In the United States, male-to-male sexual contact remains the highest risk group for HIV infection comprising 82% of all newly diagnosed infections among adult and adolescent males in 2015.1 The United States' National HIV/AIDS Strategy has set a goal of reducing new HIV infections by 25% by 2020. To accomplish this goal, the strategy has recommended increasing variation in evidence-based approaches to prevention to “ensure the most effective prevention strategies are prioritized and widely implemented.”2
There are a number of HIV prevention strategies that are employed by men who have sex with men (MSM) including behavioral strategies (condom use, monogamy/abstinence seroadaptive strategies), and chemoprophylactic strategies (preexposure prophylaxis, postexposure prophylaxis, treatment as prevention). Among behavioral strategies, consistent condom use remains the primary HIV prevention message targeted to MSM; however, only 25%–28% of MSM report using condoms consistently.3–5 In the absence of consistent condom use, HIV seroadaptive practices may be used to reduce the risk of transmission or acquisition of HIV. Seroadaptive behaviors (ie, serosorting and seropositioning) have been shown to be both intentional and used more consistently than condoms.4 The Amsterdam Cohort Study, a longitudinal study exploring the efficacy of serosorting strategies, found that 11% of MSM practiced serosorting and that both consistent condom use and serosorting independently reduced the rate of HIV infection compared with those who used neither approach.6 Treatment as prevention (TasP), a chemoprophylactic prevention strategy, reduces plasma viral load of HIV-positive individuals below the limit of detection, which leads to a decrease in transmission risk. HIV Prevention Trials Network Study 052 demonstrated a 96% reduction in HIV transmission among serodiscordant couples where the HIV-positive partner had suppressed viral load.7 Despite the wide usage of these strategies, HIV infection rates among adult and adolescent males in the United States attributed to male-to-male sexual contact have remained stable from 2010 to 2014.1
The most novel strategy is the use of Truvada (Gilead Sciences, Foster City, CA) as HIV preexposure prophylaxis (PrEP), approved by the United States Food and Drug Administration (FDA) in July of 2012. The FDA reviewed evidence from several clinical trials conducted from 2010 to 2011, which demonstrated proof of concept for PrEP. Each of these studies identified adherence as a primary obstacle to efficacy,8–11 In the original clinical trials when analyzed per protocol, PrEP use resulted in 85%–92% reduction in infections,10,11 and subsequent observational studies have reported high adherence and efficacy of PrEP,12–15 but overall low uptake. Shortly after approval, both the World Health Organization and the Centers for Disease Control and Prevention (CDC) recommended the use of PrEP among “high risk” MSM in conjunction with other prevention strategies such as condom use, abstinence, monogamy, or treatment as prevention (TasP).16,17 Despite the wealth of evidence in favor of PrEP, uptake of the drug has been low. According to data from the National Pharmacy Database, which represents approximately 39% of pharmacies in the United States, 8512 individuals were prescribed PrEP by March of 2015.18 A 2015 Morbidity and Mortality Weekly Review shows that 24.7% of MSM in the United States without HIV infection, aged 18–59 years, with at least 1 sex partner in the past year, had indications for PrEP, including having at least 2 sex partners in the past 12 months, condomless anal intercourse (AI) and/or the diagnosis of an STI within the last 12 months.19
Considering the wide range of HIV prevention strategies employed by MSM and the interactions between strategies, analytic approaches are needed to assess their individual and joint contributions in the prevention of transmission or acquisition of HIV, acknowledging that multiple prevention strategies may be used simultaneously. Simulations, and especially agent-based models, can naturally be explored using a counterfactual framework, providing keen insights into the effects of the addition, subtraction, or combination of interventions that cannot be explored experimentally.20 Indeed, recently, several agent-based models have examined the use of PrEP on the epidemic among MSM with varying targeting strategies. Jenness et al21 modeled scale-up of PrEP according to the Centers for Disease Control and Prevention “high risk” MSM guidelines, and noted that at ambitious coverage levels (≥40%), a substantial number of infections can be averted (33% or more) in a decade. Other simulations have shown similar U.S. population benefits postscale-up.22–25 While these models considered multiple prevention strategies, their focus on PrEP meant the separate contributions from other strategies were not assessed. As such, considering PrEP efficacy in the context of condom use, seroadaptive behaviors, and TasP, is critical for determining the population-level impact of PrEP on HIV incidence relative to other prevention methods.26
This study explores the impact of increasing proportions of PrEP uptake on HIV infection, alone or in concert with other prevention strategies, on HIV incidence among a hypothetical closed cohort of urban MSM in the United States. Model parameters are largely drawn from literature on MSM in the United States. It is a complex systems approach using a counterfactual, agent-based simulation model on a closed population of high-risk MSM. The aim of this study is to explore how varying levels of PrEP uptake proportionally reduce HIV infections in the context of TasP, condom use, and seroadaption. Our approach to modeling represents a worst-case scenario for HIV infection and a best-case scenario for PrEP efficacy to achieve estimates that represent the most benefit we could achieve with a scale-up of PrEP.
We developed a counterfactual, agent-based, dynamic model for sexual behavior and HIV infection with 4 components: (1) baseline data generation, (2) testing and treatment, (3) sexual partner selection, and (4) infection determination. Components 2–4 were repeated over the course of 365 “days” to represent a full calendar year. Baseline characteristics, described below, were generated for a closed cohort of 10,000 high-risk MSM who were then followed over the course of 365 days as they engaged in sexual activity with one another. For the purposes of this study, we defined “high-risk MSM” as men who were at high risk for either HIV infection or transmission, had at least 1 sex partner in the year, engaged in condomless AI (with the exception of the addition of condom use as an intervention), and did not have a regular, monogamous partner. We repeated this model 6 times increasing PrEP uptake among HIV-negative MSM from 0% to 1%, 5%, 10%, 15%, 20%, or 25% of the population and modeled the effect of PrEP on HIV incidence. Each of these runs represents the same population with the same baseline characteristics and the same sexual network at each time point. We then repeat these 7 models to include other HIV-prevention strategies including TasP, condom use, seroadaption, and all combinations thereof. The resulting HIV incidences from each set of models represent what would have happened if these strategies were employed as well. Given the susceptibility of our models to network characteristics, we repeated these models using a total of 20 sets of baseline and sexual network data to obtain variance estimates that arise from the stochastic processes of data generation. For example, if one generation of data, by random selection, has a super-spreader of HIV (an HIV-infected, but untested insertive partner with a large number of sexual partners at baseline), this may result in a higher incidence for that generation of data.
Baseline Data Generation
Baseline characteristics were generated on the basis of recently published data on HIV infections, sexual behaviors, HIV testing, and both biological and social determinants of HIV infection among MSM (Table 1). With the exception of number of sex partners, all characteristics were derived from a random uniform distribution.
In short, 19% of the population was randomly assigned to be HIV-positive, 44.0%, of whom were unaware of their HIV status, and 36.9% of those assigned to be HIV-negative were determined to not having been tested for HIV in the past year.27 Based on 1980 data of circumcision among boys in the United States, we determined that 64.7% of the population was circumcised.28 We used a gamma distribution to represent the number of sexual partnerships over the course of the year using the methods developed by Omori et al29 and the mean and standard deviation of number of annual sex partners from the Project Male-Call study.30 The distribution had a shape of 0.5 and a scale of 10 and 1 sex partner was added to the resulting distribution to ensure that all individuals had at least 1 sex partner. The resulting distribution had a median of 3 sex partners with a minimum of 1 and a maximum around 110 sex partners.
With respect to sexual positioning, most MSM who engage in AI act as both insertive and receptive partners31 and among those who do identify as having a sexual positioning preference (“top” or “bottom”), these identities do not necessarily indicate insertive- or receptive-only positioning.32,33 As such, we generated a random uniform variable to represent probability of insertive or receptive anal intercourse, described later.
We determined future HIV testing behaviors on the basis of population-based data from the Kaiser Family Foundation.5 This study showed that among HIV-negative MSM, 19% reported being tested in the past 6 months, 11% in the past 6–12 months, 36% more than 12 months before, and 30% had never been tested. As such, we determined that 30% of our population would not be tested, 19% would be tested within the first 110 days of our simulation (representing those who get tested frequently), 11% would be tested within the second 110 days (representing those who get tested once per year), and 36% would be tested within the third 110 days (representing those who infrequently get tested). The exact day of HIV testing was randomly chosen for each individual within the specified time period.
We also randomly assigned individuals into the various prevention scenarios: for TasP scenarios, 42% of HIV-diagnosed individuals were assigned to be ARV-suppressed at baseline.34 For condom-use scenarios, 25% of the population was randomly assigned to use condoms 100% of the time, 24% was assigned to never use condoms, and 21%, 16%, and 14% were assigned to use condoms 75%, 50%, and 25% of the time, respectively.5 Seroadaptive practices were determined on the basis of knowledge of HIV status.35
All characteristics were independent of one another, with the exception of HIV testing (those who already knew their HIV status were not tested for HIV further) and assignment of PrEP and TasP (only available to those who were HIV-negative and HIV-positive and aware of their status, respectively).
We determined the probability of seeking a sex partner as the ratio of the remaining number of sex partners an individual would have in a given year and the number of days remaining in the year. This ratio was compared with a randomly generated uniform variable, such that those whose ratio was greater than the randomly generated variable were eligible to be matched to a sex partner. Eligible men were randomly sorted and matched on a 1:1 basis. Any individual not matched was removed and considered as having failed to find a sex partner. Since we only explored casual partnerships, sexual partnerships did not extend beyond any given day, although individuals were not constrained from matching with previous partners.
Once matched to a partner, sexual positioning for anal intercourse was determined on the basis of each individual's sexual positioning preference. The partner with the larger sexual positioning preference variable was determined to be the insertive partner. If the difference between the 2 random numbers was less than or equal to 0.15, then both partners acted in both positioning roles. Finally, exposure and infection were evaluated among serodiscordant dyads.
All dyads engaged in AI and only AI was assessed for HIV transmission. Risk of HIV infection was determined on the basis of the positioning of the HIV-uninfected partner and the transmission probability of the infected partner. We considered 2 levels of transmission probability based on viral load: primary HIV infection and chronic HIV infection. At the start of this simulation, all HIV-infected individuals are considered to have chronic-levels of HIV infection. Transmission probability is based on the simulation work of Punyacharoensin et al.36 HIV infection was assessed by comparing the transmission probability of infection with a random variable generated for each dyad. If the randomly generated value fell below the transmission probabilities, the HIV-negative individual was infected. In the event that both partners acted in receptive and insertive roles, exposure during the receptive act was assessed first. If this did not result in an infection, a second random exposure variable was generated and the exposure for the insertive act assessed.
For those who were infected, we evaluated whether or not the infection would have been prevented by each of the prevention measures independently, depending on the scenario being evaluated and the use of that prevention measure by either or both partners. In our scenario with 0% of the population randomized to PrEP and with no other prevention strategies employed, no infections were prevented. Otherwise, each prevention strategy was used to assess whether or not an infection occurred only if that prevention strategy was being employed in that analysis.
If the infected partner was assigned to PrEP, or the HIV-positive partner had ARV-suppressed viral load, then the infection was prevented. Condom use was assessed for each partner by comparing the probability of condom use with a variable generated randomly from the uniform distribution for each partner. If that random variable fell below the probability of condom use for either partner, then a condom was used. If a condom was used, we allowed for a 29.5% condom failure rate.37 Under any seroadaptive scenario, we assumed 100% HIV disclosure on the basis of a partner's knowledge of his HIV status. For those dyads with a successful transmission, we assessed the compatibility of serosorting strategies to determine serocompatibility and sexual positioning. If the assigned sexual positioning was not compatible with seropositioning strategies, then exposure was reassessed using the preferred sexual position. Further, if the serostatus of either partner was incompatible with the other, and either partner engaged in serosorting, then it was assumed that sex did not occur between the 2 and infection did not occur. Importantly, these individuals were not reassigned to a new partner, they simply dissolved the existing partnership without engaging in sex. If an infection was not prevented, then an individual's HIV status was changed and probability of transmission set to acute HIV infection. Infections prevented by each scenario were recorded using counter variables for each individual prevention method and overall.
Testing and Treatment
HIV testing, antiretroviral assignment, and viral load determination were assessed before partner matching on each individual day. Any individual randomly assigned to be tested at time t had their HIV status revealed, and any seroadaptive strategy was updated to reflect the individual's knowledge of HIV status. Under any TasP scenario, those who were diagnosed with HIV had a 42% probability of being assigned antiretroviral therapy (ART). Individuals moved between HIV transmission probabilities such that 30 days after infection, transmission probability was reduced to chronic HIV levels and 30 days after the initiation of antiretroviral therapy, individuals were considered virally suppressed.
We conducted 2 sensitivity analyses exploring the effects of the scale of the sex partner distribution including a scale of 5 (maximum of 30 sex partners) and a scale of 15 (maximum of 180 partners) to determine the influence of the number of sex partners on the sexual network. We also conducted 3 additional analyses varying the baseline HIV prevalence from 19% to 1%, 5%, and 10% (Supplemental Digital Content, http://links.lww.com/QAI/B89).
All modeling was conducted using SAS 9.2 (Cary, NC) and verified by an independent model construction using R 3.2.1 (R Foundation for Statistical Computing, Vienna, Austria). Source code in R can be accessed here https://doi.org/10.5281/zenodo.846313.
A mean of 103.2 (95% CI: 99.7 to 106.7) infections occurred in 10,000 subjects over the course of a year with no prevention methods (Fig. 1). PrEP assigned to 1% of the HIV-negative population prevented 1.6% of infections, while 30.7% of infections were prevented when 25% were assigned to PrEP. In the absence of PrEP, TasP alone prevented 24.7% (22.6–26.8) of infections, seroadaptive strategies alone prevented 37.7% (36.0–39.5), condom use alone prevented 48.8% (47.3–50.3), and all 3 strategies prevented 71.7% (70.5–72.9) of infections (Table 2). Under each scenario, increasing PrEP resulted in an increased number of infections prevented, but the proportion of infections prevented from PrEP-alone diminished with the addition of other prevention methods. For example, when paired with condom use, 25% PrEP resulted in a 12.4% reduction in infections (a 2.5-fold reduction) compared with the 30.7% prevented by 25% of the population assigned to PrEP alone. Using both seroadaption and condom use, 25% PrEP had a 5.05-fold reduction in preventions (30.7% vs. 6.6%) (Fig. 2).
Table 3 describes the mean proportion of individuals using various combinations of prevention methods across strata of baseline HIV status and infection/transmission for scenarios including 25% PrEP uptake and all prevention methods. On average, 16 individuals seroconverted over the course of this scenario and 15 individuals who were HIV-positive at baseline transmitted HIV. Those who seroconverted were more likely to be a receptive partner than those who did not (33.9% vs. 25.1%), were less likely to always use condoms (18.3% vs. 24.8%), more likely to not employ any prevention strategy (19.5% vs. 10.1%), more likely to use condoms alone (54.7% vs. 32.2%), and had more mean sex partners (14.8 vs. 5.9). Those who transmitted an infection were more likely to be an insertive partner than those who did not (32.8% vs. 24.7%), less likely to always use condoms (13.9% vs. 24.7%), more likely to not employ any prevention strategy (20.8% vs. 11.7%), more likely to use condoms alone (55.7% vs. 36.8%), and had more mean sex partners (14.4 vs. 5.9).
As expected, our sensitivity analyses changing the scale of the number of sex partners resulted in different mean number of infections (a smaller scale resulted in a lower incidence and a larger scale resulted in a higher incidence). Similarly, reducing the baseline HIV prevalence also reduced the mean number of infections and these numbers increased proportionally as the prevalence increased. However, the proportion of infections prevented from each strategy, as well as the proportion of infections prevented from PrEP alone, remained the same, despite changes in prevalence or contacts (data not shown).
Our simulation is a novel way to experimentally explore HIV transmission and effects of concomitant engagement in various prevention strategies by modeling currently available statistics on behavior of high-risk, urban, MSM in the United States. We found that PrEP at 25% uptake among high-risk MSM resulted in the prevention of 30% of infections if no other prevention strategies are employed. As would be expected, the independent effect of PrEP is reduced in the presence of TasP, condom use, and seroadaptive behaviors, where PrEP at 25% uptake would further reduce infections by an additional 5.1%. Thus, there is a need to significantly increase PrEP uptake among MSM to see a significant effect in reducing the incidence of HIV among MSM.
Approximately 25% of HIV-negative men in the United States have indications for PrEP,19 yet rates of PrEP uptake, although increasing, have remained low.38,39 As shown by Jenness et al,21 PrEP uptake above 40% among MSM would reduce HIV incidence by 33% or more over 10 years. Achieving these levels of PrEP protection will be a monumental task requiring the reduction of some significant barriers to PrEP use including access to health insurance, knowledge and awareness of PrEP, likelihood of receiving a prescription for PrEP, and adherence to PrEP.40 Kelley et al have developed a PrEP Continuum of Care framework to determine the effect of these barriers on PrEP protection and applied it to a cohort of Atlanta-based high-risk MSM enrolled in project InvoleMENt. The authors determined that PrEP protection would be achieved by 15.2% of their cohort including 12.3% of Black MSM and 17.8% of White MSM.40 Another study of 128 Black and Latino Los Angeles-based MSM recruited from April to May 2014 showed that 60% of enrollees were aware of PrEP and among those who were aware, 40% felt that they would be able to access PrEP. The study found that Black MSM were more aware of PrEP than Latino MSM, but there was no difference in whether or not they believed they would be able to access PrEP. Among those who reported having been tested for HIV in the past 12 months, 86% felt that they had access to PrEP compared with 22% of those who had not been tested.41 These studies demonstrate the need for measures to circumvent the barriers presented in the PrEP Care Continuum to achieve levels of PrEP uptake that will impact the epidemic.
Regardless of levels of uptake, PrEP may have changed the landscape of sexual risk behaviors among MSM, and several studies have explored how sexual risk behaviors have changed among MSM taking PrEP.42 Some of these studies have found increases in risky sexual behavior13,14,43–45 including decreases in condom use,13,15,46–49 while others have found no change in behavior,11,15,47,50 an increase or no change in condom use,11,14,50 and 2 studies found no change in the number of sex partners.48,49 Several studies have reported decreased anxiety associated with fears of sexual acquisition of HIV,43–46,51 and an increased willingness to engage in sex with HIV-positive partners.46 Some studies have indicated a change in sexual positioning where MSM who engaged in 100% insertive anal intercourse were willing to explore receptive anal intercourse after initiating PrEP, signaling a change in seropositioning practices.11,13–15,51 One recent qualitative study showed that 16.7% of PrEP users reported sleeping only with other men who reported using PrEP, suggesting a new form of sexual selection that the authors term “PrEP Serosorting.”46 It would be interesting to study changes in sexual behavior and decision making among those who are not taking PrEP. We would expect to find PrEP and TasP serosorting among this group, which may result in decreased condom use and abandonment of seroadaptive strategies.
Agent-based simulation models, such as this, are subject to limitations. First, simulations necessarily rely heavily on observational, surveillance, and empiric data for their model assumptions. This simulation does not represent a heterogenous population of MSM, since we only consider high-risk MSM. Certainly the formation of monogamous partnerships, the inclusion of individuals who abstain from sex entirely, or who do not have AI have a significant effect on the incidence of HIV and in sexual decision making in the MSM population.
Second, we acknowledge the limitations resulting from our model assumptions that could result in biased estimates in our simulation. It is plausible to believe that PrEP, TasP, condom use, and seroadaptive strategies are not independent from one another as assumed in this simulation; however, without data on how these strategies are employed in conjunction, we had to assume independence and allow their interactions to be stochastic. PrEP use may result in the abandonment of the prevention strategies we have explored in our model, but since we assume both 100% adherence to and 100% effectiveness of PrEP, these behavioral changes would not bias our results.
Lastly, all sexual partnerships in our simulation are chosen at random, as the addition of factors related to assortative mixing, such as age and race, were outside the scope of our analysis. Black MSM are at a higher risk of HIV infection compared with White MSM,52,53 despite reporting fewer risky sexual behaviors.53–59 Sexual network effects resulting from assortative mixing have been posited as a cause of this disparity.58,60–62 Recent work by Goodreau et al63 determined that race-assortative mixing alone did not explain the excess prevalence of HIV among black MSM and concluded that the disparity may be better explained by a misclassification of sexual risk behaviors among black MSM, although another study found that misclassification alone would not explain the difference.64 Additionally, Black MSM experience lower levels of medical care and insurance coverage, are less aware of their HIV status, and, among those who are HIV positive, are less likely to be linked to and retained in HIV care, resulting in a decreased probability of being virally suppressed through TasP.34 This lack of medical care, combined with fewer clinical indications for PrEP, may explain the underusage of PrEP among Black MSM.65–67 With the combined effects of an increased probability of HIV infection among Black MSM, a decreased probability of awareness of HIV infection, and a decreased probability of viral suppression or PrEP use, we anticipate that including assortative mixing on the basis of race would lead to fewer infections being prevented by PrEP, seroadaption, and TasP.
Despite these limitations, our simulation model is reflective of a high-risk sexual network leading to high levels of HIV infection: a large population of MSM having casual, condomless AI. Further, our assumption about 100% adherence to PrEP and 100% efficacy assumes a most optimistic scenario for PrEP. Therefore, our estimates represent the most benefit we can likely see in the reduction of HIV incidence as a result of scaling up PrEP among MSM. Using population-level data on condom use and TasP strengthens assumptions on how multiple prevention strategies interact with one another. Further, the counterfactual basis of our analysis means that our simulation can be compared across each type of intervention. The usefulness of this simulation is in the exploration of how various HIV prevention strategies interact with one another in a controlled setting. Diez-Roux explains, “The beauty of systems modelling is that it can help us understand the plausible implications of the knowledge that we have and how pieces may act together in ways that we might not have predicted from our understanding of each component separately.”68
With the overarching goal of ending the HIV epidemic, public health practitioners require evidence-based data on the expected population-level impact of prevention strategies. Simulations studies such as ours provide these data. A common thread through HIV simulation studies that have focused on PrEP has been that individuals either know their HIV status or have access to testing. As the care continuum begins with knowledge of HIV status, a direct policy implication of our work is to ensure early and equitable access to testing. HIV is not a diagnosis limited to the traditional “gay neighborhoods,” and the decision where to deploy HIV testing services can have a large impact on identifying infected individuals.69 To address this, future work should focus on modeling disparities in access to care and testing the effects of policies that increase HIV testing, linkage to care, and access to PrEP and TasP services. To this end, and in light of the substantial barriers to PreP uptake, public health resources should not necessarily be diverted from one prevention strategy to another. The complementary role of each prevention strategy in controlling horizontal transmission among those with access to each precludes prioritizing one strategy over another until availability is more ubiquitous. In the meantime, the multipronged approach currently in place needs to be promoted to all individuals who engage in high-risk sex. Future work in this area should explore the changing landscape of sexual risk behavior among MSM and explore the concomitant use of various prevention methods to assess nonindependent effects on infection prevention.
1. Centers for Disease Control and Prevention. Diagnoses of HIV Infection in the United States and Dependent Areas, 2015. Vol 27. Atlanta, GA: Division of HIV/AIDS Prevention, National Center for HIV/AIDS, Viral Hepatitis, STD, and TB Prevention, Centers for Disease Control and Prevention (CDC), U.S. Department of Health and Human Services; 2016.
2. Office of Disease Prevention and Health Promotion. HIV. Healthy People 2020; 2016. Available at: https://www.healthypeople.gov/2020/topics-objectives/topic/hiv
. Accessed June 4, 2017.
3. McFarland W, Chen YH, Raymond HF, et al. HIV seroadaptation among individuals, within sexual dyads, and by sexual episodes, men who have sex with men, San Francisco, 2008. AIDS Care. 2011;23:261–268.
4. McFarland W, Chen YH, Nguyen B, et al. Behavior, intention or chance? A longitudinal study of HIV seroadaptive behaviors, abstinence and condom use. AIDS Behav. 2012;16:121–131.
5. Hamel L, Firth J, Hoff T, et al. HIV/AIDS in the Lives of Gay and Bisexual Men in The United States; 2014. Available at: http://kff.org/hivaids/report/hivaids-in-the-lives-of-gay-and-bisexual-men-in-the-united-states/view/footnotes/
. Accessed January 15, 2016.
6. van den Boom W, Konings R, Davidovich U, et al. Is serosorting effective in reducing the risk of HIV infection among men who have sex with men with casual sex partners? J Acquir Immune Defic Syndr. 2014;65:375–379.
7. Cohen MS, McCauley M, Gamble TR. HIV treatment as prevention and HPTN 052. Curr Opin HIV AIDS. 2012;7:99–105.
8. Marrazzo JM, Ramjee G, Richardson BA, et al. Tenofovir-based preexposure prophylaxis for HIV infection among African women. N Engl J Med. 2015;372:509–518.
9. Van Damme L, Corneli A, Ahmed K, et al. Preexposure prophylaxis for HIV infection among African women. N Engl J Med. 2012;367:411–422.
10. Baeten JM, Donnell D, Ndase P, et al. Antiretroviral prophylaxis for HIV prevention
in heterosexual men and women. N Engl J Med. 2012;367:399–410.
11. Grant RM, Lama JR, Anderson PL, et al. Preexposure chemoprophylaxis for HIV prevention
in men who have sex with men. N Engl J Med. 2010;363:2587–2599.
12. Grant RM, Anderson PL, McMahan V, et al. Uptake of pre-exposure prophylaxis, sexual practices, and HIV incidence in men and transgender women who have sex with men: a cohort study. Lancet Infect Dis. 2014;14:820–829.
13. McCormack S, Dunn DT, Desai M, et al. Pre-exposure prophylaxis to prevent the acquisition of HIV-1 infection (PROUD): effectiveness results from the pilot phase of a pragmatic open-label randomised trial. Lancet. 2016;387:53–60.
14. Molina JM, Capitant C, Spire B, et al. On-Demand preexposure prophylaxis in men at high risk for HIV-1 infection. N Engl J Med. 2015;373:2237–2246.
15. Volk JE, Marcus JL, Phengrasamy T, et al. No new HIV infections with increasing use of HIV preexposure prophylaxis in a clinical practice setting. Clin Infect Dis. 2015;61:1601–1603.
16. World Health Organization. Guidance on Oral Pre-exposure Prophylaxis (PrEP) for Serodiscordant Couples, Men and Transgender Women Who Have Sex With Men at High Risk of HIV: Recommendations for Use in the Context of Demonstration Projects. Geneva, Switzerland: World Health Organization; 2012:2012.
17. Centers for Disease C, Prevention. Interim guidance for clinicians considering the use of preexposure prophylaxis for the prevention of HIV infection in heterosexually active adults. MMWR Morb Mortal Wkly Rep. 2012;61:586–589.
18. Grant RM. Dissemination of PrEP Innovations. Paper presented at: Controlling the HIV Epidemic with Antiretrovirals: Having the Courage of our Convictions; January 10, 2015; Paris, France.
19. Smith DK, Van Handel M, Wolitski RJ, et al. Vital signs: estimated percentages and numbers of adults with indications for preexposure prophylaxis to prevent HIV acquisition–United States, 2015. MMWR Morb Mortal Wkly Rep. 2015;64:1291–1295.
20. Hernan MA. Invited commentary: agent-based models for causal inference-reweighting data and theory in epidemiology. Am J Epidemiol. 2015;181:103–105.
21. Jenness SM, Goodreau SM, Rosenberg E, et al. Impact of the centers for disease control's HIV preexposure prophylaxis guidelines for men who have sex with men in the United States. J Infect Dis. 2016;214:1800–1807.
22. Carnegie NB, Goodreau SM, Liu A, et al. Targeting pre-exposure prophylaxis among men who have sex with men in the United States and Peru: partnership types, contact rates, and sexual role. J Acquir Immune Defic Syndr. 2015;69:119–125.
23. Escobar E, Durgham R, Dammann O, et al. Agent-based computational model of the prevalence of gonococcal infections after the implementation of HIV pre-exposure prophylaxis guidelines. Online J Public Health Inform. 2015;7:e224.
24. Jenness SM, Sharma A, Goodreau SM, et al. Individual HIV risk versus population impact of risk compensation after HIV preexposure prophylaxis initiation among men who have sex with men. PLoS One. 2017;12:e0169484.
25. Kasaie P, Pennington J, Shah MS, et al. The impact of preexposure prophylaxis among men who have sex with men: an individual-based model. J Acquir Immune Defic Syndr. 2017;75:175–183.
26. Caceres CF, O'Reilly KR, Mayer KH, et al. PrEP implementation: moving from trials to policy and practice. J Int AIDS Soc. 2015;18(4 suppl 3):20222.
27. Centers for Disease Control and Prevention. Prevalence and awareness of HIV infection among men who have sex with men—21 cities, United States, 2008. MMWR Morb Mortal Wkly Rep. 2010;59:1201–1207.
28. Laumann EO, Masi CM, Zuckerman EW. Circumcision in the United States. Prevalence, prophylactic effects, and sexual practice. JAMA. 1997;277:1052–1057.
29. Omori R, Chemaitelly H, Abu-Raddad LJ. Dynamics of non-cohabiting sex partnering in sub-Saharan Africa: a modelling study with implications for HIV transmission. Sex Transm Infect. 2015;91:451–457.
30. Van de Ven P, Rodden P, Crawford J, et al. Of older Homosexually active men. J Sex Res. 1997;34:349–360.
31. Doll LS, Petersen LR, White CR, et al.; The Blood Donor Study G. Homosexually and nonhomosexually identified men who have sex with men: a behavioral comparison. J Sex Res. 1992;29:1–14.
32. Hart TA, Wolitski RJ, Purcell DW, et al.; Seropositive Urban Men's Study T. Sexual behavior among HIV-positive men who have sex with men: what's in a label? J Sex Res. 2003;40:179–188.
33. Wegesin DJ, Meyer-Bahlburg HFL. Top/bottom self-label, anal sex practices, HIV risk and gender role identity in gay men in New York city. J Psychol Hum Sex. 2000;12:43–62.
34. Singh S, Bradley H, Hu X, et al. Men living with diagnosed HIV who have sex with men: progress along the continuum of HIV care–United States, 2010. MMWR Morb Mortal Wkly Rep. 2014;63:829–833.
35. Snowden JM, Wei C, McFarland W, et al. Prevalence, correlates and trends in seroadaptive behaviours among men who have sex with men from serial cross-sectional surveillance in San Francisco, 2004–2011. Sex Transm Infect. 2014;90:498–504.
36. Punyacharoensin N, Edmunds WJ, De Angelis D, et al. Modelling the HIV epidemic among MSM in the United Kingdom: quantifying the contributions to HIV transmission to better inform prevention initiatives. AIDS. 2015;29:339–349.
37. Smith DK, Herbst JH, Zhang X, et al. Condom effectiveness for HIV prevention
by consistency of use among men who have sex with men in the United States. J Acquir Immune Defic Syndr. 2015;68:337–344.
38. Laufer FN, O'Connell DA, Feldman I, et al.; MPS. Vital signs: increased medicaid prescriptions for preexposure prophylaxis against HIV infection–New York, 2012–2015. MMWR Morb Mortal Wkly Rep. 2015;64:1296–1301.
39. Wu H, Mendoza MC, Huang YA, et al. Uptake of HIV preexposure prophylaxis among commercially insured persons-United States, 2010–2014. Clin Infect Dis. 2017;64:144–149.
40. Kelley CF, Kahle E, Siegler A, et al. Applying a PrEP continuum of care for men who have sex with men in Atlanta, Georgia. Clin Infect Dis. 2015;61:1590–1597.
41. Joseph Davey D, Bustamante MJ, Wang D, et al. PrEP continuum of care for MSM in Atlanta and Los Angeles county. Clin Infect Dis. 2016;62:402–403.
42. Freeborn K, Portillo CJ. Does pre-exposure prophylaxis (PrEP) for HIV prevention
in men who have sex with men (MSM) change risk behavior? A systematic review. J Clin Nurs. 2017. doi: 10.1111/jocn.13990.
43. Brooks RA, Landovitz RJ, Kaplan RL, et al. Sexual risk behaviors and acceptability of HIV pre-exposure prophylaxis among HIV-negative gay and bisexual men in serodiscordant relationships: a mixed methods study. AIDS Patient Care STDS. 2012;26:87–94.
44. Golub SA, Kowalczyk W, Weinberger CL, et al. Preexposure prophylaxis and predicted condom use among high-risk men who have sex with men. J Acquir Immune Defic Syndr. 2010;54:548–555.
45. Hoff CC, Chakravarty D, Bircher AE, et al. Attitudes towards PrEP and anticipated condom use among concordant HIV-negative and HIV-discordant male couples. AIDS Patient Care STDS. 2015;29:408–417.
46. Storholm ED, Volk JE, Marcus JL, et al. Risk perception, sexual behaviors, and PrEP adherence among Substance-using men who have sex with men: a qualitative study. Prev Sci. 2017;18:737–747.
47. Liu AY, Cohen SE, Vittinghoff E, et al. Preexposure prophylaxis for HIV infection integrated with municipal- and community-based sexual health services. JAMA Intern Med. 2016;176:75–84.
48. Molina JM, Charreau I, Spire B, et al. Efficacy, safety, and effect on sexual behaviour of on-demand pre-exposure prophylaxis for HIV in men who have sex with men: an observational cohort study. Lancet HIV. 2017;4:e402–e410.
49. Oldenburg CE, Nunn AS, Montgomery M, et al. Behavioral changes following uptake of HIV pre-exposure prophylaxis among men who have sex with men in a clinical setting. AIDS Behav. 2017. doi: 10.1007/s10461-017-1701-1.
50. Marcus JL, Glidden DV, Mayer KH, et al. No evidence of sexual risk compensation in the iPrEx trial of daily oral HIV preexposure prophylaxis. PLoS One. 2013;8:e81997.
51. Carlo Hojilla J, Koester KA, Cohen SE, et al. Sexual behavior, risk compensation, and HIV prevention
strategies among participants in the San Francisco PrEP demonstration project: a qualitative analysis of counseling notes. AIDS Behav. 2016;20:1461–1469.
52. Sullivan PS, Rosenberg ES, Sanchez TH, et al. Explaining racial disparities in HIV incidence in black and white men who have sex with men in Atlanta, GA: a prospective observational cohort study. Ann Epidemiol. 2015;25:445–454.
53. Millett GA, Peterson JL, Flores SA, et al. Comparisons of disparities and risks of HIV infection in black and other men who have sex with men in Canada, UK, and USA: a meta-analysis. Lancet. 2012;380:341–348.
54. Berry M, Raymond HF, McFarland W. Same race and older partner selection may explain higher HIV prevalence among black men who have sex with men. AIDS. 2007;21:2349–2350.
55. Bingham TA, Harawa NT, Johnson DF, et al. The effect of partner characteristics on HIV infection among African American men who have sex with men in the Young Men's Survey, Los Angeles, 1999–2000. AIDS Educ Prev. 2003;15(1 suppl A):39–52.
56. Harawa NT, Greenland S, Bingham TA, et al. Associations of race/ethnicity with HIV prevalence and HIV-related behaviors among young men who have sex with men in 7 urban centers in the United States. J Acquir Immune Defic Syndr. 2004;35:526–536.
57. Magnus M, Kuo I, Phillips G II, et al. Elevated HIV prevalence despite lower rates of sexual risk behaviors among black men in the district of Columbia who have sex with men. AIDS Patient Care STDS. 2010;24:615–622.
58. Sullivan PS, Peterson J, Rosenberg ES, et al. Understanding racial HIV/STI disparities in black and white men who have sex with men: a multilevel approach. PLoS One. 2014;9:e90514.
59. Rosenberg ES, Sullivan PS, Dinenno EA, et al. Number of casual male sexual partners and associated factors among men who have sex with men: results from the national HIV behavioral surveillance system. BMC Public Health. 2011;11:189.
60. Mustanski B, Birkett M, Kuhns LM, et al. The role of Geographic and network factors in racial disparities in HIV among young men who have sex with men: an egocentric network study. AIDS Behav. 2015;19:1037–1047.
61. Tieu HV, Nandi V, Hoover DR, et al. Do sexual networks of men who have sex with men in New York city differ by race/ethnicity? AIDS Patient Care STDS. 2016;30:39–47.
62. Millett GA, Peterson JL, Wolitski RJ, et al. Greater risk for HIV infection of black men who have sex with men: a critical literature review. Am J Public Health. 2006;96:1007–1019.
63. Goodreau SM, Rosenberg ES, Jenness SM, et al. Sources of racial disparities in HIV prevalence in men who have sex with men in Atlanta, GA, USA: a modelling study. Lancet HIV. 2017;4:e311–e320.
64. Goldstein ND, Burstyn I, Welles SL. Bayesian approaches to racial disparities in HIV risk estimation among men who have sex with men. Epidemiology. 2017;28:215–220.
65. Hoots BE, Finlayson T, Nerlander L, et al.; National HIVBSSG. Willingness to take, use of, and indications for pre-exposure prophylaxis among men who have sex with men-20 US cities, 2014. Clin Infect Dis. 2016;63:672–677.
66. Snowden JM, Chen YH, McFarland W, et al. Prevalence and characteristics of users of pre-exposure prophylaxis (PrEP) among men who have sex with men, San Francisco, 2014 in a cross-sectional survey: implications for disparities. Sex Transm Infect. 2017;93:52–55.
67. Rolle CP, Rosenberg ES, Siegler AJ, et al. Challenges in translating PrEP interest into uptake in an observational study of young black MSM. J Acquir Immune Defic Syndr. 2017;76:250–258.
68. Diez Roux AV. Invited commentary: the virtual epidemiologist-promise and peril. Am J Epidemiol. 2015;181:100–102.
69. Gonsalves GS, Crawford FW, Cleary PD, et al. An adaptive approach to locating mobile HIV testing services. Med Decis Making. 2017. doi: 10.1177/0272989X17716431.