Introduction

India has a largely heterosexual HIV epidemic, which is concentrated among various core populations at high risk of HIV [1], [2]. However, Men having Sex with Men (MSM), a recognised core group, are also a powerful driving force [1]–[4]. In 2011, the average national HIV prevalence among MSM was 4.43% and it was greater than 5% in nine Indian states [5]. In contrast however, the estimated adult HIV prevalence among males in general population, who are expected to contribute to the heterosexual epidemic, was much lower (0.32%) [6].

India's National AIDS Control programme (NACP) follows a strategy of targeted programmatic prevention interventions to control the epidemic. MSM are a core population subgroup for these interventions as well as HIV surveillance [1]–[4], [7]. Indian MSM include, self-identified gay men (western acculturated) as well as kothi (receptive male partner), panthi or danga (penetrative male partner), double-deckers (men with penetrative and receptive behaviours), and alis or hijras (transgendered people) [3], [8]. HIV transmission risks among MSM vary with the type of sexual behaviour with men [3].

India's NACP recognises MSM as, ‘all men who have sex with other men as a matter of preference or practice regardless of their sexual identity or orientation and irrespective of whether they have sex with women or not’ [3]. However, the ‘Targeted Intervention (TI) program’ which is a key thrust of NACP [4] focuses on ‘high risk MSM’ – i.e. primarily those who are self-identified, practice receptive anal sex and have multiple sexual partners [3].

It is accepted that the prevailing social stigma and legal environment can lead to ‘covert’ behaviour among MSM [9], [10]. It is recognised that many MSM in India do have sex with women (MSMW) [9]. Even though same sex activity is not uncommon in India [11], traditional culture creates pressure to marry. Thus, many MSM irrespective of the way they identify themselves in terms of sexual behaviour or gender are likely to be married [12]. Data from the early years of the Indian epidemic showed that among male STI clinic attendees in Pune (1993–2002), 6.6% reported having sex with men of whom 35.7% were married [13]. Studies from different parts of India with varying sampling techniques have reported that between 11–60% of MSM were married [9], [10], [13]–[17]. Additionally, many authors hypothesize that since ‘married MSM’ are perceived as heterosexuals, they may be isolated from other MSM and therefore less likely to receive prevention intervention in community based settings [14]. Thus they could have more risky sexual encounters with other men, as shown in a study from Chennai [14], [15], [18]. It is also important to note that MSMW have been reported to be less likely to disclose their bisexuality to partners of both sexes.

Although the predominant female partner was the wife (76%), MSM have also reported sex with women outside of marriage including commercial sex workers (13%) [9]. The nationally representative Behavioural Surveillance Survey (2006), revealed a high proportion of MSM reporting sex with a woman in the last six months (30.4%–54.8%) [17]. In the population based Integrated Biological and Behavioural Assessment (IBBA-2006) among MSM in four Indian states, a wide range of MSM reported regular female partners (3.9%–42.7%). The monthly average of 2.1–3.5 female sexual contacts reported by these MSM gives a snapshot of the overlap between heterosexual and same-sex behaviour [16]. A very high HIV prevalence (15.9%) was seen among those who self identified as ‘bisexuals’ in the four state IBBA survey [16].

In a study in Mumbai, married MSM and those with partners of both genders were more likely to be HIV infected [14]. Similarly in south India, the ‘proportion married’ was higher among HIV infected MSM [19].

High rates of STI and HIV and lower rates of condom use, are other factors that add to the potential of men who have sex with men and women in bridging the high HIV prevalent MSM epidemic and the lower-level heterosexual epidemic in general population [9], [10], [13], [16], [17], [20]–[22].

Thus, MSMW in India, with their high HIV prevalence and low condom use with female partners are becoming increasingly recognised as an important bridge population. Their impact on heterosexual HIV transmission dynamics and risk to women who may be married and monogamous needs to be taken into consideration while planning interventions among MSM.

Prevention intervention methods among MSM at ‘Targeted Intervention (TI)’ delivery sites primarily focus on behaviour change communication for safer sex with male partners [4]. Currently interventions focus on ‘high risk MSM’. It is challenging to identify MSMW at these places too and ‘bisexual behaviour’ is difficult to address in such a scenario.

In India, NACP has been conducting frequent and periodic (every one or two years) unlinked anonymous surveillance called HIV Sentinel Surveillance (HSS) since 1998. HSS is designed to measure the level and trends of HIV prevalence in different sub-populations of interest (e.g. MSM, IDU, FSW, ANC attendees) [5], [23]. The most recent HSS among MSM in 2011, collected limited behavioural data along with HIV testing after informed consent, as part of second generation surveillance [5]. This was conducted through organizations providing targeted interventions.

We analyzed data of HSS among MSM in the four Indian states of Goa, Maharashtra (MH), Gujarat, Madhya Pradesh (MP). The specific objectives of our analysis were to identify the prevalence and predictors of bisexual behaviour among MSM attending TI sites and to understand the risk factors for HIV infection among them.

Material and Methods

Results

In all 4682 MSM were tested for HIV under sentinel surveillance conducted at 19 sites in 4 states of west-central region of India (Goa, Gujarat, Maharashtra and Madhya Pradesh). All study sites either achieved 100% (n = 12) or over 90% (n = 7) of the expected sample of 250 and were included in analysis.

The mean age of MSM was 28 years (SD = 6.4) which did not substantially differ across states. While 23% had studied beyond matriculation, 5% were illiterate. MSM were predominantly urban residents with about 13% from rural regions. Table 1 shows the state wise differences in characteristics of MSM.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Demographic and socio-economic characteristics of MSM in states in west-central India (Year 2011).

https://doi.org/10.1371/journal.pone.0107439.t001

About 26% of respondents overall reported being in government or private service while the proportion of MSM who reported working as hotel staff and truck drivers/helpers was 4% and 2% respectively. A considerable proportion of MSM in Madhya Pradesh (29%) reported being students or unemployed which was higher than in other states (student: 2–8%, unemployed: 1–9%). Median duration of stay at current place was reported as 23 years with IQR 15–28. About 483 MSM (10.3%) respondents from two study sites had been sampled consecutively. Among these approximately 10% came to service point to seek medical care including STD treatment while 29% had come to collect condoms and 60% for other reasons which included mostly meeting/event/official work or casually with friend.

Behaviourally 51% reported themselves as receptive only, 21% as penetrative only and 28% as both. The proportion of MSM practising only ‘receptive’ anal sex differed somewhat across other states, the lowest was in Goa (34%). The proportion of MSM practising only penetrative behaviour, was lowest in MP (10%) and highest in MH (27%). Nearly half the respondents had exchanged money for sex and 36% reported only receiving money for sex. Overall 6.8% were HIV infected. MSM who had ever exchanged money for sex had a significantly higher HIV prevalence of 6.6% as compared to 5.9% in those who had never done so (p = 0.02); while prevalence among those who ever sold sex (only sold or bought and sold) was 6.3% and those who ever bought sex (only bought or bought and sold) was 5.1%.

Forty four percent of MSM reported having sex with women in last 6 months and were classified as MSMW. Prevalence of HIV among MSMW was 5.3%. Anal sex positioning with male partners among MSMW and MSMO were similar. Maximum MSMW reported receptive only (35.6%), while only penetrators formed 30.9% and both types were 33.5%.

On multivariate analysis, MSMW were found to be independently associated with being older; 26–30 years [AOR = 3.1, 95% CI (2.7, 3.7)], >30 years [AOR = 6.5, 95% CI (5.5, 7.7)]. MSM who reported penetrative behaviour alone with other men were nearly six times more likely to be MSMW [AOR = 5.8, 95% CI (4.8, 7.0), p<0.01]; while MSM reporting both penetrative and receptive behaviour were nearly thrice as likely to have sex with women [AOR = 2.7, 95% CI (2.3, 3.1), p<0.01]. Those who both ‘paid’ and ‘received’ money for sex [AOR = 0.49, 95% CI (0.38, 0.62)] were significantly less likely to be bisexual (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Predictors of bisexual behaviour among MSM in west-central India (Year 2011).

https://doi.org/10.1371/journal.pone.0107439.t002

Multivariate analysis showed that paying for sex [paid money: AOR = 1.5, 95% CI (1.03, 2.2)]; receptive MSM behaviour [AOR = 1.8, 95% CI (1.2, 2.7)] and older age; 26–30 years, [AOR = 1.5, 95% CI (1.1, 2.0)], >30 years [AOR = 1.9, 95% CI (1.4, 2.6)] showed independent association with HIV infection (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Predictors of HIV infection among MSM in west-central India (Year 2011).

https://doi.org/10.1371/journal.pone.0107439.t003

Discussion

In our analysis of HSS data from 4682 MSM sampled in western and central India, we highlight a high proportion of bisexual behaviour and identify the key determinants for the same. Our findings also emphasize the importance of transactional sex and HIV prevalence in this population. India's HIV program utilizes programmatic data for improvement of services and prevention efforts. Our findings, can guide prevention programming for MSM in India's recently initiated NACP Phase IV [4], [26].

Close to a half (44%) of the MSM attending TI sites have sex with women as well. The HIV prevalence in this population (5.3%) is higher than the overall national prevalence for MSM and is of concern due to transmission risks to men as well as women partners. Of importance, is our finding that, MSM who practice penetrative behaviour with other men, are nearly six times more likely to have sex with women, as compared to Kothi (AOR 5.8). Practicing both penetrative and receptive behaviour with men, also emerged as being significantly predictive of bisexual behaviour (AOR 2.7). That our data highlight penetrative behaviour with other men, as a very significant predictor for bisexual behaviour is probably not entirely unexpected. However, of note is the finding that this remains significant even when men also practice receptive behaviour with other men.

The study also highlights that increasing age was significantly associated with having sex with women. MSM older than 30 years, were six and half times more likely and MSM between 25–30 years of age nearly thrice as likely to be behaviourally bisexual, as compared to MSM below 25 years. A recent analysis from south India which reported that ‘typically more insertive MSM are older’ has provided context to other anecdotal reports that sexual behaviours do change over time and with increasing age [27]. Our data from other parts of India (western and central) too support these reports. This therefore raises further questions about determinants and origins of ‘change in behaviour to being bisexual’ among older MSM. We suggest that these questions should be a focus for future qualitative research in the Indian context.

In contrast to the majority of Indian studies which focused on ‘Married MSM’ [7], [13], [14], our study focuses on behaviourally bisexual men (44% MSMW). The proportions are similar to the south-Indian study (2003–04), which showed that behaviourally bisexual MSM were more than those currently married (41%); highlighting that MSM need not be married to have sex with other women [9], [16], [20].

There is a slowly increasing recognition of the importance of this subset of the MSM population for India's HIV epidemic. However, the main challenge has been ‘how to identify behaviourally bisexual men in the community for prevention interventions?’ [15], [19]. Our study highlights that, the high proportion of MSMW already attending targeted intervention programs are a prevention opportunity, which should not be missed. Findings on determinants of bisexual behaviour obtained from this study will help as a starting point to identify MSM with female partners. In this context, we caution that this needs to be accompanied by development of focused and appropriate behaviour change communication messaging, addressing the bridging potential of MSMW. Current guidelines for ‘Peer Educators’, which do not include specific counseling messages about ‘risk to female partners’, ‘safer anal/vaginal sex with females’ and ‘disclosure to and communication with females’ may thus require enhancement to address this prevention need [28].

Our study also found a higher HIV prevalence in MSM population in all four states in the west-central region than the national average [5]. This is higher than the prevalence in the core group for heterosexual epidemic i.e. FSW. We would also like to highlight that MSMW were highly represented in the states of Maharashtra (43%) and Gujarat (54%). In these two states, the HIV prevalence among women attending antenatal clinics is higher than national average and these states would be useful models to study the intersection of the MSM and heterosexual epidemics in India. Additionally in this study we report data about MSMW from low HIV prevalence state of Madhya Pradesh (25%) for the first time which will be useful for modeling and estimation purposes.

This high prevalence of both HIV and bisexual behaviour thus adds further emphasis to prevailing data [21] that this core risk group also functions as a bridging population. Addressing this bridging potential is of immense importance if the current successes in India's prevention programming are to be maintained.

In this study, individuals identifying themselves as hijra or transgender who are traditionally likely to undertake sex work were excluded. In spite of this, transactional sex was very common with nearly half the respondents ever having exchanged money for sex, while about a third (36%) had ever received money for sex. This group showed much higher prevalence and risk of HIV infection. We also found that MSM who both ‘paid’ and ‘received’ money for sex were less likely to be bisexual. A possible explanation for this could be that these men may engage primarily in transactional sex and therefore are most likely to preferentially practice receptive sex.

These data suggest that strategies specifically addressing the socio-economic contextual factors surrounding transactional sex among MSM should be a key focal point in NACP IV. Formative research to tease out the underlying constructs of transactional sex for development of interventions would be productive.

Our analysis of sentinel surveillance data does have some limitations. The surveillance occurred at TI service points, which included active MSM visible in the hot spots. Since they were recruited at a time when the programmatic focus was on ‘higher-risk’ MSM, they may not be entirely representative of the whole population. Respondents had already been coming to the TI sites and they self-identified themselves by behavioural typology (kothi, panthi or double-decker) which were mutually exclusive. This however may not be representative of the entire MSM population in their regions especially the hidden population or those who are not self-identified MSM. To reach hidden MSM, sampling strategies like respondent driven sampling can be implemented in the future and may even be considered for enhancing recruitment and registration at TI sites [29]–[31].

Although there was some difference in the sampling strategy between states, majority were sampled by random method. Sensitivity analysis by sampling methods did not show any remarkable difference in the results. This strengthens our assumption that sampling method may not have introduced significant bias for the purpose of this analysis for predictors of bisexual behaviour. Similarly analysis by individual state or by eliminating one state at a time from the model has proved the robustness of results.

Conclusions

In light of the 17.7 million sexually active males who are estimated to be same-sex oriented and engaging in same sex behaviour in India [26], our findings highlight the importance of focusing interventions in a subpopulation that is more likely to act as bridge to the low or no-risk women in the general population.

Our study shows, that a substantial proportion of ‘high risk’ MSM who receive services from TI sites have sex with men as well as women and the easy opportunity for intervention in this group should be capitalized upon. We emphasize, that these findings will be useful to the program which is largely TI based. Predictors of bisexuality identified in this analysis will help to devise prevention interventions focused on MSMW within the existing targeted intervention programs. We highlight that an additional approach could be to ensure that these prevention messages addressing protection of partners of any gender as an intervention are also targeted towards self- identified MSM who are older as well as show penetrative behaviour with other men especially those already attending TI sites.

Acknowledgments

The authors gratefully acknowledge the contributions of Chitra Kadu, Rajesh Yadav, Mycal Pereira, Rajendra Yelagate, Varsha Kale, Sangita Kulkarni, Kavita Vishwakarma, Samadhan Kumbhar from NARI, the monitors and implementing team members from the states for their inputs in the surveillance. We also thank the respondents for consenting to be part of this national sentinel surveillance and gratefully acknowledge their contribution.