Secondary Logo

Journal Logo

Supplement Article

Using Microsimulation Modeling to Inform EHE Implementation Strategies in Los Angeles County

Drabo, Emmanuel F. PhDa; Moucheraud, Corrina ScDb,c; Nguyen, Anthony MEngd; Garland, Wendy H. MPHe; Holloway, Ian W. PhDc,f; Leibowitz, Arleen PhDc,g; Suen, Sze-chuan PhDd

Author Information
JAIDS Journal of Acquired Immune Deficiency Syndromes: July 1, 2022 - Volume 90 - Issue S1 - p S167-S176
doi: 10.1097/QAI.0000000000002977


  • Evidence-based innovation: Pre-exposure prophylaxis (PrEP).
  • Innovation recipients: Men who have sex with men.
  • Setting: Los Angeles County (LAC).
  • Implementation gap: PrEP is currently being offered in LAC but there are substantial disparities in uptake. Understanding tradeoffs between the health benefits and equity impacts of PrEP implementation strategies is critical to optimize the benefits of PrEP but have not been sufficiently examined.
  • Primary research goal: To help select PrEP implementation strategies that optimize the health and racial and ethnic equity impacts of PrEP in LAC, using microsimulation models.
  • Implementation strategies: To tailor strategies (4 alternative PrEP implementation strategies, including Status-quo and 3 strategies involving coverage with 9000 additional PrEP units annually, above the Status-quo coverage level).


A goal of Ending the HIV Epidemic: A Plan for America (EHE) is to accelerate uptake of and retention in pre-exposure prophylaxis (PrEP) in communities most affected by HIV,1 given the critical role of PrEP in curbing HIV transmission. As one of the 48 highly impacted counties, Los Angeles County (LAC) received supplemental funding to implement a local EHE plan.2 In this plan, the Division of HIV and STD Programs (DHSP) at the LAC Department of Public Health emphasizes that health disparities stemming from structural racism, social inequity, and economic inequality have contributed to uneven progress to end HIV.3 DHSP also recognizes that successful implementation of the plan requires consideration of the impact of these complex issues on awareness, access, and utilization of HIV prevention and treatment services.

DHSP supports several endeavors to increase access to and use of PrEP for those at highest risk of HIV4 but utilization is low and persistently differential across racial and ethnic groups. Among men who have sex with men (MSM) with indications for PrEP, 22% of Black MSM (BMSM), 29% of Latino MSM (LMSM) and 36% of White MSM (WMSM) had used PrEP in 2017.2,5 Thus, LAC efforts could benefit from information about the aggregate and distributional health impacts of different implementation strategies for increasing PrEP use. This requires understanding and closing gaps in PrEP care for racial and ethnic minority MSM.6 Implementation science (IS) can inform such guidance.7–11

Mathematical models—which are abstract representations of real world phenomena that can be used to make predictions and test assumptions (particularly in situations when answering questions in the real world is not possible)—have played an important role in translating HIV research into health policies.6,12–22 However models remain under-utilized in the context of IS for the HIV response to provide valuable insights into the potential costs, health and equity impacts of scaling up alternative HIV-prevention strategies and guide the design and implementation of policy solutions.12,23–26

One important family of mathematical models are microsimulation models, which are individual-based, state-transition models that track individuals as they change health states and strata, by simulating their probabilities of experiencing specific events (eg, infection, diagnosis, disease progression, treatment initiation and discontinuation, death, etc). These probabilities can vary by the attributes of the simulated individual (eg, age, race and ethnicity, etc), thus allowing for variability in outcomes by personal characteristics. As such, microsimulation models address many limitations of deterministic cohort models27–34 including their ability to estimate how demographic, behavioral, and policy changes might differentially affect individual trajectories and outcomes.

To aid DHSP in operationalizing its EHE plan, we developed a microsimulation model to compare the potential population-level impact of 4 hypothetical PrEP implementation strategies in LAC, to answer: which strategy would avert the most infections? Which would be most impactful for addressing inequities in HIV burden? We simulated hypothetical scenarios of PrEP coverage and allocation strategies across select racial and ethnic groups comprising the MSM population in LAC, as prior studies have indicated that race and ethnicity are strongly associated with uptake of and retention in PrEP among PrEP-eligible MSM in the U.S.35 While our simulated strategies do not represent proposed public health or implementation strategies, they permit analytic clarity for quantifying the impacts of interventions designed with explicit HIV disparity reduction goals. Findings from this study can therefore inform more actionable implementation strategies that meet current HIV disparity reduction goals both locally and nationally. This article describes how mathematical modeling, and microsimulation specifically, can serve as an effective tool for generating robust evidence for planning implementation of public health policies, particularly when developed in conjunction with stakeholders.36


We developed a microsimulation model of HIV transmission and progression that tracks MSM in LAC across different health states, including the uninfected and the infected (ie, CD4 ≥ 500, 200 ≤ CD4 < 500, CD4 < 200) states, stratified by diagnosis status (ie, aware vs unaware of HIV+ status), PrEP use, ART use, viral suppression status (ie, HIV-1 RNA < 200 copies/mL), age group (ie, 15–29, 30–49, 50–64, ≥ 65 years), and race and ethnicity (White, Black, and Latino), to estimate HIV-related health outcomes (incidence and prevalence of both diagnosed and undiagnosed HIV, viral suppression, and mortality from HIV/AIDS complications) under alternate implementation strategies. While Asian, Native American and other racial and ethnic minority MSM groups are critical to our understanding of the HIV epidemic, we were unable to examine specific strategies for these groups for LAC due to small sample sizes. Accordingly, we omitted these groups from this analysis. We followed the process described in the logic model in Figure 1 to develop and apply our model, illustrated in the flow diagram in Figure 2.

Logic model of the development of a microsimulation model of HIV transmission among MSM in LAC.
Flow diagram of the microsimulation model of HIV transmission among MSM in LAC. Abbreviations: HIV, human immunodeficiency virus. Notes: This simplified flow diagrams is a representation of the various health states of the model, as well as transitions between them. The figure captures disease and treatment progression for individuals in one age and racial and ethnic group demographic pair,
, where
denotes the set of all age groups,
denotes the set of all race and ethnic groups, and
denotes the set of all pairwise combinations of the age groups and race and ethnicity groups. All other pairs of age group and race and ethnicity are modeled, but not represented in the flow diagram for simplicity. Each circle represents a health state in the model:
represents susceptible individuals not treated with PrEP, while
denote, respectively the state of undiagnosed infected individuals (thus unaware of their HIV serostatus) in stage 1 (CD4
), stage 2 (
), and stage 3 (AIDS;
) of HIV infection. The superscripts
, and
denote, respectively, treatment with PrEP, diagnosed, and virally suppressed. The subscripts
index the demographic subgroup and period of observation, respectively. The solid arrows connecting the circles represent possible transitions between health states within any particular pair of age group and race and ethnic group. The dashed lines represent transitions out of a given health state due to mortality from HIV or AIDS-related complications and/or other causes of death.

Establishing Scope and Involving Stakeholders

Stakeholder engagement in modeling is critical to define scope and ensure relevance to decision-making.37 We defined the scope of the analysis through discussions between academic researchers and LAC DHSP and organizations like the LA LGBT Center. These discussions revealed the importance of focusing on the MSM population (which accounted for 85% of new HIV diagnoses in LAC in 20183) and incorporating variability in HIV burden. Although we lacked detailed data to stratify by geography, we jointly determined that race and ethnicity information on the health and equity impacts of alternative PrEP implementation strategies would still be useful in informing PrEP prioritization in the LAC's health districts, as they are highly segregated by racial and ethnic composition. In February 2021, our group developed and published an online infographic38 to illustrate the potential impacts of PrEP on HIV outcomes for different racial and ethnic groups in LAC, with significant interpretation and framing input from the California's “End The Epidemics” Statewide Working Group.39

Drawing from the insights from the stakeholders and evidence from the published clinical and epidemiological literature, we developed a logic model to link EHE with outcomes relevant to stakeholders (Fig. 1): inputs (eg, funding and data) provide the foundation for EHE activities in LAC including implementation of “treatment as prevention” and PrEP strategies that ultimately affect group- and population-level HIV outcomes (eg, incidence, viral suppression, deaths).

Model Development

We developed a base model characterizing the current epidemic, based on epidemic surveillance and sexual mixing patterns data specific to LAC, and other published data. We initialized the model with data on the MSM population, stratified by HIV serostatus awareness, in year 2011. Each year, new sexually active men entered the model as uninfected (susceptible) at age 15. Once in the model, they initiate and discontinue PrEP and ART at specified rates. Each year, uninfected individuals face a non-zero risk of HIV infection, through unprotected sexual contacts with other infected MSM, depending on use of PrEP and the infectivity of their sexual partners. Sexual partners' infectivity depends on their HIV serostatus, ART use and viral suppression status. Finally, sexual mixing patterns vary by sociodemographic characteristics of age and race and ethnicity. Model parameters were sourced from DHSP and the LA LGBT Center data, the clinical literature, the natural history of HIV progression, and expert opinions (Table 1). Full methodological details on the development, calibration, and validation of the model are provided in the Supplemental Digital Content,

TABLE 1. - Selected Model Input Parameters in 2011
Parameter Total Black Latino White Source
Total LAC MSM population, N 251,521 68,69
PLWH, % total MSM 18.30 Calculated 70
Infected (PLWH) MSM, % total PLWH 19.00 43.00 38.00 *
Uninfected MSM, % total uninfected 10.00 57.00 33.00 71,72
Undiagnosed PLWH, % total PLWH 13.50 73
Viral suppression, % diagnosed PLWH
 15+ y 44.00 56.00 59.00 *
 15–29 yrs 40.00 *
 30–49 yrs 54.00 *
 50–64 yrs 62.00 *
 65+ y 63.00 *
Age group, % total PLWH
 15–29 yrs 11.00 *
 30–49 yrs 58.00 *
 50–64 yrs 28.00 *
 65+ y 3.00 *
Infection stage, % undiagnosed PLWH
 CD4 ≥ 500 41.30 74
 200 ≤ CD4 < 500 50.30 74
 CD4 < 200 8.40 74
Infection stage, % diagnosed PLWH
 CD4 ≥ 500 29.00 *
 15–29 yrs 33.90 47.10 22.90
 30–49 yrs 30.00 43.70 18.50
 50–64 yrs 12.50 6.30 6.50
 65+ y 1.00 0.70 0.80
 200 ≤ CD4 < 500 34.00 *
 15–29 yrs 34.40 56.00 23.00
 30–49 yrs 30.20 54.00 18.40
 50–64 yrs 10.60 5.10 5.50
 65+ y 1.10 0.40 0.40
 CD4 < 200 37.00 *
 15–29 yrs 95.90 98.20 97.90
 30–49 yrs 96.80 98.40 98.30
 50–64 yrs 92.70 97.40 96.90
 65+ y 28.00 1.41 16.60
Annual PrEP uptake rate, %
 2012–2013 0.04 75,76
 2014–2016 0.48 75,76
 2017–present 2.41 75,76
PrEP discontinuation, % 59.00 77,78
A complete list of the model's input parameters is provided in the supplementary digital content.
*HIV surveillance data from LAC department of public health.
Calculated using HIV surveillance data from LAC department of public health.
PLWH, persons living with HIV.


In addition to Status-quo PrEP allocation, which uniformly allocates PrEP to all at-risk MSM, we considered 3 hypothetical strategies to add 9000 PrEP units annually (ie, the quantity needed to achieve complete PrEP coverage of the highest-incidence group: BMSM): (1) equal PrEP allocation to each racial and ethnic group, (2) allocation of PrEP proportional to HIV prevalence of each racial and ethnic group, and (3) and allocation of PrEP proportional to incident HIV diagnosis rates of each racial and ethnic group (strategy 3). These scenarios are neither a comprehensive set nor proposed implementation strategies. Instead, they provide a framework for quantifying the impacts of racial and ethnic equity-based PrEP implementation strategies on HIV outcomes, under various definitions of what may be considered an equitable PrEP allocation.

Establishing Impact

The model estimates outcomes for each simulated individual under each scenario. We calculated population-level outcomes by aggregating individual-level outcomes by demographic subgroups (age, race and ethnicity) over a 15-year period (2021–2035).

The impact (net effect) of each strategy was calculated as the difference between the estimated outcome under each strategy and the same outcome under Status-quo; the main outcome of interest for this analysis is the cumulative new HIV infections averted.

We estimated each strategy's equity impact using both absolute and relative measures of inequality. We used differences in HIV incidence rates (IRs) between groups (BMSM and LMSM, compared to WMSM) as our absolute measure of inequality; and we measured relative inequality via the Gini index (G), a widely used measure of the degree of distributional inequality across population subgroups (range 0–1, details in see Figure S2, Supplemental Digital Content,–45 To the best of our knowledge, this is among its first applications to assess equity of HIV implementation strategies. A higher G indicates greater inequality wherein the population subgroup least burdened by HIV receives disproportionate total health benefits of a given intervention.40,45


Status-quo HIV prevalence (21.3% among all MSM residing in LAC in 2021) was higher among BMSM (38.1%) and WMSM (20.7%), compared to LMSM (18.3%) (Table 2). By the end of 2021, Status-quo HIV IR was 8.88 [95% confidence interval (CI): 8.48 to 9.29] per 1000 person-years (PPY). Rates were higher for BMSM (IR, 23.20; 95% CI: 21.05 to 25.52) and LMSM (IR, 8.83, 95% CI: 8.30 to 9.38), compared to WMSM (IR, 5.27; 95% CI: 4.75 to 5.83), representing a G of 0.31 (95% CI: 0.27 to 0.33). Overall and group-specific rates were similar across strategies (see Figure S1and Table S1, Supplemental Digital Content, By year 2035, LMSM accounted for the largest share of cumulative incident HIV cases under all strategies (range, 57.3%–61.4%) but HIV risk (proxied by HIV IR) remained highest for BMSM across all periods, albeit declining over time for all groups and under all strategies.

TABLE 2. - Baseline Distribution of Health and Health, and Equity Impacts of Increasing PrEP by 9000 Doses Annually for MSM in LAC, by Race and Ethnicity and by Allocation Strategy: 2021–2035
PrEP Allocation Strategy Total Race and Ethnicity
Black Latino White
Baseline distributions of the population and health (start of 2021)
 Baseline population, n (% total) 261,079 29,151 (11.2) 143,283 (54.9) 88,646 (34.0)
 Baseline HIV prevalence rate, % (95% CI) 21.3 (21.2 to 21.5) 38.1 (37.5 to 38.7) 18.3 (18.1 to 18.5) 20.7 (20.4 to 20.9)
 Rate difference, % (95% CI) 17.4 (16.7 to 18.2) 2.3 (2.0 to 2.7) Reference
 Rate ratio, RR (95% CI) 1.84 (1.80 to 1.89) 0.89 (0.87 to 0.90) Reference
 Gini index, G (95% CI) 0.28 (0.26 to 0.34)
Person-years at risk of HIV infection during 2021–2035, n (% total)
 Status-quo* 3,146,429 258,917 (8.2) 1,792,441 (57.0) 1,095,071 (34.8)
 Strategy 1 (equal allocation) 3,158,669 265,745 (20.1) 1,795,459 (56.8) 1,097,464 (34.7)
 Strategy 2 (count-based allocation) 3,156,041 263,013 (21.4) 1,795,832 (56.9) 1,097,195 (34.8)
 Strategy 3 (rate-based allocation)§ 3,161,156 270,270 (8.5) 1,794,241 (56.8) 1,096,645 (34.7)
Cumulative incident HIV infections during 2021–2035, n (% total)
 Status-quo* 24,584 5559 (22.6) 14,089 (57.3) 4936 (20.1)
 Strategy 1 (equal allocation) 22,711 4558 (20.1) 13,522 (59.5) 4631 (20.4)
 Strategy 2 (count-based allocation) 22,887 4888 (21.4) 13,334 (58.3) 4665 (20.4)
 Strategy 3 (rate-based allocation) 22,080 3758 (17.0) 13,557 (61.4) 4764 (21.6)
HIV infections averted during 2021–2035, n (% reference)
 Status-quo* Reference Reference Reference Reference
 Strategy 1 (equal allocation) 1870 (7.6) 1001 (18.0) 564 (4.0) 305 (6.2)
 Strategy 2 (count-based allocation) 1695 (6.9) 670 (12.1) 754 (5.4) 271 (5.5)
 Strategy 3 (rate-based allocation)§ 2504 (10.2) 1800 (32.4) 531 (3.8) 173 (3.5)
IRs, per 1000 person-years (PY) during 2021–2035, IR (95% CI)
 Status-quo* 7.81 (7.72 to 7.91) 21.47 (20.91 to 22.04) 7.86 (7.73 to 7.99) 4.51 (4.38 to 4.64)
 Strategy 1 (equal allocation) 7.19 (7.10 to 7.28) 17.15 (16.66 to 17.66) 7.53 (7.40 to 7.66) 4.22 (4.10 to 4.34)
 Strategy 2 (count-based allocation) 7.25 (7.16 to 7.35) 18.58 (18.07 to 19.11) 7.42 (7.30 to 7.55) 4.25 (4.13 to 4.38)
 Strategy 3 (rate-based allocation) 6.98 (6.89 to 7.08) 13.91 (13.46 to 14.36) 7.56 (7.43 to 7.68) 4.34 (4.22 to 4.47)
IR ratios during 2021–2035, IRR (95% CI)
 Status-quo* 4.76 (4.58 to 4.95) 1.74 (1.69 to 1.80) Reference
 Strategy 1 (equal allocation) 4.06 (3.90 to 4.23) 1.78 (1.73 to 1.85) Reference
 Strategy 2 (count-based allocation) 4.37 (4.20 to 4.55) 1.75 (1.69 to 1.81) Reference
 Strategy 3 (rate-based allocation) 3.20 (3.07 to 3.34) 1.74 (1.68 to 1.80) Reference
Measure of absolute inequality: differences in HIV IRs (per 1000 PY) by race and ethnicity during 2021–2035, RD (95% CI)
 Status-quo* 16.96 (16.38 to 17.54) 3.35 (3.17 to 3.53) Reference
 Strategy 1 (equal allocation) 12.93 (12.42 to 13.44) 3.31 (3.14 to 3.49) Reference
 Strategy 2 (count-based allocation) 14.33 (13.80 to 14.87) 3.17 (3.00 to 3.35) Reference
 Strategy 3 (rate-based allocation) 9.56 (9.10 to 10.02) 3.21 (3.03 to 3.39) Reference
Measure of relative inequality: G during 2021–2035, G (95% CI)
 Status-quo* 0.32 (0.28 to 0.35)
 Strategy 1 (equal allocation) 0.29 (0.25 to 0.31)
 Strategy 2 (count-based allocation) 0.30 (0.27 to 0.33)
 Strategy 3 (rate-based allocation)§ 0.24 (0.20 to 0.26)
Authors' analysis of the microsimulation model's outputs. Under each strategy, PrEP coverage is increased by 9000 units annually above the Status-quo PrEP coverage levels, from 2021 to 2035. The model's outputs may vary due to stochastic noise; hence, means were calculated using 1000 bootstrapped samples from 30 iterations. Values reported in the table are bootstrapped means. HIV infections averted under strategy was calculated as the bootstrapped means of the difference between the cumulative incidence under the Status-quo strategy and the cumulative incidence under the strategy considered, over the 2021–2035 period. These bootstrapped means may therefore not correspond exactly to the difference between the reported cumulative HIV incidence under the Status-quo strategy and each strategy.
*Status-quo refers to the allocation strategy which uniformly allocates currently available PrEP units (hence, no additional units of PrEP) to all at-risk MSM without consideration of equity.
Strategy 1 equally allocates the 9000 additional PrEP units to each racial and ethnic group.
Strategy 2 allocates the 9000 additional PrEP units to each racial and ethnic group proportionally to the count of PLWH in each group.
§Strategy 3 allocates the 9000 additional PrEP units to each racial and ethnic group proportionally to new HIV diagnosis rates in each group in 2016.
n, counts; IRR, incidence rate ratio; RD, incidence rate difference; RR, rate ratio.

Alternative PrEP coverage strategies had differential impacts across groups. Strategy 3 (allocation proportional to incident HIV diagnoses) yielded the largest reductions in new infections (10.2%, n = 2504) relative to Status-quo and 71.9% of these reductions occurred among BMSM. Relative to Status-quo, strategy 3 averted 32.4% of new infections among BMSM, compared with 3.8% and 3.5% reductions in infections among LMSM and WMSM, respectively. Strategy 1 (equal allocation) averted 1870 new infections (7.6% reduction relative to Status-quo); 53.5% of these benefits accrued to BMSM, and 30.2% to LMSM. Relative to Status-quo, strategy 1 averted 18% new infections among BMSM, 4.0% infections among LMSM, and 6.2% infections among WMSM. Strategy 2 (allocation proportional to HIV prevalence) averted 1695 new infections (6.9% reduction), with the highest share of benefits (44.5%) accruing to LMSM, followed by BMSM (39.5%). Relative to Status-quo, this strategy averted 12.1% of new infections among BMSM, and 5.5% and 5.4% among WMSM and LMSM, respectively. Together, these results suggest that strategy 3 produces the largest health impacts and may potentially yield the highest equity impact.

Relative to Status-quo, all strategies reduced BMSM-WMSM and BMSM-LMSM disparities in HIV burden, but not LMSM-WMSM disparities (incidence rate ratio ranging from 1.74 to 1.78 across all strategies). The largest reductions occurred under strategy 3 for BMSM, and strategy 2 for LMSM, with respective absolute risk differences of 9.56 (95% CI: 9.10 to 10.02) and 3.17 (95% CI: 3.00 to 3.35) per 1000 PPY, relative to WMSM. The Gini indices (G) suggest—except for strategy 3 which significantly reduced HIV inequalities relative to baseline and yielded the most equitable distribution of PrEP's health benefits (cumulative G, 0.24; 95% CI: 0.20 to 0.26)—no strategy significantly impacted inequalities (see Figure S2, Supplemental Digital Content, The equity impact of strategy 3 emerged as early as in year 2023 (see Figure S3, Supplemental Digital Content, when IRs for BMSM declined from 23.25 to 15.91 per 1000 PPY (see Figure S1, Supplemental Digital Content,

To identify the optimal strategy, we ranked policies in terms of their health and equity impacts. Relative to Status-quo, strategy 3 produced the largest health benefits (10.2% infections averted) and reductions in inequalities in the distribution of these health impacts (0.08-unit reduction in the G, 95% CI: 0.02 to 0.14), making it the dominant strategy. Strategy 2 was strongly dominated by strategy 1, producing more new infections (175 cases) but without significantly exacerbating inequalities in the distribution of these health harms.


Current HIV prevention strategies have not substantially lowered HIV IRs in the U.S. since 2015. This is a clear opportunity for IS research: there are numerous underutilized efficacious and cost-effective prevention methods, so, implementation strategies could improve outcomes. Modeling approaches can inform the optimal implementation of such interventions and provide more nuanced analyses of potential tradeoffs between health, cost and equity impacts of alternative strategies—as illustrated in this study.

While models have previously examined PrEP distribution, very few models were explicitly designed to examine health equity impacts, and with an IS lens. Notable exceptions include equity-based PrEP care analyses by Jenness et al and Goedel et al 6,20; our work is closest to this literature, and contributes to it by focusing on the LAC MSM population and measuring inequality via the composite G. We leveraged the methodological and statistical rigor of mathematical modeling, critical insights from relevant stakeholders, and high-quality local data, to address policy-relevant questions about areas of opportunity and the relative impact of alternative implementation strategies.

Large-scale evaluations of the strategies explored in this study would be infeasible for logistical and ethical reasons. We thus aimed to demonstrate how modeling can serve a unique and important role in the IS methods “toolbox,” by describing how a rigorously constructed, contextually informed, policy-relevant microsimulation model can compare implementation strategies in order to achieve policy goals and maximize population health. By simulating HIV outcomes among different groups (eg, age, race and ethnicity), we examined potential tradeoffs between health and equity impacts of alternative PrEP allocation strategies among MSM in LAC, the population most at-risk of HIV infection. A strategy that allocates additional PrEP proportionally to incident HIV diagnosis rates, across racial and ethnic groups, dominates other strategies considered and would reduce cumulative incident infections by 10.2% and lessen racial and ethnic inequalities in HIV burden (0.08-unit reductions in G). We found little evidence of significant trade-offs between reducing infections and improving equity across racial and ethnic groups. Through a global look at outcomes, simulation models can therefore inform optimal and more equitable decision-making and increase buy-in among various interest groups.

Prior studies reported on profound racial and ethnic disparities in the uptake of and retention on PrEP, and the consequences of these inequities on exacerbating existing inequities in HIV incidence among racial and ethnic groups in the U.S.46 Our findings suggest that equity-based PrEP implementation strategies can reduce racial and ethnic inequities in HIV by closing each of the 5 critical gaps along the PrEP care-continuum. Specifically, strategies focused on identifying and addressing structural barriers to the uptake of and retention in PrEP among BMSM and LMSM are likely to be of the greatest value.47–50 This could be achieved, for example, by incentivizing and equipping PrEP providers with the necessary skills and resources for making improvements in PrEP awareness, access, prescription, adherence, and retention for BMSM and LMSM, both individually and collectively.6 Incentive mechanisms could include pay-for-performance approaches (ie, value-based health care), which are already being used in LAC and at PrEP Centers of Excellence, to incentivize linkage to PrEP and PrEP-related services across different geographies, venues, and populations.

While we did not model geographically-stratified strategies, our findings can still guide PrEP allocation strategies and target prevention services to potential hotspot areas including geographies with high HIV infection risk, given that a substantial number of persons at high risk for HIV transmission (eg, BMSM) live in neighborhoods with no proximate PrEP provider51 and the persistent gentrification of many U.S. cities and neighborhoods.52 This approach must also consider how transportation burden affects access in “PrEP deserts”; community-based implementation strategies may particularly hold potential for reaching these diverse groups: for example, removing the requirement of physician prescription, delivering PrEP through nontraditional outlets like barber shops and via telemedicine as has been implemented on a temporary basis during the COVID-19 pandemic. A third strategy might target persisting financial barriers: while the Affordable Care Act significantly expanded insurance coverage in the U.S., many non-WMSM in need of PrEP services remain, or are at high risk of becoming, uninsured. Expanding and promoting insurance coverage and robust PrEP cost assistance programs (like the California Pre-Exposure Prophylaxis Assistance Program53) could significantly improve HIV outcomes and reduce disparities. The recent federal requirement that health insurance companies cover PrEP/PrEP-related services with no cost sharing is a promising step.54

This study has several limitations. First, not all key stakeholders were involved in developing the model; we have begun conversations for follow-on studies to reflect their diverse constituencies and perspectives. Second, disparities can be measured in a variety of ways. The G can be critiqued for lacking the subgroup consistency property found in other measures of inequality, including the Atkinson and Kolm indices.55–57 In future analyses, we will examine additional equity measures. Third, the model did not include Asian, Native American and other racial and ethnic minority MSM groups which are an important part of understanding the HIV epidemic in LAC. More comprehensive data for these important groups are needed and would have permitted a more nuanced analysis of the health and equity impacts of alternative PrEP allocation strategies. Fourth, our model did not account for COVID-19, as the evidence on its effects are not yet robust. There is growing evidence that the pandemic disrupted PrEP services among other prevention activities58,59; and local and statewide COVID-19 physical distancing measures may have also induced greater demand for supportive services (eg, housing, food, financial support) disproportionately across communities.58,60–63 On the other hand, some have reported reductions in risky sexual behaviors during the pandemic.64–66 Whether and how these changes have affected HIV outcomes in different groups remains to be determined. We plan to incorporate emerging evidence on the impacts of COVID-19 on HIV into the model. Fifth, due paucity of data, we were unable to model migrations and adaptive behavioral responses, capture sociodemographic heterogeneities in PrEP uptake, retention and adherence, and track all critical domains of the social determinants of health and health behaviors relevant to HIV risk. Thus, we only characterized heterogeneity in sexual mixing patterns by age, race and ethnicity. Lastly, we parameterized our model with data drawn from disparate sources. As data were not always stratified by relevant sociodemographic characteristics of interest, adjustments were made, sometimes relying on simplifying assumptions. When local estimates were unavailable, we used estimates from other populations (eg, national estimates), adding more uncertainties to the model. For example, as the LA LGBT Center sample may not be representative of the LAC MSM population, our mixing parameters estimated from these data may have limited applicability for characterizing sexual behaviors of this population. We conducted sensitivity analyses to assess the impacts of uncertainty in and assumptions about select critical model parameters such as the sexual mixing parameters which characterize transmission. We assessed alternative assumptions about the mixing patterns (eg, homogeneous vs assortative) by age, race and ethnicity and found that results were sensitive to assumptions about mixing patterns, although general trends remained consistent.67 We did not conduct more comprehensive multivariate sensitivity analyses with all parameters, to avoid making further assumptions about the correlation structures of our parameters.

Despite these limitations, this microsimulation model provides information that policymakers can use in discussing efficiency and equity trade-offs. Also, although most models in HIV research have focused on estimating the cost-effectiveness and impact of scaling up interventions, our model takes an explicitly IS lens to questions of policy implementation. By capturing subpopulation dynamics otherwise computationally difficult to include in other types of models, microsimulation models, such as the one we have developed, can be powerful tools for shaping local, national and global HIV prevention policies, by informing pre-implementation planning and implementation, guiding strategic resource allocation decisions, and supporting advocacy. To that end, our academic and public health partners at the LAC-DHSP plan to present these findings to the local Ryan White planning body to inform the development of the Integrated HIV Prevention Plan for LAC. Feedback from these discussions will be incorporated into the model, to enhance its continued relevance to local planning and decision-making. With support from California HIV/AIDS Research Program, we are also partnering with the California State Office of AIDS to use microsimulation models to understand HIV transmissions in San Diego and San Francisco counties and guide locally relevant implementation strategies to stakeholders in those jurisdictions.

There has been much discussion of disparate health outcomes experienced by diverse groups, but little ability to forecast the distribution of likely HIV outcomes resulting from policy implementation across subpopulations. Simulation models, such as the one described here, can provide precisely this type of implementation guidance, by explicitly considering equity and the needs of key populations who remain at greater risk for poor HIV-related outcomes due to social, economic and demographic factors. Guiding principles of IS and collaboration between researchers, public health officials and community stakeholders can ensure that implementation and equity considerations hold weight in shaping the development of robust models, generating relevant results, and ensuring input throughout. It is therefore critical now, more than ever, to strengthen the integration of mathematical modeling into IS, in order to optimally guide the implementation of evidence-based public health interventions.


The authors would like to acknowledge the contributions of the following individuals and groups without whom this project would not be possible: (1) Mario J. Pérez of the Division of HIV and STD Programs, County of Los Angeles, Department of Public Health; (2) Division of HIV and STD Programs staff that collect, manage, analyze and provide oversight for HIV case reporting and HIV surveillance in Los Angeles County including Yuwen, Yu, Virginia Hu and Zhijuan Sheng; and (3) the community of persons living with HIV in Los Angeles County whose aggregated health information were utilized. The authors are also grateful to David W. Dowdy, Associate Professor at Johns Hopkins Bloomberg School of Public Health and other affiliates of JHU-CFAR, for their critical reviews of earlier drafts of the manuscript. The authors thank Citina Liang for assistance with writing code for the Gini calculations, and Elizabeth Wu for assistance with literature reviews and illustrations. The authors are indebted to all the anonymous reviewers and the JAIDS editorial team for their constructive feedbacks on earlier drafts of this article.


1. What is “ending the HIV epidemic: a plan for America?” 2021. Available at: Accessed May 17, 2021.
2. Los Angeles County Department of Public Health, Los Angeles County Commission on HIV. Ending the HIV Epidemic in Los Angeles County. 2020. Available at: Accessed May 17, 2021.
3. Division of HIV and STD Programs, Department of Public Health, County of Los Angeles. HIV Surveillance Annual Report, 2019. 2021. Available at: Accessed June 12 2021.
4. Los Angeles County Department of Public Health, Los Angeles County Commission on HIV. Los Angeles County HIV/AIDS Strategy for 2020 and Beyond. 2018. Available at: Accessed June 12, 2021.
5. Sey EK, Ma Y; Los Angeles County Department of Public Health, Division of HIV and STD Programs. National HIV Behavioral Surveillance Among Men Who Have Sex With Men (MSM). NHBS-MSM5, 2017. 2018. Available at: Accessed June 9, 2021.
6. Jenness SM, Maloney KM, Smith DK, 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.
7. Shangani S, Bhaskar N, Richmond N, et al. A systematic review of early adoption of implementation science for HIV prevention or treatment in the United States. AIDS. 2021;35:177–191.
8. Cox J, Gutner C, Kronfli N, et al. A need for implementation science to optimise the use of evidence-based interventions in HIV care: a systematic literature review. PLoS One. 2019;14:e0220060.
9. Glasgow RE, Eckstein ET, Elzarrad MK. Implementation science perspectives and opportunities for HIV/AIDS research: integrating science, practice, and policy. J Acquir Immune Defic Syndr. 2013;63(suppl 1):S26–S31.
10. Smith JD, Li DH, Hirschhorn LR, et al. Landscape of HIV implementation research funded by the national institutes of health: a mapping review of project abstracts. AIDS Behav. 2020;24:1903–1911.
11. Odeny TA, Padian N, Doherty MC, et al. Definitions of implementation science in HIV/AIDS. Lancet HIV. 2015;2:e178–e180.
12. Eaton JW, Menzies NA, Stover J, et al. Health benefits, costs, and cost-effectiveness of earlier eligibility for adult antiretroviral therapy and expanded treatment coverage: a combined analysis of 12 mathematical models. Lancet Glob Health. 2014;2:e23–e34.
13. Hallett TB, Baeten JM, Heffron R, et al. Optimal uses of antiretrovirals for prevention in HIV-1 serodiscordant heterosexual couples in South Africa: a modelling study. PLoS Med. 2011;8:e1001123.
14. HIV Modelling Consortium Treatment as Prevention Editorial Writing Group. HIV treatment as prevention: models, data, and questions--towards evidence-based decision-making. PLoS Med. 2012;9:e1001259.
15. Walensky RP, Ross EL, Kumarasamy N, et al. Cost-effectiveness of HIV treatment as prevention in serodiscordant couples. N Engl J Med. 2013;369:1715–1725.
16. Ying R, Barnabas RV, Williams BG. Modeling the implementation of universal coverage for HIV treatment as prevention and its impact on the HIV epidemic. Curr HIV/AIDS Rep. 2014;11:459–467.
17. National Research Council. Improving Information for Social Policy Decisions—the Uses of Microsimulation Modeling: Volume I, Review and Recommendations. Washington, DC: The National Academies Press; 1991:360.
18. Krebs E, Nosyk B. Cost-effectiveness analysis in implementation science: a research agenda and call for wider application. Curr HIV/AIDS Rep. 2021;18:176–185.
19. Rutter CM, Zaslavsky AM, Feuer EJ. Dynamic microsimulation models for health outcomes: a review. Med Decis Making. 2011;31:10–18.
20. Goedel WC, Bessey S, Lurie MN, et al. Projecting the impact of equity-based preexposure prophylaxis implementation on racial disparities in HIV incidence among MSM. AIDS. 2020;34:1509–1517.
21. Drabo EF, Hay JW, Vardavas R, et al. A cost-effectiveness analysis of preexposure prophylaxis for the prevention of HIV among Los Angeles County men who have sex with men. Clin Infect Dis. 2016;63:1495–1504.
22. Sood N, Wagner Z, Jaycocks A, et al. Test-and-treat in Los Angeles: a mathematical model of the effects of test-and-treat for the population of men who have sex with men in Los Angeles County. Clin Infect Dis. 2013;56:1789–1796.
23. Hontelez JA, Chang AY, Ogbuoji O, et al. Changing HIV treatment eligibility under health system constraints in sub-Saharan Africa: investment needs, population health gains, and cost-effectiveness. AIDS. 2016;30:2341–2350.
24. Kazemian P, Costantini S, Kumarasamy N, et al. The cost-effectiveness of human immunodeficiency virus (HIV) preexposure prophylaxis and HIV testing strategies in high-risk groups in India. Clin Infect Dis. 2020;70:633–642.
25. Kessler J, Nucifora K, Li L, et al. Impact and cost-effectiveness of hypothetical strategies to enhance retention in care within HIV treatment programs in east africa. Value Health. 2015;18:946–955.
26. Marshall BD, Friedman SR, Monteiro JF, et al. Prevention and treatment produced large decreases in HIV incidence in a model of people who inject drugs. Health Aff (Millwood). 2014;33:401–409.
27. Caro JJ, Briggs AH, Siebert U, et al. Modeling good research practices--overview: a report of the ISPOR-SMDM modeling good research practices task force-1. Med Decis Making. 2012;32:667–677.
28. Kreke JE, Schaefer AJ, Roberts MS. Simulation and critical care modeling. Curr Opin Crit Care. 2004;10:395–398.
29. Hunink MM, Weinstein MC, Wittenberg E, et al. Decision Making in Health and Medicine: Integrating Evidence and Values. Cambridge, UK: Cambridge University Press; 2014.
30. Lay-Yee R, Cotterell G. The role of microsimulation in the development of public policy. In: Janssen M, Wimmer MA, Deljoo A, eds. Public Administration and Information Technology, Vol. 10. Heidelberg, New York, London: Springer, Cham; 2015:305–320.
31. Siebert U, Alagoz O, Bayoumi AM, et al. State-transition modeling: a report of the ISPOR-SMDM modeling good research practices task force-3. Med Decis Making. 2012;32:690–700.
32. Zucchelli E, Jones AM, Rice N. The evaluation of health policies through microsimulation methods. Health Econom Data Group (HEDG) Working Pap. 2010;10:2.
33. Brennan A, Chick SE, Davies R. A taxonomy of model structures for economic evaluation of health technologies. Health Econ. 2006;15:1295–1310.
34. Basu S. Modeling Public Health and Healthcare Systems. Oxford, UK: Oxford University Press; 2018.
35. Chan PA, Mena L, Patel R, et al. Retention in care outcomes for HIV pre-exposure prophylaxis implementation programmes among men who have sex with men in three US cities. J Int AIDS Soc. 2016;19:20903.
36. Emmons KM, Chambers DA. Policy implementation science - an unexplored strategy to address social determinants of health. Ethn Dis Winter. 2021;31:133–138.
37. Njeuhmeli E, Schnure M, Vazzano A, et al. Using mathematical modeling to inform health policy: a case study from voluntary medical male circumcision scale-up in eastern and southern Africa and proposed framework for success. PLoS One. 2019;14:e0213605.
38. California HIV/AIDS Policy Research Centers. Equity considerations for expanding PrEP use for MSM in Los Angeles county. 2021. Available at: Accessed Sep 22, 2021.
39. California HIV/AIDS Policy Research Centers. End the epidemics. 2021. Available at: Accessed October 22, 2021.
40. Abeles J, Conway DJ. The Gini coefficient as a useful measure of malaria inequality among populations. Malar J. 2020;19:444.
41. Measuring health inequalities: Gini coefficient and concentration index. Epidemiol Bull. 2001;22:3–4.
42. Steinbeis F, Gotham D, von Philipsborn P, et al. Quantifying changes in global health inequality: the Gini and slope inequality indices applied to the global burden of disease data, 1990-2017. BMJ Glob Health. 2019;4:e001500.
43. Wagstaff A, Paci P, van Doorslaer E. On the measurement of inequalities in health. Soc Sci Med. 1991;33:545–557.
44. Hicks DA. The inequality-adjusted human development index: a constructive proposal. World Development. 1997;25:1283–1298.
45. Mody A, Pfeifauf K, Geng EH. Using lorenz curves to measure racial inequities in COVID-19 testing. JAMA Netw Open. 2021;4:e2032696.
46. Kanny D, Jeffries IVWL, Chapin-Bardales J, et al. Racial/ethnic disparities in HIV preexposure prophylaxis among men who have sex with men—23 urban areas, 2017. Morbidity Mortality Weekly Rep. 2019;68:801.
47. Rolle C-P, 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.
48. Chan PA, Mena L, Patel R, et al. Retention in care outcomes for HIV pre‐exposure prophylaxis implementation programmes among men who have sex with men in three US cities. J Int AIDS Soc. 2016;19:20903.
49. Serota DP, Rosenberg ES, Lockard AM, et al. Beyond the biomedical: preexposure prophylaxis failures in a cohort of young black men who have sex with men in Atlanta, Georgia. Clin Infect Dis. 2018;67:965–970.
50. Doblecki-Lewis S, Liu A, Feaster D, et al. Healthcare access and PrEP continuation in san Francisco and miami after the US PrEP demo project. J Acquir Immune Defic Syndr. 2017;74:531–538.
51. Siegler AJ, Bratcher A, Weiss KM. Geographic access to preexposure prophylaxis clinics among men who have sex with men in the United States. Am J Public Health. 2019;109:1216–1223.
52. Menendian S, Gailes A, Gambhir S. The Roots of Structural Racism: Twenty-First Century Racial Residential Segregation in the United States. Othering and Belonging Institute; 2021. Available at: Accessed Sep 22, 2021.
53. National Alliance of State & Territorial AIDS Directors (NASTAD). The California pre-exposure prophylaxis Assistance program (PrEP-AP). 2021. Available at: Accessed October 22, 2021.
54. Keith K. New guidance on PrEP: support services must Be covered without cost-sharing. Health Aff Blog. 2021;2021.
55. Atkinson AB. On the measurement of inequality. J Econ Theor. 1970;2:244–263.
56. Kolm S-C. Unequal inequalities. I. J Econ Theor. 1976;12:416–442.
57. Kolm S-C. Unequal inequalities. II. J Econ Theor. 1976;13:82–111.
58. Hogan AB, Jewell BL, Sherrard-Smith E, et al. Potential impact of the COVID-19 pandemic on HIV, tuberculosis, and malaria in low-income and middle-income countries: a modelling study. Lancet Glob Health. 2020;8:e1132–e1141.
59. California HIV/AIDS Policy Research Centers. COVID-19 organizational health survey. 2021. Available at: Accessed October 22, 2021.
60. Silhol R, Geidelberg L, Mitchell KM, et al. Assessing the potential impact of disruptions due to COVID-19 on HIV among key and lower-risk populations in the largest cities of Cameroon and Benin. J Acquir Immune Defic Syndr. 2021;87:899–911.
61. Santos GM, Ackerman B, Rao A, et al. Economic, mental health, HIV prevention and HIV treatment impacts of COVID-19 and the COVID-19 response on a global sample of cisgender gay men and other men who have sex with men. AIDS Behav. 2021;25:311–321.
62. Rao A, Rucinski K, Jarrett BA, et al. Perceived interruptions to HIV prevention and treatment services associated with COVID-19 for gay, bisexual, and other men who have sex with men in 20 countries. J Acquir Immune Defic Syndr. 2021;87:644–651.
63. Reitsma MB, Claypool AL, Vargo J, et al. Racial/ethnic disparities in COVID-19 exposure risk, testing, and cases at the subcounty level in California. Health Aff (Millwood). 2021;40:870–878.
64. Sanchez TH, Zlotorzynska M, Rai M, et al. Characterizing the impact of COVID-19 on men who have sex with men across the United States in april, 2020. AIDS Behav. 2020;24:2024–2032.
65. Shilo G, Mor Z. COVID-19 and the changes in the sexual behavior of men who have sex with men: results of an online survey. J Sex Med. 2020;17:1827–1834.
66. Hyndman I, Nugent D, Whitlock GG, et al. COVID-19 restrictions and changing sexual behaviours in HIV-negative MSM at high risk of HIV infection in London, UK. Sex Transm Infect. 2021;97:521–524.
67. Nguyen A, Drabo EF, Garland W, et al. Impact and equality of expanding pre-exposure prophylaxis (PrEP) for men who have sex with men in Los Angeles county. medRxiv. 2021. Available at:
68. Lieb S, Fallon SJ, Friedman SR, et al. Statewide estimation of racial/ethnic populations of men who have sex with men in the U.S. Public Health Rep. 2011;126:60–72.
69. Grey JA, Bernstein KT, Sullivan PS, et al. Estimating the population sizes of men who have sex with men in US states and counties using data from the American Community Survey. JMIR Public Health Surveill. 2016;2:e14.
70. Centers for Disease Control and Prevention. HIV Infection Risk, Prevention, and Testing Behaviors Among Men Who Have Sex With Men—National HIV Behavioral Surveillance, 23 U.S. Cities, 2017. HIV Surveillance Special Report 22. 2019. Available at: Accessed May 17, 2021.
71. United States Census Bureau. QuickFacts: Los Angeles county. 2021. Available at: Accessed October 8, 2021.
72. Statistical atlas. Race and ethnicity in Los Angeles county, California (county). 2021. Available at: Accessed October 8, 2021.
73. Hall HI, An Q, Tang T, et al. Prevalence of diagnosed and undiagnosed HIV infection-United States, 2008–2012. MMWR Morbidity mortality weekly Rep. 2015;64:657.
74. Khurana N, Yaylali E, Farnham PG, 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.
75. Harawa N, McBride S, Leibowitz A, et al. 2018. Available at: Accessed May 17, 2021.
76. Los Angeles County Department of Public Health. Los Angeles county HIV/AIDS strategy for 2020 and beyond October 8, 2021. Available at: Accessed October 8, 2021.
77. Shover CL, Javanbakht M, Shoptaw S, et al. High Discontinuation of Pre-exposure Prophylaxis within Six Months of Initiation [#1009]. Paper presented at: Conference on Retroviruses and Opportunistic Infections. Boston; 2018.
78. Shover CL, Shoptaw S, Javanbakht M, et al. Mind the gaps: prescription coverage and HIV incidence among patients receiving pre-exposure prophylaxis from a large federally qualified health center in Los Angeles, California. AIDS Behav. 2019;23:2730–2740.

HIV; AIDS; PrEP; pre-exposure prophylaxis; implementation science; microsimulation model; equity

Supplemental Digital Content

Copyright © 2022 Wolters Kluwer Health, Inc. All rights reserved.