State-sector MSM-targeted sexual health services have existed inSouthAfrica since 2009 when the Anova Health Institute’s Health4Men programme was launched as a collaborative project with the Department of Health with support and funding from PEPFAR/USAID.  The lon- gest running of these services is the Ivan Toms Centre for Men’s Health (ITCMH) inCapeTown which functions as a centre of excellence for developing and delivering locally-appropri- ate MSM sexual health care. This clinic has provided services to more than 6,500 MSM to date. [18,19] The clinic is a primary-level, sexual health and wellness service focusing on HIV and STI diagnosis, treatment and prevention and includes biomedical, psychosocial, educational and community components. The ITCMH is situated alongside major car, taxi and train com- muter routes and attracts MSM clients with a range of sexual identities from a wide geographic catchment area surrounding CapeTown. Since the clinic attracts clients from such a diverse spectrum of MSM in different geographic areas, andwith diverse socio-economic status and cultural backgrounds, it is likely that MSM attending the clinic are broadly representative of urban and peri-urban MSM inand around the City ofCapeTown.
In terms of control variables, CAPS collected a range of information on sexual behaviors/partners, HIV knowledge, and socioeconomic indicators that could jointly influence perceptions of HIV risk and reported condom use and knowledge about the protective benefits of circumcision. Regarding sexual behaviors/ partners, we included binary measures of whether the respondent ever had a sexually transmitted disease, reported that either they or their last partner had concurrent partners, and was currently married, as well as a continuous measure of age at first sex. We also included a measure of HIV knowledge and whether or not the respondent knew someone who died of HIV. The HIV knowledge measure counted correct answers to whether HIV can be transmitted via food prepared by someone with HIV/AIDS, by being coughed or sneezed on, from mother to child, and whether it is possible for a healthy-looking person to have HIV. Furthermore we created an indicator of conspiracy beliefs (equal to 1 if the respondent agreed that HIV was created by humans, AIDS was created by scientists in America, or that AIDS was invented to kill black people), which could influence both risk perceptions and Circumcision and Risk Compensation
The iPrEx study (ClinicalTrials.gov: NCT00458393) was approved by the Committee on Human Research at the University of California, San Francisco, as well as local institutional review boards (IRBs) at each study site: Comite´ Institucional de Bioe´tica, Asociacio´n Civil Impacta Salud y Educacio´n, Lima, Peru; Universidad San Francisco de Quito, IRB #1, Quito, Ecuador; Fenway Community Health Institu- tional Review Board, Boston, MA; Comissa˜o de E ´ tica para Ana´lise de Projetos de Pesquisa, CAPPesq Hospital das Clı´nicas da Faculdade de Medicina da USP, Sa˜o Paulo, Brazil; Comiteˆ de E ´ tica em Pesquisa, Hospital Universitario Clementino Fraga Filho/Universidade Federal de Rio de Janeiro, Rio de Janeiro, Brazil; Comiteˆ de E ´ tica em Pesquisa do Instituto de Pesquisa Clı´nica Evandro Chagas, Rio de Janeiro, Brazil; National IRB: Comissa˜o Nacional de E ´ tica em Pesquisa – CONEP, Ministe´rio da Sau´de, Brası´lia, Brazil; University ofCapeTown Research Ethics Committee, CapeTown, SouthAfrica; Human Experi- mentation Committee, Research Institute for Health Sciences, Chiang Mai, Thailand; Ethical Review Committee for Research in Human Subjects, Department of Medical Services, Ministry of Public Health, Nonthaburi, Thailand; Research Ethics Commit- tee, Faculty of Medicine, Chiang Mai University, Chiang Mai, Thailand. Written informed consent was obtained from each participant prior to enrollment in the study.
Multiple logistic regression was used to examine the association between the outcome vari- ables (suicidal ideation and attempted suicide) and perceived discrimination under consider- ation of potential confounders and covariates. The prevalence of the outcome was compared between the categories of each variable using Chi-square tests for categorical and t-tests for continuous variables. Variables presenting a p-value .20 in the bivariate analysis were included in the multiple logistic regressions. The final models were compiled using backward stepwise elimination with removing variables presenting a p-value > .10 from the saturated model. Following this approach, three multiple logistic regression models were computed in primary analysis. The first model explored the association between suicidal ideation and per- ceived discrimination among all participants. The second model explored the association between suicidal ideation and the extent of perceived discrimination among the participants who had perceived discrimination on the basis of sexual orientation. The third model explored the association between attempted suicide and perceived discrimination among all partici- pants. In secondary analysis, the same multiple logistic regression models were computed sepa- rately for MSM and TG, resulting in six additional models. All final models contained no variables with missing data and the variance inflation factors (VIF) indicated no problematic multicollinearity (all VIF-values 1.29). The final models were further tested for interactions. None of the interaction terms were found to be significant, thus they were excluded from the final models. All statistical tests were based on a significance level of a p-value .05 and confi- dence intervals (CI) were set at 95% confidence level.
This is to our knowledge the first systematic literature re- view of methods used to sample most-at-risk populations of FSW and MSM. We identified 268 published articles from 53 countries or regions and 11 recruitment methods. Over 427,000 participants were surveyed in these 268 stud- ies. Sampling methods we classified as semi-probabilistic (internet, TLS and RDS) were used in 59 % of the retrieved studies. These results are consistent with prior studies find- ing that web-based surveys, TLS and RDS methods have been used more extensively in health research in the past years [28–30]. The increase in the use of semi-probabilistic methods might be associated with the 2005 proposal of United Nations General Assembly Special Session 2 (UNGASS) that proposed a new set of indicators according to the level of epidemics in countries: generalized or Table 3 Study populations by categories
Overall, HIV rates were substantial, and risks for HIV infection from sexwith both were menand women were common. The participants were generally young, though there was a significant association between HIV and age. Excluding the few men above the age of 49, overall more than one-third (35.7%, 95%CI 26.3– 46.4) of MSM between the ages of 30–49 were HIV infected. These data suggest that this is not a new epidemic of HIV among African MSM which is spreading more rapidly among younger MSM, as has been seen observed among MSM in other settings such as Russia . Because younger men were much less likely to be HIV infected, prevention programs targeting younger MSM in these populations could have marked potential for avoiding future infections. All possible combinations of biomedical and behav- ioural interventions need to be evaluated including those directed at MSM who are already HIV seropositive. While very little is known about the benefit of targeted HIV prevention programming among MSM inAfrica, in other contexts these approaches are known to be very effective in decreasing unprotected anal intercourse (UAI) [18,19]. Prevention research and optimization of existing prevention tools for MSM are a clear public health priority for Southern Africa.
All interviews were transcribed and translated into English when necessary. An analysis team consisting of the four social scientists who interviewed participants and an additional senior social scientist coded each transcript. Guidelines that included three distinct stages (sampling and design, theme development, and theme validation and code use) were used for thematic coding as the primary analytic strategy with an emphasis on descriptive thematic coding. After reading two transcripts, the analysis team members collaboratively developed a codebook of themes based on the interview topics as well as those emerging from the data. Two more transcripts were then reviewed to include additional topic areas and themes. This process was repeated until a sample of 12 transcripts had been reviewed and the codebook reached a stage where no new themes or topic areas emerged. To ensure inter-rater consistency, the analysis team compared their individual coding of the same transcripts. All transcripts were then coded using the final version of the codebook and merged using NVivo software (version 2.0, QSR International Pty. Ltd, Victoria, Australia) before themes were summarized across respondents. After coding, the merged project was transferred to NVivo version 8 for analysis. Analysis focused on identifying the dominant and the range of explanations for sexual identity, sexual behavior, motivations for preventive behaviors and comparisons across clients. Multiple (monthly during analysis) interactive discussions were held with the analysis team and senior researchers to validate data interpretations and resolve any interpretation discrepancies.
ranging questionnaire was used. This involved a series of questions about sexual practices and other risk behaviors related to HIV, such as use of injecting drugs. However, the 1998 sample was not representative of the Brazilian young male population, as only areas with excessive AIDS incidence were selected for study. In that year, blood sam- ples were also collected for HIV testing. The results were compared with a control group composed of randomly se- lected conscripts in Rio de Janeiro and São Paulo. 12
Independent variables included: age (based on self-reported year of birth), education level (secondary school or lower, high school/post-secondary education/vocational school or college, or university degree/higher), migrant status (based on country of birth and country of residence: native, emigrant, immigrant or visitor), ‘outness’ (the extent to which participants reported being open about their sexual attraction towards menwith others: being out to ‘less than half’ or ‘out to the majority’), overall per- ceived attitude towards gay or bisexual people at work/school and amongst parents/friends/acquaintances (positive, neutral or negative attitude), HIV testing in the last 12 months and results known (no or yes), knowledge of own HIV status (using both self-reported status and status based on laboratory results: newly diagnosed, negative test result, already known), sex role at last anal sex (insertive, receptive, versatile), number of substances (type speci ﬁed in the questionnaire) used at last anal sex (0, 1–2, 2+), frequency of visits to gay venues during last 3 months where sex-on-premises is possible (0 ‘no’, 1–3 ‘low’ 3+ ‘high’), currently having sexwith women (no or yes), serostatus com- munication at last anal intercourse (successful, unsuccessful; this constructed variable distinguishes between successful serostatus disclosure [i.e. a communication that establishes HIV serostatus concordance or discordance, including unilateral HIV infection disclosure], and unsuccessful serostatus disclosure [i.e. a commu- nication where either none or only one of the involved partners disclosed his serostatus, with the exception of unilateral HIV infection disclosure]), see. 23
The 92-item online survey took approximately 45 minutes and included a collection of items from previous research on sexual risk behaviors, Internet use, as well as a number of novel items related to use of social networking technologies. Items focused on demographics (age, gender (to ensure that all participants reported being men), race/ethnicity, income, and education); Internet and social media usage; and sexual health behaviors (see Participant Questionnaire). Prior to responding to items related to social media use, participants were given a definition of social networking sites along with a list of examples sites (such as Facebook and Myspace). Internet and social media usage items focused on the amount of time spent using the Internet and social media (in hours per day, days per week); reasons for using these technologies (e.g., news, dating, finding sex partners); and comfort when using these technologies to talk about sexual risk behaviors. For example, after a description of the difference between general Internet sites and social networking sites, participants were asked about their use of social networking sites for seeking sex, ‘‘In the past 3 months, how many sexual partners have you met on the Internet/social networking sites?’’ They were not asked to differentiate between general social networking websites and websites that were designed specifically for dating or seeking sex. Sexual risk behavior-related items focused on number and gender of sexual partners (from online and offline sources), sexual behaviors, and number of times exchanging sex for food, drugs, or a place to stay. For example, participants were asked, ‘‘In the past 3 months, have you exchanged sex for food, drugs, or a place to stay?’’
Specimens were self-collected vaginal swabs, anorectal swabs and urine. Trained STI nurses asked patients to take a self-collected anorectal swab and provided them with verbal instruc- tions and a diagram (i.e. insert the swab 2.5 cm into the rectum, rotate for 5–10 seconds, and place the swab in the capped tube). Specimens were processed at the department of Medical Microbiology at Maastricht University Medical Center + (Maastricht, The Netherlands) using nucleic acid amplification tests (NAAT) [polymerase chain reaction PCR; Cobas Amplicor until 2012 and afterward Cobas 4800, both from Roche Diagnostics, San Francisco, CA]. Serum was tested for Treponema pallidum hemagglutination antigen (TPHA) and human immunodeficiency virus (HIV) (anti-HIV[1/2], Axsym; Abbott Laboratories, Chicago, IL). Reactive samples were confirmed using Western blot (HIVblot 2.2; Genelabs Diagnostics, Sci- ence Park, Singapore), according to the manufacturer’s protocol. In addition to testing, trained study nurses took a standardised medical and sexual history at each consult, which included demographic data, self-reported symptoms, sexual behaviour in the preceding six months and antibiotic use in the past month. Anal symptoms (proctitis) included rectal discharge, bleeding, pain, redness, burning sensation or itching. All data were registered in an electronic patient registry.
The substitution rates of the different clusters were estimated using the BEAST software and implementing an MCMC method . The GTR+I+Г4 nucleotide substitution model and coa- lescent Bayesian skyline model were incorporated in the MCMC method . A relaxed molecular clock model with uncorrelated lognormal distribution was used to infer the time- scaled maximum clade credibility phylogenies . Multiple independent MCMC runs were performed and assessed for consistency. The MCMC analyses were combined to give a total chain length of 0.5-4x10 7 steps with sampling every 5,000 steps. The first 10% of the states of each chain were discarded as burn-in. Ten thousand trees were then sampled to estimate the evolutionary rate using LogCombiner v1.8.0. Convergence of relevant parameters was assessed by effective sample sizes over 200 in Tracer v1.5 (http://tree.bio.ed.ac.uk/software/tracer/).
GMSM who were recruited through a single-website survey differed considerably from the population estimates. This sample contained a higher proportion ofmen aged over 45, was less ethnically diverse, and included a greater proportion ofmenwith high-school only education. These men were less likely to have ever undergone HIV testing, and were more likely to be unaware of their HIV status. This method produced a sample ofmenwho appeared to be less connected to the gay community, andwho were more likely to report no sexual contact with either a causal or regular partner in the past six months. In comparison, the online sample recruited through multiple websites was more consistent with population estimates. Overall, few socio-demographic differ- ences were noted, although this sample contained more variation in sexual identification. These men tended towards spending less time with gay friends than the overall population, and had fewer gay friends. These findings are consistent with the profiles described in two previous Australian studies the Private Lives-2  and e-Male [22,23], which both reported similar patterns of socio-demographic characteristics, sexual identification and HIV testing history. Few behavioural differences were noted in the PASH sample compared to population estimates, although this sample underestimated the proportion of UAIR as well as the proportion ofmenwho did not engage in any anal intercourse with a casual partner.
This study describes the prevalence of HIV-1 subtype B and F1 dual infectionsin recently infected Brazilian MSM. Among the 41 subjects studied, 12.2% were posi- tive for both subtypes B (from previous study) and F1 proviral DNA (current study). These results are not sur- prising because both viral subtypes and recombinants are widely circulating in Brazil, which is a country that offers an excellent setting for such studies. It is probable that our results have underestimated the true rate of dual infection in this group. The most likely explanation for underestimation is that some isolates could have been undetected by our specific PCR screening method because of a mismatch at the primer binding sites, low proviral load, or that the subclade F1 isolates were main- tained in another reservoir other than the CD4-positive compartment that was sampled in the peripheral blood. Additionally, our method only detects dual infection of subtypes B and F1 when the pol-IN region is subclade F1. We could have missed some instances of co- infection if recombination had happened, and the pol-IN
To estimate the divergence times of the respective subtype B and CRF01_AE transmission clusters, the Bayesian coalescent- based relaxed molecular clock model was performed in BEAST 1.7 . The uncorrelated lognormal model  nested in general time-reversible (GTR) nucleotide substitution model  with a proportion of invariant sites and four rate categories of gamma- distribution model ofamong site rate heterogeneity  was employed to estimate viral phylogenies, nucleotide substitution rates and to date the time of the most recent common ancestor (tMRCAs) for the respective subtype B and CRF01_AE transmis- sion clusters. As for the coalescent priors, different parametric demographic models namely, constant population size, exponen- nested in general time-reversible (GTR) nucleotide substitution model and a proportion of invariant sites. The Markov chain Monte Carlo (MCMC) analysis was computed for 50 million states sampled every 10,000 states and output was assessed for convergence by means of effective sampling size (ESS) after a 10% burn-in. Transmission clusters are defined based on strong statistical supports generated at the internal nodes of the maximum likelihood and MCC tree reconstructions (bootstrap values of more than 90% and posterior probability of 1, respectively). A total of 12 monophyletic transmission clusters of different sizes (2–7 sequences) were determined, of which 6 clusters (B.1– B.6) were found within the subtype B lineages (blue branches) and another 6 clusters (AE.1– AE.6) were identified within the CRF01_AE lineages (red branches). The mean tMRCA and 95% highest posterior distribution (HPD) for each cluster are indicated in parentheses. In addition, 25 single unique lineages (green branches) involving subtype B and CRF01_AE are indicated in the phylogenetic analysis. The scale bar indicates the time in years and the alphabet at the tip of each branch represents the ethnicity of the subject, namely Malay (M), Chinese (C), Indian (I) and others (O). (B) Bayesian skyline plots (BSPs) generated from 35 HIV-1 subtype B and 35 CRF01_AE heterochronously sampled pro-rt gene from the MSM population in Kuala Lumpur. The origin and changes in effective population size through time for subtype B and CRF01_AE in the country were estimated. The 95% HPD of the effective population size is indicated in dashed lines. (C) Relative posterior probability distribution of the tMRCAs for the respective subtype B and CRF01_AE transmission clusters.
Therefore, it can be concluded from the data used – and its limitations – that the vulnerability of MSM to AIDS remains at high levels. The causes of this vulnerability, which may include the inadequate prevention of HIV infection in the past, should be a motive for future studies. The late diagnosis of the infection or disease, due to denial or lack of adequate health orientation, is another possibility. In addition, as this epidemic is oldest among the MSM population, some members of this population are probably in treatment for longer periods of time and as such, closer to exhausting the repertoire of existent therapies. In accordance with the results of the present study, some subpopulations of MSM, like exclusive homosexuals and bisexu- als, present diverse RR. The cause of this difference remains to be determined, as does the tendency for the increase or lower drop of RR among bisexuals in relation to exclusive homosexuals. The subpopulations of transvestites and commercial sex workers are other segments of the MSM population in which the RR may have a differentiated behavior.
Given that HIV is still incurable, non-availability of a safe and effective vaccine, limited access to anti-retroviral treatment (ART) and declining international fund, till date, preventive approaches through behavioral interventions (BIs) and voluntary counseling and testing (VCT) remain the cornerstone of HIV control [1–4]. Researchers have argued that a considerable reduction in HIV transmission could be successfully achieved through a combined effect of wide spread and sustained behavioral changes by appropriate counseling of a good number of potentially high risk individuals [4, 5]. Analysisof data from three countries around the world having different cultures and diverse HIV epidemics: Uganda, Senegal and Thailand, clearly indicated that behavioral changes were successful in some containment of HIV epidemic [6–10]. Likewise, VCT has been also effective in motivating people to change their risky behaviors and previous meta-analyses demonstrated that VCT recipients were less likely to engage in unsafe sex than VCT non-recipients [11, 12]. While the UNAIDS report 2012  revealed an increasing trend towards uptake of HIV testing globally, it was estimated that only half of all people living with HIV (PLHIV) knew their HIV sero-status and probably HIV prevention programs were not adequately reaching the population groups at highest risk. Globally, population at highest risk like female sex workers (FSW), menwhohavesexwithmen (MSM) and injecting drug users (IDUs) continue to be disproportionately affected by HIV [1, 13, 14]. It was estimated that odds of having HIV infection among MSM on average was 13 times more compared to general population in any capital city of the world . Although prior research indicated that counseling and BIs were effective in reducing high risk sexual behavior among MSM [15–17], worldwide the median coverage of risk reduction programs among MSM was estimated to be 55%, only few countries reported 75% consistent condom use and knowledge about their own HIV sero- status was reported to be exceptionally low among MSM .
Recent studies have indicated that pre-existing humoral immunity to HAdV-5 is associated with reduced immunogenicity to vaccina- tion with recombinant Ad5 vaccines. In the Step Vaccine Study, a phase II test-of-concept study trial of the Merck HIV-1 gag/pol/nef vaccine, menwhohavesexwithmen (MSM) with previous exposure to HAdV-5 through natural infection appear to have had at least transiently increased risk of acquiring HIV after vaccination [15,20]. The mechanism(s) underlying this observation remain unknown, and may include a spurious association because of the small sample size of the trial, or true biological phenomena such as upregulation of HIV-susceptible target cells within mucosal tissues, and interactions between HAdV-5 immune complexes and dendritic cells resulting in enhanced T cell infection . Recent studies also have demon- strated cross-reactive T cell responses to many human adenoviruses  (and unpublished data). These observations draw attention to the potentially important role of natural adenovirus infectionsin influencing responses to recombinant Ad vector vaccines, and highlight the need to more clearly define the epidemiology and virology of natural human adenovirus infections.
Furthermore, in multivariate analysis, older age ( 40 years) was another risk factor of HIV infection in both samples. This finding is consistent with previous studies that showed older MSM to be more at risk for HIV infection than young MSM . Many older MSM were unaware of high-risk sexual behaviours, or they thought that their beha- viours were of no risk. A study showed that the knowledge level regarding the specific behaviours leading to the trans- mission of HIV among older adults was significantly lower than among younger adults . Without HIV/AIDS educa- tional and prevention programmes tailored to older audi- ences, it is likely that many older adults are unwilling to attend or accept HIV education, prevention and intervention activities. Therefore, older MSM perceive themselves to be at low risk for HIV infection, which leads them to have a greater chance of becoming infected with HIV.
Among the different methods that we applied, wisdom of the Crowds was relatively a new method. The estimates that provided by this method had the biggest range and maximum esti- mate of the MSM size. This also has been reported in the study of PSE of MSM in Ghana . The wide range could be due to misinterpretation of the question by some participants whom have reported their own personal network, rather than the overall size of MSM community in Tbilisi. Others might have reported a huge unbelievable number of MSM as their own desire to show that such behaviors (having sexual contact with another men) are not anymore uncom- mon. Tbilisi is a big capital city, and MSM might not know or contact with the whole commu- nity of MSM living in this big cosmopolitan area. This method may provides more accurate estimates when the member of the community of target population, e.g. MSM, are visible to each other, have gatherings and social events and connected as one solid community. It also provides more precise estimates with bigger sample size; which was not the case for our study. While in MSM population size estimation study in Nairobi, Kenya , WOC produced the lowest plausible estimates; In contrast, in our study, WOC method yielded the high estimates. This telling us that the direction of bias using WOC is not predictable. However, since the esti- mated number of WOC was in range with estimates from other method (some of the multipli- ers), we decided to include this in the overall population size estimation of MSM; as presented in the result, even if we would have excluded the WOC estimates from the combined overall estimates, the overall size of MSM decreased, but not that much, an ensuring finding that the median is a robust estimator.