More than 30 years since the first reported cases of AIDS, HIV infection continues at epidemic proportions in the United States. The Centers for Disease Control and Prevention (CDC) estimates that approximately 50 000 new HIV infections occur each year and 1.2 million Americans are living with HIV [1,2]. Young MSM remain disproportionately affected by HIV, and this disparity continues to grow. Based on incidence surveillance, between 2006 and 2009, there was a 21% increase in new infections for people aged 13–29 years, driven by a 34% increase among MSM in this age group. This risk was especially notable for young MSM from minority racial or ethnic groups. During 2006 and 2009, there were more new HIV infections among young, black MSM than among any other age or racial group of MSM and there was a 48% increase over this time period in HIV incidence for this population. Among Hispanic MSM, the largest number of new infections occurred among 13–29-year-olds [2].

The burden of undiagnosed infection is particularly high for young MSM. MacKellar et al. [3] found that 77% of HIV-infected 15–29-year-old MSM (and 91% of HIV-infected young black MSM) tested in six US cities were unaware of their infection. Increasing the percentage of HIV-infected persons who are aware of their serostatus is a central tenet of HIV prevention and care, as persons who know that they are HIV infected can implement behavioral changes and begin clinical treatment that optimize clinical outcomes and reduce the risk of transmission to others [4–6].

The research to date highlights the growing burden of HIV on young MSM, particularly young, minority MSM. Although estimates of HIV incidence are an essential element for monitoring the epidemic, estimates of HIV incidence rates among young MSM in the United States remain limited. Data from HIV case surveillance can provide an estimate of the number of incident infections among this population; however, as the size of the population of young MSM is difficult to assess, these data cannot be used to calculate an incidence rate.

In addition, the characteristics and behaviors associated with HIV infection and diagnosis in this population require further investigation. Previous studies have identified individual-level factors that put young MSM at increased risk for acquiring HIV [7,8]. However, individual-level risk behaviors have not adequately explained the HIV-related racial disparities between black and white MSM [9,10]. Although partner-level factors may partially explain disparities, there has been limited research on the role of partner characteristics and relationship dynamics in young MSM’s risk for HIV [8]. Moreover, sociocultural and structural determinants of health, including community, political, educational, and economic factors are now understood to play a role in HIV risk [11].

In this study, we used data from the second round of the National HIV Behavioral Surveillance System (NHBS) among MSM (NHBS-MSM2), conducted during 2008, to estimate HIV prevalence, annual HIV incidence density, and factors associated with being HIV-infected among MSM aged 18–24 years.

In total, 28 468 persons were approached for participation at 626 venues; 12 474 (44%) persons were screened for participation, 11 074 (89%) of whom were eligible for the survey. A total of 1400 were not eligible: 1138 lived outside the MSA, 45 were aged less than 18 years, 71 were previous participants, 91 did not identify themselves as men, and 83 were not able to provide their consent to the survey (e.g. men who were intoxicated or who did not speak either English or Spanish well); exclusion categories were not mutually exclusive. Among the persons who were eligible, 10 645 (86%) completed the survey with valid responses. Of these, 9342 (88%) consented to HIV testing and had a valid test result, 8153 (87%) of whom reported male–male sex during the past 12 months. Of these, 1889 (23%) were aged 18–24 years and are included in this analysis. There were 203 individuals who were aged 18–24 years and eligible for the study who were excluded from this analysis because they did not have a complete survey or a valid HIV test result. Those excluded were more likely to be non-Hispanic black (P = 0.009).

Mean age was 21.4 years. Of the young men, 36% identified as non-Hispanic black, 29% as Hispanic, and 27% as non-Hispanic white (Table 1). Most (90%) young men reported completing at least a high school degree and almost half (47%) reported less than $20 000 in annual household income. The majority (76%) of young men identified as gay or homosexual. Most (65%) of the young men were recruited in bars and dance clubs.

During the past 12 months, almost half (46%) of participants reported using marijuana, over a quarter reported using stimulants [cocaine, ecstasy, poppers (amyl nitrate), or methamphetamine], whereas few (2%) reported injection drug use. A majority of participants had health insurance (60%), visited a healthcare provider during the past 12 months (75%), and reported being tested for HIV during the past 12 months (67%).

Mean age of estimated anal sexual debut with a male partner was 17.8 years (interquartile range, 16.5–19.5). Both current age of participant and age at first anal sex with a man were highest for whites. Of 1889 young MSM, 198 (10.5%) had a positive HIV test; 6.2% of whites, 16.5% of blacks, 6.9% of Hispanics, and 11.0% of MSM of other races tested positive for HIV infection. Overall, the estimated annual HIV incidence density was 2.9%. Incidence was 1.6% for whites, 5.1% for blacks, 1.9% for Hispanics, and 2.9% for MSM of other races (Table 2). By region, the incidence was 3.0% in cities in the Northeast, 3.8% in cities in the South, 3.0% in cities in the Midwest, 1.8% in cities in the West, and 1.2% in city of San Juan the US Territory of Puerto Rico.

Table 3 presents characteristics of the most recent sexual encounter with a man, stratified by whether he was a main or casual partner. Of all young MSM, 956 (52%) reported that their last partner was a main partner, and 870 (48%) a casual partner. Forty-two percent of casual partners compared to 31% of main partners were 5 or more years older. Forty-eight percent of casual partners and 23% of main partners were of unknown HIV status. Nineteen percent of main partnerships and 62% of casual partnerships were less than 3 months in duration. Concurrent sexual relationships were more likely to occur with casual partners. Among men with casual partners, 51% reported that during the time that they were having a sexual relationship with their last partner, they were also having sex with other people and 62% reported that their most recent partner definitely or probably had sex with other people during the same time. For men with main partners, the percentages were 28 and 29, respectively. Young MSM were significantly more likely to report anal sex or unprotected anal sex with a main partner and less likely to report use of drugs or alcohol with main partners, compared with their casual partners.

To identify risk factors that may contribute to HIV acquisition among young MSM, we excluded those who reported being previously diagnosed with HIV and analyzed characteristics associated with being HIV-infected in the remaining sample (Table 4). Of these, 136 (7.4%) young MSM tested positive for HIV infection. In multivariable analysis, young MSM were less likely to be HIV infected if they were of ages 18–19 [adjusted prevalence ratio (aPR) 0.39, confidence interval (CI) 0.22–0.70] compared to ages 23–24. Young MSM were more likely to be HIV infected if they were black (aPR 3.08, 95% CI 1.89–5.00) or of other race/ethnicity (aPR 2.02, CI 1.06–3.86) compared to white; had less than a high school degree (aPR 1.87, CI 1.08–3.23) compared to some college or more; used both alcohol and drugs before or during last sex (aPR 1.97, 95% CI 1.20–3.55); had their most recent HIV test more than 12 months ago (aPR 1.57, CI 1.18–2.10) compared to less than 12 months ago; or reported a visit to a medical provider in the past 12 months (aPR 1.84, CI 1.21–2.81).

Of the 136 young MSM who were HIV infected, 16 (11%) had never been tested, 32 (24%) had their most recent test more than 12 months ago, and 87 (64%) had been tested in the past 12 months.

HIV prevalence and estimated annual incidence density for young MSM were high. These findings were particularly pronounced for young black MSM, who had an HIV prevalence of 16.5% and an estimated annual incidence of 5.1%. This is comparable to the 5.9% HIV incidence found among 18–30-year-old black MSM in a recent longitudinal study conducted in six cities [20] and the 6.4% incidence found among a young sample of black MSM enrolled in a prospective HIV/STI incidence cohort study in Atlanta [21]. The similarity of these cohort study HIV incidence estimates to the incidence density calculated for the current study provides validation for an approach to estimating incidence that is less complex and more cost-effective.

The high prevalence and estimated annual incidence density among MSM aged 18–24 in our study underscores the need for this group to be considered a priority for HIV prevention. CDC currently recommends the use of a High-Impact Prevention approach to reduce new HIV infections. This approach uses a combination of targeted, scientifically proven interventions that are cost-effective and scalable [22]. For young MSM, this would include reaching the population with proven approaches such as condom distribution and behavioral interventions before they become infected [8,23]. In addition, it is imperative to implement effective strategies among HIV-infected young MSM to prevent transmission of HIV to others [8]. We found that one-third of sexually active young MSM had not been tested in the past 12 months; this suggests that efforts to improve the uptake of current CDC recommendations [24] that all sexually active MSM test for HIVat least annually, need to be expanded. However, the proportion reporting testing for HIV in the past year is higher than for MSM of other age groups [25]. Even among those young MSM who were HIV infected at the time of interview, 41% reported a negative HIV test result in the past year, supporting previous data suggesting that more frequent testing, such as every 3–6 months, of some groups of MSM may be warranted [26].

Similar to previous research [27–30], our analysis found that young MSM were more likely to have unprotected anal intercourse (UAI) with a main partner than a casual partner. This pattern may be linked to characteristics of main partnerships, such as greater trust and familiarity, the perception that condoms interfere with intimacy, and in some cases, a representation of relationship power differentials [29,31,32]. We also found that almost half of relationships with main partners were less than 6 months in duration. This is consistent with other studies [29,33] suggesting a tendency for serial monogamy among young MSM, increasing the likelihood of main partner change. The short duration of relationships combined with the higher occurrence of UAI could substantially increase HIV risk. Studies have suggested that the majority of HIV transmissions among MSM, and particularly young MSM, occur between main partners [33,34]. Our findings support the continuing examination of the association between partnership characteristics and HIV risk and suggest the importance of developing relationship-level HIV prevention interventions for young MSM.

Whereas UAI was more common with main partners, those who had a casual partner were more likely to use drugs or alcohol before or during the last sexual encounter, and an association was found between combined drug and alcohol use and being HIV infected. Although not examined in this analysis, previous research has found a relationship between the use of drugs and alcohol and sexual risk-taking [7,8]. Together, these findings emphasize the need for HIV prevention strategies aimed at young MSM to address the association between substance use and sexual risk-taking [7].

A high percentage of young men with casual partners reported that their partner had concurrent sexual relationships. Although research on concurrency among MSM is rare [35,36], mathematical modeling studies have suggested that differential rates of concurrency across populations may explain population differences in HIV prevalence, as well as racial disparities in HIV observed within the United States [37–40].

We also found that sociodemographic variables (including race and education status) were more strongly associated with HIV infection than were individual risk behaviors, emphasizing the continued need to look beyond individual-level factors to explain disparities in HIV infection [9,10]. Likewise, interventions focusing on individual behaviors need to be paired with community-level interventions that raise awareness about HIV and improve the community response to the epidemic; structural interventions that address social and economic disparities of HIV infection are also needed. As there remains a dearth of HIV prevention interventions targeting young MSM [41], it is imperative that future efforts be directed toward developing, testing, and implementing efficacious interventions for this population.

Among respondents who did not previously report being HIV infected, those who tested HIV positive during NHBS were more likely than those who tested HIV negative to have seen a healthcare provider in the past year. It is possible that some of these individuals saw a healthcare provider for symptoms related to their HIV infection or another sexually transmitted disease (STD), but were not offered HIV testing as part of their visit. A goal of the National HIV/AIDS Strategy is to increase the proportion of persons who are aware of their status [42], and CDC recommends HIV screening as part of regular healthcare in clinical settings [24].

This study has several limitations. First, participants were recruited at venues in 21 US cities with high AIDS prevalence and are not representative of all young MSM. Second, the data in this study are not weighted to account for variations in venue attendance or likelihood of being selected to participate in the survey. Third, because of the sensitive nature of HIV status, some participants who had previously been diagnosed with HIV infection may not have reported their positive HIV status, resulting in their inclusion in the analysis presented in Table 4. Fourth, all of our survey data were self-reported and may, therefore, have associated biases. Finally, our calculations of estimated annual incidence density are likely an under-estimate. Because we do not have the actual date of seroconversion, we calculated person-years at risk using the date of first positive HIV test. This likely inflates the denominator for this calculation and results in our estimate being a conservative one.