|Home | About | Journals | Submit | Contact Us | Français|
Jessie L. Juusola, Department of Management Science and Engineering, Huang Engineering Center Suite 263, 475 Via Ortega, Stanford, CA 94305
Margaret L. Brandeau, Department of Management Science and Engineering, Huang Engineering Center Room 262, 475 Via Ortega, Stanford, CA 94305
Douglas K. Owens, Center for Primary Care and Outcomes Research, 117 Encina Commons, Stanford, CA 94305
Eran Bendavid, Division of General Internal Medicine, 251 Campus Drive, MSOB x332, Stanford, CA 94306
In a recent randomized controlled trial, daily oral preexposure chemoprophylaxis (PrEP) was shown to be effective for HIV prevention in men who have sex with men (MSM). The United States Centers for Disease Control and Prevention (CDC) recently provided interim guidance for PrEP use among MSM who are at high risk for acquiring HIV. Previous studies failed to reach a consistent estimate of its cost-effectiveness.
To estimate the effectiveness and cost-effectiveness of PrEP in MSM in the United States.
Dynamic model of HIV transmission and progression combined with a detailed economic analysis.
MSM aged 13–64 in the United States.
We evaluated PrEP for the general MSM population and for high-risk MSM. We assumed that PrEP reduces infection risk by 44%, based on clinical trial results.
New HIV infections, discounted quality-adjusted life-years (QALYs) and costs, and incremental cost-effectiveness ratios.
If PrEP is initiated in 20% of MSM in the United States, we estimate a 13% reduction in new HIV infections and a gain of 550,166 QALYs over 20 years at a cost of $172,091/QALY gained. Initiating PrEP in a larger proportion of MSM averts more infections but at increasing cost per QALY gained ($216,480/QALY gained when 100% of MSM receive PrEP). Using PrEP only in high-risk MSM can improve its cost-effectiveness. PrEP costs approximately $50,000/QALY gained for MSM with 5 annual partners on average. PrEP for all high-risk MSM for 20 years leads to $75 billion in healthcare-related costs incremental to the status quo and costs $600,000 per HIV infection averted, compared with incremental costs of $95 billion and $2 million per infection averted for 20% coverage of all MSM.
PrEP use in the general MSM population costs less than $100,000/QALY gained if the daily cost of antiretroviral drugs for PrEP is less than $15 or if PrEP efficacy is greater than 75%.
When examining PrEP use in high-risk MSM, we did not model mixing between low- and high-risk MSM because of lack of data on mixing patterns.
Use of PrEP for HIV prevention in the general MSM population could prevent a substantial number of HIV infections but is expensive. PrEP use in high-risk MSM compares favorably to other interventions considered cost-effective, but could result in annual expenditures on PrEP of over $4 billion.
Men who have sex with men (MSM) are an important group to reach with programs to prevent human immunodeficiency virus (HIV) infection. Of the estimated 48,000–56,000 annual new HIV infections in the United States, 56–61% occur among MSM (1, 2). In 2010, results from a clinical trial suggested that antiretroviral drugs (ARVs) can be effectively used as preexposure prophylaxis (PrEP) for HIV prevention among uninfected MSM (3). The Preexposure Prophylaxis Initiative (iPrEx) study showed that in MSM, daily tenofovir and emtricitabine (TDF/FTC) reduced HIV incidence by 44% overall, and by 73% among MSM who reported high adherence (3). In 2011, the Centers for Disease Control and Prevention (CDC) published interim guidance for prescribing TDF/FTC as PrEP for MSM (4). This guidance highlights the importance of regular monitoring for adverse effects from PrEP, HIV testing, and counseling to encourage adherence and risk reduction.
While PrEP has been shown to be effective at preventing HIV acquisition, its costs are considerable. Prior studies on the cost-effectiveness of PrEP in MSM in the United States have reached inconsistent conclusions (5, 6). Paltiel et al. (5) estimated that PrEP costs $298,000 per quality-adjusted life year (QALY) gained among high-risk MSM, while Desai et al. (6) estimated that a PrEP program targeting high-risk men in New York City would cost $32,000/QALY gained. We examine the effectiveness and cost-effectiveness of PrEP strategies for MSM in the United States using the newly available clinical trial data, CDC interim guidance, and an epidemic modeling framework combined with a detailed economic analysis.
We adapted a previously published deterministic dynamic compartmental model of HIV transmission and progression to assess the effectiveness and cost-effectiveness of PrEP for HIV prevention in MSM (7). We calibrated the model to estimates of HIV incidence among MSM aged 13–64 in the United States (Appendix) (1, 2), and estimated HIV prevalence, incidence, QALYs, and healthcare costs of various PrEP strategies over a 20-year time horizon. We assumed a societal perspective and discounted costs and QALYs at 3% annually (8). Key model parameters are presented in Table 1. The Appendix provides additional model details.
We divided the population by HIV infection status, awareness of HIV status, and PrEP use. Infected individuals were additionally defined by HIV disease stage and treatment status (Appendix Figure A1). Initial HIV prevalence was 12.3% and annual incidence was 0.8%, representing an average across the United States (2, 9–12). Individuals entered the model at age 13 and were followed for 20 years or until age 65. Modeling MSM aged 13–64 is consistent with CDC recommendations for routine HIV screening (62). Mortality comprised HIV-related and non-HIV-related deaths.
We modeled HIV transmission via homosexual contact based on number of male sexual partners (28, 29) and average condom use (31). We assumed proportional mixing in selecting an HIV-infected partner; that is, individuals were more likely to partner with MSM who had many partners than with those who had few partners. The probability of HIV transmission between sero-discordant partners depended on the infected individual’s disease stage and antiretroviral therapy (ART) use, as well as PrEP status of the uninfected individual (25, 26).
HIV disease progression was based on the average time in each of four disease stages: acute infection, asymptomatic HIV, symptomatic HIV, and AIDS. Progression rates were based on HIV natural history and ART status (20–22, 35).
We assumed that HIV-infected individuals with a CD4 cell count of 0.350 × 109 cells/L or lower were offered ART (63). Given recent guidelines recommending ART initiation at CD4 cell counts great than 0.350 × 109 cells/L, we examined the effects of earlier ART initiation in sensitivity analysis (64). The benefits of ART and suppression of viral replication included improved quality of life and reduced disease transmission, progression, and mortality (22, 35). We assumed a 90% reduction in sexual infectivity due to ART used for treatment of HIV infection in our base case and varied this in sensitivity analysis (22, 26, 37, 38).
To be treated, HIV-infected individuals must be identified as infected. We estimated that 67% of MSM are currently screened annually using antibody tests (28, 45). Counseling accompanying HIV testing has been found to reduce risky sexual behavior (49). Accordingly, we assumed a 20% reduction in risky behavior for both infected and uninfected individuals after HIV screening (22, 35, 49).
We considered two strategies: PrEP for the general MSM population and PrEP for high-risk MSM. We chose a representative example of a high-risk population, and we varied risk levels in sensitivity analysis. We compared PrEP strategies to the status quo of no PrEP use. We assumed that MSM receiving PrEP would begin immediately and would remain on PrEP for the 20-year time horizon or until aging out of the model.
We based our PrEP protocol on the CDC interim guidance on PrEP use in uninfected MSM (4). MSM were initiated on PrEP after a negative HIV antibody test, adequate calculated creatinine clearance, and testing for sexually transmitted infections. The PrEP regimen included daily TDF/FTC and physician visits 5 times per year, or every 2–3 months, where HIV-negative status was confirmed with an antibody test and risk-reduction counseling and condoms were provided. Additionally, sexually transmitted infection testing was performed every 6 months and renal function was tested annually. Individuals who became HIV-infected while on PrEP, once infection was detected, were provided with appropriate counseling and discontinued PrEP.
We assumed TDF/FTC reduces the probability of acquiring HIV by 44%, based on the overall reduction in incidence seen in iPrEx study subjects (3). PrEP was more effective in preventing HIV infection in iPrEx subgroups with higher adherence, and incidence reduction in study subjects who reported or exhibited pill use on at least 90% of days was 73% (3). Accordingly, we varied PrEP efficacy in sensitivity analysis. Although behavioral disinhibition is a concern with PrEP use, there is no conclusive evidence regarding the effect of PrEP on sexual risk behavior (65, 66). Hence, in our base case we assumed no change in risky behavior from counseling or risk compensation. In sensitivity analysis we examined the impact of changes in number of sexual partners and condom use as a result of PrEP.
Various studies have assessed willingness and likelihood to use PrEP in surveyed MSM populations with varying conclusions (67–70). The percentage of MSM who will ultimately use PrEP is unknown and will depend on factors such as public health campaigns and access to healthcare. We evaluated initiating 5–100% of the HIV-negative MSM population on PrEP (Appendix), but we focus results on 20%, 50%, and 100% of uninfected MSM initiating PrEP to illustrate the differences in effectiveness and cost-effectiveness as the percentage of MSM on PrEP increases.
We simulated the population over time and calculated discounted costs and QALYs for each PrEP use scenario. We estimated quality of life for each health state and adjusted the utilities based on the average age of the modeled population (22, 35, 71). We assumed no reduction in quality of life from PrEP, as clinical trials have shown minimal side effects from TDF/FTC (3, 72). In sensitivity analysis, we evaluated the impact of decreased quality of life while on PrEP, as study participants on PrEP were more likely than those on placebo to experience minor side effects such as nausea (3, 72).
We included costs associated with medical care in each health state, PrEP, and HIV testing, counseling, and diagnosis to calculate total health-related costs. Baseline medical costs, HIV-related healthcare costs (with costs of associated co-morbidities), cost of ART, and costs of counseling were estimated from the published literature (22, 53). Costs of ARVs for PrEP were estimated as the average monthly wholesale price of TDF/FTC adjusted to reflect rebates and retail pharmacy dispensing fees (5). Costs of non-ARV components of the PrEP protocol and the HIV testing protocol were obtained from the Centers for Medicare and Medicaid Services 2010 fee schedules (59). We also included discounted health-related costs and QALYs for the remaining lifetime of the population in the model at the end of the time horizon and for individuals who matured out of the modeled population.
The National Institute of Health and the Department of Veterans Affairs supported this study. None of the funding sources had any role in the design, conduct, or reporting of the study.
In the absence of PrEP use, we estimate that 491,784 new HIV infections will occur among MSM in the United States in the next 20 years (Table 2). Use of PrEP can substantially reduce this incidence. Initiating 20% of the MSM population in the United States on PrEP will reduce the number of infections over 20 years by 62,759 (13% of the projected total) and yield 550,166 incremental QALYs (Table 2). Initiating a greater proportion of the MSM population on PrEP yields greater health benefits. With 50% of MSM using PrEP, 143,291 new infections (29%) are averted and 1.3 million QALYs are gained incremental to the status quo. If 100% of MSM use PrEP, 249,156 (51%) infections will be averted, HIV prevalence in MSM will drop to 6.4% by the end of 20 years (Appendix Figure A2a), and more than 2 million QALYs will be gained.
In the first year of PrEP use, the percentage reduction in annual HIV incidence is similar to the efficacy of PrEP scaled by the percentage of MSM using PrEP. For example, with 20% of MSM using PrEP, incidence is reduced by 10% in the first year (approximately equal to 44% efficacy multiplied by 20% on PrEP) (Appendix Figure A2b). With 50% and 100% of MSM on PrEP, incidence is reduced by 24% and 45% in the first year, respectively. However, averting secondary transmissions causes the percentage reduction in HIV incidence to increase over time. After 20 years, annual HIV incidence is reduced by 17% when 20% of MSM use PrEP, by 37% when 50% of MSM use PrEP, and by 60% when all MSM use PrEP.
PrEP is effective at preventing HIV in MSM but is an expensive intervention. Initiating PrEP in 20% of MSM has an incremental cost-effectiveness ratio of $172,091/QALY gained compared to the status quo (Table 2, Figure 1). As the percentage of MSM initiated on PrEP increases, the incremental cost-effectiveness ratio also increases because of diminishing benefits from additional PrEP. Initiating 50% of MSM on PrEP costs $188,421/QALY gained compared to the status quo, or $201,012 compared to initiating 20% on PrEP, while initiating 100% of MSM on PrEP costs $216,480/QALY gained, or $253,645 compared to initiating 50% on PrEP.
The costs associated with PrEP are substantial. PrEP use in 20% of MSM over 20 years generates an incremental $95 billion in healthcare-related costs compared to the status quo ($98 billion for PrEP minus $3 billion in savings in HIV care), or nearly $2 million per HIV infection averted (Table 2). These costs include all medical costs (HIV- and non-HIV-related) over the lifetime of the cohort. Antiretrovirals and monitoring associated with PrEP for 20% of MSM cost $98 billion over 20 years, or an average of $4.9 billion per year (Appendix Figure A3a). Costs increase as a higher percentage of MSM are initiated on PrEP. If 100% of MSM initiate PrEP for 20 years, healthcare-related costs increase by $480 billion.
Targeting high-risk MSM, rather than offering PrEP to the general MSM population, can improve the cost-effectiveness of PrEP. We estimated that 20% of MSM are high risk, with 5 annual partners on average, initial HIV prevalence of 20%, and initial annual incidence of 2.3%, for a representative high-risk population (28, 31, 45). PrEP use in 100% of these high-risk MSM costs $52,443/QALY gained relative to the status quo (Table 3). As with the general MSM population, there are diminishing returns associated with increasing the percentage of high-risk MSM on PrEP: if 20% of high-risk MSM use PrEP, the intervention costs $40,279/QALY gained relative to the status quo (Figure 1); with 50% access, PrEP costs $44,556/QALY gained relative to the status quo. PrEP is effective at preventing HIV infections in these scenarios, preventing 13%, 29%, and 52% of new infections in high-risk MSM with 20%, 50%, and 100% coverage, respectively (Table 3).
Use of PrEP in high-risk MSM costs less than does PrEP for all MSM. If all high-risk MSM initiate PrEP for a 20-year period, total healthcare-related costs incremental to the status quo are $75.5 billion, or approximately $600,000 per HIV infection averted. ARVs and monitoring for PrEP cost $85.2 billion over 20 years, or an average of $4.3 billion per year, while HIV treatment and care are associated with savings of nearly $10 billion due to fewer future infections (Table 3, Appendix Figure A3b). If fewer high-risk MSM use PrEP, as would likely be the case in practice, PrEP costs are lower. With 50% of high-risk MSM initiating PrEP, the costs of PrEP over 20 years are $41.9 billion, with total healthcare-related costs incremental to the status quo of $36.4 billion. With only 20% of high-risk MSM initiating PrEP, the costs of PrEP over 20 years are $16.6 billion, or an average of $828 million per year, with total healthcare-related costs incremental to the status quo of $14.2 billion. The cost per infection averted decreases at lower coverage levels of high-risk MSM, to $460,000 per infection averted at 20% coverage of high-risk MSM.
In sensitivity analysis, we found that PrEP cost, efficacy, and impact on quality of life considerably affected the cost-effectiveness of PrEP. One common concern with PrEP is resulting behavioral disinhibition (65, 66), but we found that the effectiveness and cost-effectiveness of PrEP are not substantially impacted by moderate changes in number of sexual partners and/or condom use induced by PrEP use (Appendix). We also found that initiating ART at CD4 cell counts of 0.500 × 109 cells/L, as opposed to at 0.350 × 109 cells/L as in our base case, did not qualitatively change the effectiveness and cost-effectiveness of PrEP (Appendix). The daily cost of TDF/FTC and its efficacy are key drivers of the cost-effectiveness of PrEP. PrEP use in 20% of the general MSM population costs less than $100,000/QALY gained if its daily cost drops below $15 or if PrEP is more than 75% effective at reducing risk of HIV acquisition, which could possibly be attained in highly adherent patients. The daily cost would have to nearly double in order for PrEP for high-risk MSM to cost more than $100,000/QALY gained. While in our base case we assumed that PrEP had no impact on quality of life, some patients may experience side effects. We found that PrEP must reduce the quality of life by more than 8% on average for the cost of PrEP for all high-risk MSM to rise above $100,000/QALY gained.
Since risk behaviors and HIV prevalence vary across subpopulations of MSM, we examined a range of risk levels in sensitivity analysis (Appendix). In a high-risk group with 5 annual partners on average, PrEP for all high-risk MSM costs less than $100,000/QALY gained as long as initial prevalence in the group is 8% or higher. For a medium-risk group with initial prevalence of 15% and 4 annual partners on average, PrEP for 100% of the group would cost approximately $100,000/QALY gained.
Both the cost and efficacy, and thus the cost-effectiveness, of PrEP depend on patient adherence. The iPrEx finding of 44% efficacy corresponded to approximately half of the study subjects reporting or exhibiting high adherence, as defined by pill use on at least 90% of days, but average patient adherence is unknown (3). For MSM with lower adherence, the cost of PrEP will likely be lower than in our base case, which includes the costs of daily pills, since non-adherent MSM will refill their ARVs less often. If MSM on PrEP on average take their daily pills only 50% of the time and efficacy is still 44%, PrEP for high-risk MSM costs $25,165/QALY gained, as compared to $52,443/QALY gained with the full daily pill cost. If average pill use is 75%, PrEP for high-risk MSM costs $38,804/QALY gained. If all MSM are highly adherent to PrEP, such that efficacy is 73% and the full daily pill cost is realized, PrEP for high-risk MSM costs $35,080/QALY gained. Figure 2b illustrates how the cost-effectiveness of PrEP for high-risk MSM varies as a function of cost and efficacy. Since average pill use will likely be less than daily in practice, use of PrEP in high-risk MSM likely costs less than $50,000/QALY gained. Figure 2a illustrates this relationship for the general MSM population.
Our analysis demonstrates that the use of PrEP for HIV prevention in MSM could have an important impact on the HIV epidemic in the United States. We show that even if only 20% of the MSM population in the United States were to initiate PrEP, more than 60,000 new HIV infections could be prevented over the next 20 years. However, PrEP is a costly intervention, and use in high-risk MSM will be integral to producing health benefits in an economically efficient manner (73). Initiating PrEP in 20% of the overall MSM population costs $172,091/QALY gained. In contrast, targeting PrEP to all high-risk MSM, whom we estimate comprise 20% of the overall MSM population, costs approximately $50,000/QALY gained and prevents a greater number of infections. If the daily cost of ARVs for PrEP decreased to $15, PrEP could be more cost-effective in a broader population of MSM, but at current costs, use of PrEP only in high-risk MSM is attractive. While the cost per QALY gained of PrEP in high-risk MSM is comparable to other interventions considered cost-effective, initiation of PrEP in all high-risk MSM would cost more than $4 billion per year for ARVs and monitoring.
Despite the good value provided by PrEP use in high-risk MSM, the budgetary impact will be large. Affordability of PrEP is uncertain, particularly in view of other competing priorities such as providing ART to infected individuals and new prevention technologies. Other interventions for MSM also provide good value but at lower expenditures, such as symptom-based viral load testing for acute HIV infection (7). In addition, if PrEP were implemented, many practical issues remain to be resolved, such as who would prescribe and who would pay for PrEP.
Because PrEP provides the most value in reducing HIV transmission when used in high-risk MSM, efficient clinical use of PrEP will depend on clinicians’ abilities to identify high-risk MSM. In MSM, number of partners and consistency of condom use are two key drivers of risk of HIV acquisition, and our results suggest that use of PrEP in MSM with multiple sexual partners can be cost-effective. To stratify patients by risk, a clinician can ask a patient how many partners he had in the past year. Additionally asking about condom use with non-primary partners could strengthen the risk assessment and inform the decision to offer PrEP to the patient. Assessing risk from these parameters is not uncommon with HIV prevention interventions (42, 62). Future studies on risk stratification of MSM could provide further guidance to clinicians in identifying high-risk MSM.
Our analysis finds that PrEP provides better value (73) when a smaller percentage of MSM initiate PrEP than when all MSM use PrEP. This phenomenon of diminishing returns is seen in other prevention programs as they are scaled up (74, 75). Secondary transmission benefits lead to substantial numbers of HIV infections averted even at low PrEP coverage. Thus, PrEP can have a clinically significant – and efficient – impact at relatively low coverage levels.
We considered daily oral PrEP. The efficacy of intermittent PrEP is currently being tested in clinical trials. If PrEP use only before high-risk behavior is shown to be effective at preventing HIV acquisition, this could substantially lower the cost of the ARVs associated with PrEP. Our examination of the cost-effectiveness of daily PrEP as a function of cost and efficacy can help estimate thresholds of cost and efficacy for intermittent PrEP to be cost-effective.
Previous studies reached different conclusions about the cost-effectiveness of PrEP in MSM. Paltiel et al. (5) found that PrEP use for MSM cost $298,000/QALY gained in a high-risk population with 1.6% annual HIV incidence, while Desai et al. (6) found that PrEP cost $32,000/QALY gained for high-risk MSM in New York City with 1.35% annual incidence. While Paltiel et al. suggest that PrEP is only cost-effective in very high incidence subgroups, we find lower incremental cost-effectiveness ratios for PrEP for high-risk MSM under a broader set of risk assumptions. Our results differ in part because of the extent to which each model captures the full benefits from reduced transmission due to PrEP and the length of time MSM remain on PrEP. While Paltiel et al. modeled PrEP efficacy as a flat percentage reduction in incidence and kept MSM on PrEP for their entire lifetime, we captured the dynamic effects of reduced transmission on incidence and we assumed that MSM would discontinue PrEP after 20 years or at age 65. When we match Paltiel et al.’s assumptions as closely as possible given our different modeling frameworks, we arrive at a similar conclusion, that PrEP is only cost-effective for the highest-risk subgroup of MSM. With our base-case model assumptions, our findings are more similar to those of Desai et al. (6). We both find that PrEP costs less than $100,000/QALY gained for high-risk MSM with annual incidence above 1.3%. A strength of our analysis is that we capture longer-term benefits of reduced transmission and incorporate a detailed economic analysis along with the epidemic modeling.
Our analysis has several limitations. The cost-effectiveness of offering PrEP to all MSM will depend on the propensity of MSM to use PrEP by risk level. For example, if high-risk MSM are more likely than low-risk MSM to use PrEP, offering PrEP to all MSM may be cost-effective, as in practice it will be as if PrEP is targeted to high-risk MSM. If, conversely, low-risk MSM are more likely than high-risk MSM to use PrEP, perhaps because they are more risk-averse, PrEP for the general MSM population may be even less cost-effective than in our base case. Second, high-risk MSM may mix with low-risk MSM and have networks of sexual partners, which we did not model when examining PrEP for high-risk MSM because of lack of data on mixing patterns. Consequently, we cannot directly compare strategies of PrEP for all MSM with PrEP for high-risk MSM. However, we can infer that PrEP for all MSM would be even less cost-effective when compared to PrEP for high-risk MSM than when compared to the status quo. With regard to sexual networks, to the extent that PrEP could be targeted to MSM centrally located in the networks, PrEP would be even more cost-effective. Third, we did not include in our model the possibility that PrEP may facilitate the development of drug-resistant HIV, since the iPrEx study found that no resistance developed in study subjects who acquired HIV during the trial (3). In sensitivity analysis, we estimated the effects of resistance in terms of the efficacy and cost of future treatment for those who acquire HIV while taking PrEP and found that resistance did not largely affect results. Fourth, we did not include potential renal impairment associated with PrEP. Since we included monitoring for elevated creatinine levels for patients on PrEP, per CDC guidelines, and TDF/FTC was not found to create long-term renal impairment in iPrEx participants (3), we do not anticipate this exclusion to impact results. Finally, we did not model HIV risk and benefits for MSM who also have female partners or who are injection drug users.
Our analysis demonstrates that PrEP use in MSM has the potential to prevent a considerable number of new HIV infections. However, PrEP use in the general MSM population is expensive. Use of PrEP in high-risk MSM provides substantial health benefits at a lower cost, although its budgetary impact is still sizeable. PrEP use in 100% of the estimated 20% of MSM who are at high risk of acquiring HIV infection averts more than twice as many infections as does PrEP use in 20% of the general MSM population, and provides better value. These findings can help inform clinicians’ decisions about whom to offer PrEP and policymakers’ decisions about recommendations for the use of PrEP.
This work was supported by Grant Number R01-DA15612 from the National Institute on Drug Abuse. Dr. Owens is supported by the Department of Veterans Affairs. Dr. Bendavid is supported by the National Institute of Allergy and Infectious Diseases (K01-AI084582).
Primary Funding Source: National Institute on Drug Abuse, Department of Veterans Affairs, National Institute of Allergy and Infectious Diseases, and Stanford University.