Young gay men continue to engage in high rates of unprotected anal intercourse, and are becoming infected with HIV at alarming rates in the USA and many other countries . In a study of gay men aged 18 to 25 years, in three mid-sized West Coast communities, 43% reported engaging in unprotected anal intercourse during the past 6 months . A recent seroprevalence study of 2100 young gay men aged 15 to 22 years conducted by the Centers for Disease Control and Prevention, Atlanta, Georgia, USA, in six urban counties across the USA found a median HIV prevalence of 7% . Additional studies of young gay men have found seroprevalence rates of 9.4% in San Francisco  and 9% in New York City .
HIV-prevention specialists have recognized the need to implement programs specifically tailored to young gay men's needs, but there is disagreement about what are the most effective HIV-prevention approaches for this population. One issue of considerable debate recently is whether HIV status should be a central feature in the design of prevention programs. In particular, some have argued that prevention interventions specifically targeting HIV-negative men are necessary because of the unique stresses and emotional issues HIV-negative men confront in AIDS-ravaged communities [6,7]. Although these ideas have generated considerable discussion within the HIV-prevention field and a number of community-based organizations have begun incorporating them into their prevention programs , they have not been evaluated using empirical research. In particular, there is a lack of data comparing the sexual behavior patterns of HIV-negative, HIV-positive and untested men, and little information on the specific prevention needs of men of different HIV statuses. In addition, since one cannot assume that an individual's perception of his HIV status is an accurate representation of his actual HIV status, programs targeting men by their self-identified HIV status may be problematic.
In this study, our objectives were to: (i) assess the degree to which an individual's perceived HIV status matches his actual HIV status; (ii) compare the sexual behavior patterns of young gay and bisexual men who perceive themselves to be HIV-negative, perceive themselves to be HIV-positive, or do not know their status; and (iii) examine whether factors related to engaging in unprotected anal intercourse vary according to perceived HIV status. This information will help guide decision-making regarding the degree to which HIV status should be considered in designing the most effective HIV-prevention interventions for gay and bisexual men.
Participants and procedures
The data for this study come from the first wave of the San Francisco Young Men's Health Study (SFYMHS). The SFYMHS is a survey of HIV infection and risk behavior based on a multi-stage probability sample of single men, aged 18–29 years, residing in households from the 21 census tracts in San Francisco with the highest cumulative number of AIDS cases in 1992 . The SFYMHS was modeled after the San Francisco Men's Health Study , which was a multi-stage probability sample drawn in 1984 of single men aged 25–54 years residing in 19 census tracts with the highest AIDS incidence in San Francisco at that time. All 19 census tracts in San Francisco Men's Health Study were included in the 21 census tracts sampled for the SFYMHS. Two additional census tracts were included to increase the potential number of men under age 30 in the sample population.
All households in the 21 census tracts were listed, and a multi-stage sample was drawn of 6671 addresses of which 6186 proved to be eligible households. The sampling fractions were varied within blocks from one in four to one in eight households to oversample areas with higher concentrations of young men. Each house-hold sampled was approached to enumerate the residents and determine their eligibility for study. Unmarried, English-speaking men between 18 and 29 years of age at time of contact were eligible.
Enumeration of 5801 (94%) households yielded 1387 eligible young single men of whom 1076 (78%) agreed to participate and were interviewed in their home. Two hundred and twenty-seven (16%) refused participation, 39 (3%) moved after enumeration, 31 (2%) were excluded because of a language barrier and 14 (1%) were never found at home. Among those interviewed, 408 (38%) either identified their sexual orientation as gay/bisexual or reported engaging in sexual behavior with a male in the past 12 months. The final sample for the analyses presented here thus comprises 408 men.
The men were interviewed in their homes and asked to give a sample of peripheral blood which was obtained on filter paper by a finger stick; eighty-nine per cent (n = 364) agreed to do so. Samples were tested for HIV antibodies at the California Department of Health Services Viral and Rickettsial Disease laboratory (Berkeley, California, USA) with enzyme-linked immunosorbent assay (ELISA; Organon Teknika, Durham, North Carolina, USA) and specimens positive on ELISA were confirmed with Western blot (Organon Teknika).
The interview protocol assessed a wide variety of topics related to HIV and sexual behavior. For this paper, we examined five main sets of variables, which were assessed with single items or brief scales:
I. Demographics: Participants reported their age, ethnic identification, education and income.
II. Perceived HIV status: Participants were asked whether they had ever been tested for HIV antibodies, and if so, the results.
III. Sexual behaviors: Participants reported the numbers of males and females with whom they had sex during the past 12 months, and indicated whether they had engaged in a variety of sexual behaviors with men and with women during the previous 12 months. Participants also rated the frequency with which they engaged in sex while under the influence of drugs or alcohol during the past 2 months, using a 5-point scale which ranged from ‘never’ to ‘almost all the time’.
IV. Integration into gay community: The degree to which the men were ‘out of the closet’ about their homosexuality was assessed by asking participants to indicate, on a 5-point Likert scale, ‘how open’ or ‘out of the closet’ [they were] about their sexual relationships with other men or about [their] sexual orientation (ranging from ‘not out to anyone’ to ‘out to almost everyone’). Participants were asked to rate on a five-point scale how many of their friends were gay or bisexual (ranging from ‘almost none or none’ to ‘almost all or all’). Using a six-point scale, participants indicated the degree to which they had frequented gay bars during the past 6 months (ranging from ‘never’ to ‘several times a week’). Participants also indicated the number of people with HIV or AIDS that they had known personally. Participants were asked whether they were currently involved in a boyfriend/lover relationship.
V. Psychosexual factors: Using four-point Likert scales that ranged from ‘disagree strongly’ to ‘agree strongly’, participants rated the degree to which they agreed with a variety of HIV-related attitude statements, including: degree of enjoyment of safe sex (three items, Cronbach alpha = 0.70), interpersonal barriers to safe sex (two items, Cronbach alpha = 0.84), sexual communication skills (one item), perceived social norms regarding safe sex (one item), social support for safe sex (one item) and sexual impulsivity (one item). Although we acknowledge that single-item measures are not ideal, the time constraints of the interview did not permit longer scales to be used. Each of the items used had demonstrated construct and predictive validity in previous investigations .
The men ranged in age from 18 to 29 years old, with a mean of 25.83 years (SD = 2.52). The sample was 77% white, 8% Hispanic/Latino descent, 7% Asian/Pacific Islander, 5% African American, 2% native American, and 2% ‘Other’. Forty-four per cent were college graduates (including 11% who had gone to graduate school); 22% were currently attending school, either full-time (10%) or part-time (12%).
Eighty-four per cent described themselves as gay, 14% as bisexual, and 2% labeled themselves heterosexual. Eighty-two per cent described themselves as ‘out of the closet’ to most or almost everyone they knew, and 18% described themselves as out to half or less than half of the people they knew. Sixty-seven per cent reported going to gay bars a few times a month or more often, with 48% reporting going to gay bars at least once a week. Sixty-one per cent reported that more than half of their friends were gay or bisexual. The vast majority of the men (87%) reported they had personally known someone who was HIV-positive; seventy-seven per cent reported knowing someone with AIDS.
Forty-eight per cent of the men reported currently being in a primary relationship with a male. These relationships ranged from 1 month to 12 years, with a median length of 12 months and a mode of 1 month. Six per cent of the men reported being in a primary relationship with a woman, ranging from 2 months to 7 years, with a median of 13 months and a mode of 3 months.
HIV status: perceived versus actual
Eighty-four per cent of the men (n = 342) had been tested for HIV antibodies prior to their interview. Of those men, 17% (n = 57) reported they were HIV-positive, and 83% (n = 285) reported they were HIV-negative. For the 364 men who provided a blood sample for HIV-testing, we were able to compare the men's perceptions of their status with their actual HIV status (see Table 1). A total of 18.7% of the men in this sample were found to be HIV-positive. Eleven men or 4% of those who perceived themselves to be HIV-negative were in fact found to be HIV-positive, as were six (10%) of the untested men. Thus, 25% of the men who were HIV-positive in this sample did not know it.
We next used analyses of variance and χ2 tests to examine differences in demographic and psychosocial variables between men of each of the following four HIV status groups: (i) men who reported they were HIV-positive (n = 57); (ii) men who reported they were HIV-negative and tested HIV-negative on our antibody test (n = 244); (iii) men who reported they were HIV-negative but tested HIV-positive on our antibody test (n = 11); and (iv) men who were previously untested and stated they did not know their HIV status (n = 66). Table 2 presents the group comparisons.
The men who knew they were HIV-positive were somewhat older, knew more people with AIDS and HIV-positive individuals, had the greatest percentage of gay or bisexual friends, had the most male sex partners, and were least likely to have had sex with women in the past 12 months. In contrast, the men who perceived themselves to be HIV-negative but were actually HIV-positive were more likely to be from an ethnic minority group (14.8% of colored men misperceived their serostatus compared with 1.4% of the white men) and were less educated, but reported somewhat higher incomes than the other groups. The previously untested men appeared to be the least integrated into the gay community; they were most likely to identify as bisexual, had the smallest percentage of gay friends, were least ‘out of the closet’, knew the fewest people with AIDS/HIV-positive people, had the fewest male sex partners, and were most likely to have had sex with women in the past 12 months.
Ninety-five per cent of the men reported having had sex with a male in the past 12 months. The men who knew they were HIV-positive reported the greatest number of male sex partners in the past year (median = 8.5, range 0–300) compared with men who perceived themselves to be HIV-negative (median = 5, range 0–260) and untested men (median = 2, range 0–100).
Table 3 presents the percentages of men who reported engaging in each of the various sexual activities with men during the preceding 12 months, for the sample as a whole and for each perceived HIV-status group separately. Overall, 37% of the men reported engaging in unprotected anal intercourse during the past year; however, this differed with perceived HIV status. Fifty-nine per cent of the men who perceived themselves to be HIV-positive reported engaging in unprotected anal intercourse, compared with 35% of the men who perceived themselves to be HIV-negative and 28% of the men who did not know their status. In addition, of the 17 men who did not know they were HIV-positive, 47% reported engaging in unprotected anal inter-course (data not shown). For the men who perceived themselves to be HIV-positive, the rates of unprotected receptive intercourse were over twice as high (56%) as those for the HIV-negative (24%) and untested men (21%), but the HIV-positive men also reported as much unprotected insertive (30%) — with and without ejaculation — as did the other two groups (28% for HIV-negative and 25% for untested). Nine per cent of the HIV-positive men reported unprotected insertive inter-course with ejaculation. A similar trend was found with oral sex with ejaculation: HIV-positive men reported almost twice as much receptive oral sex with ejaculation without condom (35%) compared with the other groups (19% for HIV-negative and 15% for untested), but fairly similar rates of insertive oral sex with ejaculation (21% versus 25% and 22%).
Since we assessed sexual behaviors during the past year, it is possible that a number of the HIV-positive men may have reported behavior that occurred prior to their knowledge of their seropositivity. For example, they may have reported behavior that occurred 6 months ago but only found out they were HIV-positive 2 months ago. Therefore, in order to examine more conservatively the degree to which HIV-positive men who knew they were HIV-positive engaged in unprotected anal intercourse, we removed from the analyses eight HIV-positive men who did not know their status for at least 1 year prior to their interview. Even after this correction, the percentages of HIV-positive men engaging in unprotected anal intercourse remained virtually the same: 57%.
Data on characteristics of the men's sex partners were not collected, but whether the men were involved in a boyfriend relationship and, if so, the HIV status of their partner was assessed. High percentages of men who had themselves been tested for HIV reported that they knew their boyfriend's HIV status (80% of the HIV-negative men and 83% of the HIV-positive men reported knowing their boyfriends' HIV status). In contrast, among the untested men, only 46% reported that they knew their boyfriend's HIV status. HIV-positive men were twice as likely to report that they knew their boyfriend was HIV-positive (28%) compared with HIV-negative men and untested men (both 14%). This suggests that at least some of the unprotected sex reported by HIV-positive men may have occurred with an HIV-positive partner, although it was not possible to determine precisely which partners the respondents engaged in unprotected sex with.
Predictors of risk-taking
Logistic regression analyses were used to examine whether the factors associated with engaging in unprotected anal intercourse differed for men who perceived themselves to be HIV-negative, perceived themselves to be HIV-positive, or were untested. For each HIVstatus group, bivariate logistic regressions were first computed for each predictor variable. Variables which were significant at the bivariate level (P < 0.05) were then included in a multivariate logistic regression for each HIV-status group, using forward stepwise inclusion with a P value of 0.10 for entry. The results are presented in Table 4.
For HIV-negative men, there were a number of factors that predicted engaging in unprotected anal intercourse at the bivariate level, including sexual impulsivity (i.e., ‘having trouble being safe when you're really turned on’), feeling less enjoyment of safe sex, having a boyfriend, a greater number of male sex partners, higher percentage of friends who are gay or bisexual, having sex while high on drugs or alcohol, not having sex with women, and problems communicating to sex partners that you want to have safe sex. When those variables were entered in a multivariate stepwise logistic regression, the variables which entered as significant predictors were in this order: sexual impulsivity, feeling less enjoyment of safe sex, not having sex with women, having a boyfriend, having sex while high on drugs or alcohol, and number of male sex partners.
Most of the significant bivariate predictors of unprotected anal intercourse for HIV-negative men were also predictive of unprotected anal sex for HIV-positive men, including being highly aroused, having sex while high on drugs or alcohol, problems communicating to sex partners that you want to have safe sex, feeling less enjoyment of safe sex, and a greater number of male sex partners. There were a few differences, however. Having a boyfriend, which was a strong predictor of unprotected sex for HIV-negatives, was not related to unprotected sex for HIV-positive men. Likewise, having sex with women was not a significant predictor for HIV-positive men, of whom only 4% had sex with a woman. In contrast, having interpersonal barriers was significantly predictive of unprotected sex for HIV-positive men but was only marginally so for HIV-negative men. When the significant bivariate variables were entered in a multivariate stepwise logistic regression, the following variables entered as significant predictors in this order: being turned on, having sex while high on alcohol or drugs, and communication problems.
For untested men, there were fewer significant predictors of unprotected anal intercourse. Those that emerged suggest that a greater degree of gay involvement, which reflected in being more ‘out of the closet’ about one's homosexuality, going more to gay bars and having a boyfriend, was associated with unprotected sex. In contrast, having sex with women predicted safe sex. In the multivariate analysis, only being ‘out of the closet’ emerged as a significant predictor.
When we repeated the logistic regressions to examine unprotected receptive and insertive intercourse separately, the results were basically consistent with those reported here for any unprotected intercourse.
Given the 18.7% HIV prevalence rate found in this population-based sample of young gay and bisexual men, the fact that 37% of the respondents reported engaging in unprotected anal intercourse during the past year is cause for great concern. The high rate of unprotected intercourse among HIV-positive men (59%) is especially disturbing, but we need more information to understand it. Most importantly, the men's partners were not known. It is possible that the HIV-positive men were selectively choosing to have unprotected sex with other HIV-positive men, where there is no risk of HIV transmission. We found that HIV-positive men were somewhat more likely to report that their boyfriend was also HIV-positive compared with HIV-negative and untested men, but unfortunately we do not have data on characteristics of each of the men's sex partners or the types of sex engaged in with them . Future research that examines the contexts and motivations associated with unprotected sex among HIV-positive men is greatly needed. Ironically, the current trend in prevention campaigns for gay men is toward special programs for HIV-negative men. Our findings clearly show that prevention programs must not ignore HIV-positive men. In addition, medical information about the dangers of reinfection with HIV (e.g., the possibility of transmitting strains of HIV that are more virulent or resistant to medications) is needed in order to evaluate the health risks of unprotected sex among HIV-positive men.
We found that a significant number of men were not aware of their HIV status. Twenty-five per cent of the men who tested HIV-positive in this study did not know they were HIV-positive, therefore HIV-prevention programs that target men by their self-identified HIV status may be problematic. Likewise, prevention approaches which encourage men to ask prospective partners about their HIV status may be misguided and dangerous. Our finding that a disproportionate number of the men who did not know they were HIV-positive were from an ethnic minority group suggests that increased efforts to understand and address HIV-testing issues among young gay and bisexual men of color are imperative.
Most of the predictors of unprotected sex were the same for both HIV-positive and HIV-negative men. The common predictors for both HIV-positive and HIV-negative men were: sexual impulsivity, having sex while high on alcohol or drugs, decreased enjoyment of safe sex, poor sexual communication skills, and high numbers of male sex partners. This suggests that the most effective and economical approach for prevention campaigns may be to focus on these core issues for all young gay men, i.e., dealing with tendencies toward unsafe sex when highly aroused, addressing substance-use issues, increasing the enjoyment value of safe sex, and improving sexual communication skills. Whether the approaches used to address these common issues should be differentially tailored to be maximally effective for HIV-negative and HIV-positive men is a matter for future research.
There were some differences in predictors for HIV-negative and HIV-positive men, however, suggesting that prevention programs may be enhanced by incorporating some HIV status-specific components. For example, issues regarding having a boyfriend may be especially important for HIV-negative men for whom involvement with a boyfriend may foster an unrealistic sense of safety and romanticism. Interestingly, having a boyfriend was not related to unprotected sex for HIV-positive men. It may be that the partner's HIV status is critical here. HIV-positive men may be less likely to have unprotected sex with a partner they know is HIV-negative, but more prone to do so with a boyfriend who is also HIV-positive. In addition, for HIV-positive men, interpersonal barriers may be particularly critical since orchestrating condom use may be complicated by concerns about revealing one's HIV status and possibly losing one's partner.
Obviously it is possible that there are important differences in the correlates of unprotected sex between HIV-negative and HIV-positive men that were not assessed in our survey, or that there are differences that are not amenable to assessment using a survey methodology. For example, issues of a more existential or emotional nature may influence differentially the sexual behavior patterns of HIV-negative and HIV-positive men. Qualitative research may be valuable in identifying unique issues for HIV-negative, HIV-positive, and untested men beyond those that were found in this study.
Interestingly, the men who had not taken the HIV-antibody test prior to their participation in this study showed a different pattern of risk behavior than the men who had been tested previously. The untested men had the fewest male sex partners and engaged in the least amount of unprotected anal sex (28%), compared with the HIV-positive (59%) and HIV-negative (35%) men. In addition, a much greater percentage of the untested men were bisexual (41%) than were the HIV-negative (16%) and HIV-positive men (4%). Having sex with women was found to predict safer sex with men for both untested and HIV-negative men. For many untested men, the decision not to be tested may actually be a realistic appraisal of their risk for HIV. Many of the untested men have less sex with men and may be more cautious when having sex with men than the men who had been tested, who were more gay-identified and more involved in the gay community. Since the untested men who were more ‘out of the closet’ and more involved in the gay community were more likely to have unprotected sex, it may be informative to examine that subgroup of untested men separately.
This study has focused on young gay men in San Francisco. Whether these findings would generalize to older gay men and gay men living in lower HIV-prevalence areas are questions for future research. Nonetheless, our findings clearly demonstrate that HIV-prevention efforts for young gay men must continue to be an urgent public health priority.