gixhuru E, Omar A, Musyoki H, Graham SM, Soc. 2015;18:20226. et al. The Green shoots of a novel training programme: progress and identified 11. van der Elst EM, Gichuru E, Omar A, Kanungi J, Duby Z, Midoun M, et al. key actions to providing services to MSM at kenyan health facilities. J Int AIDS Experiences of Kenyan healthcare workers providing services to men who have Society. 2015;18:20226. sex with men: qualitative findings from a sensitivity training programme. J Int 27. Braun V, Clarke V. Using thematic analysis in psychology. Qual Res Psychol. AIDS Soc. 2013;16:18741. 2006;3(2):77–101. 12. Micheni M, Kombo BK, van der Elst EM, Simoni JM, Operario D, San- 28. NACC. Kenya AIDS response progress report 2018. 2020. Available from: ders EJ, et al. Health provider perspectives on determinants of ART adherence https://www.lvcthealth.org/wp-content/uploads/2018/11/KARPR-Report_2018. among MSM in Coastal Kenya. AIDS Patient Care STDs. 2017;31(3):113–21. pdf 13. Savage A, Lumbasi L. The impact of decentralization in Kenya - Aidlink. Trin- 29. NACC. Kenya AIDS Strategic Framework 2014/15 - 2018/19. 2019;84. ity College Dublin. Masters in Development Practice. 2016 [cited 2020 Aug 30. Rimal RN, Real K. How behaviors are influenced by perceived norms a 10]. Available from: http://aidlink.ie/wp-content/uploads/2016/08/THE-IMPACT- test of the theory of normative social behavior. Commun Res. 2005;32 OF-DECENTRALIZATION-IN-KENYA (3):389–414. 14. Onyango PO. Devolution made simple: a popular version of the County 31. Rondinelli D, Nellis J. Assessing decentralization policies in developing coun- governance [cited 2020 Aug 10]. Available from: http://www.feskenya.org/med tries: a case for cautious optimism. Develop Policy Rev. 1986;4:3–23. ia/publications/Devolution%20Booklet%20for%20web.pdf 32. Havlir DV, Balzer LB, Charlebois ED, Clark TD, Kwarisiima D, Ayieko J, 15. Nyikuri M, Tsofa B, Barasa E, Okoth P, Molyneux S. Crises and resilience at et al. HIV testing and treatment with the use of a community health approach the frontline- public health facility managers under devolution in sub county on in rural Africa. N Engl J Med. 2019;381(3):219–29. the Kenyan Coast. PLoS One. 2015;10:e0144768. 33. van der Elst EM, Mahmoud S, Oduor C, Chirro O, Ibrahim F, Kombo B, 16. McCollum R, Theobald S, Otiso L, Martineau T, Karuga R, Barasa E, et al. et al. Peer-led oral HIV-self testing finds undiagnosed HIV among MSM in Mal- Priority setting for health in the context of devolution in Kenya: implications for indi, Kenya. Conference on Retroviruses and Opportunistic Infections, Seattle, health equity and community-based primary care. Health Policy Plan. 2018;33 USA; 2017. Abstract/893; available online in the Special Issue of Topics in (6):729–42. Antiviral Medicine. 17. Tsofa B, Molyneux S, Gilson L, Goodman C. How does decentralisation affect 34. KLRC. Constitution of Kenya; Part 2. Rights and fundamental freedoms; 43. health sector planning and financial management? a case study of early effects of Economic and social rightsChapter Four - The Bill of Rights. 2013. devolution in Kilifi County, Kenya. Int J Equity Health. 2017;16(1):151. 35. Kenya Constitution 2013. The Bill of Rights. Rights and fundamental free- 18. Barasa E, Molyneux S, English M, Cleary S. Setting healthcare priorities at doms; 43. Economic and social Rights;Part 2(Chapter Four). the macro and meso levels: a framework for evaluation. Int J Health Policy 36. Beyrer C. Section 377: Why sodomy statutes matter. J Int AIDS Soc. Manag. 2015;4(11):719–32. 2019;22:e25285.

132 Engagement of African men and transgender women who have sex with men in HIV research Journal of the International AIDS Society 2020, 23 (Suppl S6)

AUTHOR INDEX

A K O eno, F.O. 51 O eno-Nyunya, B. 51 Adebajo, S. 70 Kabu , R. 91 Ake, J.A. 40 Kagaba, A. 80 P Allen, S. 80 Kamba, D. 59 Kanyemba, B. 59 Palanee-Phillips, T. 91 B Karita, E. 80 Palmer, S. 106 Kayode, B.O. 40, 70 Panchia, R. 59 Bailey, R.C. 51 Ket, J.C. 106 Phaswana-Mafuya, N. 6, 17 Baral, S.D. 6, 17, 40, 70, 80, 120 Ketende, S. 80 Bekker, L-G. 1, 6, 17 Kga tswe, L.B. 6, 17 R Bourne, A. 91 Kiirika, L. 126 Kimani, J. 91 Radebe, O. 59 C Kokogho, A. 70 Ramadhani, H.O. 40, 70 Kombo, B. 126 Reynolds, D. 59 Kunzweiler, C. 51 Charurat, M.E. 40, 70 Robb, M.L. 40 Chege, W. 59 Chen, Y.Q. 59 L S Crowell, T.A. 1, 40, 70 Cummings, V. 59 Liestman, B. 80 Sanchez, T.H. 6, 17 Lucas, J. 59 Sanders, E.J. 1, 30, 106, 126 D Lyons, C.E. 80 Sandfort, T.G.M. 59 Schim van der Loeﬀ , M.F. 106 Dadabhai, S. 59 M Schnabel, D. 59 de Bree, G.J. 106 Shoyemi, E. 70 Dijkstra, M. 106 Makofane, K. 120 Siegler, A.J. 6, 17 Djomand, G. 51 Marais, A. 59 Smith, A.D. 91 Dominguez, K. 6, 17, 59 Mazzei, A. 80 S rra , M. 59 Mbeda, C. 59 Sugarman, J. 59 McKinstry, L.A. 59 Sullivan, P.S. 6, 17, 80 E McNaghten, A.D. 6, 17 Mehta, S.D. 51 Ekeh, C. 70 T Mitchell, A. 40 Eshleman, S.H. 59 Mohamed, K. 30 Tenza, S. 91 Mudza, R. 126 Thiongʼo, A.N. 30 F Mugwaneza, P. 80 Twahirwa Rwema, J.O. 80 Mwambi, J. 30 Fast, P.E. 1 Fawzy, M. 59 V Fearon, E. 91 N Valencia, R. 6, 17 Ndembi, N. 40, 70 van der Elst, E.M. 30, 106, 120, 126 G Nemande, S. 120 Nowak, R.G. 40, 70 Gichuru, E. 30, 106 Nsanzimana, S. 80 W Graham, S.M. 30, 51, 126 Nutland, W. 91 Guo, X. 59 Nyombayire, J. 80 Wahome, E.W. 30, 106 Walimbwa, J. 106, 120 Weatherburn, P. 91 H O Hamilton, E.L. 59 Oduor, T. 30 Y Herbst, S. 80 Ogendo, A. 59 Okall, D.O. 51 Yah, C.S. 6, 17 J Olawore, O. 80 Onguso, J.M. 126 Z Jahangir, N. 126 Ononaku, U. 70 Jones, J. 6, 17 Operario, D. 126 Zahn, R. 6, 17 Zulu, F. 59

133

Editors Journal Information Editors-in-Chief: Kenneth H Mayer (United States) About the journal Annette Sohn (Thailand) The Journal of the International AIDS Society, the offi cial journal of the Society, provides a peer- Executive Editor: reviewed, open access forum for essential and innovative HIV research, across all disciplines. All Marlène Bras (Switzerland) articles published by the Journal of the International AIDS Society are freely accessible online. Deputy Editors: The editorial decisions are made independently by the journal’s Editors-in-Chief. Laith Abu-Raddad (Qatar) Jenny Anderson (Australia) Website: www.jiasociety.org Benjamin Bavinton (Australia) eISSN: 1758-2652 Carol Camlin (United States) Morna Cornell (South Africa) Contact details Claudia Cortés (Chile) Trevor Crowell (United States) Editorial offi ce: Alison Drake (United States) Avenue de France, 23 Matthew Fox (United States) CH-1202 Geneva Omar Galárraga (United States) Switzerland Kimberly Green (Vietnam) Email: [email protected] Anna Grimsrud (South Africa) Tel: +41 (0) 22 710 0800 Renee Heff ron (United States) Martin Holt (Australia) Publisher Mina Hosseinipour (United States) Rami Kantor (United States) The Journal of the International AIDS Society is published by John Wiley & Sons Ltd on behalf Sheri Lippman (United States) of the International AIDS Society Carmen Logie (Canada) John Wiley & Sons Ltd Nyaradzo Mgodi (Zimbabwe) Matthew Mimiaga (United States) 9600 Garsington Road Kenneth Ngure (Kenya) Oxford, OX4 2DQ UK Sophie Pascoe (South Africa) Telephone: +44 1865 776868 Nittaya Phanuphak (Thailand) Email: [email protected] Colette Smith (United Kingdom) Sunil Solomon (United States) Production Editor Jonathan Stadler (South Africa) Junko Tanuma (Japan) Jayanthi Adilingam (email: [email protected]) Elona Toska (South Africa) Lara Vojnov (Switzerland) Abstracting and Indexing Services Iryna Zablotska (Australia) The Journal of the International AIDS Society is indexed in a variety of databases including Managing Editor: PubMed, PubMed Central, MEDLINE, Science Citation Index Expanded and Google Scholar. Elisa de Castro Alvarez (Switzerland) The 2019 Journal Impact Factor is 5.553, Journal Citation Reports (Clarivate Analytics, 2020). Editorial Assistant: Karoline Soerensen (Switzerland) Advertising, sponsorship and donations Editorial Board Please contact the editorial offi ce if you are interested in advertising on our journal’s website. Quarraisha Abdool Karim (South Africa) We also gladly receive inquiries on sponsorship and donations to support open access Laith J Abu-Raddad (Qatar) publications from authors in low- and middle-income countries. Adaora Adimora (United States) Joseph Amon (United States) Supplements Judith D Auerbach (United States) The Journal of the International AIDS Society publishes supplements and thematic series on Linda-Gail Bekker (South Africa) its own initiative or based on proposals by external organizations or authors. Inquiries can be Sarah Bernays (Australia) sent to the editorial offi ce at [email protected]. Chris Beyrer (United States) Carlos Cáceres (Peru) All articles submitted for publication in supplements are subject to peer review. Published Andrea Ciaranello (United States) supplements are freely accessible online and can also be produced in print. Elisabeth Connick (United States) Mark Cotton (South Africa) Disclaimer Diana Dickinson (Botswana) The Publisher, International AIDS Society and Editors cannot be held responsible for Sergii Dvoriak (Ukraine) errors or any consequences arising from the use of information contained in this journal; Paul Flowers (United Kingdom) the views and opinions expressed do not necessarily refl ect those of the Publisher, Nathan Ford (Switzerland) International AIDS Society and Editors, neither does the publication of advertisements Omar Galárraga (United States) constitute any endorsement by the Publisher, International AIDS Society and Editors of Elvin Geng (United States) the products advertised. Beatriz Grinsztejn (Brazil) Huldrych Günthard (Switzerland) Diane Havlir (United States) Copyright and Copying Amy Justice (United States) The content in this supplement is published under the Creative Commons Attribution license Adeeba Kamarulzaman (Malaysia) (“CC-BY”). The license allows third parties to share the published work (copy, distribute, Rami Kantor (United States) transmit) and to adapt it under the condition that the authors are given credit, and that in Sukhontha Kongsin (Thailand) the event of reuse or distribution, the terms of this license are made clear. Authors retain Nagalingeswaran Kumarasamy (India) the copyright of their articles, with fi rst publication rights granted to the Journal of the Mathias Lichterfeld (United States) International AIDS Society. Kathleen MacQueen (United States) Navid Madani (United States) Wiley’s Corporate Citizenship Initiative Paula Munderi (Uganda) Denis Nash (United States) Wiley’s Corporate Citizenship Initiative seeks to address the environmental, social, economic, Christy E Newman (Australia) and ethical challenges faced in our business and which are important to our diverse stakeholder Richard Parker (Brazil) groups. Since launching the initiative, we have focused on sharing our content with those in need, Praphan Phanuphak (Thailand) enhancing community philanthropy, reducing our carbon impact, creating global guidelines and Tonia Poteat (United States) best practices for paper use, establishing a vendor code of ethics, and engaging our colleagues Anton Pozniak (United Kingdom) and other stakeholders in our eff orts. Follow our progress at www.wiley.com/go/citizenship. Linda Richter (South Africa) Jürgen Rockstroh (Germany) Research4Life Sean Rourke (Canada) Gabriella Scarlatti (Italy) Wiley is a founding member of the UN-backed HINARI, AGORA, and OARE initiatives. They Lorraine Sherr (United Kingdom) are now collectively known as Research4Life, making online scientifi c content available free or Colette Smith (United Kingdom) at nominal cost to researchers in developing countries. Tim Spelman (Australia) Please visit Wiley’s Content Access – Corporate Citizenship site: Steff anie Strathdee (United States) www.wiley.com/WileyCDA/Section/id-390082.html Francois Venter (South Africa) Sten Vermund (United States) Alan Whiteside (South Africa) Iryna Zablotska (Australia)