Abstract
We conducted a respondent-driven sampling survey (N = 215) to characterize correlates of risk for HIV infection among gay and bisexual men in Kampala, Uganda. We used RDSAT software to produce population estimates for measures and created exportable weights for multivariable analysis. Overall, 60.5% of gay/bi men identify as gay and 39.5% as bisexual; 91.6% are Ugandans. Unprotected receptive anal intercourse (URAI) was associated with identifying as gay, being younger and having had an HIV test in the past 6 months. Perceptions of being low risk to acquire or transmit HIV infection were paradoxically associated with higher likelihood of URAI. Programs to address risk of HIV infection among gay and bisexual men in Kampala need to address perceptions of risk among gay identified men.
Keywords: Men who have sex with men, Uganda, Africa, HIV, Risk behavior, Correlates, Gay and bisexual men
Introduction
Men who have sex with men (MSM) worldwide are disproportionately affected by HIV, with Africa being no exception. In fact, MSM in low- and middle-income countries, including many of those in Africa, have a higher risk of HIV infection than the general population (Baral et al. 2007). Compounding the risk of HIV infection among MSM is the difficulty involved in reaching this population for basic epidemiological studies. In many African countries, MSM are highly stigmatized, and in many cases homosexuality is illegal and can result in imprisonment [Ugandan Penal Code (http://www.mask.org.za/index.php?page=uganda), Sections 21 and 140]. As a result, little information has been gathered about risk behavior among MSM in African countries. This situation has recently begun to change with a number of studies being carried out in Africa (Cáceres et al. 2008). Basic data are needed not only for advocacy but also for correctly prioritizing and creating HIV prevention intervention programs. We therefore conducted a survey of gay and bisexual identified MSM in Kampala using respondent driven sampling (RDS), a method in wide use globally to reach hidden populations (Heckathorn 1997; Magnani et al. 2005; Malekinejad et al. 2008). The basic demographics and indicators of risk among gay and bisexual men in Kampla have been reported previously (Kajubi et al. 2007). Techniques to analyze RDS-generated data have evolved since the initial publication allowing us to examine multiple correlates of unprotected receptive anal intercourse (Heckathorn 2007).
Methods
Setting and Study Population
Our study sampled men who self-identified as gay or bisexual and who lived in Kampala, Uganda. Kampala, the capital and largest urban area of Uganda, is home to approximately 1.2 million persons (Uganda Bureau of Statistics 2002). This study received approval from Makerere University and the University of California San Francisco’s institutional review boards as well as the Uganda National Council for Science and Technology.
Sampling Design and Recruitment
We conducted recruitment using RDS, a sampling method widely used to sample hidden populations (Heckathorn 1997; Magnani et al. 2005). Recruitment for this study has been described in detail previously (Kajubi et al. 2007). Ten initial seeds were chosen with respect to diversity (age, educational attainment, income and sexual identity). Interested potential participants contacted study staff via a mobile phone number or in person. Subsequently they were screened for eligibility by the Principal Investigator. Holding a study recruitment coupon, being 18 years old or older, self-identifying as gay or bisexual (that is identifying as a “kuchu” in Luganda) and not having previously participated in the survey were the only recruitment criteria. After eligibility was determined, verbal consent was obtained using a standard consent script. Face-to-face interviews were conducted in private locations by trained staff members. At the completion of the interview, participants were given 4,000 shillings (approximately $2USD in 2004) to defray the cost of transportation. Participants also received referrals and information regarding HIV prevention.
Measures
The behavioral survey used in this study was based on standard behavioral surveillance surveys used throughout the developing world (Family Health International 2000) and on those employed among high-risk populations in the United States (Lansky et al. 2007). In addition, we asked a detailed series of partner-by-partner sexual behavior questions for up to five of the respondents’ most recent partners in the past 6 months. Partners assessed included male and female partners. This sexual activity “matrix” has been helpful in describing and quantifying detailed sexual behaviors and partnerships (Berry et al. 2007, 2008). Attitudes and context of condom use were asked about all partners.
Data Analysis
RDSAT statistical software (www.respondentdrivensampling.org) was used to produce population estimates for behavioral measures by adjusting for network size and recruitment patterns. These adjustments are only made on the non-seed observations. Behaviors were then analyzed in relation to unprotected receptive anal intercourse (URAI) as our main outcome of interest. Adjusted point estimates and corresponding confidence intervals were examined. If point estimates differed and their confidence intervals were non-overlapping, we report this as a strong possibility of being associated with our outcome of interest. If at least one point estimate was outside the confidence interval of the other estimate, we report this as suggestive of a possible association. Variables that were determined to be strongly or potentially associated with URAI were then included in multivariable logistic regression.
RDSAT also provides individual weights for use in multivariable analyzes. Following methods outlined by Heckathorn (2007), we exported individual weights produced by RDSAT for our outcome variable and used them in multivariable analyzes using the survey logistic procedure in Statistical Analysis Software [SAS] version 9, SAS Institute Inc., Cary, NC.
Results
Recruitment was conducted for 8 weeks, September through October 2004. Recruitment produced a crude sample of 224 gay and bisexual men. In adjusted analyzes based on 215 non-seed respondents, 60.5% of gay/bi men identify as gay and 39.5% as bisexual. Fifty percent of gay/bi men have completed secondary school or higher and 91.6% are Ugandan nationals. Only 9.3% are unemployed. About 23.7% have tested for HIV in the past 6 months while just 12.2% perceive themselves to be at high risk for HIV infection. Twenty-one percent have any unprotected receptive anal intercourse in a recent 6 months period, and 39.2% have ever had a sexually transmitted disease [STD] (Table 1). Of all partnerships reported, 22% were with female partners.
Table 1.
Crude and adjusted population estimates of characteristics, gay and bisexual men, Kampala, Uganda
| Variable | Crude % (N) | Adjusted % (95% CI) |
|---|---|---|
| Sexual identity | ||
| Gay | 63.7 (137) | 60.5 (47.6–70.4) |
| Bisexual | 36.3 (78) | 39.5 (29.7–52.4) |
| Highest education level attained | ||
| None | 1.8 (4) | 1.3 (0.1–2.6) |
| Some primary school | 17.5 (40) | 17.7 (8.4–27.0) |
| Completed primary | 26.8 (58) | 30.0 (19.8–43.2) |
| Completed secondary school | 17.1 (38) | 24.0 (13.1–34.3) |
| Some tertiary school | 19.3 (43) | 16.8 (11.3–28.0) |
| Completed tertiary school | 17.5 (36) | 10.3 (5.8–15.0) |
| Occupation/employment | ||
| Student | 30.7 (66) | 29.5 (20.0–40.8) |
| Professional | 24.7 (53) | 18.0 (10.1–24.2) |
| Retail | 11.2 (24) | 23.6 (9.9–38.6) |
| Trade | 24.2 (52) | 18.8 (12.0–27.4) |
| Other | 0.5 (1) | 0.7 (0–2.0) |
| Unemployed | 8.4 (18) | 9.3 (3.7–17.8) |
| Missing | 0.5 (1) | 0.1 (0–0.3) |
| National origin | ||
| Ugandan | 91.6 (157) | 91.6 (85.8–97.8) |
| Non-Ugandan African | 7.4 (16) | 7.9 (1.9–13.9) |
| Non-African | 0.9 (2) | 0.5 (0–1.0) |
| Age group (years) | ||
| 18–20 | 29.3 (63) | 44.5 (30.5–54.5) |
| 21–25 | 37.7 (81) | 30.7 (22.8–40.7) |
| 26–30 | 22.3 (48) | 18.6 (11.5–28.8) |
| 31–35 | 7.4 (16) | 4.1 (2.3–7.2) |
| 36–40 | 1.4 (3) | 0.7 (0–1.7) |
| 41+ | 1.9 (4) | 1.5 (0.1–2.2) |
| URAI with male partners | 24.2 (52) | 21.2 (15.4–30.2) |
| UIAI with male partners | 17.5 (36) | 19.2 (10.0–31.1) |
| Partners | ||
| 0 | 15.9 (34) | 25.7 (11.8–39.7) |
| 1 | 31.8 (68) | 34.4 (23.3–44.5) |
| 2 | 22.9 (49) | 19.7 (13.5–29.1) |
| 3–5 | 24.3 (52) | 16.1 (11.4–22.7) |
| 6–9 | 3.3 (7) | 2.3 (0.7–4.5) |
| 10+ | 1.9 (4) | 1.7 (0.1–4.8) |
| STD ever | 46.0 (99) | 39.2 (29.5–49.3) |
| Foreign partner ever | 44.1 (90) | 24.2 (17.8–33.7) |
| HIV test past 6 months | 29.8 (64) | 23.7 (15.5–33.3) |
| Ever forced to have sex | 15.0 (32) | 11.3 (7.0–15.0) |
| Out | 59.1 (127) | 59.4 (49.1–70.2) |
| Self perception of risk for HIV infection | ||
| None | 8.9 (19) | 6.4 (3.2–11.0) |
| Low | 13.1 (28) | 13.3 (7.0–21.0) |
| Somewhat high | 65.3 (139) | 68.1 (58.6–78.2) |
| High | 12.7 (27) | 12.2 (5.4–19.0) |
In bivariate analysis URAI appears to be associated with having had an HIV test in the past 6 months, having ever been forced to have sex, identifying as gay, having multiple partners in the past 6 months, being in the “heat of the moment” during sex, dislike of condoms (by both respondent and partners), condom availability and respondents not thinking they could get or pass on HIV (Table 2).
Table 2.
Prevalence of unprotected receptive anal intercourse (URAI) in the past 6 months by characteristics of gay and bisexual men, Kampala Uganda, 2004
| Variable | Prevalence of unprotected receptive anal intercourse |
||
|---|---|---|---|
| Crude % (N) | Adjusted % | Adjusted 95% CI | |
| Ever STD | |||
| Yes | 27.3 (27) | 20.2 | 12.2–28.9 |
| No | 21.0 (25) | 21.5 | 12.6–36.7 |
| Foreign partner | |||
| Yes | 27.8 (25) | 25.8 | 15.9–42.4 |
| No | 20.2 (23) | 20.0 | 10.4–25.3 |
| HIV test past 6 months | |||
| Yes | 26.6 (17) | 30.2 | 15.6–55.8 |
| No | 23.2 (35) | 18.6 | 11.5–27.1 |
| Ever forced to have sex | |||
| Yes | 43.8 (14) | 41.5 | 25.4–63.5 |
| No | 20.9 (38) | 18.5 | 11.9–28.3 |
| Out | |||
| Yes | 29.1 (37) | 18.9 | 11.7–30.1 |
| No | 17.0 (15) | 25.1 | 12.4–39.7 |
| Gay identified | |||
| Yes | 27.0 (37) | 26.7 | 16.6–40.7 |
| No | 19.2 (15) | 11.8 | 6.8–20.3 |
| Partners past 6 months | |||
| 0 | 0 | – | – |
| 1 | 20.6 (14) | 26.8 | 11.4–43.2 |
| 2 | 34.7 (17) | 39.6 | 23.3–54.0 |
| 3–5 | 30.8 (16) | 22.2 | 13.8–39.4 |
| 6–9 | 28.6 (2) | 9.9 | 0–41.2 |
| 10+ | 75.0 (3) | 12.5 | 1.0–100.0 |
| Multiple partners past 6 months | |||
| Yes | 33.9 (38) | 30.5 | 22.3–43.1 |
| No | 13.6 (14) | 15.2 | 6.1–26.6 |
| Sell sex | |||
| Yes | 30.5 (18) | 17.5 | 8.4–31.5 |
| No | 21.8 (34) | 23.0 | 14.7–33.7 |
| Self perception of risk for HIV infection | |||
| None | 31.6 (6) | 15.3 | 4.8–33.4 |
| Low | 39.3 (11) | 45.2 | 23.6–68.6 |
| Somewhat high | 20.9 (29) | 17.4 | 11.2–29.2 |
| High | 18.5 (5) | 24.8 | 5.0–56.8 |
| Heat of the moment | |||
| Yes | 54.9 (39) | 45.7 | 35.0–64.0 |
| No | 9.0 (13) | 9.4 | 3.7–20.0 |
| Respondent does not like condoms | |||
| Yes | 46.1 (35) | 41.5 | 28.7–56.7 |
| No | 12.2 (17) | 9.2 | 4.6–17.8 |
| Partner(s) does not like condoms | |||
| Yes | 42.7 (32) | 43.3 | 30.5–58.1 |
| No | 30.7 (43) | 8.8 | 5.0–15.8 |
| Condoms available | |||
| Yes | 44.4 (16) | 44.7 | 25.9–64.5 |
| No | 20.1 (36) | 16.8 | 10.9–25.6 |
| Partner(s) was HIV – | |||
| Yes | 55.1 (34) | 45.9 | 33.3–63.4 |
| No | 12.8 (18) | 10.9 | 5.0–20.3 |
| Respondent thought partner(s) were at low risk for HIV infection | |||
| Yes | 52.9 (36) | 45.1 | 31.2–61.8 |
| No | 21.8 (32) | 10.2 | 5.1–21.6 |
| Mutually faithful sexual relationship | |||
| Yes | 50.0 (38) | 42.5 | 32.1–58.7 |
| No | 10.1 (14) | 12.6 | 4.7–22.9 |
| Respondent and partner(s) had same HIV status | |||
| Yes | 54.4 (37) | 44.8 | 32.2–62.8 |
| No | 10.2 (15) | 10.6 | 4.2–21.0 |
| Respondent did not think he could get or pass on HIV | |||
| Yes | 50.0 (29) | 51.5 | 37.6–70.8 |
| No | 14.6 (23) | 10.1 | 5.9–16.2 |
Using weighted logistic regression with the variables associated with URAI in bivariate analyzes while controlling for age and level of educational attainment, having an HIV test in the past 6 months (adjusted odds ratio [AOR] 2.81, 95% confidence interval [CI] 1.2–7.4) being gay-identified (AOR 9.92, 95% CI 3.2–30.2), being in the heat of the moment (AOR 5.72, 95% CI 2.2–15.2) and not thinking he or his partners could get or pass on the HIV virus (AOR 12.25, 95% CI 4.5–32.9) were significantly associated with URAI. Age (AOR 0.91, 95% CI 0.8–1.0) appears to have an significant inverse relationship with URAI. Variables with significant association with URAI, adjusted for all other variables included in the model, are shown in Table 3.
Table 3.
Correlates of unprotected receptive anal intercourse, gay and bisexual men, Kampala, Uganda
| Variable | AOR | 95% CI | Wald Chi square |
|---|---|---|---|
| Age | 0.91 | 0.8–1.0 | 5.9* |
| HIV test past 6 months | 2.1 | 1.1–7.4 | 4.4* |
| Gay identified | 9.92 | 3.2–30.2 | 16.3** |
| Heat of the moment | 5.72 | 2.2–15.2 | 12.3** |
| Did not think he or his partners could get/pass on the HIV virus | 12.25 | 4.5–32.9 | 24.5** |
P < 0.05
P < 0.001
Discussion
We were able to characterize basic correlates of sexual risk among gay and bisexual men in Kampala, Uganda. Those men identifying as gay and holding the perception that one is at low risk for HIV acquisition or transmission stand out in our analysis. These factors must be addressed when formulating prevention approaches for this population. Interestingly, ever having a sexually transmitted infection, being “out”, availability of condoms and selling sex were not associated with URAI.
There are limitations to our investigation. First, as this was a first attempt at sampling gay and bisexual men in this setting, we chose not to undertake HIV testing. Serological results would have greatly strengthened this analysis. Additionally we also did not ask for a self-report of HIV status. Furthermore, while measures of UIAI and URAI focused on male partners, measures of condom attitudes and context were asked about all partners. Future studies will need to ask more detailed questions about these domains in relation to the gender of partners.
Despite these limitations, this investigation serves as a starting point for further studies of gay and bisexual men at risk for HIV infection. Interventions must include gay-appropriate messages that address the potential of HIV infection among this population to reverse the perception of being at low risk for HIV infection. Furthermore, these interventions may do well to focus on younger MSM who may have limited understanding of the risks of URAI with men, particularly in the context of a generalized epidemic where having women partners is perceived as a greater risk. Finally, further investigation of the association of recent HIV testing with URAI is needed as frequent HIV testing, with repeated HIV-negative results, may be seen as reinforcing risk behavior.
References
- Baral S, Sifakis F, Cleghorn F, Beyrer C. Elevated risk for HIV infection among men who have sex with men in low- and middle-income countries 2000–2006: A systematic review. PLoS Medicine. 2007;4(12):e339. doi: 10.1371/journal.pmed.0040339. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Berry M, Raymond HF, McFarland W. Same race and older partnering may explain differences in HIV prevalence among men who have sex with men. AIDS (London, England) 2007;21(17):2349–2350. doi: 10.1097/QAD.0b013e3282f12f41. [DOI] [PubMed] [Google Scholar]
- Berry M, Raymond HF, McFarland W. The Internet, HIV serosorting and transmission risk among men who have sex with men, San Francisco. AIDS (London, England) 2008;22(6):787–789. doi: 10.1097/QAD.0b013e3282f55559. [DOI] [PubMed] [Google Scholar]
- Cáceres CF, Konda K, Segura ER, Lyerla R. Epidemiology of male same-sex behaviour and associated sexual health indicators in low- and middle-income countries: 2003–2007 estimates. Sexually Transmitted Infections. 2008;84(Suppl 1):i49–i56. doi: 10.1136/sti.2008.030569. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Family Health International. Behavioral Surveillance Surveys. Arlington, Virginia: Family Health International; 2000. [Google Scholar]
- Heckathorn DD. Respondent-driven sampling: A new approach to the study of hidden populations. Social Problems. 1997;44:174–199. doi: 10.1525/sp.1997.44.2.03x0221m. [DOI] [Google Scholar]
- Heckathorn DD. Extensions of respondent driven sampling: Analyzing continuous variable and controlling for differential recruitment. Sociological Methodology. 2007;37(1):151–207. doi: 10.1111/j.1467-9531.2007.00188.x. [DOI] [Google Scholar]
- Kajubi P, Kamya MR, Raymond HF, Chen S, Rutherford GW, Mandel JS, et al. Gay and bisexual men in Kampala, Uganda. AIDS and Behavior. 2007;12(3):492–504. doi: 10.1007/s10461-007-9323-7. [DOI] [PubMed] [Google Scholar]
- Lansky A, Sullivan PS, Gallagher KM, Fleming PL. HIV behavioral surveillance in the U.S.: A conceptual framework. Public Health Reports. 2007;122:16–23. doi: 10.1177/00333549071220S104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Magnani R, Sabin K, Saidel T, Heckathorn D. Review of sampling hard-to-reach and hidden populations for HIV surveillance. AIDS (London, England) 2005;19:S67–S72. doi: 10.1097/01.aids.0000172879.20628.e1. [DOI] [PubMed] [Google Scholar]
- Malekinejad M, Johnston LG, Kendall C, Kerr LR, Rifkin MR, Rutherford GW. Using respondent-driven sampling methodology for HIV biological and behavioral surveillance in international settings: A systematic review. AIDS and Behavior. 2008;12(4 Suppl):S105–S130. doi: 10.1007/s10461-008-9421-1. [DOI] [PubMed] [Google Scholar]
- Uganda Bureau of Statistics. [Accessed 26 March 2007];Uganda Population and Housing Census. 2002 Available at: http://www.ubos.org/2002%20Census%20Final%20Reportdoc.pdf.