Background

HIV in men who have sex with men (MSM) influences the overall HIV epidemic in both low and high income countries, which is related to high per-act and per-partner transmission probability of HIV transmission in receptive anal sex [14]. According to the sentinel surveillance of the National AIDS Control Organization of India, HIV prevalence in MSM in 2008–2009 was estimated to be 7%, which is about 20 times higher than the overall national adult HIV prevalence rate [5].

Various MSM programmes were started in India in the last decade to reduce HIV including in Andhra Pradesh which is one of the high HIV prevalence states in India [6, 7]. Avahan, the India AIDS Initiative, funded by the Bill and Melinda Gates Foundation was one among these [8]. The Avahan MSM programmes in Andhra Pradesh started between October 2005 and April 2006 [9]. Avahan intervention was conceived as a focused prevention program, offering a standardized package of proven prevention interventions for high-risk groups, such as community participation, clinical services for STI and counselling, peer lead outreach and commodity distribution [8]. Two rounds of integrated biological and behavioural assessment (IBBA) surveys were conducted among MSM in Andhra Pradesh in April-June 2006 and April-June 2009 to assess the effect of intervention programme [9]. The first IBBA round was within six months of start of the Avahan MSM programmes, and therefore can be considered as baseline for this intervention.

Previous studies have suggested that many MSM in India have a mix of paid, occasional and regular male and female partners [1015]. Consistent condom use by MSM in Andhra Pradesh has been previously reported to be low with both male and female partners [11, 14, 16], though there has been a suggestion of increase in condom use with paid partners [17]. In order to understand how the MSM intervention programmes can be improved further, this paper reports the changes in consistent condom use by MSM with various types of partners from 2006 to 2009 in Andhra Pradesh from the two IBBA rounds, and the associations between various factors and inconsistent condom use for sex with regular partners as well as HIV in MSM.

Methods

Multiple rounds of IBBA surveys were designed to evaluate the impact of Avahan programme on populations at high risk as well as provide data on changes in risk behaviours and prevalence of STI including HIV. Both rounds of IBBA covered different categories of high risk population, including MSM, female sex workers and their clients. The operational definition used for MSM in both surveys was “any male or hijra (eunuch) aged 18 years or older who had any type of sex with another male in the last one month” [9, 18].

Survey procedures

Two independent cross-sectional IBBA surveys were implemented in Andhra Pradesh as part of an evaluation of the Avahan initiative among high risk group [9, 18]. These surveys were carried out among MSM in four districts in Andhra Pradesh, namely East Godavari, Guntur, Hyderabad and Visakhapatnam. The IBBA survey in Andhra Pradesh was led by the National Institute of Nutrition, and the field work was done by a hired research agency AC Nielsen ORG MARG Private Ltd for both rounds under the guidance of the National Institute of Nutrition. The National AIDS Research Institute coordinated the conduct of the IBBA survey at the national level and FHI 360 provided technical assistance for implementing the IBBA. The details of study design are available elsewhere [9, 18].

Relevant for this report, training was given to the survey team about survey protocol, questionnaire administration, sample collection and transport of biological samples. Written informed consent was obtained from all participants. A sample size of 400 self-identified MSM per district was interviewed face-to-face and biological specimen collected with a two-stage cluster sampling design. Fixed-location and time-location clusters were the sampling strategy in East Godavari district. In the other three districts, only time-location clusters were considered and recruitment of high-risk MSM for the survey was predominantly from public places. Data were collected from a total of 1,621 and 1,608 sampled self-reported MSM from Andhra Pradesh in round 1 and round 2 of IBBA surveys, respectively. Both rounds of IBBAs collected detailed behavioural information including socio-demographic characteristics, sexual history, practices, type of sex partners, condom use, exposure to Avahan or other HIV/AIDS prevention initiatives, and biological specimens to test for HIV and STIs. Standardized laboratory methods were used for HIV testing. Double-data entry of district-level datasets was conducted using CSPro software (U.S. Census Bureau, Washington DC) for both rounds of IBBA. Centralized data management was done for IBBA round one by the National Institute for Epidemiology, Chennai and decentralized data management was done in round two in each state.

The protocol for the IBBA surveys was approved by the Health Ministry Screening Committee and the Indian Council of Medical Research, and ethics approval was given by the FHI Protection of Human Subjects Committee [19].

Statistical analysis

We used data on condom use by MSM with different partners in each round of IBBA survey to measure the change in consistent condom use between 2006 and 2009. Data are presented for regular partners and partners who were paid for sex. Data are not presented for partners who paid for sex because the variable used for this in the two IBBA rounds was different which did not allow direct comparison. We defined consistent condom use if condom was used every time during sex over the past one month with different sex partners. The choice of past one month was based on the availability of this variable in both IBBA rounds for all types of partners that we considered, whereas condom use data over longer duration was not available in the first round of IBBA for partners who were paid. Appropriate sample weights were used for estimating consistent condom use and HIV prevalence among MSM; the 95% confidence intervals were calculated taking into account the design effect of the cluster sampling strategy [20].

A detailed analysis of the variables associated with inconsistent condom use and with HIV was performed for the second round of IBBA data using multivariate analysis. We used multiple logistic regression models to assess the association of socio-demographic variables and behavioural variables with inconsistent condom use by MSM with regular male and female sex partners. We included these variables in the models: age, education, marital status, ever received cash or kind for sex, paid to have anal intercourse with a male or hijra ever, circumcision, duration of sex with male, received condoms from the peers or outreach workers of the NGO/programme, and membership in community-based organization. We used a similar approach for assessing the associations with HIV in MSM. In addition to the previously mentioned variables, we also assessed the association of inconsistent condom use with regular male partner with HIV.

A total of 13 respondents, 5 in IBBA round 1 and 8 in round 2, did not report any male sex partner, and were therefore not considered for the analysis of MSM in this paper. Analysis was done using SPSS 17.0 (IBM SPSS statistics standard, USA) and STATA 11.2 (StataCorp, USA) software.

Results

Sample characteristics

The distribution of the demographic and sex behaviour characteristics of MSM in the two rounds of IBBA surveys are presented in Table 1. The proportion of literate and never married MSM was slightly higher in the second IBBA round. As compared with the first round, in the second round a lower proportion of MSM reported paid male sex partners while a slightly higher proportion reported occasional male sex partners, and a substantially lower proportion reported a paid female sex partner. The proportion of MSM who reported membership in community based organization jumped from 2.9% in first IBBA round to 65.8% in the second round, as might be expected after the Avahan initiation.

Table 1 Characteristics of MSM in the two IBBA rounds

Consistent condom use with different sex partners

The proportion of MSM who reported consistent condom use for each sex episode over the past one month with paid male partners improved considerably from 19.5% (95% CI 14.9-24.1, design effect 2.1) in IBBA round 1 to 93.8% (95% CI 90.0–97.6, design effect 11.6) in IBBA round 2, and a similar trend was seen for occasional male partners as well (Figure 1). Even though there was a substantial increase in consistent condom use with regular male partners from 8.4% (95% CI 6.5-10.3, design effect 1.3) in round 1 to 75.8% (95% CI 69.7-81.9, design effect 4.9) in round 2, a quarter of the MSM were still not using condoms consistently with their regular male partners. A significant increase was also seen in consistent condom use by MSM with female paid partners from 25.9% (95% CI 20.7-31.1, design effect 2.1) in round 1 to 94.5% (95% CI 79.1-100, design effect 10.6) in round 2. However, only 15.7% (95% CI 7.8-23.6, design effect 8.7) MSM reported using condoms consistently over the past one month with their regular female partner in IBBA round 2. Among the 223 MSM who reported inconsistent condom use with their regular male partner, 81 (36.2%) also had a regular female partner of whom 78 (96.9%) did not use condoms consistently with her. On the other hand, among the 699 MSM who reported consistent condom use with their regular male partner, 272 (39%) also had a regular female partner of whom 211 (77.5%) did not use condoms consistently with her.

Figure 1
figure 1

Consistent condom use every time in the last one month by MSM with different sex partners in the two IBBA rounds. The truncated bars indicate 95% confidence intervals. Regular male sexual partner means lover/boyfriend. Occasional male partner means non-regular and non-paid male sex partner. Paid male partner means paid to have anal sex with male/hijra. Regular female sexual partner means spouse/lover/girlfriend. Paid female partnr means paid to have sex with female.

It is interesting to note that in the IBBA round 2, of the 702 MSM who used condom consistently with their regular male partner over the last one month, 681 (97%) also reported using condom consistently with their regular male partner over the last six months, indicating that these proportions were not very different between one and six months. Similarly, of the 111 MSM who used condom consistently with their regular female partner over the last one month, 100 (90.1%) also reported using condom consistently with their regular female partner over the last six months.

Associations with inconsistent condom use with regular male partner

Of the 1,600 MSM in IBBA round 2, 979 (61.2%) reported having had a regular male sex partner. Among them 926 (94.5%) ever had anal sex with him, of which 223 (24.2%) did not use condoms for over the past one month. Table 2 shows the variables associated with inconsistent use of condom by MSM with their regular male partners. MSM who received condoms from peers or outreach workers of Avahan NGO had lower odds of inconsistent condom use with their regular male partners (odds ratio 0.3; 95% CI 0.1-0.5). The following MSM were more likely to use condoms inconsistently with their regular male partners: those who never received cash or kind for sex (odds ratio 2.4; 95% CI 1.6-3.5), those with 6 or less years of duration of sex with males (odds ratio 2.2; 95% CI 1.3-3.6), and those aged 25–34 years (odds ratio 1.9; 95% CI 1.2-3.1).

Table 2 Association of socio-demographic and behavioural variables with inconsistent condom use by MSM over the past one month with their regular male and female sex partners, using multiple logistic regression

Associations with inconsistent condom use with regular female partner

Of the 1,600 MSM in IBBA round 2, 706 (44.1%) reported having had a regular female sex partner and ever had sex with her. Of these, 84% did not use condom for each sex episode over the past one month with their regular female partner. Table 2 shows the association of inconsistent condom use by MSM with their regular female partner over the past one month. The strongest variable associated with inconsistent condom use was MSM who were ever married with female (odds ratio 51.6; 95% CI 18.4-144.5). The other significant associations with inconsistent condom use were with never received cash or kind for sex (odds ratio 5.0; 95% CI 2.3-11.1), age 25–34 years (odds ratio 4.8; 95% CI 1.3-17.2), and membership in an MSM community-based organization (odds ratio 3.8; 95% CI 1.5-9.6). In contrast, a higher proportion of MSM who had membership in an MSM community-based organisation used condoms consistently with their non-regular male partners and regular male partner as compared with other MSM (85.6% versus 77% [p = 0.034] and 79.8% versus 63.3% [p < 0.001], respectively). Of the 1,040 MSM who reported membership in a community-based organization, 440 (42.3%) reported being ever married, whereas among the other 552 MSM only 144 (26.1%) reported being ever married.

Associations with HIV

The HIV sero-prevalence among MSM reported in the two IBBA rounds was not significantly different: IBBA round 1 (17.3%, 95% CI 14.1-20.5; design effect 3.1) and IBBA round 2 (19%, 95% CI 15.4-22.6; design effect 3.4). The district-wise HIV prevalence reported in IBBA round 2 ranged from 4.9% in Visakhapatnam to 28.9% in Hyderabad. Table 3 shows that the MSM who used condoms inconsistently with regular male sex partner had higher odds of HIV (odds ratio 1.8; 95% CI 1.2-2.7). The other significant associations for being HIV positive were more than 6 years duration of sex with male (odds ratio 2.6; 95% CI 1.8-3.9), received condoms from peers or outreach workers of the Non-Avahan NGO/programme (odds ratio 2.1; 95% CI 1.2-3.9), never married MSM (odds ratio 1.7; 95% CI 1.2-2.4), and illiterate MSM (odds ratio 1.4; 95% CI 1.0 -1.9).

Table 3 Association of socio-demographic and behavioural variables with HIV, using multiple logistic regression

Discussion

This analysis of data from the two rounds of IBBA in Andhra Pradesh reveals that consistent condom use reported by MSM over the past one month increased for all types of sex partners during the three year period from 2006, corresponding to the start of Avahan intervention, to 2009. While the consistent condom use rate by MSM with paid male and female partners reached 94% in 2009, this rate was lower with their regular male partners at 76% and dismally low with their regular female partners at 16%. The MSM who used condoms inconsistently with their regular male partner were also more likely to use condoms inconsistently with their regular female partner. We found a positive association of HIV among MSM with inconsistent condom use with regular male partners. This does not necessarily imply that MSM get HIV from their regular partners, rather that this association likely reflects broader risk exposure of MSM who use condoms inconsistently with regular male partners. While some of the inconsistent condom use by MSM with their regular male partners could be explained by ‘negotiated safety’ between them, generally the inconsistent condom use by MSM with their regular female and male partners needs attention due to their potential to transmit HIV.

Of the MSM in IBBA round 2, 37% were married. Married MSM had very high odds of inconsistent condom use with their regular female partner. Some other studies from India have also reported a very high proportion of unprotected vaginal sex by MSM with their wives [15, 17, 2123]. It is plausible that the high level of inconsistent condom use by MSM with their regular female partner, often their wife, is due to concealment of their sex with men from their female partner.

The MSM who received condoms from the Avahan peers or outreach workers had the least odds of inconsistent condom use with their regular male partner. While the much higher proportions of consistent condom use by MSM with their various kinds of partners in 2009 as compared with 2006 is suggestive of a beneficial effect of Avahan in Andhra Pradesh, this cannot be firmly concluded from the data analysed in this paper as there could also have been other contributors to this observed increase. However, two other analyses of the population level effect of Avahan across six states have revealed that Avahan had a beneficial effect in preventing HIV in several states of India including Andhra Pradesh [24, 25].

We found a positive association between inconsistent condom use with regular female partner and membership of MSM in community-based organisations. This contrasted with a more consistent condom use by these MSM with their regular and non-regular male partners. While the proportion of ever-married MSM among those with community-based organisation membership was higher than among those without this membership, both this membership and being married had significantly higher odds of inconsistent condom use with their regular female partner in multivariate analysis. This peculiar relation between membership and inconsistent condom use probably reflects some other underlying characteristics of the MSM sample in IBBA second round who had membership of community-based organisations.

The MSM who had never received cash or kind for sex had higher odds of inconsistent condom use with both male and female regular partners. This possibly reflects that MSM who are sex workers are more likely to be exposed to the MSM intervention programmes, indicating that the intervention programmes also need specific efforts to reach MSM who are not sex workers.

MSM who had had sex with males for 6 or less years were less likely to use condoms consistently with their regular male partner, indicating that it would be useful for MSM intervention programmes to try to reach out to early-stage MSMs. On the other hand, MSM who had had sex with males for more 6 years had higher odds of being HIV positive, which likely reflects the cumulative risk of HIV among them and also the possibility that in the past they may not have used condoms as consistently as they were using at the time of the survey.

A significant limitation of this study is that causality between the Avahan intervention and the increase in condom use by MSM observed after its roll-out cannot be ascertained in this pre- and post-intervention study design. In addition, self-reported data about condom use by MSM may be influence by social desirability bias, that is, MSM may have become more aware by the second IBBA round that condom use is expected and therefore they may have over-reported it. Also, there were some socio-demographic differences in the MSM samples of the two IBBA rounds, which may have influenced to some degree the differences observed in the two rounds. On the whole, however, the data presented in this paper indicate the possibility of a beneficial effect of Avahan, which is supported by two other analyses of HIV prevention at the population level that could be attributed to Avahan [24, 25].

Conclusions

Condom use by MSM increased markedly after implementation of Avahan intervention in Andhra Pradesh state in southern India, indicating a beneficial effect, though a causal association cannot be firmly assessed with the available data. However, the continuing very low rate of consistent condom use by MSM with their regular female partners, and to some degree a low rate of consistent condom use with their regular male partners, need particular attention in HIV prevention programmes for MSM in India.