Heterosexual intercourse represents the main mode of HIV transmission in most developing countries . However, in some Latin American and Asian countries [2,3], men who have sex with men (MSM) make up a disproportionately high number of reported AIDS cases. As of May 2000, MSM accounted for 42% of the total cumulative AIDS cases reported in the Andean Region (Peru, Colombia, Venezuela, Ecuador, and Bolivia) and had the highest prevalence rates . The true proportion of AIDS cases attributable to same sex contact among men is likely to be higher given the strong social stigma against homosexuality, as reflected by the substantial proportion of MSM in Latin cultures including Peru, who self-identify as ‘heterosexual’ [4–6]. In Lima, a representative sample of men undergoing routine pre-employment physical examination, found that 11% reported previous sex with men, and 87% of these men also reported sex with women . This current study and our study in the Dominican Republic  show that MSM comprise highly distinct subpopulations, based on their social strata, marital status, sexual identities, and communities of reference. Many of these populations reported distinctive sexual practices and sexual networks.
The declining male-to-female ratio of reported AIDS cases, together with the increasing number and visibility of women living with HIV infection in Latin America, have been characterized as ‘heterosexualization’ of the epidemic in the region. However, in Central America and the Andean Region, the epidemic still disproportionately affects traditional high-prevalence groups such as MSM, and sexual ‘bridging’ from bisexual men to women may account for a substantial proportion of ongoing heterosexual transmission, similar to injecting drug use contributing to transmission of HIV to heterosexuals in North America. A study of pregnant women presenting to a public maternity hospital in Lima found that most HIV-infected women were probably infected by their stable partners, and 27% of the HIV-seropositive male partners of those HIV-infected pregnant women reported sexual contact with other men, which was highly associated with HIV infection [odds ratio (OR), 6.9%; 95% confidence interval (CI), 1.4–35.5] .
The present study was undertaken to: (i) determine the prevalence of and risk factors associated with HIV infection and syphilis seroreactivity; (ii) use the less sensitive enzyme immunoassay (EIA) to estimate the incidence of HIV infection; (iii) characterize the behaviors of men who reported sex with both men and women in the past year (`bridgers'); and (iv) describe the diversity of self-reported sexual identity among MSM in Lima, Peru.
Study population and design
In 1996, men over 17 years of age who reported sexual contact with another man during the previous year were eligible to participate, regardless of prior HIV testing history or HIV serostatus. A team of 30 experienced outreach workers/peer educators, representing the different subgroups of the diverse MSM community in Lima, recruited potential participants, through street-based outreach in settings where MSM meet, snowball referrals (i.e., referral by an individual already participating in the study), referrals from community facilities that provide services for MSM, posters and flyers in MSM-oriented businesses, and announcements in the press.
This study was approved by the Human Subjects Review Committee of the University of Washington and by the National AIDS and STD Control Program of Peru. Participants gave written informed consent. Trained counselors used a standardized questionnaire administered through face-to-face interviews. Participants underwent serologic testing for syphilis and HIV and genital and rectal examination. Men with symptomatic bacterial sexually transmitted diseases (STD) were provided syndromic treatment at no charge.
Antibodies to Treponema pallidum were measured by the quantitative Venereal Disease Research Laboratory (VDRL) test with confirmatory testing provided by T. pallidum hemagglutination (TPHA). Syphilis seroreactivity was defined as VDRL titer ≥ 1:1 and reactive TPHA. Antibodies to HIV were detected by ELISA (Abbott Laboratories, Chicago, Illinois, USA) and repeatedly reactive sera were confirmed using Western blot (Cambridge Biotech, Worcester, Massachusetts, USA). The less sensitive EIA, utilizing methods previously described , was performed to detect persons with recent HIV infection among those who had positive HIV serologies. This less sensitive EIA has been validated only for HIV subtype B; 720 of 733 (98.2%) serum samples from HIV-infected persons from Peru, Colombia, Ecuador, Bolivia, Chile, and Paraguay were subtype B .
Bivariate comparisons of selected variables used t tests for continuous variables and chi-squared tests or Fisher's exact tests for categorical variables. Variables found to be significantly associated (P ≤ 0.05) or with an OR < 0.75 or > 1.5 were entered stepwise into a multivariate logistic regression model to identify independent predictors. The appropriate functional forms of the covariates in the logistic regression model were determined by generalized additive modeling . Influence diagnostics, such as drop-1 case, and robust methods were used to examine for outliers and influential observations . HIV incidence and CI were estimated using the formula described previously by Janssen et al. .
Of 451 men who agreed to participate, over 90% were recruited by one-on-one street-based outreach or ‘snowball’ referral. Median age of participants was 25 years (range, 18–56 years), and the majority (91%) were single and of low socioeconomic status (median income of $US100 per month). One-half reported fewer than 11 years of formal education.
Demographics, behavior characteristics, and HIV and syphilis seroreactivity according to self-perceived sexual identity
Of the 442 participants who responded to the question regarding sexual orientation, 6.3% self-identified as heterosexual, 22.9% as bisexual, 60% as homosexual and 10.9% as cross-dresser (Table 1). Interestingly, men from all four subgroups of sexual identity acknowledged ever having had sex with women, ranging from 12.5% of cross-dressers to 85.1% of bisexuals. A substantial proportion (46.4% of self-identified heterosexuals and 37.6% of bisexuals) reported two or more female partners in the last year. Of those who reported having sex with at least one woman in the last year, only a minority reported always using condoms: 10.5% of heterosexuals, 20.0% of bisexuals, and 18.5% of homosexuals.
All four subgroups reported high numbers (five or more) of male partners in the past year, ranging from 17.9% of heterosexuals to 66.7% of cross-dressers, and commonly reported insertive anal sex during the past year. However, among those who reported insertive anal sex in the last year, always using condoms was low overall and varied from 19.9% among homosexuals to 53.8% among heterosexuals. Receptive anal sex during the past year was reported by a majority of homosexual men, cross-dressers, and bisexual men (89.8%, 100%, and 52.5%, respectively). Notably, 28.6% of self-identified heterosexual men also reported receptive anal sex in the past year. Participants in all four subgroups reported receiving sex for money, with the highest proportions among heterosexuals (42.9%) and cross-dressers (68.8%).
HIV infection and syphilis seroreactivity according to sexual identity
HIV and syphilis seroreactivity were detected in 82 (18.5%) and 71 (16.0%) of 444 participants in whom sera was available, respectively. Less than half (43.7%) of men reported having been tested for HIV antibodies prior to the study, and importantly, only eight men who tested HIV positive reported having previously received a positive result. HIV seroprevalence varied significantly among sexual identity groups (Table 1). Cross-dressers were more likely to be HIV positive than either homosexuals (OR, 2.3; 95% CI, 1.0–4.8) or bisexuals (OR, 2.7; 95% CI, 1.1–6.5). Based on the proportion of HIV-positive men who were non-reactive on the less sensitive EIA, the estimated overall HIV seroincidence was 11.2% per year (95% CI, 4.8–23.6). Syphilis seroprevalence rates paralleled HIV infection rates by self-identified sexual orientation. Cross-dressers were more likely to be syphilis reactive than homosexuals (OR, 6.8; 95% CI, 3.2–14.1) or bisexuals (OR, 7.8; 95% CI, 3.1–19.7). Cross-dressers also had the highest prevalence (28.9%) of VDRL titers indicating recent acquisition or active syphilis (≥ 1:16).
Risk markers and risk factors associated with HIV infection
Variables associated with HIV serostatus by bivariate analysis included self-identification as a cross-dresser (compared to self-identified heterosexual men) (OR, 6.5; 95% CI, 1.4–31.1); receptive anal sex with six or more men in the last year (OR, 3.9; 95% CI, 1.8–8.7); history of a STD in the past year (OR, 2.0; 95% CI, 1.2–3.4); lack of circumcision (OR, 4.1; 95% CI, 1.0–17.5); and reactive syphilis serology (OR, 3.5; 95% CI, 2.0–6.2).
In the multivariate logistic regression analysis, adjusting for other variables, HIV infection was independently associated with receptive anal sex during the last year (OR, 2.2; 95% CI, 1.0–4.6), reactive syphilis serology (OR, 2.9; 95% CI, 1.7–5.1), and history of STD in the past year (OR, 1.8; 95% CI, 1.1–3.0). Lack of circumcision did not remain as a significant correlate in the multivariate model.
In the multivariate logistic regression analysis, syphilis seroreactivity was independently associated with lower educational level (< 11 years of school; OR, 4.3; 95% CI, 1.2–16.7), receptive anal intercourse during the last year (OR, 3.6; 95% CI, 1.2–10.4), history of rectal discharge (OR, 2.6; 95% CI, 1.2–5.3), HIV positivity (OR, 2.5; 95% CI, 1.3–4.6), and self-identification as a cross-dresser (OR, 3.2; 95% CI, 1.5–6.9).
Characteristics related to MSM who were potential ‘bridgers'
Of the 451 participants, 334 (74.1%) reported sex exclusively with men (`non-bridgers') and 117 (25.9%) acknowledged sex with both men and women (`bridgers') during the previous year (Table 2). Sixty-four (54.7%) ‘bridgers’ reported two or more female sexual partners in the last year, 20 (17.1%) reported always using condoms, and 50 (42.7%) reported never using condoms when having sex with women during the last year (data not shown). Importantly, with respect to transmission risk to women, ‘bridgers’ reported less frequent condom use during the last episode of vaginal sex (36.8%) than during the last episode of insertive (45.9%) or receptive anal sex (43.3%) with men.
Although ‘bridgers’ were less likely to use condoms with female partners, they had lower prevalence of HIV and syphilis than ‘non-bridgers'. HIV infection rates were 11.1% among ‘bridgers’ and 21.1% among ‘non-bridgers’ (OR, 2.1; 95% CI, 1.1–5). All 13 of the HIV-seropositive ‘bridgers’ and 68 out of 69 HIV-seropositive ‘non-bridgers’ had sera tested by the less sensitive EIA, yielding estimated HIV seroincidence of 7.9% per year (95% CI, 0–22.1) among ‘bridgers’ and 12.5% per year (95% CI, 4.4–24.9) among ‘non-bridgers'. Syphilis reactivity (OR, 1.7; 95% CI, 0.83–3.3) and serologic indication of more recent or active syphilis (VDRL > 1:16) (OR, 2.4; 95% CI, 0.91–10) were more common among ‘non-bridgers’ than ‘bridgers'.
`Bridgers’ reported similar numbers of male sexual partners as ‘non-bridgers'. Consistent condom use during insertive sex in the last year was reported by only 25.5% and 21.3% of ‘bridgers’ and ‘non-bridgers', respectively, with comparable rates of consistent condom use during receptive anal sex. In the multivariate model, ‘bridgers’ were more likely than ‘non-bridgers’ to report insertive anal intercourse in the last year (OR, 2.3; 95% CI, 1.3–4.1), to currently have a regular sexual partner (OR, 1.6; 95% CI, 1.0–2.6), and to have reported a STD ever (OR, 2.1; 95% CI, 1.3–3.5). ‘Bridgers’ were less likely than ‘non-bridgers’ to report receptive anal sex (OR, 0.20; 95% CI, 0.10–0.3) and less likely to be HIV positive (OR, 0.50; 95% CI, 0.20–0.90).
This study is the first to correlate risk behavior, particularly bisexuality, with HIV and syphilis rates among a large and diverse cross-sectional sample of MSM in Lima, Peru. We found HIV and syphilis rates were significantly different among Peruvian MSM, with the highest prevalence rates among cross-dressers and homosexuals. Not surprisingly, a high proportion of MSM who self-identified as heterosexual or bisexual reported both male and female partners. We termed men who reported sex with both men and women in the last year ‘bridgers’ and found that they reported more insertive anal sex than ‘non-bridgers', had comparable numbers of male partners, and a majority of their recent sex with both male and female sexual partners was unprotected. The cross-dressers interviewed in this study reported large numbers of male partners, and receptive as well as insertive anal sex. Based on a previous study of MSM in Peru , male partners of cross-dressers are largely heterosexually-identified and also represent an important potential bridge group for spread of HIV and syphilis to the general heterosexual population.
`Bridge populations’ have been described in transmission of HIV and other STD. Using sexual network analysis, Morris and colleagues  reported that heterosexual Thai men who frequent female commercial sex workers contribute disproportionately to the spread of HIV and other STD to women. Gorbach et al.  described the high prevalence of ‘active bridging’ (defined as unprotected sex with female sex workers and low risk female partners) among heterosexual men in Cambodia and found an association between bridging and STD symptoms.
In other Latin cultures, a substantial proportion of MSM have been reported to self-identify as heterosexual or bisexual, rather than homosexual [4,5,16]. Our behavioral eligibility criteria included sexual behavior in the last year rather than self-reported sexual identity, thus allowing us to gather information on MSM who traditionally might have been excluded if we had enrolled only men who self-identify as homosexual or bisexual.
Previous studies have reported HIV prevalence among MSM in Lima of 11.1% in 1985  and 6.5% in 1988 , but were limited by small sample sizes. The current study used more peer outreach-based recruitment strategies, which successfully recruited a diverse sample of study participants and provides an assessment of estimated HIV incidence and prevalence of this diverse sample of MSM in Lima. The estimated HIV seroincidence of 11.2% per year suggests that MSM are being infected at high rate even if the lower bound of the 95% CI (4.8% per year) is closer to the actual HIV incidence.
Not surprisingly, anal sex was a risk factor for being HIV-infected, independent of other variables. Reactive syphilis serology and a reported history of STD ever were also independently associated with HIV seropositivity suggesting that genital ulcer disease and other STD may be important risk factors for HIV transmission among MSM in Peru. While genital ulcer disease is a significant risk factor for heterosexual transmission of HIV, especially in the developing world, few studies have examined this association among MSM, after controlling for sexual behavior. Syphilis seropositivity was found to be a significant correlate of prevalent HIV infection in our study of MSM in the Dominican Republic (adjusted OR, 4.0; 95% CI, 1.4–11.3), after adjusting for sexual behavior . Serologic evidence of syphilis or herpes simplex virus infection and history of syphilis or anogenital herpes have also been independently associated with HIV infection among MSM in studies undertaken in the USA and the Netherlands [19–21].
This study has inherent biases and limitations. First, the recruitment methods relied on convenience sampling and may not be truly representative of the MSM population in Lima. Second, the cross-sectional design of this study makes it impossible to infer temporal relationships of reported biologic or behavioral risk factors with HIV seropositivity. Third, the sample size of 442 has limited power to estimate HIV incidence  as reflected in the wide confidence intervals. However, even the lower bound of the 95% CI for estimated HIV incidence (4.8% per year) is two-fold higher than that observed in contemporaneous cohorts of MSM in the USA [22,23]. While these data were collected in 1996, our recent cohort studies of MSM in 1999–2000 document similar HIV and syphilis prevalence and HIV incidence rates, even after antiretroviral agents have been introduced to a small number of patients in Peru .
In summary, this study has identified behavioral profiles among MSM in Peru that are associated with substantially different risks of HIV and syphilis, low rates of and inconsistent condom usage, and potential sexual ‘bridging’ between MSM and the general population of women. The shape and scope of the HIV epidemic is greatly influenced by the degree of mixing between people with different prevalences of disease and risk behavior. ‘Disassortative’ mixing between differing risk groups leads to the largest spread  of HIV infection and other STD. The most efficient strategies for reducing the spread of the disease are to prevent transmission among those with the highest rates of partner change and to reduce unprotected and disassortative sexual mixing. Prevention of infection among those with the highest rates of partner change has a multiplier effect in terms of preventing many more subsequent, secondary infections .
The different behavioral profiles among MSM in Latin America offer both a challenge and an opportunity for innovative prevention and intervention strategies. Prevention strategies should be developed and evaluated to address the dynamic and varied sexual behavior profiles of MSM. Non-stigmatizing prevention efforts should target the most marginalized and vulnerable MSM, such as cross-dressers, who have the highest rate of HIV and syphilis. Prevention services should be coordinated for heterosexuals and MSM to reduce HIV and syphilis transmission rates in bisexual men. Prevention strategies directed at STD control among MSM may help reduce HIV transmission to both their male and female partners.
We thank B. Courtois, M. Chion, M. Chiappe, J. Alva, Z. Maza, M. Lacherre, R. Villegas, M. Laguna, A. Caviedes, A. Soto, M. Guevara, R. Medina, L. Portugal, J. Asencios, M. Kahata, C. Cespedes, E. Garrido, M. Reyes, W. Whittington, R. Nelson, J. Hughes, K. Thomas and the peer educators who assisted with various aspects of this project.
1. Quinn TC. Global burden of the HIV pandemic. Lancet 1996, 348: 99–106.
2. Centers for Disease Control and Prevention. Statistics from the Centers for Disease Control and Prevention. AIDS 1996, 10: 117–119.
3. Pan American Health Organization. AIDS Surveillance in the Americas.
Biannual Report, May 2000.
4. Daniel H, Parker R. Sexuality, Politics and AIDS in Brazil.
Washington, DC: Falmer Press; 1993.
5. Tabet SR, deMoya EA, Holmes KK. et al
. Sexual behaviors and risk factors
for HIV infection among men who have sex with men in the Dominican Republic. AIDS 1996, 10: 201–206.
6. Mexican male bisexuality.
In:Bisexualities: Theory and Research
. Edited by Klein F, Wolf T. New York: Harrington Press; 1985: 75–85.
7. Sanchez J, Gotuzzo E, Escamilla J. et al
. Gender differences in sexual practices and sexually transmitted infections among adults in Lima, Peru. Am J Public Health 1996, 86: 1098–1107.
8. Johnson Kay, Alarcon JO, Rodriguez C, et al.Sexual Networks of Pregnant Women with HIV Infection in Lima, Peru.
Presented at XI Latin American Congress on Sexually Transmitted Diseases/V Pan American Conference on AIDS
. Lima, December 1997.
9. Janssen RS, Statten GA, Stramer SL. et al
. New testing strategy to detect early HIV-1 infection for use in incidence estimates and for clinical and prevention purposes. JAMA 1998, 280: 42–48.
10. Russell KL, Carcamo C, Watts DM. et al
. Emerging genetic diversity of HIV-1 in South America. AIDS 2000, 18: 1785–1791.
11. Hastie T, Tibshirani R. Generalized Additive Models.
New York: Chapman and Hall; 1990.
12. Hosmer DW, Lemeshow S. Applied Logistic Regression.
New York: John Wiley & Sons Inc.; 1989.
13. Cáceres C. Male bisexuality in Peru and the prevention of AIDS.
In:Bisexualities and AIDS - International Perspectives.
Edited by Aggleton P. Bristol, PA: Taylor and Francis; 1996.
14. Morris M, Podhisita C, Wawer M. et al
. Bridge populations in the spread of HIV/AIDS in Thailand. AIDS 1996, 10: 1265–1271.
15. Gorbach PM, Sopheab H, Phalla T. et al
. Sexual bridging
by Cambodian men: potential importance for general population spread of STD
and HIV epidemics. Sex Transm Dis 2000, 27: 320–326.
16. Morse EV, Simon PM, Osofsky HJ. et al
. The male street prostitute: a vector for transmission of HIV infection into the heterosexual world. Soc Sci Med 1991, 32: 535–539.
17. Rojas G, Gotuzzo E, Yi A, Koster F. Acquired immunodeficiency syndrome in Peru. Ann Intern Med 1986, 105: 465–466.
18. Cáceres CP, Gotuzzo E, Wignall S. et al
. Sexual behavior
and frequency of antibodies to Type 1 Human Immunodeficiency Virus (HIV-1) in a group of Peruvian male homosexuals. Bull Pan Am Health Organ 1991, 25: 306–319.
19. Kuiken CL, van Griensven GJ, de Vroome EM. et al
. Risk factors
and changes in sexual behavior
in male homosexuals who seroconverted for human immunodeficiency virus antibodies. Am J Epidemiol 1990, 132: 523–531.
20. Stamm WE, Handsfield HH, Rompalo AM. et al
. The association between genital ulcer disease and acquisition of HIV infection in homosexual men
. JAMA 1988, 260: 1429–1433.
21. Holmberg SD, Stewart JA, Gerber AR. et al
. Prior herpes simplex virus type 2 infection as a risk factor for HIV infection. JAMA 1988, 259: 1048–1050.
22. Seage GR 3rd, Holte SE, Metzger D. et al
. Are US populations appropriate for trials of human immunodeficiency virus vaccine? The HIVNET Vaccine Preparedness Study. Am J Epidemiol 2001, 153: 619–627.
23. Buchbinder SP, Douglas JM, McKirnan DJ, Judson FN, Katz MH, MacQueen KM. Feasibility of human immunodeficiency virus vaccine trials in homosexual men
in the United States; risk behavior, seroincidence, and willingness to participate. J Infect Dis 1996, 174: 954–961.
24. Lucchetti A, Sanchez J, Collis T, et al.Bacterial STDs and HSV-2 acquisition among MSM HIV seroconverters in Peru.The 1st IAS Conference on Pathogenesis and Treatment
. Buenos Aires, July 2001 [abstract 370].
25. Garnett GP, Anderson RM. Factors controlling the spread of HIV in heterosexual communities in developing countries: patterns of mixing between different age and sexual activity classes. Philos Trans R Soc Lond B Biol Sci 1993, 342: 137–159.
26. World Bank. Confronting AIDS: Public priorities in a global epidemic.
New York: Oxford University Press; 1997.