Secondary Logo

Journal Logo


A modeling framework to inform preexposure prophylaxis initiation and retention scale-up in the context of ‘Getting to Zero’ initiatives

Khanna, Aditya S.a,b,*; Schneider, John A.a,b,*; Collier, Nicholsonc; Ozik, Jonathanc; Issema, Rodala,b; di Paola, Angelad; Skwara, Abigaila,b; Ramachandran, Arthia,b; Webb, Jeannettea,b; Brewer, Russella,b; Cunningham, Williame; Hilliard, Charlesf; Ramani, Santhoshinia,b; Fujimoto, Kayod; Harawa, Ninaf,g for the BARS Study Group and Getting to Zero IL Research Evaluation and Data (RED) Committee

Author Information
doi: 10.1097/QAD.0000000000002290



Getting to Zero (GTZ) and other HIV elimination initiatives have gained momentum in the United States following the UNAIDS strategic plan in 2010 [1]. At least 14 jurisdictions are developing GTZ plans, and others are reporting new initiatives [2]. These initiatives, which typically focus on HIV elimination within local jurisdictions, have been driven by a plateauing domestic HIV incidence combined with the growing use of biomedical prevention modalities such as Treatment as Prevention, and especially, preexposure prophylaxis (PrEP). The majority of initiatives have established HIV incidence reduction targets and action plans that highlight specific strategies to be implemented within a defined period.

GTZ Illinois was conceptualized in 2016 by a group of 12 organizational partners from health departments, advocacy groups, academia, service providers, and community organizations. Following the presentation of a simpler incidence forecasting model [3,4], a major theme emerging from GTZ Illinois assessments is that the overall declines in HIV incidence in Illinois have not been experienced equally by subpopulations; younger (18–34 years) Black gay, bisexual and other MSM (YBMSM) have experienced relatively stable incidence rates [5,6]. These observations support the development of a model that focuses on YBMSM. Emerging phylogenetic evidence also indicates that YBMSM are involved in ongoing HIV transmission in other sociodemographic groups, including cisgender and transgender women of color [7,8]. A successful PrEP intervention modeled with YBMSM would, therefore, indicate a greater likelihood for a successful wider statewide GTZ strategy.

Based on preliminary modeling work, GTZ Illinois set benchmarks of a 20% increase in PrEP and antiretroviral therapy (ART) initiation by 2030 to achieve a ‘functional zero’, that is, getting to fewer than 200 new infections annually [3]. These benchmarks were primarily based on simple forecasting models that analyzed trends in incidence data [3,4]. PrEP implementation at the population level, however, is complicated and requires systematic strategy development that employs several facets of implementation science [9,10]. These include ‘scaling out’ specific PrEP implementation strategies, focusing on specific populations and synthesizing locally sourced data to parameterize complex forecasting models [11]. PrEP implementation decisionmaking for actual programs and policies, therefore, requires a model that can incorporate the dynamics of standalone interventions or combinations of distinct interventions; an assessment of candidate PrEP implementation interventions, informed by input from expert stakeholders, and examination of the impact of such interventions on HIV incidence. Increasing PrEP uptake is a complex goal, and requires attention to PrEP continuum components, such as initiation and retention. Local couples-based testing initiatives [12] and social network awareness programs [13] for instance, are examples of strategies supported by local health departments that can be considered for identifying serodiscordant partnerships or influential people who can convince their social network members to initiate PrEP.

In this article, we introduce an agent-based network model (ABNM) [14–16] – parameterized with data largely collected in Illinois – designed to inform GTZ efforts, that examines the effectiveness of PrEP initiation and retention interventions for YBMSM. We focus on YBMSM exclusively, in contrast to models that have focused on addressing racial disparities [17–19], because of feedback from GTZ stakeholders on the importance of developing interventions for YBMSM, a population that represents a unique mix of intersectional identities. The ABNM approach used here addresses the limitations of simpler forecasting approaches considered thus far by GTZ Illinois [3,4] and has the flexibility to be applied to other jurisdictions where initiatives are being developed.


Agent-based network model development

The ABNMs presented in this study were designed to examine PrEP interventions for YBMSM in Illinois. These ABNMs combined sexual network structure with a number of processes that impact HIV transmission (described below). The sexual network structure was estimated using exponential random graph models [20]; this approach is consistent with the methodology developed in previous work [15,21–23]. The ABNMs presented here were implemented using the statnet[24,25] suite of packages in the R programming language to simulate dynamic networks. The Agent-based model (ABM) components were developed with the C++-based Repast High Performance Computing ABM toolkit [26,27]. Parameters and computer code to reproduce results are available in a public GitHub repository [28].

Demographic, network, behavioral, and biological data

Data sources used to parameterize the ABNM were selected based upon a hierarchy of quality with a focus on systematically sampled data that were representative of YBMSM in Illinois. Local data sources included cohort data on Chicago YBMSM from ‘uConnect’ [29,30] and the Young Men's Affiliation Project [31]; both studies recruited participants in Chicago from 2013 to 2016 using systematic sampling schemes. Additional data on YBMSM were obtained from the 2014 cycle of the National HIV Behavioral Surveillance survey [32] in the Chicago Metropolitan Statistical Area. Other local and national sources, described below, were included where representative data from Illinois were not available. All procedures and protocols were approved by relevant institutional review boards.

Baseline model

Baseline HIV transmission was simulated to capture existing epidemic features among adolescents and young adults (age 18–34 years), populated with 10 000 individuals at the start of the dynamic simulations. The model was calibrated using published HIV incidence and prevalence estimates. Simulations proceeded in daily time steps. The key model parameters are listed in Table 1[33–47]. The substantive model components included arrivals, departures, dynamic sexual network structure, the temporal evolution of CD4+ cell counts and HIV RNA (viral load), HIV testing and diagnosis, dynamics of ART and PrEP use, external HIV infections, and HIV transmission dynamics. These processes are described in greater detail in Section A.4 of the Appendix, and represent the control setting without specific GTZ interventions.

Table 1:
Parameters to model HIV transmission among young Black MSM, Illinois.

Modeling candidate preexposure prophylaxis interventions

Candidate PrEP implementation strategies were selected from ongoing interventions supported by health departments and organized across two axes: Axis I, PrEP continuum stage (initiation and retention); and Axis II, sex network targeted (serodiscordant partners and network position). For Axis I, the focus is on PrEP initiation and retention. PrEP initiation was modeled by scaling up the proportion of HIV-negative YBMSM initiating PrEP. PrEP retention interventions were modeled by increasing the mean duration of PrEP use. For PrEP initiation, HIV-negative individuals not on PrEP were randomly selected to initiate PrEP until the target scale-up was achieved. PrEP initiation probabilities were set to achieve uptake levels of approximately 20, 30, 40, 50, and 60%, measured in terms of percentage of HIV-negatives on PrEP; the scenario corresponding to each PrEP initiation probability is labeled with the corresponding approximate uptake level that the probability is set to achieve. At baseline, PrEP uptake was set to approximately 13%, consistent with empirical data [29,30]. Average PrEP retention was varied between our base assumption (12 months on average) and an average retention of 4 years, at 6-month increments.

Axis II interventions were motivated by the importance of incorporating factors that move beyond the individual, such as social network structure, that are important drivers of HIV [14,48,49]. Two network-based interventions are considered here, one that prioritized more proximal network components such as HIV-negative individuals within serodiscordant partnerships, given the success of serodiscordant targeting through couples testing and partner services type interventions [50,51]. A second Axis II intervention that considered larger network structures was also modeled. In this intervention, HIV-negative individuals who were in critical network positions were selected [13,52]. (Background details behind the selection of these specific interventions is provided in Section A.7.1 in the Appendix,

For serodiscordant partner interventions, HIV-negative main partners of HIV-positive individuals were selected for PrEP initiation, beyond the number that would be selected as per the baseline rate. A variant of this intervention where main and casual seronegative partners were selected was also modeled. This selection mechanism was implemented by computing a selection probability for HIV-negative persons in serodiscordant partnerships, calculated as the number of individuals required to make up the difference between baseline and the target PrEP uptake divided by the total number of HIV-negative individuals. Thus, HIV-negative individuals in serodiscordant partnerships had an additional selection probability over and above the selection probability determined by the baseline rates.

The second Axis II intervention type – network interventions – was modeled by selecting individuals who were in critical network positions, as measured by their degree centrality and eigenvector centrality computed on the sex network. Degree is a count of the number of sexual partnerships, and eigenvector centrality indicates individuals who are influential in a given network [53] (technical definitions of the two network position algorithms are provided in Appendix Section A.7.2, Both measures were computed daily for each individual in the simulated population and included both main and casual partnerships. The HIV-negative individuals with the top 10% degree scores and the HIV-negative individuals with the top 10% eigenvector scores were selected for PrEP initiation. These individuals are selected by defining a selection probability which is calculated as the number of individuals required to make up the difference in the target and baseline levels of PrEP uptake divided by the total number of HIV-negative individuals. Thus, HIV-negative individuals with the top 10% degree and eigenvector scores had an additional selection probability over and above the selection probability determined by the baseline. (Sensitivity analyses that varied the proportion of top scorers are described in Section A.7.3 of the Appendix,

Finally, a combination scenario where initiation probability under random selection and mean retention period were simultaneously increased was also considered. All Axes I and II interventions were simulated for 10 years. PrEP initiation probabilities were uniformly increased over the first 5 years, and held constant for the next five, to achieve the uptake levels described above. The PrEP inititation probabilities were gradually scaled up to better approximate real world implementation.


The control setting and intervention were each simulated 30 times. The primary outcome was the HIV incidence rate 10 years after the start of implementation, averaged over the 30 model simulations. The annual incidence rate in the 10th year for the serodiscordant couples’ and network interventions were compared with the random scale-up scenario, which served as a ‘control’ setting.


Tables 2 and 3 provide the mean 10th year incidence rate (per 100 person years) and the mean number of HIV infections in the 10th year, at the target levels of initiation and retention periods described above. Figure 1 displays the average annual incidence rates, with color bands that demonstrate the standard error across the 30 simulations for a given set of parameters, with color bands that demonstrate the standard error, for each of the scenarios. The top and bottom panels of Fig. 1 display simulated data for interventions that increase PrEP initiation and retention respectively.

Table 2:
(a) Mean (standard error) population HIV incidence at 10 years among young Black MSM with varying preexposure prophylaxis initiation probabilities across several intervention scenarios.
Table 3:
Mean population HIV incidence and number of new infections at 10 years among young Black MSM with increasing preexposure prophylaxis retention duration.
Fig. 1:
Simulated annual mean incidence rates and preexposure prophylaxis uptake levels, under several intervention strategies to increase preexposure prophylaxis initiation (panels a and b) to 30% while preexposure prophylaxis retention is held at base levels*, and increasing preexposure prophylaxis retention to a range of levels (panels c and d), while preexposure prophylaxis initiation is held constant at base levels**.*In the base case, preexposure prophylaxis retention is 12 months on average. **In the base case, preexposure prophylaxis initiation is 12.7% of HIV-negatives for persons of 26 years old or less and 14.7% of HIV-negatives for persons more than 26 years old.

HIV incidence at 10 years declined considerably as PrEP uptake increased across all intervention scenarios. Prioritizing HIV-negative YBMSM in serodiscordant main and serodiscordant main and casual partnerships resulted in the largest declines in the 10th year incidence rate, relative to other interventions. In early intervention years, HIV incidence declines were most noticeable when those in serodiscordant main partnerships were selected (2.75 per 100 person years at 30% PrEP initiation) with target levels of PrEP initiation set to 30% or below. As PrEP initiation increased to 40% (or above), HIV incidence declined more rapidly in the scenario where PrEP was provided to MSM in serodiscordant main and casual partnerships. Note also that when PrEP initiation probabilities were set to achieve uptake of 30% (or more) of HIV-negatives, the pool of available main serodiscordant partners was exhausted, and the final PrEP uptake achieved was about 25% (Fig. 1). This ‘saturation’ effect was not seen when both main and casual serodiscordant partners were selected for the intervention. Even with 25% uptake, selecting main serodiscordant couples produced consistently lower incidences than other interventions, which did not saturate. It should also be noted that even in scenarios without such a saturation effect, the attrition of PrEP users due to age-specific mortality and imperfect adherence, resulted in simulated uptake levels that were close, but not equal, to the uptake levels defined in the six PrEP initiation scenarios.

A visualization of the YBMSM sexual network at time 0, prior to the start of the interventions, is given in Fig. 2. Note that this cross-sectional network on a given day is mostly composed of distinct small clusters. The sexual network-based interventions that utilized this network structure also had a substantial, though smaller impact, on HIV incidence than interventions that focused on serodiscordant partners. Selecting HIV-negative YBMSM with more sex partners, as measured by degree centrality, generally outperformed selecting individuals based upon influence (as measured by eigenvector centrality).

Fig. 2:
Composition of steady-state sexual network clusters at t = 0.Main partnerships are displayed in blue, and casual partnerships are displayed in red.

Table 3 provides the 10th year incidence for the intervention where PrEP retention was increased. Increasing PrEP retention also demonstrated substantial reductions, with increasing duration from 12 to 24 months having similar HIV incidence reduction as increasing PrEP initiation by approximately 20% (Tables 1 and 2). Even under the most effective interventions, however, the annual number of new HIV infections after 10 years was close to 200. At a given level of PrEP uptake, increasing PrEP initiation and retention in combination produced a similar incidence decline as increasing either alone (refer to Appendix Section A.7.4, Note that when PrEP initiation and retention are simultaneously increased, under some conditions, PrEP uptake can rise to levels that might be unrealistically high; our interpretation, however, was based on combination interventions that produced the uptake levels close to the uptake targets described above.


Primary findings

In this article, ABNMs revealed findings about several GTZ PrEP implementation interventions. First, increasing PrEP initiation by 20% from baseline uptake modestly decreased HIV incidence and is unlikely to have a major impact on GTZ goals. Second, increasing PrEP initiation and retention had similar effects on HIV incidence; increasing PrEP retention three-fold from baseline to 36 months produced a similar average 10th year incidence as increasing initiation roughly three-fold from baseline to 40%. Third, PrEP interventions that prioritized HIV-negative YBMSM in serodiscordant partnerships could be highly effective in decreasing HIV incidence. Fourth, among interventions that leverage network structure information, prioritizing individuals with more sex partners performed better than interventions targeting persons in influential positions. Finally, results indicate that increasing both initiation and retention is comparable with increasing either one exclusively to the same PrEP uptake level; no synergistic effect between retention and initiation was observed.

Interpretation of primary findings

The conclusions presented here are consistent with previous modeling studies that have suggested the effectiveness of PrEP scale-up in reducing HIV incidence worldwide [54–59]. Targeted PrEP initiation approaches, that prioritize individuals for PrEP based on their partner characteristics, have been modeled before [60–62]. Among studies that focus on MSM, however, ours is one of the first that considers YBMSM exclusively – a population that represents a unique mix of intersectional identities that include age, race and sexual minority statuses. This is in contrast to recent modeling work that has focused on addressing racial disparities [17–19] in the PrEP continuum among MSM in the United States. The dynamics of PrEP initiation and retention modeled here were in response to the expressed needs of GTZ Illinois stakeholders, who have found it necessary to increase both; whereas models examining novel technologies for PrEP delivery [63] and the impact of increasing adherence [64,65] have been developed, models for increasing initiation and retention have received less attention. Our finding that increasing PrEP initiation by 20% would have limited impact among YBMSM was likely driven by low PrEP retention [39]; even the highest PrEP initiation scenarios considered here with random selection did not reduce the number of new HIV infections annually below 200. It is worth noting that initiation and retention represent critically different interventions and may not be equivalent in terms of implementation complexity or cost. While increasing retention three-fold would require intensive peri-clinic interventions, increasing PrEP initiation would require making PrEP broadly accessible and easy to initiate. Substantial HIV incidence reduction was observed in the serodiscordant couples intervention; PrEP scale-up to at least 50% uptake achieved the GTZ target of a functional zero HIV incidence. Serodiscordant interventions that included main partners only had larger HIV incidence reduction effects with modest scale-up, which is unsurprising given the higher frequencies of condomless anal sex within these partnerships [66]. As PrEP for serodiscordant couples was scaled up, however, initiating PrEP within both main and casual partnerships was important in incidence reduction. This finding is reasonable given that the pool of HIV-negative individuals in main serodiscordant partnerships were quickly saturated at the population level as PrEP scale-up targets were increased.

Our finding that that of the two network interventions considered here, degree outperformed eigenvector centrality, is not surprising because degree directly measures the number of sexual partnerships on any given day, and would be expected to have a greater protective effect. Eigenvector centrality, which reflects larger network structures, provides a measure of social influence [53]. It is likely that the sparse, disconnected structure of the cross-sectional sexual network (Fig. 2) was another reason why degree centrality was generally more effective than eigenvector centrality. A previous modeling study has found a similarly high impact of prioritizing individuals with a high number of sexual partners for PrEP [67]. Regardless, neither network intervention had as large an impact as prioritizing serodiscordant partners. Both network and serodiscordant partner interventions have received public health support in the form of diffusion-of-information type interventions that utilize social network structure [13], and partner services interventions that focus on initiating PrEP for HIV-negative individuals in serodiscordant partnerships [68].


There are several limitations in this study. A major limitation in the study is that HIV infections among YBMSM, attributable to Black MSM older than 35 years of age and other populations, were not explicitly modeled, but computed as a proportion of the overall HIV incidence among YBMSM. The model design thus treated transmission among YBMSM as primary effects and transmission attributable to non-YBMSM as secondary. The model was designed as such because YBMSM have experienced relatively stable HIV incidence rates [5,6], even as overall HIV incidence in Illinois has declined, and a systematic strategy focused on reducing new HIV infections among YBMSM is needed. Moreover, external infections from women [7] and non-Black MSM [69] were not included due to evidence that very few infections among YBMSM come from either of these populations [7]. This model design, however, limits its generalizability to broader populations. Explicitly including a broader age range of Black MSM and other populations would be a useful next step to aid the design of comprehensive GTZ strategies. In addition, modeled interventions for serodiscordant couples did not account for possible misclassification of HIV serostatus, which can occur during acute infection or when HIV test results are unknown to either partner. Conversations on HIV testing and status disclosure within main and casual partnerships are also necessary to make this model more realistic. Finally, while our baseline model did include treatment-as-prevention as a consequence of adherence to ART regimens, we did not explicitly consider interventions that expand treatment-as-prevention. Modeling combination ART and PrEP interventions is an important next step that may be needed to achieve GTZ targets.

In addition, we note several real-world implementation challenges. First, the random implementation of PrEP served as a ‘control’ scenario, which allowed us to assess the decline in incidence attributable to a specific serodiscordant partnership or network intervention. We considered a random implementation of PrEP as the control, and not specific PrEP use guidelines (which have been modeled previously [16]), because of evidence on the limited utility of such guidelines for PrEP provision among Black MSM [70,71]. Expanding serodiscordant partner interventions in the real world is challenging given complexities around the determination of serodiscordance within casual or exchange partnerships. Serodiscordant interventions may thus require the use of digital extraction and analyses from hook-up apps or other social media, which could also be challenging to implement. Third, among the two network interventions considered here, degree interventions might be easier to implement, as they directly measure the number of sexual partnerships, whereas eigenvector centrality measures influence, and is harder to determine. Fourth, combination PrEP initiation and retention interventions are likely more feasible given lower levels of increase required in each and should be further explored. Finally, explicit considerations of implementation costs in future work may be helpful to GTZ stakeholders.


In sum, comprehensive models that have the capacity to include multiple biological, social, and network interventions are needed to inform GTZ decision-making for the development of systematic HIV elimination initiatives [16,17,72]. We present an agent-based network model with the goal of developing an implementation science framework to inform GTZ initiatives in the United States, which require examination of downstream intervention effects, and may therefore be less amenable to empirical trials. We found that the goal of fewer than 200 new infections annually – that is, a ‘functional zero’ – was achieved in a decade by increasing PrEP uptake to at least 50% of HIV-negatives when new initiators were selected from HIV-negative individuals in serodiscordant partnerships. Rolling out a PrEP intervention targeted toward serodiscordant couples, however, might face implementation impediments that prevent scale-up to sufficiently high levels. Thus, to accomplish GTZ targets, other synergistic interventions, particularly scaling up ART use, and addressing psychosocial and structural barriers, will need to be considered. Our group, and others, are conducting modeling and empirical studies to design such interventions.


The authors are grateful for the input provided by Getting to Zero Illinois with particular thanks to leadership from John Peller, David Kern, Toyin Adeyemi, Eduardo Alvarez and Erik Glenn. The authors also acknowledge helpful feedback from researchers at the Chicago Center for HIV Elimination, particularly Anna Hotton and Babak Mahdavi Ardestani. In addition, we acknowledge the guidance on implementation science provided by Hendricks Brown and the Center for Prevention Implementation Methodologies (Ce-PIM). This work was completed with resources provided by The University of Chicago Research Computing Center and support from NIH grants R01 DA 039934, P30 AI 117943, and P30 DA 027828. Members of the Getting to Zero Illinois Research Evaluation and Data (RED) committee include: Jessica Ridgway, Diana Lemos, Gregory Phillips, Nanette Benbow, Stephanie Schuette, Christina Hayford, Gary Beringer, Moira McNulty, Roger Fierro, Amy Johnson, and Peter Lindeman. The authors also acknowledge the editor and two anonymous reviewers for suggesting revisions to strengthen the article.

Author contributions: J.A.S., K.F., N.H., W.C., C.H., and A.S.K. conceptualized the study design. A.S.K. led the modeling team, consisting of N.C., J.O., A.S., and A.R. The modeling team coded the model, generated data, analyzed simulated data, and produced the figures and tables. Input data were analyzed by A.S.K., A.d.P., and R.I. A.S.K. and J.A.S. wrote the first draft of the article. All authors contributed to the study design, data interpretation, writing, and revision of the article and the Appendix, All authors have read and approved the text as submitted to AIDS.

Research support: this work was supported by NIH R01 DA 039934, P30 AI 117943, P30 DA 027828, NIH U01DA036267, NIMH; P30 MH 58107-21, California HIV/AIDS Research Program (CHRP; OS17-LA-003) and NIH/NIDA (1R01MD011773).

Conflicts of interest

There are no conflicts of interest.


1. Joint United Nations Programme on HIV/AIDS (UNAIDS). Getting to zero: 2011–2015 strategy. 2010, UNAIDS. Available at [Accessed 19 July 2019]
2. National Alliance of State and Territorial AIDS Directors (NASTAD). Ending the HIV epidemic: jurisdictional plans. Published online by National Alliance of State and Territorial AIDS Directors (NASTAD) at [Accessed 19 July 2019]
3. Getting to Zero Exploratory Workgroup. Getting to zero: a framework to eliminate HIV in Illinois. Published online by the Gettig to Zero Exploratory Workgroup. 2017. Available at [Accessed 19 July 2019]
4. Khanna AS, Skwara A, Collier N, Ozik J, Issema R, Di Paola A, et al. Getting to zero: triangulating HIV incidence predictions from micro-level network modeling and macro-level incidence projections. Washington, DC: Society for Prevention Research; 2018, [Accessed 19 July 2019]
5. Illinois Department of Public Health. Data and statistics: HIV fact sheets: men who have sex with men. 2016, [Accessed 19 July 2019]
6. Illinois Department of Public Health. Data and statistics: HIV fact sheet: blacks. 2014. [Accessed 19 July 2019]
7. Oster AM, Wertheim JO, Hernandez AL, Cheryl M, Ocfemia B, Saduvala N, et al. Using molecular HIV surveillance data to understand transmission between subpopulations in the United States. J Acquir Immune Defic Syndr 2015; 70:444–451.
8. Ragonnet-Cronin M, Hué S, Hodcroft EB, Tostevin A, Dunn D, Fawcett T, et al. Nondisclosed men who have sex with men in UK HIV transmission networks: phylogenetic analysis of surveillance data. Lancet HIV 2018; 5:e309–e316.
9. Chamberlain P, Brown CH, Saldana L. Observational measure of implementation progress in community based settings: the stages of implementation completion (SIC). Implement Sci 2011; 6:116.
10. Landsverk J, Brown CH, Rolls Reutz J, Palinkas L, Horwitz SM. Design elements in implementation research: a structured review of child welfare and child mental health studies. Adm Policy Ment Health 2011; 38:54–63.
11. Aarons GA, Sklar M, Mustanski B, Benbow N, Brown CH. Scaling-out evidence-based interventions to new populations or new healthcare delivery systems. Implement Sci 2017; 12:111.
12. Stephenson R, Suarez NA, Garofalo R, Hidalgo MA, Hoehnle S, Thai J, et al. Project stronger together: protocol to test a dyadic intervention to improve engagement in HIV care among sero-discordant male couples in three US cities. JMIR Res Protoc 2017; 6:e170.
13. Young LE, Schumm P, Alon L, Bouris A, Ferreira M, Hill B, et al. PrEP Chicago: a randomized controlled peer change agent intervention to promote the adoption of preexposure prophylaxis for HIV prevention among young Black men who have sex with men. Clin Trials 2018; 15:44–52.
14. Khanna A, Goodreau SM, Wohlfeiler D, Daar E, Little S, Gorbach PM. Individualized diagnosis interventions can add significant effectiveness in reducing human immunodeficiency virus incidence among men who have sex with men: insights from Southern California. Ann Epidemiol 2015; 25:1–6.
15. Goodreau SM, Carnegie NB, Vittinghoff E, Lama JR, Sanchez J, Grinsztejn B, et al. What drives the US and Peruvian HIV epidemics in men who have sex with men (MSM)?. PLoS One 2012; 7:e50522.
16. Jenness SM, Goodreau SM, Rosenberg E, Beylerian EN, Hoover KW, Smith DK, 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.
17. Goedel WC, King MRF, Lurie MN, Nunn AS, Chan PA, Marshall BDL. Effect of racial inequities in pre-exposure prophylaxis use on racial disparities in HIV incidence among men who have sex with men. JAIDS J Acquir Immune Defic Syndr 2018; 79:323–329.
18. Hamilton DT, Goodreau SM, Jenness SM, Sullivan PS, Wang LY, Dunville RL, et al. Potential impact of HIV preexposure prophylaxis among black and white adolescent sexual minority males. Am J Public Health 2018; 108:S284–S291.
19. Jenness SM, Maloney KM, Smith DK, Hoover KW, Goodreau SM, Rosenberg ES, et al. Addressing gaps in HIV preexposure prophylaxis care to reduce racial disparities in HIV incidence in the United States. Am J Epidemiol 2019; 188:743–752.
20. Robins G, Snijders T, Wang P, Handcock M, Pattison P. Recent developments in exponential random graph (p*) models for social networks. Soc Netw 2007; 29:192–215.
21. Khanna AS, Goodreau SM, Gorbach PM, Daar E, Little SJ. Modeling the impact of post-diagnosis behavior change on HIV prevalence in Southern California men who have sex with men (MSM). AIDS Behav 2014; 18:1523–1531.
22. Khanna AS, Roberts ST, Cassels S, Ying R, John-Stewart G, Goodreau SM, et al. Estimating PMTCT's impact on heterosexual HIV transmission: a mathematical modeling analysis. PLoS One 2015; 10:e0134271.
23. Roberts ST, Khanna AS, Barnabas RV, Goodreau SM, Baeten JM, Celum C, et al. Estimating the impact of universal antiretroviral therapy for HIV serodiscordant couples through home HIV testing: insights from mathematical models. J Int AIDS Soc 2016; 19:20864.
24. Handcock M, Hunter D, Butts C, Goodreau S, Krivitsky P, Morris M. Fit, simulate and diagnose exponential-family models for networks. The Statnet Project 2018 ( R package version 3.9.4, [Accessed 19 July 2019]
25. Hunter D, Handcock M, Butts C, Goodreau S, Morris M. A package to fit, simulate and diagnose exponential-family models for networks. J Statist Software 2008; 24:1–29.
26. Collier N, North M. Parallel agent-based simulation with repast for high performance computing. Simulation 2012; 89:1215–1235.
27. Collier N, Murphy JT, Ozik J, Tatara E. Repast for high performance computing. Lemont, IL: Argonne National Laboratory; 2018.
28. Khanna AS, Collier N, Ozik J. BARS: building agent-based models for racialized justice systems. 2017, [Accessed 19 January 2019]
29. Khanna AS, Michaels S, Skaathun B, Morgan E, Green K, Young L, et al. Preexposure prophylaxis awareness and use in a population-based sample of young black men who have sex with men. JAMA Intern Med 2016; 176:136–138.
30. Schneider J, Cornwell B, Jonas A, Lancki N, Behler R, Skaathun B, et al. Network dynamics of HIV risk and prevention in a population-based cohort of young Black men who have sex with men. Netw Sci 2017; 5:1–29.
31. Kuhns LM, Hotton AL, Schneider J, Garofalo R, Fujimoto K. Use of preexposure prophylaxis (PrEP) in young men who have sex with men is associated with race, sexual risk behavior and peer network size. AIDS Behav 2017; 21:1376–1382.
32. Paz-Bailey G, Mendoza MCB, Finlayson T, Wejnert C, Le B, Rose C, et al. Trends in condom use among MSM in the United States. AIDS 2016; 30:1985–1990.
33. Waldo CR, McFarland W, Katz MH, MacKellar D, Valleroy LA. Very young gay and bisexual men are at risk for HIV infection: the San Francisco Bay Area Young Men's Survey II. J Acquir Immune Defic Syndr 2000; 24:168–174.
34. Centers for Disease Control and PreventionCDC wonder. 2017; [Accessed 5 December 2017].
35. Branson BM, Stekler JD. Detection of acute HIV infection: we can’t close the window. J Infect Dis 2012; 205:521–524.
36. Fairall LR, Bachmann MO, Louwagie GM, van Vuuren C, Chikobvu P, Steyn D, et al. Effectiveness of antiretroviral treatment in a South African program: a cohort study. Arch Intern Med 2008; 168:86–93.
37. Bussmann H, Wester CW, Ndwapi N, Grundmann N, Gaolathe T, Puvimanasinghe J, et al. Five-year outcomes of initial patients treated in Botswana's National Antiretroviral Treatment Program. AIDS 2008; 22:2303–2311.
38. Panel on Antiretroviral Guidelines for Adults and Adolescents. Guidelines for the Use of Antiretroviral Agents in HIV-1-Infected Adults and Adolescents. [Accessed 19 July 2019]
39. Rusie LK, Orengo C, Burrell D, Ramachandran A, Houlberg M, Keglovitz K, et al. Preexposure prophylaxis initiation and retention in care over 5 years, 2012–2017: are quarterly visits too much?. Clin Infect Dis 2018; 67:283–287.
40. Liu AY, Cohen SE, Vittinghoff E, Anderson PL, Doblecki-Lewis S, Bacon O, et al. Preexposure prophylaxis for HIV infection integrated with municipal- and community-based sexual health services. JAMA Intern Med 2016; 176:75–84.
41. Jenness SM, Weiss KM, Goodreau SM, Gift T, Chesson H, Hoover KW, et al. Incidence of gonorrhea and chlamydia following HIV preexposure prophylaxis among men who have sex with men: a modeling study. Clin Infect Dis 2017; 65:712–718.
42. Grant RM, Anderson PL, McMahan V, Liu A, Amico KR, Mehrotra M, et al. Uptake of preexposure 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.
43. Witt MD, Lewis RJ, Rieg G, Seaberg EC, Rinaldo CR, Thio CL. Predictors of the isolated hepatitis B core antibody pattern in HIV-infected and -uninfected men in the multicenter AIDS cohort study. Clin Infect Dis 2013; 56:606–612.
44. Little SJ, McLean AR, Spina CA, Richman DD, Havlir DV. Viral dynamics of acute HIV-1 infection. J Exp Med 1999; 190:841–850.
45. Buchbinder SP, Katz MH, Hessol NA, O’Malley PM, Holmberg SD. Long-term HIV-1 infection without immunologic progression. AIDS 1994; 8:1123–1128.
46. Pilcher CD, Price MA, Hoffman IF, Galvin S, Martinson FEA, Kazembe PN, et al. Frequent detection of acute primary HIV infection in men in Malawi. AIDS 2004; 18:517–524.
47. Hughes JP, Baeten JM, Lingappa JR, Magaret AS, Wald A, de Bruyn G, et al. Determinants of per-coital-act HIV-1 infectivity among African HIV-1-serodiscordant couples. J Infect Dis 2012; 205:358–365.
48. Beyrer C, Baral SD, van Griensven F, Goodreau SM, Chariyalertsak S, Wirtz AL, et al. Global epidemiology of HIV infection in men who have sex with men. Lancet 2012; 380:367–377.
49. Sullivan PS, Carballo-Diéguez A, Coates T, Goodreau SM, McGowan I, Sanders EJ, et al. Successes and challenges of HIV prevention in men who have sex with men. Lancet 2012; 380:388–399.
50. Sullivan PS, White D, Rosenberg ES, Barnes J, Jones J, Dasgupta S, et al. Safety and acceptability of couples HIV testing and counseling for US men who have sex with men. J Int Assoc Provid AIDS Care 2014; 13:135–144.
51. Katz DA, Hogben M, Dooley SW, Golden MR. Increasing public health partner services for human immunodeficiency virus: results of a second national survey. Sex Transm Dis 2010; 37:469–475.
52. Khanna AS, Goodreau SM, Michaels S, Schneider JA. Using partially observed Facebook networks to develop peer-based HIV prevention interventions: a Chicago case study (Preprint). J Med Internet Res 2018; 20:e11652.
53. Borgatti SP. Centrality and network flow. Soc Netw 2005; 27:55–71.
54. Vissers DCJ, Voeten HACM, Nagelkerke NJD, Habbema JDF, de Vlas SJ. The impact of pre-exposure prophylaxis (PrEP) on HIV epidemics in Africa and India: a simulation study. PLoS One 2008; 3:e2077.
55. Nichols BE, Boucher CAB, van Dijk JH, Thuma PE, Nouwen JL, Baltussen R, et al. Cost-effectiveness of pre-exposure prophylaxis (PrEP) in preventing HIV-1 infections in rural Zambia: a modeling study. PLoS One 2013; 8:e59549.
56. Gomez GB, Borquez A, Case KK, Wheelock A, Vassall A, Hankins C. The cost and impact of scaling up preexposure prophylaxis for HIV prevention: a systematic review of cost-effectiveness modelling studies. PLoS Med 2013; 10:e1001401.
57. Brookmeyer R, Boren D, Baral SD, Bekker L-G, Phaswana-Mafuya N, Beyrer C, et al. Combination HIV prevention among MSM in South Africa: results from agent-based modeling. PLoS One 2014; 9:e112668.
58. Ying R, Sharma M, Heffron R, Celum CL, Baeten M, Katabira E, et al. Cost-effectiveness of preexposure prophylaxis targeted to high-risk serodiscordant couples as a bridge to sustained ART use in Kampala, Uganda. J Int AIDS Soc 2015; 18:20013.
59. Schackman BR, Eggman AA. Cost-effectiveness of preexposure prophylaxis for HIV. Curr Opin HIV AIDS 2012; 7:587–592.
60. Cremin I, Morales F, Jewell L, O’Reilly KR, Hallett TB. Seasonal PrEP for partners of migrant miners in southern Mozambique: a highly focused PrEP intervention. J Int AIDS Soc 2016; 18:19946.
61. Cremin I, McKinnon L, Kimani J, Cherutich P, Gakii G, Muriuki F, et al. PrEP for key populations in combination HIV prevention in Nairobi: a mathematical modelling study. Lancet HIV 2017; 4:e214–e222.
62. McGillen JB, Anderson S-J, Hallett TB. PrEP as a feature in the optimal landscape of combination HIV prevention in sub-Saharan Africa. J Int AIDS Soc 2016; 19:21104.
63. Marshall BDL, Goedel WC, King MRF, Singleton A, Durham DP, Chan PA, et al. Potential effectiveness of long-acting injectable preexposure prophylaxis for HIV prevention in men who have sex with men: a modelling study. Lancet HIV 2018; 5:e498–e505.
64. Khurana N, Yaylali E, Farnham PG, Hicks KA, Allaire BT, Jacobson E, et al. Impact of improved HIV care and treatment on PrEP effectiveness in the United States, 2016–2020. J Acquir Immune Defic Syndr 2018; 78:399–405.
65. MacFadden DR, Tan DH, Mishra S. Optimizing HIV preexposure prophylaxis implementation among men who have sex with men in a large urban centre: a dynamic modelling study. J Int AIDS Soc 2016; 19:20791.
66. Sullivan PS, Salazar L, Buchbinder S, Sanchez TH. Estimating the proportion of HIV transmissions from main sex partners among men who have sex with men in five US cities. AIDS 2009; 23:1153–1162.
67. Gantenberg JR, King M, Montgomery MC, Galárraga O, Prosperi M, Chan PA, et al. Improving the impact of HIV preexposure prophylaxis implementation in small urban centers among men who have sex with men: an agent-based modelling study. PLoS One 2018; 13:e0199915.
68. Bhatia R, Modali L, Lowther M, Glick N, Bell M, Rowan S, et al. Outcomes of preexposure prophylaxis referrals from public STI clinics and implications for the preexposure prophylaxis continuum. Sex Transm Dis 2018; 45:50–55.
69. Panneer N, Whiteside YO, France AM, Zhang T, Wertheim JO, Oster AM. Temporal changes in HIV transmission patterns among young men who have sex with men, United States. Conf. Retroviruses Opportunistic Infect. (CROI). Sess. P-B1. 2017. [Accessed 19 July 2019]
70. Sullivan PS, Rosenberg ES, Sanchez TH, Kelley CF, Luisi N, Cooper HL, 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.
71. Lancki N, Almirol E, Alon L, McNulty M, Schneider JA. Preexposure prophylaxis guidelines have low sensitivity for identifying seroconverters in a sample of young Black MSM in Chicago. AIDS 2018; 32:383–392.
72. Jenness SM, Sharma A, Goodreau SM, Rosenberg ES, Weiss KM, Hoover KW, 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.

* Aditya S. Khanna and John A. Schneider contributed equally to the article.


computer simulation; data mining; HIV infections; preexposure prophylaxis; preventive medicine; sexual and sex minorities

Supplemental Digital Content

Copyright © 2019 Wolters Kluwer Health, Inc.