NHBS monitors HIV-associated behaviors and HIV prevalence in 20 cities with high AIDS burden among 3 populations at high risk for infection: MSM, people who inject drugs, and heterosexual adults at increased risk for HIV infection [8]. Cross-sectional data reported in this analysis are from MSM recruited for interviews and HIV testing through venue-based sampling (VBS) in 2014. Characteristics of this sample have been previously published [9]. NHBS VBS procedures have also been previously published and are briefly summarized here [8, 10]. First, staff identified venues frequented by MSM (eg, bars, clubs, gyms, parks, street locations, and social organizations) and days and times when men frequented those venues. Venues in which 50% of men attending were MSM were eligible for inclusion. Site staff determined venue eligibility through secondary data review, interviews, focus groups, or observations. Second, venues and corresponding day/time periods were selected randomly for recruitment events. Third, staff members systematically approached men at recruitment events to screen for eligibility (aged ≥18 years, lived in a participating city, ever had sex with another man, and able to complete the interview in English or Spanish). Interviews were conducted by trained interviewers using a standardized questionnaire. NHBS activities were approved by local institutional review boards (IRBs) in each participating city. NHBS activities were determined to be research in which the CDC was not directly engaged and did not require review by the CDC IRB.

MSM with complete and valid interview data who tested HIV negative in NHBS and who were currently sexually active, defined as having ≥1 male partner in the past 12 months, were included in analyses. Validity was assessed by the interviewer’s confidence in the respondent’s answers. Interviewers received in-person training on administering the questionnaire, and interviews they marked invalid (16 total interviews [0.2%]) were excluded from analysis. The 3 main outcomes were self-reported willingness to take PrEP, PrEP use in the past 12 months, and indications for PrEP. Willingness to take PrEP was ascertained from the question, “Would you be willing to take anti-HIV medicines every day to lower your chances of getting HIV?” PrEP use in the past 12 months was ascertained from the question, “In the past 12 months, have you taken anti-HIV medicines before sex because you thought it would keep you from getting HIV?” Indications for PrEP were adapted from the CDC clinical guidelines and were defined as (1) having ≥2 male partners in the past 12 months and either reporting anal sex without a condom or a bacterial sexually transmitted disease (STD; ie, chlamydia, gonorrhea, or syphilis) in the past 12 months, or (2) having 1 main, HIV-infected male partner in the past 12 months.

Bivariate analyses were conducted to explore differences by demographic characteristics (race, age, education, income, and insurance) for the 3 outcomes. We also examined differences by sexual behaviors for willingness to take PrEP and PrEP use, but not for indications for PrEP, as sexual behaviors were used to define this variable. Adjusted prevalence ratios and 95% confidence intervals were calculated from log-linked Poisson regression models with generalized estimating equations. Separate models were built for each covariate and outcome combination. All models were clustered on VBS recruitment event. Covariates associated with the outcomes in bivariate analyses with P values <.10 were considered as confounders in the multivariable models and backward elimination was used to reduce models. Willingness to take PrEP models was adjusted for age and city. PrEP use models were adjusted for race/ethnicity, health insurance, and city. PrEP indication models were adjusted for race/ethnicity, age, and city. Analyses were conducted using SAS software, version 9.3 (SAS Institute Inc, Cary, North Carolina).

Of 9640 sexually active MSM who consented to an HIV test as part of their participation in NHBS, 6847 (71%) tested negative and were eligible for inclusion in analyses. After excluding those with missing data (231 MSM missing data on PrEP willingness, 131 additional MSM missing data on risks used to assess indications for PrEP, and 2 additional MSM missing data on PrEP use), 6483 MSM were left for analysis.

Whereas more than half of MSM were willing to take PrEP, only about 4% were actually using PrEP in 2014. There was no difference in willingness to take PrEP between black and white MSM. Willingness was associated with riskier sexual behaviors, as was PrEP use. PrEP use, although low, was higher among white MSM compared with black MSM and among those with greater education and income levels. White MSM were also more likely to have an indication for PrEP, as were those in younger age groups. Young, black MSM, despite being at higher risk of HIV acquisition, were less likely to have an indication for PrEP compared with young MSM of other races/ethnicities.

Our findings on willingness to use PrEP are consistent with a recent review of PrEP acceptability as measured by willingness [11]. Of 8 studies that offered numerical findings on willingness, all reported levels of 50% or higher. While one study included in the review found no association between willingness and risk behavior, 4 studies reported an association between willingness to take PrEP and unprotected anal intercourse and an additional study reported an association between willingness and inconsistent condom use.

PrEP use was most common among San Francisco and Washington, D.C., participants, which may be due to their participation in the PrEP Demonstration Project, an open-label PrEP delivery project for MSM and transgender women [12]. The project was launched in late 2012 and completed enrollment in 2014. Although only a few hundred MSM were enrolled at each site (300 in San Francisco, 157 in Miami, and 100 in Washington, D.C.), the Demonstration Project likely increased demand for PrEP beyond the participants, as both San Francisco and Washington, D.C., had enrollment waitlists throughout the project [13]. Miami is also a PrEP Demonstration Project city, but had a much lower prevalence of PrEP use among participants in NHBS (1.6%). This may be due to less demand for PrEP in Miami compared with the other 2 demonstration cities or to the fact that MSM in Miami were less likely to have an indication for PrEP (25% of those approached for the PrEP Demonstration Project were ineligible for behavioral or medical reasons in Miami compared with 8% in San Francisco and 12% in Washington, D.C.) [13]. In our analysis, MSM in Miami were also less likely to have an indication for PrEP (55% compared with 69% in both San Francisco and Washington, D.C.; data not shown).

A recent CDC Vital Signs publication that estimated PrEP indications according to the 2014 guidelines using nationally representative data found that approximately 25% of adult, sexually active MSM would have an indication for PrEP to prevent HIV acquisition [14]. The estimate of indications for PrEP in NHBS was more than double this estimate, at 57%. This difference is likely due to NHBS operating in large urban areas with high AIDS prevalence and to the sampling method for MSM, which recruits venue-using MSM who may report more risky sexual behaviors compared with the general population of MSM. Other reasons for the lower percentage indicated in the nationally representative sample used in the Vital Signs analysis could be the lack of data on MSM with an HIV-infected sex partner or the lack of inclusion of MSM who inject drugs, both of which likely led to a lower percentage having indications for PrEP compared to our analysis. Most MSM (95%) recruited for a web-based HIV Behavioral Surveillance System in cities that also conduct NHBS reported attending a venue in the past 12 months and could have been potentially recruited to participate in NHBS [15]; we therefore believe NHBS is fairly representative of MSM in core urban areas, according to this comparison. Because program planning for PrEP rollout is currently occurring in these core urban areas, the NHBS estimate of indications for PrEP may be useful for planning purposes.

Our findings on the disparity in PrEP indications by race/ethnicity are consistent with data from an observational cohort of HIV-uninfected MSM in Atlanta, which found that 65% of black MSM and 88% of white MSM for whom behavioral data were available met indications for PrEP based on the criteria in the guidelines [15]. Black MSM were less likely to have indications for PrEP in both this analysis and our analysis because they reported fewer risk behaviors. Because the increased risk of HIV acquisition for black MSM is associated with social determinants such as poverty and racial discrimination rather than higher-risk sexual behaviors [15, 16], current clinical guidelines will not indicate PrEP for some MSM at highest risk. While the guidelines recommend that clinicians consider offering PrEP as one prevention option to “those whose sexual or injection behaviors and epidemiologic context place them at substantial risk of acquiring HIV infection” (emphasis added) [7], epidemiologic context is not taken into account in the operationalization of the guidelines.

A 2012 mathematical model of PrEP impact found that a strategy that targets PrEP to the 20% of MSM considered to be at highest risk for HIV infection prevented twice as many infections over 20 years and at better economic value than a strategy that provides PrEP to 20% of all HIV-uninfected MSM regardless of risk [17]. In the model, the highest-risk group was defined by behavioral risk factors (5 annual partners on average) as well as population-level factors (initial HIV prevalence of 20% and initial annual incidence of 2.3%). Educating providers about how to consider incidence and prevalence data on the MSM subpopulations most affected by HIV in their communities together with behavioral indications for PrEP could ensure that those at highest risk are offered PrEP.

In addition to being less likely to have indications for PrEP when visiting a healthcare provider, it is important to note that significant barriers to accessing healthcare, and therefore PrEP, exist for black MSM. Some of the racial disparity in healthcare access is attributable to black MSM having lower prevalence of health insurance compared with white MSM [15, 18]. If PrEP is then offered more frequently to those who have insurance and access healthcare, the existing differences by race/ethnicity in PrEP use and ultimately HIV incidence may potentially widen [4]. Ongoing changes in the US healthcare system through the Patient Protection and Affordable Care Act of 2010 offer opportunities to improve use of clinical preventive services by MSM and to potentially reduce the disparity in healthcare access [19]. In the meantime, clinicians and MSM should be informed of the medication assistance program and the drug copay program that provide PrEP to uninsured or underinsured individuals, respectively (https://start.truvada.com/paying-for-truvada) [20].

Our analyses are subject to several limitations. First, not all of the 2014 PrEP guidelines could be matched directly to variables collected in NHBS. We used having ≥2 partners in the past 12 months as a proxy for being in a nonmonogamous relationship, but these multiple partnerships may not have overlapped in time and may therefore have been monogamous. Thus, NHBS data may have overestimated indication for PrEP. Similarly, NHBS variables often ask about behavior in the past 12 months, whereas the PrEP guidelines specify a time period of the past 6 months, also leading to an overestimation of indications for PrEP. Second, NHBS is not a nationally representative sample, so results may not be generalizable to all MSM in participating cities or to all cities. Third, our findings are based on self-reported data and might be subject to social desirability and recall bias. Social desirability may lead to overestimation of prevalence estimates for willingness to take PrEP and PrEP use, while the direction of bias due to recall error is unknown. Finally, data were not weighted to account for the complex sampling methodology used to recruit MSM. Point estimates may therefore be biased by over- or underrepresented subgroups of the population.

PrEP use was not common among MSM in NHBS in 2014, although a racial disparity exists between black and white MSM with respect to PrEP use in our sample. This disparity may be attributable to racial differences in healthcare access rather than a lack of interest in PrEP as both black and white MSM reported high willingness to take PrEP. To prevent racial disparities in PrEP use, it will be important for clinicians to consider epidemiologic context in addition to risk behaviors when considering PrEP indications. This will increase the likelihood that PrEP is offered to those at highest risk of HIV acquisition and that the greatest impact on HIV incidence is achieved.