Adherence to quarterly HIV prevention services and its impact on HIV incidence in men who have sex with men in West Africa (CohMSM ANRS 12324 – Expertise France)

Background Access to tailored HIV prevention services remains limited for West African MSM. We assessed adherence to quarterly HIV prevention services and its impact on HIV incidence in MSM followed up in four cities in Burkina Faso, Côte d’Ivoire, Mali, and Togo. Methods We performed a prospective cohort study between 2015 and 2018. HIV-negative MSM aged over 18 benefited from quarterly medical visits which included a clinical examination, HIV testing, screening and treatment for other sexually transmitted infections, peer-led counselling and support, and the provision of condoms and lubricants. Determinants of adherence to quarterly follow-up visits and incident HIV infections were identified using generalized estimating equation models and Cox proportional hazard models, respectively. Results 618 MSM were followed up for a median time of 20.0 months (interquartile range 15.2–26.3). Overall adherence to quarterly follow-up visits was 76.5% (95% confidence interval [CI] 75.1–77.8), ranging from 66.8% in Abidjan to 87.3% in Lomé (p < 0.001). 78 incident HIV infections occurred during a total follow-up time of 780.8 person-years, giving an overall incidence of 10.0 per 100 person-years (95% CI 8.0–12.5). Adherence to quarterly follow-up visits was not associated with the risk of incident HIV infection (adjusted hazard ratio 0.80, 95% CI 0.44–1.44, p = 0.545). Conclusions Strengthening HIV prevention services among MSM in West Africa, including the use of PrEP, will be critical for controlling the epidemic, not only in this key population but also in the general population. Quarterly follow-up of MSM, which is essential for PrEP delivery, appears feasible. Trial registration ClinicalTrials.gov, number NCT02626286 (December 10, 2015).


Background
Men who have sex with men (MSM) are disproportionately affected by HIV in Africa and elsewhere [1][2][3]. In 2018, their risk of acquiring HIV was 22 times higher than that of all male adults. In Western and Central Africa, the HIV epidemic is concentrated in key populations (including MSM, sex workers, and people who inject drugs), which together with their sexual partners represented 64% of all new HIV infections in 2018 [4]. Furthermore, median HIV prevalence in 16 West African countries was 13.7% among MSM compared with 1.5% in the general population. In addition, HIV infection in African MSM contributes to the dynamic of the epidemic [5,6] because 50-90% of MSM also have sex with women [7,8].
Since the early 2010s, the World Health Organization (WHO) has recommended targeting MSM as part of the response to the HIV epidemic [9,10]. More specifically, recommendations encourage healthcare decision-makers to promote routine HIV testing and counselling in this population. MSM are also encouraged to perform HIV retesting at least annually and every three months if they engage in high-risk sexual behaviours. Since 2014, the WHO has also recommended developing comprehensive prevention and care services which not only include the abovementioned HIV testing and counselling but also pre-exposure prophylaxis (PrEP), linkage and enrolment in care, HIV treatment and care, prevention and management of co-infections and other co-morbidities, as well as sexual and reproductive health interventions [11].
Risk factors for HIV acquisition in MSM include high per-act transmission probability of receptive anal sex, high rate of sexually transmitted infections (STI), high frequency of condomless anal sex, high number of male partners, and involvement in large sexual networks [2,12,13]. Moreover, MSM in Africa face social and legal barriers (i.e., homophobia, stigmatization, discrimination, violence, and penalization of same-sex relationships) that hinder their access to HIV prevention and care services [14,15]. Unfortunately, HIV services tailored to their needs are limited and are mostly offered by communitybased organizations [16,17]. Furthermore, in general, HIV-negative MSM come to prevention services when they have a health need (e.g., STI) or for specific needs (e.g., condoms and lubricants).
The CohMSM study was designed to assess the feasibility and interest of implementing quarterly HIV prevention and care services in MSM in West Africa. In the present analysis, we assessed adherence to quarterly prevention services and its impact on HIV incidence in MSM followed up in four West African countries. Our study hypothesis was that high adherence to quarterly prevention services would be associated with lower HIV incidence.

Study design, setting and participants
A prospective cohort study was performed between June 2015 and January 2018 in Abidjan (Côte d'Ivoire), Bamako (Mali), Lomé (Togo), and Ouagadougou (Burkina Faso). MSM were enrolled and followed up in community-based clinics already providing MSMspecific prevention, care, and support (Clinique de Confiance in Abidjan, Clinique des Halles in Bamako, Centre Lucia in Lomé, and Centre Oasis in Ouagadougou). MSM were eligible if they were aged 18 or over, reported at least one episode of anal intercourse with another man in the previous three months, and were HIV negative (status confirmed at study enrolment). At enrolment and during the quarterly follow-up visits, participants benefited from a clinical examination, HIV testing, screening and treatment for other STI, personalized peer-led counselling and support, and the provision of condoms and lubricants. PrEP was not available. MSM who seroconverted during follow-up were invited to initiate antiretroviral therapy (ART) immediately. Participants could also attend the clinics at any time according to their needs. All services were free of charge. Participants were compensated 3000 Francs CFA (approximately US$5) for transport costs for each scheduled follow-up visit. Socio-demographic and behavioural data were collected at enrolment and every six months thereafter using a standardized face-to-face questionnaire administered by trained research assistants. Finally, with their consent, peer-educators could contact the participants by phone if they were 15 days late for their scheduled visits. The study was discontinued when PrEP was added to the CohMSM cohort. Specifically, the study was performed from June 2015 to January 2018 in Bamako, from October 2015 to January 2018 in Abidjan, from February 2016 to November 2017 in Ouagadougou, and from June 2016 to November 2017 in Lomé.

Laboratory procedures
Screening for HIV was performed according to national algorithms [18][19][20][21]. All four study cities first used the Determine HIV 1/2 assay (Abbott Laboratories, Chiba, Japan). Positive results were confirmed using the Bioline HIV-1/2 3.0 assay (SD, Gyeonggi-do, Republic of Korea) in Abidjan, Bamako, and Ouagadougou, or the First Response HIV-1/2 assay (Premier Medical Corporation, Mumbai, India) in Lomé. Samples with discordant results were tested a third time using the HIV 1/2 Stat-Pak assay (Chembio Diagnostics, New York, USA) in Abidjan, the First Response HIV-1/2 assay (Premier Medical Corporation, Mumbai, India) in Bamako, the Inno-Lia HIV I/II Score assay (Fujirebio, Zwijnaarde, Belgium) in Lomé, or a Western Blot assay in Ouagadougou.

Adherence to quarterly follow-up visits
Adherence to quarterly follow-up visits was defined as the proportion of scheduled visits which MSM actually attended (i.e., the number of attended visits divided by the number of scheduled visits). A visit was considered attended when it was carried out at the scheduled date plus or minus 45 days [22]. The period of participation for each participant (hereafter called "observation time") ended at study discontinuation, first HIV positive screening test, or death.

HIV incidence
Incident HIV infection was defined as a HIV infection detected during follow-up. The date of HIV infection was estimated as the midpoint between the date of the last negative screening test and the date of the first positive one. HIV incidence was calculated per 100 personyears. Follow-up time was calculated from enrolment to HIV infection or the last participant's HIV screening test.

Explanatory variables
Potential explanatory variables considered in this analysis included: i) socio-demographic characteristics: city, age, education (primary school or higher versus never attended school or koranic school), marital status (married or in free union versus single/divorced/separated/ widowed), ii) psychosocial characteristics: self-defined sexual orientation (homosexual or gay/heterosexual/bisexual versus transsexual/transgender), self-identified gender (much more a woman/both a man and a woman/neither man nor woman versus a man or a boy), sexual attraction (to men and women/to women versus to men), received psychological support (yes versus no); iii) sexual behaviours in the previous 6 months: condom use during insertive anal sex (inconsistent, no insertive anal sex versus consistent), condom use during receptive anal sex (inconsistent, no receptive anal sex versus consistent), received payment (whether financial or other) for transactional sex with male partners (sometimes, always versus never), provided payment (whether financial or other) for transactional sex with male partners (sometimes, always versus never), number of male sexual partners (1-5 versus ≥ 6); iv) clinical data: history of HIV screening (no versus yes), STI (other than HIV) symptoms (yes versus no).

Statistical analysis
Adherence to quarterly follow-up visits was calculated overall and according to each study city. The 95% confidence intervals (CI) of adherence were computed using the binomial method. The evolution of adherence during follow-up was assessed using the χ 2 test for trend.
Determinants of adherence to quarterly follow-up visits were identified using generalized estimating equation models which provide population-averaged estimates while controlling for the correlation of repeated measures for the same individual. Independent variables associated with adherence with a p-value < 0.20 in univariate analyses were selected for the complete multivariate model. A backward elimination procedure based on the quasi-likelihood Akaike's information criterion was used to determine the final multivariate model.
Cumulative hazards of incident HIV infections were estimated by Kaplan-Meier survival curves and compared between the study cities using the log-rank test. Given the fact that the proportional hazards hypothesis using the Schoenfeld residuals was verified for the most important covariates (i.e., adherence to quarterly followup visits and study city), the determinants of incident HIV infections were investigated using Cox regression models. Independent variables associated with incident HIV infections with a p-value < 0.20 in univariate analyses were specified in the complete multivariate model. A manual backward selection based on the loglikelihood method was used to determine the final multivariate model.
For the analyses of both outcomes, we used timeconstant and time-dependent variables. The former were collected at enrolment and included the study city, age, educational level, marital status, and history of HIV screening. The latter included self-defined sexual orientation, self-identified gender, sexual attraction, condom use, sexual behaviours, psychological support, and STI (i.e., other than HIV) symptoms.
For all calculations, statistical significance was defined with a p-value < 0.05. All statistical analyses were performed using Stata software (version 15; Stata Corp LP, College Station, Texas). The CohMSM study is registered with ClinicalTrials.gov, number NCT02626286). Adherence to quarterly follow-up visits was highest at month 3 (82.2%) and lowest at month 24 (64.5%; Fig. 1). Overall, it decreased significantly over time (p < 0.001). The decrease was significant in Bamako and Abidjan (p < 0.001 for both) but not in Ouagadougou (p = 0.887) and Lomé (p = 0.730).

Discussion
This multi-country study conducted in MSM living in West Africa showed good overall adherence to quarterly HIV prevention services, which included HIV testing and counselling, screening and treatment for other STI, and the provision of condoms and lubricants. However, we recorded a decrease in adherence over time as well as differences between the study cities (Abidjan, Bamako, Lomé, and Ouagadougou). Moreover, adherence had no significant impact on HIV incidence, which was high in the study population. The good overall adherence to quarterly follow-up visits by HIV-negative individuals is encouraging for the implementation of prevention programmes, as MSM commitment is crucial for repeated HIV testing and counselling, screening and treatment for other STI, as well as PrEP (not examined here). Importantly, repeated HIV testing was well accepted by participants and was performed systematically at each visit. The good adherence observed in the study was likely related to the favourable study context, specifically the fact that the study clinics were MSM-friendly, and that peereducators were very involved in enrolment and retention of MSM in the programme, as well as counselling and psychosocial support [23]. Adherence support for MSM included a reminder telephone call 15 days after an outstanding follow-up visit. This measure was inspired by routine practice in the study clinics, in which medical teams call HIV-positive patients if they are late for scheduled visits. Mobile phones are now widely used throughout Sub-Saharan Africa, and just as is the case for ART, they constitute a useful and affordable tool for maintaining adherence to prevention services. The compensation of US$5 for transport costs for each follow-up visit certainly encouraged adherence. Although the use of financial compensation for routine medical interventions has been debated, its effectiveness has been shown in different contexts [24,25].
However, two of our findings on adherence to quarterly HIV prevention services call for caution. First, despite having the same study procedures, lower adherence was observed in Abidjan than in Bamako and Lomé, which suggests that the local context (e.g., difficulty and cost of transportation, reluctance/willingness to come to clinics, and spatial and temporal organization of clinics) had an impact. This underlines the need for additional measures adapted to the local context in terms of adherence support for MSM and solutions to overcome organizational constraints in clinics. Second, the observed decrease of adherence over time could be problematic for long-term public health programmes. However, the relatively small proportion of participants lost to follow-up shows that most of the missed visits were accounted for MSM still in the programme. We hypothesize that long-term adherence to follow-up will be better in MSM using PrEP who will need to come to medical visits for supplies.
Despite the good adherence to quarterly prevention services, HIV incidence in our study was far higher than the WHO-recommended threshold of 3 per 100 person-  [13,[27][28][29]. Although HIV incidence did not differ statistically between the four study cities, our data confirm that it was especially high in Abidjan, reflecting findings in CohMSM's pilot study in 2013-2015 (15.9 per 100 person-years, 95% CI 7.6-33.4) [30]. In a previous analysis, we found that MSM at greater risk of  exposure to HIV infection decreased their risky sexual behaviours during their follow-up in CohMSM, suggesting a potential positive effect of the quarterly prevention services on those who need it most [31]. However, the high HIV incidence and the lack of association between adherence to these services and incident HIV infections observed in the present analysis strongly suggest that this intervention alone is insufficient to significantly reduce the burden of the epidemic in this population. This may be due to the fact that most HIV infections occurred in the first months of follow-up while changes in sexual behaviours require more time and are rarely optimal. These data confirm the need to strengthen HIV prevention services through the use of PrEP. In our study, some participants did not attend the quarterly follow-up visits because they had moved elsewhere, either temporarily or permanently. Only when participants moved from one study city to another -which was quite common in the study population -could they continue their follow-up in the second study clinic. This underscores the need to establish a network of intra-country and inter-country MSM-friendly clinics in West Africa (and elsewhere) for sustainable prevention.
A major challenge for the control of HIV infection in MSM is to reach and test those who have never been tested. In our study, these men were at greater risk of acquiring HIV infection. A recent study in neighbouring Nigeria reported a similar finding [13]. The relationship between condom use during insertive anal sex and incident HIV infections reflects existing evidence, as inconsistent condom use is a well-known risk factor and most participants who reported no insertive anal sex also reported receptive anal sex.
The main strength of this study is the fact that it was conducted in four West African countries, which allowed us to highlight differences in the outcomes between all four. However, our findings have the following limitations. First, our participants might not be fully representative of the global MSM community in the four study cities because the study was performed in MSM enrolled and followed up in MSM-friendly communitybased clinics. Second, the study was performed in the major cities of Burkina Faso, Côte d'Ivoire, Mali, and Togo and our results may not be generalizable to other contexts of these countries. Third, the duration of the study differed between countries due to the staggered start of the study and the staggered discontinuation of the study. Finally, a social desirability bias may have affected participants' responses (e.g., regarding condom use).