The US Food and Drug Administration (FDA) recently approved the use of tenofovir–emtricitabine (TDF/FTC) as pre-exposure prophylaxis (PrEP) in populations who are susceptible to HIV by sexual transmission, including MSM and serodiscordant heterosexual couples . The FDA relied primarily on the results of two international randomized trials: the iPrEX trial  conducted in MSM in North and South America, Thailand, and South Africa, which found a 44% reduction in HIV incidence with daily TDF/FTC (73% among fully adherent participants) and the Partners PrEP study  in serodiscordant heterosexual couples in Africa which found a 73% reduction in HIV incidence (100% among fully adherent participants) with the same regimen.
Additional trials are investigating the efficacy of PrEP in women in sub-Saharan Africa using a variety of dosing strategies. The CAPRISA 004 trial  conducted in South Africa found a 39% reduction in HIV incidence among high-risk women administering a 1% TDF vaginal microbicide gel before and after intercourse (50% among women who used the product at least 80% of the time). In the FEM-PrEP trial of once-daily, oral TDF/FTC, however, adherence was too low to allow assessment of efficacy . Ongoing trials of oral and vaginal microbicide PrEP (VOICE , FACTS 001 ) will provide further insights when completed.
Given the potentially high cost of providing PrEP to populations at risk for HIV infection and the possible long-term risks and benefits associated with use of antiretroviral therapy (ART) drugs in HIV-uninfected persons , cost–effectiveness models are a valuable tool to assess this new HIV prevention strategy. Cost–effectiveness models provide a comparative assessment of the value of a health intervention, measured in quality-adjusted life years (QALYs), disability-adjusted life years (DALYs), or years of life saved (YLS), compared with its cost. These models are useful in identifying health programs with greatest value to society and projecting long-term program impacts. A threshold of US$100 000 per QALY is increasingly considered to be cost effective in the USA  and has recently been used to evaluate HIV interventions [10–12]. In resource-limited countries, a threshold of one times GDP per capita (e.g., US$8200 in South Africa) is considered very cost effective and three times GDP per capita is considered cost effective . A finding of cost–effectiveness below these thresholds, however, does not imply that a program is cost saving. Cost–effectiveness is typically evaluated from the societal perspective and does not directly address the cost impact on specific budgets, which requires a budget impact analysis . Even programs that are cost-saving over the long term for society as a whole will often incur costs to specific budgets over a shorter time horizon.
We identified six recent modeling studies of PrEP that reported cost–effectiveness results: three in the USA and three in South Africa. Below, we discuss modeling considerations relevant to both USA and South African studies, separately review the findings for each country, and conclude with some suggestions for future research.
Of the six modeling studies we reviewed, two used an individual-level state transition model of HIV disease, the Cost–Effectiveness of Preventing AIDS Complications (CEPAC) model [15,16], and the remainder used HIV population-level transmission models  (Fig. 1). The advantage of an individual-level model of HIV disease is that a detailed simulation of the course of disease allows for varying individual inputs such as the time from HIV infection to detection, linkage to care, treatment efficacy, and costs of care. The disadvantage of an individual-level model is that it does not provide a population framework to project population-level benefits from reducing the risk of HIV transmission, either by reducing the likelihood of infection among HIV-uninfected individuals through prevention interventions such as PrEP or by reducing HIV viral load among HIV-infected individuals through treatment. These effects can be simulated in population-level transmission models containing compartments of infected and susceptible individuals and transmission event probability inputs. The level of detail and complexity of transmission models can vary, however, and substantial differences in results have been observed depending on model structure and assumptions .
Key inputs for all models of PrEP include the demographics of the target population (age, sex), HIV incidence in the absence of PrEP, effectiveness of PrEP in reducing incidence, duration of PrEP, PrEP side effects, HIV testing frequency with and without PrEP, and time of ART initiation with and without PrEP. As noted above, effectiveness of PrEP in clinical trials is consistently associated with adherence. Hence, embedded in any effectiveness estimate is an assumption about adherence in the target population. Although not observed in clinical trials to date, there is also a risk of behavioral disinhibition among HIV-uninfected individuals receiving PrEP whereby they increase their level of risk behavior (e.g., reduce their use of condoms) compared with the behavior that would have otherwise occurred had they not been receiving PrEP. This may be a particular consideration in practice, as opposed to clinical trials in which individuals receiving PrEP are uncertain about whether the drug they are taking is active or placebo.
Behavioral disinhibition can be modeled relatively easily, however, as reduced efficacy of PrEP. Similarly, there is a theoretical risk of drug resistance among individuals who become HIV-infected despite being on PrEP, because PrEP does not contain a fully suppressive antiretroviral ART regimen once HIV infection has been established. Although resistance has not been observed among PrEP failures in clinical trials to date, it can be modeled by reducing the efficacy of subsequent ART among a portion of PrEP failures.
Both individual-level and transmission models require ART testing and linkage to care inputs, but they have different implications for the models. In individual-level models, these parameters determine the likelihood and timing of HIV detection and treatment of individual HIV infections that occur if PrEP is not implemented versus if PrEP is implemented. In transmission models, the parameters also determine the likelihood of an HIV transmission occurring depending on the proportion of HIV-infected partners successfully linked to care and achieving suppressed HIV viral load, and hence lower infectivity. Transmission models also require initial input parameters that can change over time due to transmission dynamics (HIV prevalence) or policy decisions (uptake of other prevention activities such as condom use, timing of PrEP roll-out in the population).
COST INPUTS AND DISCOUNTING
Cost inputs should include the cost of the PrEP intervention, costs of HIV testing in the absence of PrEP, and costs of HIV care. PrEP intervention costs include not only the recommended dose of ART drugs but also laboratory monitoring (including HIV tests, metabolic tests, etc.) and clinician visits at specified frequencies including adherence monitoring or risk-reduction counseling as appropriate. Although most models assume costs based on full adherence to medication and monitoring visits, sensitivity analyses on cost assumptions can reflect potential lower utilization of these resources more consistent with expected behavior in a nonclinical trial setting. Start-up costs for PrEP can be significant, such as provider education and recruitment activities. Other than initial eligibility screening, however, these costs are not usually considered in cost–effectiveness models. They should be considered, however, in budget impact studies.
Both costs and QALYs should be discounted in accordance with accepted standards in economic evaluation of healthcare interventions, typically at a rate of 3% per year . Discounting reflects individuals’ time preference for benefits received now versus later; it is not the same as correcting for inflation (which should be conducted separately if data sources for costs are from different years) . Discounting presents a particular challenge for economic evaluation of prevention interventions such as PrEP, because spending usually occurs immediately but benefits such as costs saved and QALYs gained from reduced transmissions of HIV occur several years in the future. Discounting is one reason why it is difficult for PrEP to be cost saving even when considering a long time horizon of costs and benefits.
COST–EFFECTIVENESS OF PRE-EXPOSURE PROPHYLAXIS IN HIGH-RISK MSM IN THE UNITED STATES
Three recent studies modeled the cost–effectiveness of once-daily TDF/FTC as an HIV-prevention strategy in high-risk MSM (Table 1) [21,22,23▪▪]. Desai et al.  developed a compartmental transmission model using a 5-year time horizon to determine PrEP efficacy in urban MSM. They used published epidemiologic and survey data from New York City to inform their model and stratified the population into four age and four sexual risk classes. High-risk MSM were those who in the past 6 months reported unprotected sex with an HIV-infected person, unprotected sex in exchange for money or drugs, anonymous sex, five or more sexual or needle-sharing partners, or were diagnosed with a sexually transmitted disease. The authors assumed PrEP efficacy of 50% and coverage of 25% of high-risk MSM. This article reported the lowest base case cost–effectiveness ratio for PrEP in high-risk MSM (US$31 970 per QALY in 2007 US dollars), reflecting a 5-year cost horizon for PrEP but an assumption of lifetime cost and QALY benefit for each HIV infection prevented during the 5-year period (i.e., no subsequent HIV infection after completion of PrEP).
Paltiel et al.  (including one of the authors of this review) used the CEPAC individual-level state transition Monte Carlo simulation model of HIV disease to evaluate the cost–effectiveness of PrEP in high-risk MSM using a lifetime time horizon. Data on HIV incidence in high-risk MSM were from the HIV Network for Prevention Trials (HIVNET) Vaccine Preparedness study  and varied with age. The authors assumed PrEP efficacy of 50%, resistance occurring in individuals who became HIV-infected on PrEP and reducing subsequent ART efficacy, and that in the absence of PrEP, all high-risk MSM were screened for HIV annually. PrEP toxicity was explored in sensitivity analyses. This article reported the highest base case cost–effectiveness ratio for PrEP in high-risk MSM (US$298 000 per QALY in 2006 US dollars), reflecting lifetime use and cost of PrEP, consistent uptake of HIV testing in the absence of PrEP, and a model that does not include secondary transmission benefits of HIV infections prevented.
Most recently, Juusola et al. [23▪▪] used a compartmental transmission model that also takes into account HIV progression and ART treatment to evaluate the cost–effectiveness of PrEP in general MSM and high-risk MSM populations over a 20-year time horizon. High-risk MSM were defined as having an average of five annual partners, 20% initial prevalence of HIV, and initial annual HIV incidence of 2.3% based on survey data. They assumed PrEP efficacy of 44% based on the iPrEx study  and that 67% of MSM were screened annually in the absence of PrEP. Resistance and PrEP toxicity were explored in sensitivity analyses. This study reported base case cost–effectiveness ratios of PrEP in high-risk MSM in 2010 US dollars of US$40 279 per QALY with 20% coverage, US$44 556 per QALY with 50% coverage, and US$52 443 per QALY with 100% coverage. In contrast, cost–effectiveness ratios for PrEP in the general MSM populations were between US$172 091 and 216 480 per QALY depending on the coverage. These cost–effectiveness ratios, which fall between those reported by Desai et al. and those reported by Paltiel et al., reflect the intermediate time horizon (20 years) and coverage of HIV testing in the absence of PrEP (67%), the inclusion of secondary transmission benefits, and the highest base case incidence of HIV in high-risk MSM (see Table 1).
Despite the apparent differences in base case cost–effectiveness ratios, there are several consistent findings among the three articles. Overall, PrEP is most attractive when targeted to high-risk MSM with HIV incidence of greater than 2%, with a cost–effectiveness ratio less than US$100 000 per QALY, and less attractive in general MSM populations with incidence less than 1%, with a cost–effectiveness ratio greater than US$200 000 per QALY. High-risk MSM may be targeted based on younger age, having five or more annual partners, and not currently being screened for HIV annually. In all models, the cost–effectiveness of PrEP was very sensitive to PrEP effectiveness, which can be affected by drug characteristics, adherence, or potentially by behavioral disinhibition. Cost–effectiveness was also very sensitive to the cost of PrEP, which can be affected by drug cost as well as monitoring costs and adherence. Results were less sensitive to potential drug resistance due to PrEP and PrEP toxicity at reasonable values. The cost of implementing PrEP in high-risk MSM in the USA was only explored in the article by Juusola et al., who estimated that the cost of 20% coverage in high-risk MSM for 20 years would be US$16.6 billion and would generate US$4.4 billion in savings to the healthcare system, for a net cost of US$14.2 billion. In comparison, the annual budget for the Ryan White CARE Act that supports services for HIV-infected uninsured and underinsured individuals in the USA is US$2.3 billion .
COST–EFFECTIVENESS OF PRE-EXPOSURE PROPHYLAXIS IN SOUTH AFRICA
Three recent studies examined the cost–effectiveness of PrEP in South Africa. The purpose, interventions studied, target populations, and study designs varied widely. Pretorius et al.  used a transmission model of the generalized HIV epidemic in South Africa to evaluate the interaction between implementing PrEP and expanding ART availability in South Africa between 2014 and 2025. The model projects relatively low HIV incidence based on current ART expansion alone (0.8% by 2014) and high PrEP efficacy (90%). The authors then varied coverage assumptions for PrEP expansion in the HIV-uninfected population and ART expansion in the HIV-infected population after 2014 and found that the additional benefit of PrEP remained independent of ART expansion until the South African national ART program coverage reached three times its 2010 level, or higher coverage levels if PrEP was targeted at women age 25–35 years who were at higher risk of HIV infection. The authors did not conduct a detailed cost–effectiveness analysis but discussed the economic implications of their findings assuming an annual PrEP cost of US$150 and an annual ART cost of US$600. Estimated cost–effectiveness ratios for PrEP ranged from US$12 500 per infection avoided to more than US$20 000 per infection avoided as coverage levels of the PrEP and ART programs varied.
Hallett et al. [26▪] constructed a transmission model to examine the interactions between oral PrEP and earlier initiation of ART among HIV serodiscordant couples in South Africa, comparing results based on data from the Partners in Prevention HSV/HIV Transmission Study [27,28] to a set of assumptions about ‘more typical couples’ with less condom use and more external partners. For Partners in Prevention couples the PrEP cost per HIV infection averted was US$6000–66 000 depending on PrEP efficacy (range 50–80%) and duration of PrEP use (always, until 1 year after ART initiation in the HIV-infected partner, or until ART initiation in the HIV-infected partner). For more typical couples, the results varied from cost saving to US$21 000 per HIV infection averted. Cost per QALY outcomes are also reported, but the methods used to derive them are not described in detail.
A limitation of both of these studies is that they do not provide details on costing inputs and do not indicate whether discounting was used, and their primary cost–effectiveness results are reported as costs per HIV infection averted. A strength of both studies is that they provide an economic framework to consider the interactions between PrEP and expansion of ART coverage in South Africa on a population level.
Walensky et al. [29▪▪] used the CEPAC individual-level state transition Monte Carlo simulation model to examine the cost–effectiveness of a 1% TDF vaginal microbicide gel with 39% efficacy in South African women based on the results of the CAPRISA 004 trial. The authors assumed lifetime use of PrEP and took into account potential HIV resistance with PrEP failure and PrEP side effects. PrEP program costs included 1% TDF gel (based on US$0.32 per dose, the recommended two doses per sex act and an average 7.2 acts per month), monthly HIV testing, and bi-annual chemistry panels. In the base case, the cost–effectiveness of PrEP was US$2700 per YLS or US$1600 per YLS if HIV testing in the PrEP program occurred at a more realistic frequency of every 3 months. Results were sensitive to PrEP efficacy, costs, and annual HIV incidence in the target population, and less sensitive to HIV resistance and PrEP toxicity. In order to be cost-saving, PrEP would need to be targeted to higher risk women and have either higher efficacy or lower cost. The authors project that to achieve PrEP coverage for half of HIV-uninfected South African women for 5 years would cost US$6.0 billion. In comparison, the estimated South African government annual budget for HIV/AIDS was US$2.0 billion in 2009 . The PrEP costs may be overestimated, however, due to the assumptions of consistent use of PrEP and the fact that the model does not consider secondary HIV transmissions avoided as a result of PrEP.
Recent studies have used either individual-level state transition or population-level transmission models to demonstrate that PrEP is a cost-effective HIV prevention alternative when targeted at high-risk MSM in the USA or young South African women. The studies all highlight the importance of PrEP efficacy and cost as key drivers of cost–effectiveness. Real-world implementation considerations will have a major impact on efficacy and cost, including the level of uptake in high-risk groups, adherence, frequency of monitoring, insurance coverage and access, and duration of PrEP. In addition to examining implementation, future economic studies should include more detailed budget impact analyses that consider the potential impact of PrEP on different health system and health insurer budgets. Finally, all of the studies highlight the interactions between PrEP for HIV-uninfected individuals and ART for HIV-infected individuals. Future studies should evaluate PrEP as part of combination HIV strategies including expanded testing and access to treatment.
The authors wish to acknowledge Rochelle Walensky, MD, MPH, for assistance in preparing the conference presentation that was the basis for this review and Jared Leff, MS, for assistance in preparation of the manuscript.
Conflicts of interest
The authors report no financial conflicts of interest.
REFERENCES AND RECOMMENDED READING
Papers of particular interest, published within the annual period of review, have been highlighted as:
▪ of special interest
▪▪ of outstanding interest
Additional references related to this topic can also be found in the Current World Literature section in this issue (p. 612).
2. Grant RM, Lama JR, Anderson PL, et al. Preexposure chemoprophylaxis for HIV prevention in men who have sex with men. N Engl J Med 2010; 363:2587–2599.
4. Abdool Karim Q, Abdool Karim SS, Frohlich JA, et al. Effectiveness and safety of tenofovir gel, an antiretroviral microbicide, for the prevention of HIV infection in women. Science 2010; 329:1168–1174.
5. Van Damme L, Corneli A, Ahmed K, et al
. Preexposure prophylaxis for HIV infection among African women. N Engl J Med 2012; 367:411–422.
7. US National Institutes of Health. Safety and effectiveness of tenofovir gel in the prevention of human immunodeficiency virus (HIV-1) infection in women and the effects of tenofovir gel on the incidence of herpes simplex virus (HSV-2) infection NCT01386294. 2012. http://clinicaltrials.gov/ct2/show/NCT01386294?term=FACTS+001&rank=1
. [Accessed 20 June 2012]
8. Paxton LA, Hope T, Jaffe HW. Pre-exposure prophylaxis for HIV infection: what if it works? Lancet 2007; 370:89–93.
9. Grosse SD. Assessing cost-effectiveness in healthcare: history of the $50 000 per QALY threshold. Expert Rev Pharmacoecon Outcomes Res 2008; 8:165–178.
10. Braithwaite RS, Fiellin DA, Nucifora K, et al. Evaluating interventions to improve antiretroviral adherence: how much of an effect is required for favorable value? Value Health 2010; 13:535–542.
11. Farnham PG, Sansom SL, Hutchinson AB. How much should we pay for a new HIV diagnosis? A mathematical model of HIV screening in US clinical settings. Med Decis Making 2012; 32:459–469.
12. Gopalappa C, Farnham PG, Hutchinson AB, Sansom SL. Cost-effectiveness of the national HIV/AIDS Strategy (NHAS) goal of increasing linkage to care for HIV-infected persons. J Acquir Immune Defic Syndr 2012; 61:99–105.
14. Mauskopf JA, Sullivan SD, Annemans L, et al. Principles of good practice for budget impact analysis: report of the ISPOR Task Force on good research practices – budget impact analysis. Value Health 2007; 10:336–347.
16. Ciaranello AL, Perez F, Maruva M, et al. WHO 2010 guidelines for prevention of mother-to-child HIV transmission in Zimbabwe: modeling clinical outcomes in infants and mothers. PLoS One 2011; 6:e20224.
17. Pretorius C, Stover J, Bollinger L, et al. Evaluating the cost-effectiveness of pre-exposure prophylaxis (PrEP) and its impact on HIV-1 transmission in South Africa. PLoS One 2010; 5:e13646.
18. Baggaley RF, Fraser C. Modelling sexual transmission of HIV: testing the assumptions, validating the predictions. Curr Opin HIV AIDS 2010; 5:269–276.
19. Gold MR, Siegel JE, Russell LB, Weinstein MC. Cost effectiveness in health and medicine. Oxford, UK:Oxford University Press; 1996.
20. Drummond MF, Schulpher MJ, Torrance GW, et al. Methods for the economic evaluation of healthcare programmes. 3rd ed.Oxford, UK:Oxford University Press; 2005.
21. Desai K, Sansom SL, Ackers ML, et al. Modeling the impact of HIV chemoprophylaxis strategies among men who have sex with men in the United States: HIV infections prevented and cost-effectiveness. AIDS 2008; 22:1829–1839.
22. Paltiel AD, Freedberg KA, Scott CA, et al. HIV preexposure prophylaxis in the United States: impact on lifetime infection risk, clinical outcomes, and cost-effectiveness. Clin Infect Dis 2009; 48:806–815.
23▪▪. Juusola JL, Brandeau ML, Owens DK, Bendavid E. The cost-effectiveness of pre-exposure prophylaxis for HIV prevention in the United States in men who have sex with men. Ann Intern Med 2012; 156:541–550.
This recent study used a transmission model to conduct a cost-effectiveness analysis of PrEP in US MSM over a 20-year time horizon and concluded that PrEP is cost-effective when targeted to high-risk MSM. It clearly defines the model inputs and assumptions; a good example of a well written cost-effectiveness study using a transmission model.
24. Seage GR 3rd, Holte SE, Metzger D, et al. Are US populations appropriate for trials of human immunodeficiency virus vaccine? The HIVNET Vaccine Preparedness Study. Am J Epidemiol 2001; 153:619–627.
26▪. 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.
This is an interesting study that evaluates PrEP in HIV-1 serodiscordant couples in South Africa in the context of current ART guidelines and an earlier ART initiation strategy using a transmission model. The authors report results for two different cohorts: Partners in Prevention HSV/HIV Transmission Study participants and ‘more typical couples’ in South Africa who exhibit more high-risk behaviors.
27. Lingappa JR, Kahle E, Mugo N, et al. Characteristics of HIV-1 discordant couples enrolled in a trial of HSV-2 suppression to reduce HIV-1 transmission: the partners study. PLoS One 2009; 4:e5272.
28. Lingappa JR, Baeten JM, Wald A, et al. Daily acyclovir for HIV-1 disease progression in people dually infected with HIV-1 and herpes simplex virus type 2: a randomised placebo-controlled trial. Lancet 2010; 375:824–833.
29▪▪. Walensky RP, Park JE, Wood R, et al. The cost-effectiveness of pre-exposure prophylaxis for HIV infection in South African women. Clin Infect Dis 2012; 54:1504–1513.
Another recent, well written article that clearly defines the model inputs including incidence, efficacy, and all costs. This study used an individual-level state transition model of PrEP in young South African women and found tenofovir-based vaginal microbicide to be cost-effective but only cost-saving if targeted at high-risk women and either more effective or costing less.
30. Country Progress Report on the Declaration of Commitment on HIV/AIDS. 2010. (Accessed 14 September 2012, at http://data.unaids.org/pub/Report/2008/south_africa_2008_country_progress_report_en.pdf.)