The HIV/AIDS response in South Africa is likely the largest and most costly in the world at around USD$2 billion in 2018 with almost 80% of all spending from domestic public funds.1 The epidemic is generalized, and although incidence is decreasing overall,1 prevalence and incidence vary geographically and within population subgroups. Consequently, investment in preventative interventions has focused on certain key populations; yet, men have been described as a “blind-spot” in the HIV response.2 South African men have been identified as one of the key drivers of the country's HIV epidemic due to comparatively low levels of health-seeking behavior and age disparate relationships, alongside gender norms and inequalities that contribute to the cycle of transmission.3 Yet, unless in a serodiscordant relationship, men have been largely absent from studies on new preventative technologies.
The South African government's combination prevention strategy was recently updated to include daily oral pre-exposure prophylaxis (PrEP) at select facilities to female sex workers (FSWs), men who have sex with men (MSM), serodiscordant couples, and adolescent girls and young women.4,5 A second PrEP formulation, cabotegravir, a long-acting injectable, was in phase III trials6–9 at time of analysis to test its efficacy at preventing HIV acquisition, as an alternative to oral PrEP.6 The effectiveness of PrEP is dependent on both biological determinants of drug concentration and adherence to the product. Oral PrEP, which is required to be taken daily under current South African guidelines, has poor adherence in some studies.10–12 Good adherence may be partly attributable to convenience of use; consequently, injectable PrEP administered every few weeks may result in better adherence than a daily oral pill.13
Although the country′s HIV response is comparatively well funded, existing programs are not effectively reaching heterosexual men in the general population. Shisana et al 14 argue for an expanded definition of key populations that recognizes the varied risks faced by members of the general South African population. Although overall incidence in South Africa is decreasing,1 the 2012 survey identified urban informal areas as having the highest HIV prevalence (20%) by locality type and Black African men as having the highest prevalence by race (15%).15 Nonbiomedical HIV prevention activities have had varying success at effectively reaching men, with only 34% of men aged 15–64 years medically circumcised.16
Only providing PrEP for heterosexual men who are in serodiscordant relationships may miss others at high risk of HIV acquisition. In addition, the considerable size and potential impact of the male heterosexual population on HIV transmission in South Africa makes this population group worth investigating. To date, there has been no study assessing the potential cost-effectiveness of PrEP introduction among heterosexual men in generalized epidemics. This study addresses this gap by conducting a cost-utility analysis for the provision of oral and injectable PrEP to South African men.
Static epidemiological and costing models were developed to estimate the health system costs and impact [disability-adjusted life years (DALYs) averted] associated with use of oral and injectable PrEP by heterosexual South African men under 3 intervention scenarios: (1) provision of oral PrEP, (2) provision of injectable PrEP, and (3) provision of dual PrEP (both oral and injectable PrEP) where men's preference for one modality over the other was considered using a discrete choice experiment (DCE) (model structure in Supplement A, Supplemental Digital Content, https://links.lww.com/QAI/B447). Results for each scenario were compared with a counterfactual of current practice (no PrEP). We take a simple, transparent modeling approach and focus on the impact of PrEP modalities in one year and do not model temporal reductions in the overall level of HIV transmission due to long-term product use, which may be sensitive to assumptions around adherence and retention. We model the one-year impact of introducing each product among heterosexual men in South Africa and compare the total and incremental costs and benefits of each of the 3 introduction scenarios over the life course. Summary model inputs are listed in Table 1 with full details available in Supplement B, Supplemental Digital Content, https://links.lww.com/QAI/B447.
Product Uptake and Study Cohorts
A DCE, conducted in South Africa in 2015, assessed the preferences of a random sample of periurban men for 3 HIV prevention products (condoms, oral and injectable PrEP).17 We used DCE data17 for the 95% of men reporting sexual attraction to women to estimate uptake of daily oral PrEP and 3-monthly injectable PrEP for 2 cohorts: younger men aged 18–24 years and older men aged 25–49 years (see Supplement D, Supplemental Digital Content, https://links.lww.com/QAI/B447). These age cohorts were selected due to the availability of epidemiological incidence data for South African men in similar age groups.
Central prevalence and incidence for each cohort were taken from the 2017 HSRC survey and weighted by population.16,18,19 Prevalence and incidence for men aged 15–24 years was assumed to be equal to the incidence for the 18- to 24-year-old cohort (Table 1).16
Data from a demonstration project on the provision of PrEP among South African FSW were used to establish costs associated with the provision of PrEP.10 Costing calculations and assumptions are detailed in Supplement B, Supplemental Digital Content, https://links.lww.com/QAI/B447. In brief, we use primary cost data from a PrEP demonstration project among FSW in South Africa, take a health system perspective to include all relevant treatment and hospitalisation costs, and account for variation between first- and second-line antiretroviral therapy (ART). All cost parameters were adjusted for inflation and adjusted to 2018 USD. Lifetime costs were applied after the first year modeled and discounted at 3% based on life expectancy and adjusted for anticipated inflation using the January 2018 inflation rate.
The incremental cost-effectiveness ratios (ICERs) from each scenario were compared with a conservative cost-effectiveness threshold of USD$1175/DALY averted. This threshold was taken from the lowest estimates of Woods et al 20 following critiques of cost-effectiveness thresholds based on Gross Domestic Product (GDP).21–23
ART coverage was assumed to be unchanged from the current national coverage of 61% of HIV-positive individuals1; however, the effect of variable coverage rates on the ICER was explored in a sensitivity analysis.
Prevention Product Efficacy
The effectiveness of any HIV prevention product was estimated as the product of efficacy, correct use, and adherence (Table 1. More details in Supplement C, Supplemental Digital Content, https://links.lww.com/QAI/B447). Published evidence on product effectiveness was used to calculate central, best case, and worse case effectiveness estimates. As injectable PrEP was still in the trial stage, efficacy per dose was assumed to be similar to oral PrEP.
Calculating Protective Effect
As condoms act as a physical barrier and PrEP is pharmacological, the model assumes that protective effect of multiple products is additive. The final protective effect (P) of a PrEP product (denoted i = 1…m) under each intervention scenario (s) was determined using Equation 1 adapted from Quaife et al.17 E0U0 represents base case protection from existing condom use (U0) at current efficacy (E0), α is the estimated proportional decrease in condom use among previous condom users who now use PrEP, and PrEP efficacy (Ei,c) and uptake (Ui,c) varies between PrEP products and among condom users (c = 1) and noncondom users (c = 0). The formula is below, with further details in Supplement C, Supplemental Digital Content, https://links.lww.com/QAI/B447.
The protective effect is equivalent to the reduction in incidence resulting from the use of PrEP and any change in condom use (see Supplement E, Supplemental Digital Content, https://links.lww.com/QAI/B447). This protective effect was applied to the baseline incidence to determine the new incidence rate associated with each PrEP scenario compared with the counterfactual. The new incidence was calculated as:
Disability-Adjusted Life Years
Disability weightings associated with HIV and ART take-up were taken from the 2013 Global Burden of Disease Report (see Supplement B, Supplemental Digital Content, https://links.lww.com/QAI/B447). Years lived with disability were calculated for both cohorts, accounting for ART coverage and discounted at 3%. It was assumed that those not on treatment progressed to symptomatic HIV after the first model year while those on treatment were initiated immediately. All those with HIV were assumed to experience an AIDS health state for 2 years before death. Age-specific weighting was not used in the DALY calculation. Total DALYs averted were calculated by multiplying the number of infections averted by the intervention by the average discounted lifetime DALYs averted accounting for age and ART coverage rate.
The ICER of providing PrEP in each of the 3 intervention scenarios given a counterfactual of current practice was calculated as net costs divided by the DALYs averted per scenario. ICERs were calculated for varying incidence levels (central and bounds), and further one-way deterministic sensitivity analyses were conducted using upper and lower bound model parameters. Two-way deterministic sensitivity analyses simultaneously varied 2 parameters that caused the most variation in ICERs to determine further uncertainty (see Supplement H, Supplemental Digital Content, https://links.lww.com/QAI/B447). In addition, a threshold analysis was conducted on the ICER for each intervention scenario and each male subgroup to identify the minimum incidence required to produce a cost-effective result (see Supplement G, Supplemental Digital Content, https://links.lww.com/QAI/B447). Finally, a probabilistic sensitivity analysis sampled parameter values 1000 times with results presented as cost-effectiveness acceptability curves.
The DCE study was reviewed and approved by the University of the Witwatersrand Human Research Ethics Committee (M140614) and the Research Ethics Committee at the London School of Hygiene and Tropical Medicine (8541-2).
The DCE indicates that PrEP uptake may be higher among noncondom users (90% in younger men and 89% in older men) than condom users (59% in younger men and 56% in older men), and that oral PrEP is more preferred than injectable PrEP.
Averted Infections, DALYs, and Costs
Based on a potential susceptible nationwide population of 4 million younger men and 9 million older men, an estimated 3668 infections in younger men could be averted in 1 year through the introduction of dual PrEP while more than 4 times as many infections (16,786) could be averted with the same intervention in older men. Introducing oral PrEP could avert 24,511 DALYs in younger men and 99,548 DALYs in older men. By contrast, less DALYs were averted if only injectable PrEP was introduced (see Supplement F, Supplemental Digital Content, https://links.lww.com/QAI/B447).
Incremental Cost-Effectiveness Ratios
The analysis found that, at a central incidence estimate, all calculated ICERs were not cost-effective as a threshold of $1175/DALY averted (Fig. 1). However, for all intervention scenarios, the ICER of providing PrEP to older men was better than providing PrEP to younger men, with introducing oral PrEP to older men having the lowest ICER ($2873 per DALY averted) (Fig. 1). Interestingly, the mean ICERs did not vary much across the different intervention scenario for each age subgroup (Fig. 1).
Deterministic Sensitivity Analyses
A one-way deterministic sensitivity analysis found the model results were robust to most parameter variations, but highly sensitive to a few. The ICER for each scenario was most sensitive to HIV incidence, varying the ICER by as much as 150% for younger men and 211% for older men (Fig. 2 and see Supplement G, Supplemental Digital Content, https://links.lww.com/QAI/B447). The ICER for interventions among younger men was also sensitive to ART coverage, adherence to PrEP products, and the efficacy of PrEP products, although none of the sensitivity analyses resulted in the ICER going below the cost-effectiveness threshold, ie, making products cost-effective (Fig. 2). However, when HIV incidence was increased in older men, the ICER fell below the cost-effectiveness threshold for oral and dual PrEP intervention scenarios. The model was also highly sensitive to uncertainty around ART coverage, product efficacy, and adherence. Among older men, the model was also sensitive to age at infection. Additional two-way sensitivity analyses are presented in Supplement H, Supplemental Digital Content, https://links.lww.com/QAI/B447.
We note that the sensitivity analysis for increases in incidence is mathematically equivalent to including a multiplier for onward infections in a static model. In this case, the 45% (12%) increase in incidence assumed among younger (older) men is the same as assuming that every HIV infection averted by PrEP will also avert 0.45 (0.12) onward infections. PrEP tends toward marginal cost effectiveness at these upper bounds for older men, although not for younger men.
Probabilistic Sensitivity Analysis
Probabilistic sensitivity analysis results are displayed in the cost-effectiveness acceptability curves in Figure 3 and demonstrate that oral PrEP is more likely to be cost-effective under assumptions of high incidence, but only for those over 25 years.
Overall, results indicate that expanding PrEP to all South African men is unlikely to be cost-effective at a willingness-to-pay threshold of USD $1175/DALY averted (see Supplement J, Supplemental Digital Content, https://links.lww.com/QAI/B447). By contrast, PrEP was found elsewhere to be highly cost-effective for South African FSW and women aged 15–24 years at the same threshold.17 Moreover, as ART coverage increases with the national push toward the 90% coverage target, PrEP is likely to become even less cost-effective if current incidence remains constant (see Supplement H, Supplemental Digital Content, https://links.lww.com/QAI/B447). This mirrors the analysis in South Africa's investment case for HIV and tuberculosis which found that the expansion of PrEP to adolescent girls and young women, FSW, and serodiscordant couples was less cost-effective than alternative spending options.40
This analysis found that considerably more DALYs could be averted by extending PrEP to older men than younger men; however, this may be explained by older men having a larger population cohort, a higher incidence rate, or higher predicted PrEP uptake than younger men. In particular, incidence was an important determinant of cost-effectiveness. According to the WHO, “PrEP should be a priority for populations with an HIV incidence of about 3 per 100 person-years or higher.” 41 However, this study demonstrated that a <2% incidence in the target population could result in a cost-effective intervention (shown in sensitivity analyses of Supplement G, Supplemental Digital Content, https://links.lww.com/QAI/B447). Incidence is substantially harder to measure than prevalence, and data are not currently publicly available at granular levels, such as by age group, sex, and race for men. A study in 2017 suggests that black men in informal urban areas have higher incidence rates than their rural or white counterparts,16 but exact data for our cohorts are not available. Ideally, further studies would build more heterogeneity into the population cohort to determine whether PrEP is cost-effective to specific male population subsets which could be effectively targeted by PrEP programs, eg, age or geographical heterogeneity.
Without the DCE, uptake parameters across cohorts and products would be assumptions or based on expert opinions, which are uncertain. Inclusion of men's stated preferences considers patterns of heterogeneity in use (eg, among condom users) within the targeted population37,42 and can aid in determining the true cost implications of introducing PrEP.43 Although, using stated preference data may introduce hypothetical bias since users are not observed.43 However, systematic review evidence shows that DCEs can predict health choices with imperfect but substantive accuracy.43
Furthermore, the DCE found condom use likely to decrease with the uptake of PrEP making it important that programs continue to emphasize couse, particularly if users cannot fully adhere to PrEP. This analysis assumed high levels of adherence compared with observational studies in other populations. A study among FSW found adherence has been observed to be low (70%) and loss to follow-up high (77%) after 12 months.10 In studies with serodiscordant couples (including heterosexual men), adherence was higher at 89%,35 and in a recent study on MSM in Australia, adherence was 85%.36 This analysis assumed retention on both injectable and oral PrEP was equal, although comparatively high.
At a willingness-to-pay threshold of USD$1,175, this study does not support a policy where PrEP is expanded to heterosexual South African men; however, we acknowledge several limitations to this study. First, the choice of threshold greatly impacts recommendations. South Africa does not yet have a standard cost per DALY averted threshold, so we used the lower bound threshold from Woods et al.20 However, this threshold was calculated with a number of assumptions and sources of uncertainty, and at Wood's upper bound estimate $4714 per DALY averted,20 all PrEP scenarios for older men were cost-effective. Alternatively, the South African HIV investment case considers interventions cost-effective based on life-year saved (LYS) as benchmarked against costs of expanding UTT (∼$1000 per LYS).40 Meyer-Rath calculated that actual spending on HIV interventions in South Africa used a threshold of $547–$872 per LYS.44 There is therefore substantial uncertainty in South Africa's true willingness to pay for new health investments, which would help inform analyses such as this.
Second, this model has structural limitations as future DALYs and costs averted from preventing HIV infections by having fewer infectious people (especially younger women) in the population is not considered. This understates PrEP's cost-effectiveness. Furthermore, South African men are known to test for HIV less regularly and initiate treatment later than women,3 but this model does not consider the downstream benefits of men's PrEP-associated access to health services such as initial HIV testing, regular HIV testing and STI screening, and earlier ART initiation.34 We do not model how improved treatment regimens and adherence could reduce the proportion of infectious HIV-positive persons. Similarly, we do not model changes in other preventative interventions such as male circumcision, and further research is needed to estimate HIV incidence among circumcised and noncircumcised men. This model also does not use age-weighted DALYs which would have made cost estimates for infection averted in younger men more favorable.
As in a similar model,17 although our simple epidemiological model omits future costs and benefits of PrEP expansion, it is effective at providing transparent estimates of the individual benefits of PrEP for men, likely similar to short-term results should this intervention have been trialed.17 A dynamic transmission model would have provided better insight into the long-term implications of PrEP expansion; however, projections would be strongly reliant on uncertain assumptions around disease transmission. Furthermore, Eaton et al's.45 analysis of 12 dynamic models of ART uptake in South Africa found considerable variability in predictions and inaccurate predictions of future infections, suggesting that increased complexity is not necessarily a guarantee of accuracy.
There is uncertainty in the costs associated with the roll out of PrEP (see Supplement I, Supplemental Digital Content, https://links.lww.com/QAI/B447). Costs used in this study were drawn from cost estimates for the provision of PrEP to FSW in select government clinics. If PrEP was made available to men, it is not known where it would distributed, or how potentially lower utilization may affect costs. Older men in particular may need extra encouragement to attend services if PrEP is provided from government facilities that traditionally cater for women. In addition, injectable PrEP (cabotegravir) is still under trial,46 and as a result, the efficacy and market price of the product is yet to be determined. Even if approved, cabotegravir will remain subject to patent law and originator prices unless price negotiations are agreed upon. This may prevent the product from being available in South Africa. Finally, PrEP has known side effects 47,48; however, the DALYs and associated care costs due to taking PrEP are unknown and not factored into this model. To understand the true cost implications of expanded PrEP, additional data need to be collected to value these.
Finally, there are ethical considerations around expanding PrEP to those who may benefit. Some may argue that there is a human rights imperative to making PrEP available to anyone who wants it. This could even possibly result in the most at-risk men self-selecting to take PrEP or being more adherent to the product, which would make PrEP more cost-effectiveness than estimated here.
This is the first study known to evaluate the cost-effectiveness of expanding PrEP to heterosexual men not in serodiscordant couples in a setting when HIV is a generalized epidemic. Although our findings indicate neither oral nor injectable PrEP is cost-effective at a threshold of USD1,175/DALY averted, there may still be potential for select subpopulations such as those with high HIV incidence or low ART coverage to benefit. Future studies should explore the impact of variation in HIV risk and uptake on cost-effectiveness, particularly the role of geographical and age-related heterogeneities.
1. UNAIDS. Country: South Africa. 2018. Available at: http://www.unaids.org/en/regionscountries/countries/southafrica
. Accessed August 13, 2018.
2. On World AIDS Day, UNAIDS warns that men are less likely to access HIV treatment and more likely to die of AIDS-related illnesses. Geneva, Switzerland: UNAIDS; 2017.
3. UNAIDS. A Snapshot of Men and HIV in South Africa. Geneva, Switzerland: UNAIDS; 2017.
4. National Department of Health and Implementing Partners, PrEP Implementation Pack: South Africa 2016–2017. Pretoria, South Africa: OPTIONS Consortium and the South Africa Department of Health; 2017.
5. National Policy on HIV Pre-exposure Prophylaxis (PrEP) and Test and Treat (T&T)—Final Draft. Pretoria, South Africa: Department of Health; 2016.
6. Cabotegravir, AIDS Info: US Dept. Of Health and Human Services. 2017. Available: https://aidsinfo.nih.gov/drugs/513/cabotegravir/0/patient
. Accessed July 03, 2018.
7. Clinical trials: evaluating the safety, tolerability, and pharmacokinetics of an investigational, injectable HIV medicine (GSK1265744) in HIV-uninfected adults. AIDSinfo: US Dept Health Hum Serv; 2018. Available: https://aidsinfo.nih.gov/clinical-trials/details/NCT02178800
. Accessed August 23, 2018.
8. Clinical Trials: Safety and Efficacy Study of Injectable Cabotegravir Compared to Daily Oral Tenofovir Disoproxil Fumarate/Emtricitabine (TDF/FTC), for Pre-exposure Prophylaxis in HIV-Uninfected Cisgender Men and Transgender Women Who Have Sex with Men. AIDSinfo: US Dept of Health and Human Services; 2018. Available: https://aidsinfo.nih.gov/clinical-trials/details/NCT02720094
. Accessed August 23, 2018.
9. Clinical Trials: Evaluating the Safety and Efficacy of Long-Acting Injectable Cabotegravir Compared to Daily Oral TDF/FTC for Pre-exposure Prophylaxis in HIV-Uninfected Women. AIDSinfo: US Dept of Health and Human Services; 2018. Available: https://aidsinfo.nih.gov/clinical-trials/details/NCT03164564
. Accessed: August 23, 2018.
10. Eakle R, Gomez GB, Naicker N, et al. HIV pre-exposure prophylaxis and early antiretroviral treatment among female sex workers in South Africa: results from a prospective observational demonstration project. PLoS Med. 2017;14:e1002444.
11. Baeten JM, Donnell D, Ndase P, et al. Antiretroviral prophylaxis for HIV prevention in heterosexual men and women. N Engl J Med. 2012;367:399–410.
12. 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.
13. Glaubius RL, Hood G, Penrose KJ, et al. Cost-effectiveness
of injectable preexposure prophylaxis for HIV prevention in South Africa. Clin Infect Dis. 2016;63:539–547
14. Shisana O, Zungu N, Evans M, et al. The case for expanding the definition of “key populations” to include high-risk groups in the general population to improve targeted HIV prevention efforts. S Afr Med J. 2015;105:664–669.
15. Shisana O, Rehle T, Simbayi LC, et al. South African National HIV Prevalence, Incidence and Behaviour Survey, 2012. Cape Town, South Africa: HSRC Press; 2014.
16. HSRC, the Fifth South African National Prevalence, Incidence, Bahaviours, and Communication Survey, 2017. Cape Town, South Africa: HSRC Press; 2018.
17. Quaife M, Terris-Prestholt M, Eakle R, et al. The cost-effectiveness
of multi-purpose HIV and pregnancy prevention technologies in South Africa. J Int AIDS Soc. 2018;21:e25064.
18. Mid-year Population Estimates. Pretoria, South Africa: Statistics South Africa; 2018.
19. TBFacts.org. HIV Statistics for South Africa—Prevalence, Incidence, ARVs. 2018. Available at: https://www.tbfacts.org/hiv-statistics-south-africa/
. Accessed: April 12, 2019.
20. Woods B, Revill P, Sculpher M, et al. Country-Level Cost-Effectiveness
Thresholds: Initial Estimates and the Need for Further Research. York, United Kingdom: Centre for Health Economics, University of York; 2015.
21. Bertram MY, Lauer JA, Joncheere K, et al. Cost–effectiveness thresholds: pros and cons. Bull World Health Organ. 2016;94:925–930.
22. Shillcutt SD, Walker DG, Goodman CA, et al. Cost effectiveness in low- and middle-income countries. Pharmacoeconomics. 2009;27:903–917.
23. Marseille E, Larson B, Kazi DS, et al. Thresholds for the cost-effectiveness
of interventions: alternative approaches. Bull World Health Organ. 2015;93:118–124.
24. Medicine Price Registry, Database of Medicine Prices 12/7/18. Pretoria, South Africa: Department of Health; 2018.
25. ARV Market Report: The State of the Antiretroviral Drug Market in Low-And Middle-Income Countries. Boston, MA: Clinton Health Access Initiative; 2017.
26. Meyer-Rath G, Brennan AT, Fox MP, et al. Rates and cost of hospitalization before and after initiation of antiretroviral therapy in urban and rural settings in South Africa. J Acquir Immune Defic Syndr. 2013;62:322–328.
27. Venter F. Can We Improve First-Line and Second-Line ART? Johannesburg, South Africa: Wits Reproductive Health and HIV Institute; 2018.
28. Meyer-Rath G, Johnson LF, Pillay Y, et al. Changing the South African national antiretroviral therapy guidelines: the role of cost modelling. PLoS One. 2017;12:e0186557.
29. Oanda Currency Converter—Foreign Exchange Rates. Available at: https://www.oanda.com/currency/converter/
. Accessed: July 12, 2018.
30. Inflation.eu Worldwide Inflation Data, Historic Inflation South Africa. Available at: https://www.inflation.eu/inflation-rates/south-africa/historic-inflation/cpi-inflation-south-africa.aspx
. Accessed: July 19, 2018.
31. Gumede A. South Africa's Inflation Rate Slows to 4.4% in January. Bloomberg online; 2018.
32. Claxton K, Revill P, Sculpher M, et al. The Gates Reference Case for Economic Evaluation. York: The Bill and Melinda Gates Foundation; 2014.
33. Mortality and Causes of Death in South Africa, 2016: Findings from Death Notification. Statistical Release P0309.3. Pretoria, South Africa: Statistics South Africa; 2018:138.
34. Johnson LF, Mossong J, Dorrington RE, et al. Life expectancies of South African adults starting antiretroviral treatment: collaborative analysis of cohort studies. PLoS Med. 2013;10:e1001418.
35. Thigpen MC, Kebaabetswe PM, Paxton LA, et al. Antiretroviral preexposure prophylaxis for heterosexual HIV transmission in Botswana. N Engl J Med. 2012;367:423–434.
36. Grulich A, Guy RJ, Amin J, et al. Rapid reduction in HIV diagnoses after targeted PrEP implementation in NSW, Australia. CROI conference, Abstract 8, 2018.
37. Quaife M, Eakle R, Cabrera Escobar MA, et al. Divergent preferences for HIV prevention: a discrete choice experiment for multipurpose HIV prevention products in South Africa. Med Decis Making. 2018;38:120–133.
38. Thembisa Project: Model 3.2. 2017. Available: https://www.thembisa.org/downloads
. Accessed: July 14, 2018.
39. USAID. USAID—Condom Fact Sheet. Washington, DC: USAID; 2015.
40. South Africa HIV and TB Investment Case—Summary Report Phase 1. Department of Health and South African National AIDS Council; 2016.
41. WHO. Policy Brief: WHO Expands Recommendation on Oral Pre-exposure Prophylaxis of HIV Infection. Geneva, Switzerland: PrEP; 2015.
42. Terris-Prestholt F, Quaife M, Vickerman P. Parameterising user uptake in economic evaluations: the role of discrete choice experiments. Health Econ. 2016;25(suppl 1):116–123.
43. Quaife M, Terris-Prestholt F, Di Tanna GL, et al. How well do discrete choice experiments predict health choices? A systematic review and meta-analysis of external validity. Eur J Heal Econ. 2018;19:1053–1066.
44. Meyer-Rath G, Van Rensburg C, Larson B, et al. Revealed willingness-to-pay versus standard cost-effectiveness
thresholds: evidence from the South African HIV Investment Case. PLoS One. 2017;12:1–9.
45. Eaton JW, et al. HIV treatment as prevention: systematic comparison of mathematical models of the potential impact of antiretroviral therapy on HIV incidence in South Africa. PLoS Med. 2012;9:e1001245.
46. ViiV Healthcare Starts Third Phase III HIV Treatment Study Investigating Long-Acting Two-Drug Regimen of Cabotegravir Plus Rilpivirine. London, United Kingdom: GSK.
47. Haberer JE, Johnson LF, Salomon JA, et al. Adherence to antiretroviral prophylaxis for HIV prevention: a substudy cohort within a clinical trial of serodiscordant couples in east Africa. PLoS Med. 2013;10:e1001511.
48. Desai M, Field N, Grant R, et al. Recent advances in pre-exposure prophylaxis for HIV. BMJ. 2017;359:j5011.