DESIGNING NOVEL PUBLIC HEALTH interventions to reduce human immunodeficiency virus (HIV) transmission requires an in-depth understanding of the target population. In addition, interventions that are tailored to incorporate cultural constructs are more likely to be effective.1,2 Moreover, an accurate description of the characteristics of HIV-infected individuals is required to inform policymakers and public health officials because they allocate prevention resources.
Available evidence suggests that the epidemiology of HIV infection in the United States has shifted over the past decade. Several serosurveys have documented an increase in the number of HIV cases among non-Hispanic blacks and Hispanic/Latinos relative to whites.3–5 This pattern is particularly pronounced in the southeastern United States, which is experiencing a disproportionate increase in HIV infections.3,6
Seroepidemiologic studies have documented a change in the risk behaviors associated with new HIV infections. Although men who have sex with men (MSM) still account for the majority of reported HIV infections, the incidence of HIV is increasing among women and men who have sex with women (MSW).3 From 2000 to 2003, 80% of black women acquired HIV through heterosexual contact, whereas only 27% of black men acquired HIV through heterosexual contact; the majority of black men (54%) reported being MSM as their transmission risk.7 The current risk classification hierarchies in most states rely on self-reported risk behaviors at the time of HIV testing. Few studies have assessed the validity of these self-reported risk behaviors. Several investigators have raised the possibility that men who have sex with men and women (MSM/W) may serve as a “bridge” for infection between these groups.8–11 Despite these concerns, the behavior and characteristics of HIV-infected MSM/W has received relatively little attention.
Network analysis can help explain transmission of HIV and other sexually transmitted infections (STIs)12 and supports the use of the network perspective in understanding HIV risk behaviors.13 Network structure has been found to be an important factor in infectious disease propagation,14 and there is evidence that a core group of people is responsible for the most of the maintenance and spread of STIs, including HIV.15 Moreover, sociosexual network analysis may have potential applications for future STI interventions targeted at high-risk groups.
As part of our attempts to understand the role that MSM/W play in the spread of HIV in young adults aged 18 to 30 years in North Carolina, we determined the prevalence of MSM/W among newly diagnosed HIV-infected men between January 1, 2000, and December 31, 2004, compared social and behavioral characteristics of this group with MSM and MSW, and examined the sexual networks associated with HIV-infected college students among these groups. In addition, we compared risk behaviors reported at the initial HIV voluntary counseling and testing (VCT) session with risk behaviors reported to the disease intervention specialists (DIS) during follow-up interviews of infected individuals.
We conducted a review of the North Carolina HIV/acquired immune deficiency syndrome (AIDS) surveillance records for men ages 18 to 30 years who were newly diagnosed with HIV infection between January 1, 2000, and December 31, 2004. Included in these records were all reports and interview records filed by DIS as well as VCT site information on patients diagnosed with HIV infection and contacts of patients as reported to the DIS. In North Carolina, HIV reporting is mandatory and DIS are assigned to investigate and interview any positive HIV test result reported to the state or local health department by a medical provider or clinical laboratory. Using information from these records, males were classified as MSM/W, MSM, or MSW.
This surveillance record review was considered part of an ongoing public health epidemiologic investigation sanctioned by the North Carolina Division of Health Services and therefore exempt from Institutional Review Board approval.
We compared risk behavior and demographic information of newly diagnosed HIV-infected MSM/W with MSM, MSM/W with MSW, and MSM/W college students with MSM/W noncollege students. In bivariate analyses, we calculated test statistics (t-test) and P values (2-tailed) for continuous variables and odds ratios (ORs), 95% confidence intervals (CIs), and chi-squared P values (2-tailed) for binary variables using Epi Info 2002 (Atlanta, GA) and SAS software (version 9.1.2; Cary, NC). Given the small sample size, exact methods were used to compare proportions.
To determine factors independently associated with MSM/W compared with MSM and MSW, respectively, we constructed 2 exploratory logistic regression models. Variables that had a P value <0.15 in bivariate analyses or were determined a priori to be important confounding factors (such as number of partners, race, ethnicity, and age) were entered into a multivariable logistic regression model to determine factors independently associated with being an HIV-infected MSM/W. Backward elimination was used to simplify the models using the likelihood ratio test, and all pairwise comparisons were tested for interaction (effect measure modification). Significant interactions at the P <0.10 level were considered important, and the results of these interaction terms are therefore presented stratified.
Sexual Networks Among College Students
Previously we reported an increase of HIV infections in college students and noted that HIV-infected college males were more likely to be MSM/W.16 Therefore, in addition to the comparisons of risk behavior and demographic information, we also evaluated the impact of MSM/W on college HIV sexual networks. We constructed a sexual partner network through information obtained from DIS interview records of college males and their sexual contacts. We examined potential sexual partner network links as defined by students’ self-reports to DIS of sexual partners at their college of enrollment or sexual partners at other colleges. This was achieved through manual review of all DIS interview records of index patients diagnosed with HIV infection and their contacts from January 1, 2000, through December 31, 2004.
DIS are assigned to investigate any positive HIV test result reported to the state health department. They review medical records to obtain demographic and clinical information about the reported index patient and attempt to contact the patient to conduct a voluntary, confidential, in-depth interview.17 Interviews with infected persons are conducted using a case report form that included demographic information, place of employment and/or school attendance, risk factors, sexual partners in the 12 months before diagnosis, and social acquaintances of persons with HIV infection. The case report form also includes an area for a narrative in which the DIS can record comments from the interview that are not captured elsewhere on the form (e.g., activities reported by interviewees and places where they socialize, meet partners, and have sex). Cluster interviewing is applied to gain information regarding the index patients’ social circles, which may include additional persons exposed through unobserved sexual networks.18
Comparison of Reported Risk Behavior in College Students
We compared the self-reported sexual risk behaviors given by the male college case patients during their initial visits to the VCT sites with information about risk behaviors reported during the DIS interviews. Male college case patients who reported sex with men during both the VCT and DIS interviews were compared with those men who had different risk behaviors recorded in both interviews.
Between January 1, 2000, and December 31, 2004, 1292 men aged 18 to 30 years were reported with newly diagnosed HIV infection in North Carolina. Of 1105 (85.5%) records available for review, 1013 records had valid information about sexual partners in the 12 months before diagnosis. Of these HIV-infected men, 573 (57%), 279 (27%), and 161 (16%) classified themselves as MSM, MSW, and MSM/W, respectively, during their interviews with DIS.
Comparing MSM/W with MSM, MSM/W were more likely to be black (OR = 1.84; 95% CI = 1.21–2.79; referent = white), enrolled in college at the time of diagnosis (OR = 2.10; 95% CI = 1.39–3.18), and report having sex partners who were also college students (OR = 2.42; 95% CI = 1.56–3.74) (Table 1). MSM/W were significantly more likely than MSM to engage in sex for drugs or money (OR = 2.81; 95% CI = 1.67–4.71).
In multivariable analysis, MSM/W were significantly more likely than MSM to meet their sex partners on college campuses and to report 2 to 5, 6 to 10, or >10 sex partners in the year before diagnosis (referent = 0–1 sex partners). In this analysis, we found that the effect of having MSM/W sex partners was not the same across strata of being enrolled in college, and the effect of exchanging sex for drugs or money was not the same across race categories. Among those without MSM/W partners, students were 1.86 (1.08–3.21) times as likely to be MSM/W compared with nonstudents. Among nonstudents, those with MSM/W partners were 2.14 (1.25–3.66) times as likely to be MSM/W compared with those without MSM/W partners. The odds of being MSM/W among those who exchange sex and are white is 9.93 (95% CI = 3.24–30.49) times the odds of being MSM/W among whites who do not exchange sex. The odds of being MSM/W among those who exchange sex and are black is 3.27 (95% CI = 1.39–7.67) times the odds among whites who do not exchange sex.
Comparing MSM/W with MSW, MSM/W were more likely to be younger (OR = 6.07; 95% CI = 3.22–11.43 for age 18–22 years, referent = age 29–30 years) and black (OR = 1.90; 95% CI = 1.21–2.99; referent = white) (Table 2). MSM/W were more likely to be coinfected with syphilis (OR = 3.00; 95% CI = 1.41–6.35), to be enrolled in college at the time of their diagnosis (OR = 17.10; 95% CI = 7.10–41.23), and less likely to have been incarcerated (OR = 0.45; 95% CI = 0.28–0.72). Factors that remained significant in multivariable analysis included enrollment in college, having sex partners who were also enrolled in college, syphilis coinfection, and reporting 2 to 5, 6 to 10, or >10 sex partners in the year before diagnosis (referent = 0–1 sex partners). We found significant interactions in this model with age group and having college student sex partners and incarceration and ethnicity. The odds of being MSM/W among those who have college student sex partners and are age 18 to 22 years is 17.61 (95% CI = 2.00–154.90) times the odds of being MSM/W with no college sex partner and age 29 to 30.
Of the newly diagnosed HIV-infected males, 44 (27.3%) of the 161 MSM/W were attending college at the time of their diagnosis (Table 3). All but 2 of the college MSM/W (95.5%) were black compared with 69.2% of the noncollege MSM/W (P = 0.02). The 44 MSM/W who attended college were more likely to meet sex partners on college campuses than the 117 MSM/W who were not college students (OR = 12.66; 95% CI = 3.34–48.07). However, venues for meeting sex partners were not limited to college campuses. Only a minority of the MSM/W college students (25.0%) reported meeting sex partners at college. These students also reported meeting partners at either bars or clubs (31.8%) and/or on the Internet (27.3%). Only 2 (4.5%) of the MSM/W college students reported having sex with a person known to have HIV/AIDS compared with 30 (25.6%) of the MSM/W noncollege students (OR = 0.14; 95% CI = 0.03–0.61).
Overall, MSM/W named a total of 293 partners; 191 (65%) were male and 102 (35%) were female. Fifty-five (34%) MSM/W were found to have a total of 66 infected partners—either previous positive or new diagnosis. Of these, 54 (82%) were men, 10 (15%) were women, and for 2 (3%) individuals, there was no gender information. Of note, 6 of the infected women were Hispanic, 3 were pregnant at the time of their diagnosis, and 2 were college students.
Sexual Networks Among College Students
We examined potential sexual partner network links, as defined by HIV-infected college males self-reports to DIS of sexual partners at their college of enrollment or sexual partners at other colleges or with noncollege students. The network illustrated a densely interconnected pattern of interactions among HIV-infected college students with MSM/W occupying a central position within the network. When only considering MSM and MSW, 6 discrete networks consisting of 17 schools, 58 students, and 5 contacts of students are evident (Fig. 1A). However, when the MSM/W are included, a single network connecting 26 schools, 95 students, and 8 contacts of students is elucidated (Fig. 1B). This network accounts for 67% of the total students diagnosed with HIV over the 5-year period.
Reported Risk Behavior of College Students
As part of our attempts to characterize the reporting of risk behaviors, we compared the self-reported sexual risk behaviors given by the college case patients during their initial visits to the publicly funded VCT sites with information about risk behaviors reported during the DIS visits. For 83 of the 142 students (58%), there was VCT data available for review. Nearly 50% of these students had discrepant information between their VCT data and their DIS interview. Of the 20 individuals who described themselves as either MSW or reported having no sexual risk factors at the time of their initial VCT visit, 40% reported that they were MSM/W and 40% MSM during the follow-up visit with the DIS.
Compared with college case patients who changed the reports of their risk behaviors (16 individuals), those MSM that reported sex with men during both the VCT and DIS interviews (61 individuals) were more likely to be coinfected with syphilis (11.5% vs. 0%), meet sex partners at bars/clubs (36.1% vs. 25.0%) or on the Internet (32.8% vs. 25.0%).
Among men in North Carolina aged 18 to 30 years who were newly infected with HIV, MSM/W were more likely to be young, black, and enrolled in college at the time of their HIV diagnosis. In addition, the MSM/W were found to be significantly different than both MSM and MSW in terms of both demographics and risk-taking behaviors.
There have been mixed reports concerning the role that male bisexual behavior may play in the heterosexual transmission of HIV.8,9 Previous studies examining bisexual behavior have important limitations. First, the studies are largely cross-sectional and do not examine reported behavior over time. Second, these studies often use a wide window of time to define bisexual behavior. Moreover, previous study samples of behaviorally bisexual men have usually been comprised of gay men who also have sex with women19–21 or have aggregated MSM and MSM/W into a single group.22,23 This study is unique in that it measures self-report of sexual behavior in the 12 months before infection, thus more accurately characterizing bisexual behavior. The finding that MSM/W were more likely than both MSM and MSW to be black is in agreement with other studies that found that black MSM are more likely than MSM of other races to identify themselves as bisexual and to be bisexually active.19,24–29
Several recent reports have focused on a group of young black men who identify themselves as MSW but who also have undisclosed sexual encounters with other men. These so-called “down low” (DL) men have received relatively little scientific scrutiny, yet have been the subject of much recent media attention. These descriptions in the popular media portray DL men as secretive and, because they do not perceive themselves to be at risk for HIV infection, unreceptive to standard HIV-prevention messages.30–33
The precise prevalence of the DL phenomenon among the HIV-infected black college students in our cohort is not known and was not addressed directly in our studies. We did find evidence of probable transmission of HIV infection to 2 female partners of college MSM/W and to 12 male partners of MSM/W. This would suggest the potential for bridging transmission of HIV by MSM/W to heterosexual women, especially in the black community; however, to date, this remains to be proven. Furthermore, of the 83 college case patients with both VCT and DIS information, 3 of 8 (37%) white men did not report their MSM activity at the time of testing compared with 13 of 75 (17%) black men. Despite the small sample size, this finding underscores the point that identifying oneself as heterosexual and having sex with men is not unique to black men.20,24,34,35
Network analysis offers unique insight into the transmission of HIV and STIs among individuals and within communities.36 A network approach facilitates identification of persons in an infected individual’s social group who may benefit from diagnostic screening and selective treatment.37 Our data support the notion that college MSM/W occupy a central position in the sexual network of HIV-infected male college students. Inclusion of MSM/W formed the “bridge” that linked 6 discrete networks of schools and students to form one large, dense (26 schools) complex network of HIV-infected male students stretching across North Carolina and into surrounding states. A significant number of MSM/W college students reported sex with anonymous partners (36%), thereby preventing complete definition of the network. In addition, because we only reviewed the charts of HIV-infected males, we are unable to comment fully on the impact that this network may have in the heterosexual black community. Nevertheless, this network demonstrates numerous interconnections and underscores the importance of HIV-prevention interventions directed toward individuals as well as social circles linked through different venues, both defined and through the Internet, where there may be an increased risk for sexual transmission.
The findings in this study are subject to several limitations. Because all of the information on new HIV infections in this investigation related to new diagnoses, it is possible that case reporting could be influenced by changes in provision or uptake of HIV testing in specific populations (e.g., at campus testing sites) over time. However, for the 5 years of case data, the proportion of cases diagnosed with early HIV infection (defined as the presence of HIV ribonucleic acid [RNA] detected in plasma in the presence of a negative HIV antibody test or a documented negative HIV antibody test and a subsequent positive HIV antibody test within 6 months) has increased from 4.5% in 2002 to 9.4% in 2004, indicating ongoing HIV transmission. In addition, the retrospective nature of this study limited our ability to validate interviewees’ responses with the DIS during their interview(s). Moreover, the social and sexual networks may be incomplete and biased because HIV-infected men may occupy positions in the networks that are different from HIV-uninfected men, and contact tracing is unavoidably incomplete. DIS information and VCT site data were available for only 58% of the HIV-infected college students, and thus our findings may not be applicable to students testing at non-VCT sites such as physicians’ offices where this information is not readily accessible. Finally, it should be noted that North Carolina is one of 10 states in which HIV VCT is confidential rather than offered as anonymous or confidential.38 It is possible that concerns about lack of confidentiality among our participants led to the large number of changes in reported risk factors.
To devise successful prevention messages and direct limited resources, we must recognize differences in HIV risk behaviors and transmission patterns within populations of MSM. A fairly hidden group of MSM/W is at high risk for HIV and poses a potential threat to the wider black community. College MSM/W appear to occupy a unique, central place in the network of HIV-infected students. Sexual contact with both men and women increases the likelihood that these college students may serve as bridge contacts, responsible for transmission of HIV and other STIs between sexual networks; however, this investigation did not find evidence to support black MSM/W as the primary source of HIV infection for black women. Is HIV infection in the black community a result of a single highly interconnected epidemic among MSM, MSMW, and heterosexuals? Or are there essentially 2 loosely connected epidemics—one among MSM and one among heterosexuals? Providing answers to these questions are critical to successful prevention efforts.
Grant support HRSA Special Projects of National Significance IH97H03789-01-01 and the UNC Chapel Hill CFAR.
1. Stevenson HC, Davis G. Impact of culturally sensitive AIDS video education on the AIDS risk knowledge of African-American adolescents. AIDS Educ Prev 1994; 6:40–52.
2. Herek GM, Gillis JR, Glunt EK, Lewis J, Welton D, Capitanio JP. Culturally sensitive AIDS educational videos for African American audiences: Effects of source, message, receiver, and context. Am J Community Psychol 1998; 26:705–743.
3. Centers for Disease Control and Prevention. STD Surveillance Report. 2003.
4. Centers for Disease Control and Prevention. HIV/AIDS among African Americans.
5. Laumann EO, Youm Y. Racial/ethnic group differences in the prevalence of sexually transmitted diseases in the United States: A network explanation. Sex Transm Dis 1999; 26:250–261.
6. Karon JM, Fleming PL, Steketee RW, De Cock KM. HIV in the United States at the turn of the century: An epidemic in transition. Am J Public Health 2001; 91:1060–1068.
7. Centers for Disease Control and Prevention. Diagnoses of HIV/AIDS—32 States, 2000–2003. MMWR Morb Mortal Wkly Rep 2004; 53:1106–1110.
8. Ekstrand ML, Coates TJ, Guydish J, Hauck W, Collette L, Hulley S. Are bisexually identified men in San Francisco a common vector for spreading HIV infection to women? Am J Public Health 1994; 84:915–919.
9. Chu SY, Peterman TA, Doll LS, Buehler JW, Curran JW. AIDS in bisexual men in the United States: Epidemiology and transmission to women. Am J Public Health 1992; 82:220–224.
10. Doll LS, Beeker C. Male bisexual behavior and HIV risk in the United States: Synthesis of research with implications for behavioral interventions. AIDS Educ Prev 1996; 8:205–225.
11. Wood RW, Krueger LE, Pearlman TC, Goldbanm G. HIV transmission: Women’s risk from bisexual men. Am J Public Health 1993; 83:1757–1759.
12. Friedman SR, Neaigus A, Jose B, Curtis R, et al. Sociometric risk networks and risk for HIV infection. Am J Public Health 1997; 87:1289–1296.
13. Kelly JA, Murphy DA, Sikkema KJ, et al. Randomised, controlled, community-level HIV prevention intervention for sexual-risk behaviour among homosexual men in US cities. Lancet 1997; 350:1500–1505.
14. Potterat JJ, Rothenberg RB, Muth SQ. Network structural dynamics and infectious disease propagation. Int J STD AIDS 1999; 10:182–185.
15. Aral SO. Behavioral aspects of sexually transmitted disease: Core groups and bridge populations. Sex Transm Dis 2000; 27:327–328.
16. Hightow LB, MacDonald PDM, Pilcher CD, et al. The unexpected movement of the HIV epidemic in the southeastern United States: Transmission among college students. J Acquir Immun Defic Syndr 2005; 38:531–537.
17. Centers for Disease Control and Prevention. Increases HIV partner counseling and referral services, including partner notification—North Carolina, 2001. MMWR Morb Mortal Wkly Rep 2003; 52:1181–84.
18. Centers for Disease Control and Prevention. Program Operation Guidelines for STD Prevention. 2000.
19. Adib SM, Joseph JG, Ostrow DG, Tal M, Schwartz A. Relapse in sexual behaviour among homosexual men: A 2-year follow-up from the Chicago MACS/CCS. AIDS 1991; 5:757–760.
20. Davies PM, Hickson FC, Weatherburn P, Hunt AJ. Sex, Gay Men and AIDS. London: Taylor and Francis, 1993.
21. Stall R, Ekstrand M, Pollack L, McKusick L, Coates TJ. Relapse from safer sex: The next challenge for AIDS prevention efforts. J Acquir Immun Defic Syndr 1990; 3:1181–1187.
22. Doll L, Myers T, Kennedy M, Allman D. Bisexuality and the HIV risk: Experiences in Canada and the US. Annu Rev Sex Res 1997; 8:102–148.
23. Kalichman SC, Roffman RA, Picciano JF, Bolan M. Risk for HIV infection among bisexual men seeking HIV-prevention services and risks posed to their female partners. Health Psychol 1998; 17:320–327.
24. Montgomery JP, Mokotoff ED, Gentry AC, et al. The extent of bisexual behavior in HIV-infected men and implications for transmission to their female sex partners. AIDS Care 2003; 25:829–837.
25. Doll LS, Petersen LR, White CR, et al. Homosexually and nonhomosexually identified men who have sex with men: A behavioral comparison. J Sex Res 1992; 29:1–14.
26. Easterbrook PJ, Chmiel JS, Hoover DR, et al. Racial and ethnic differences in human immunodeficiency virus type-1 seroprevalence among homosexual and bisexual men. Am J Epidemiol 1993; 138:415–429.
27. Kramer MA, Aral SO, Curran JW. Self-reported behavior patterns of patients attending a sexually transmitted disease clinic. Am J Public Health 1980; 70:997–1000.
28. Torian LV, Makki HA, Menzies IB, et al. HIV infection in men who have sex with men, New York City Department of Health sexually transmitted disease clinics, 1990–1999: A decade of serosurveillance finds that racial disparities and associations between HIV and gonorrhea persist. Sex Transm Dis 2002; 29:73–78.
29. Millet G, Malebranche D, Mason B and Spikes P. Focusing ‘down low’: Bisexual black men, HIV risk and heterosexual transmission. J Natl Med Assoc 2005; 97:52S–59S.
30. Steinhauer J. Secrecy and stigma keep AIDS risk high for gay black men. New York Times February 11, 2001:37.
31. Denizet-Lewis B. Double lives on the down-low. New York Times Magazine August 3, 2003:28–33, 48, 52–53.
32. Vargas JA. HIV-positive without a clue: Black men’s hidden sex lives imperiling female partners. Washington Post April 3,1998:B01.
33. Harris EL, Roberts T. Passing for straight. Essence 2004; 156:161, 210.
34. Goldbaum G, Perdue T, Wolitski R, et al. Differences in risk behavior and sources of AIDS information among gay, bisexual and straight-identified men who have sex with men. AIDS Behav 1998; 2:13–21.
35. Ross MW, Essien JE, Williams ML, et al. Concordance between sexual behavior and sexual identity in street outreach samples of four racial/ethnic groups. Sex Transm Dis 2003; 30:110–113.
36. Doherty IA, Padian NS, Marlow C, Aral SO. Determinants and consequences of sexual networks as they affect the spread of sexually transmitted infections. J Infect Dis 2005; 191(suppl 1):S42–54.
37. Rothenberg R, Narramore J. Commentary: The relevance of social network concepts to sexually transmitted disease control. Sex Transm Dis 1996; 23:24–29.
38. Henry J. Kaiser Family Foundation. HIV/AIDS Policy Fact Sheet: HIV Testing in the United States. June 2004.