|Home | About | Journals | Submit | Contact Us | Français|
World Health Organization guidelines for antiretroviral treatment (ART) in resource-limited settings recommend either stavudine or tenofovir as part of initial therapy. We evaluated the clinical outcomes and cost-effectiveness of first-line ART using tenofovir in India, compared to current practice using stavudine or zidovudine.
We used a state-transition model of HIV disease to examine strategies using different nucleoside reverse transcriptase inhibitors, combined with lamivudine and nevirapine, compared to no ART: 1) stavudine; 2) stavudine, with substitution by zidovudine after six months; 3) zidovudine; 4) tenofovir. Data were from the Y.R. Gaitonde Centre for AIDS Research and Education in Chennai, India and published studies.
Discounted mean per person survival was 36.9 months (40.1 months undiscounted) with no ART, 115.5 months (145.3) with stavudine-containing ART, 115.6 months (145.5) with stavudine and six-month zidovudine substitution, 115.7 months (145.6) with zidovudine-containing ART, and 125.9 months (162.2) with initial tenofovir. Discounted lifetime medical costs were $610 with no ART and ranged from $5,560 with stavudine-containing ART to $5,720 with zidovudine-containing ART. Initial tenofovir had an incremental cost-effectiveness ratio of $670/year of life saved compared to no ART and was more economically efficient than the other regimens. Results were most sensitive to variations in the costs of first-line tenofovir, access to additional ART after failure, mean initial CD4 count, and quality of life adjustment.
Using tenofovir as part of first-line ART in India will improve survival, is cost-effective by international standards, and should be considered for initial therapy for HIV-infected patients in India.
There are approximately 2.5 million people in India living with HIV infection . In the US and Europe, first-line therapy usually includes tenofovir – a nucleotide analogue that is more effective and considerably less toxic than stavudine, which is not recommended . Current (World Health Organization) WHO guidelines for ART in resource-limited settings recommend either stavudine or tenofovir as part of initial combination therapy . In much of Asia and other resource-limited settings, the most common first-line regimens for HIV-infected patients still contain either stavudine or zidovudine, which are less costly than tenofovir [3,4]. While effective, stavudine is associated with several toxicities including lipodystrophy, neuropathy, pancreatitis, and the rarer but potentially fatal lactic acidosis; anemia is a common toxicity from zidovudine . In comparison, tenofovir is relatively well tolerated, though it may be associated with nephrotoxicity . Tenofovir has been found to be more effective than other nucleoside reverse transcriptase inhibitor (NRTI) agents . Recent price negotiations have reduced the cost of tenofovir in many developing countries, but access remains limited .
The National AIDS Control Organization of India has published a detailed plan for scaling up HIV prevention and treatment, stating a goal that all HIV-infected individuals have access to comprehensive HIV care by 2011 . As both the government and private agencies are increasing ART access throughout India, there is an opportunity to provide evidence-based guidance on the optimal selection of treatment regimens. Our objective was to project the clinical outcomes, cost, and cost-effectiveness of various first-line ART regimens in India.
We adapted a published model of HIV disease and treatment in international settings, including southern India . To evaluate the benefits and costs associated with different first-line ART regimens, we simulated four strategies for the NRTI component, along with lamivudine and nevirapine, each compared to no ART. The four ART strategies included cotrimoxazole prophylaxis and: 1) a stavudine-containing regimen with zidovudine substituted for stavudine for specified grade 3 or 4 toxicities (severe lactic acidosis, neuropathy, and lipodystrophy) (“initial stavudine”) ; 2) a stavudine strategy with zidovudine uniformly substituted after six months to minimize the risk of lipodystrophy and lactic acidosis (“initial stavudine-to-zidovudine”) ; 3) a zidovudine strategy with a switch to stavudine, if and when anemia developed (“initial zidovudine”); and 4) a tenofovir-containing regimen in which zidovudine was substituted if nephrotoxicity developed (“initial tenofovir”). We assessed the clinical and cost outcomes associated with each strategy. All patients had access to two sequential “lines” of ART, reflecting national guidelines in India . Second-line ART, which included tenofovir or zidovudine and a protease inhibitor (PI), was initiated following first-line ART failure. We summarized results using the incremental cost-effectiveness ratio: the increase in cost divided by the increase in years of life saved (YLS), or quality-adjusted life years (QALYs) saved, compared to the next lower cost alternative . Results are reported in 2005 US dollars per life year saved. The analysis was from a modified societal perspective, excluding costs of patients’ time and travel, and all costs and life years were discounted at 3% per year .
Using first-order Monte Carlo simulation, the model tracks patients selected randomly from an initial distribution of age, sex, CD4 count, HIV RNA level, and history of opportunistic infection (OI). Initiation of ART decreases HIV RNA levels and, for those who are virologically suppressed, increases CD4 counts, which then decreases the probability of OIs and mortality. (See Appendix for references and additional model details.)
The Commission on Macroeconomics and Health of the WHO has suggested that cost-effectiveness ratios below three times the per capita gross domestic product (GDP) in a country (US$1,860 in India) might be considered “cost-effective,” and ratios below the GDP itself ($620 in India) might be considered “very cost-effective .” We used the three times GDP threshold as a comparative benchmark. The study was approved by the Ethics Board at the Y.R. Gaitonde Centre for AIDS Research and Education (YRG CARE) in Chennai, India and the Partners HealthCare Human Subjects Committee, in Boston, MA, USA.
Patients in the simulated cohort had demographic and disease characteristics based on an HIV-infected cohort of approximately 7,000 patients from YRG CARE, the largest non-governmental HIV care center in India, between 1996 and 2004 (Table 1) [10,16]. Mean age was 33 years; baseline mean CD4 count was 318/μL (SD 291/μL); and 66% of the subjects were male.
In the absence of clinical trial data from India, the effectiveness of alternative ART regimens was from a published observational cohort in India and from clinical trials in the US (Table 1). For first-line regimens including either stavudine or zidovudine, we used 24-week virologic suppression of 73% (<400 copies/mL) and, for those suppressed, a CD4 count increase of 132 cells/μL at 24 weeks; once suppressed, probability of virologic failure after 48 weeks (“long-term”) was estimated at 5%/month [10,17]. For first-line tenofovir, we used 85% suppression at 24 weeks and assumed the same immunologic benefit as for the other regimens (132 cells/μL) [7,17]. Tenofovir had an estimated long-term virologic failure probability of 1%/month [7,18]. We incorporated local strategies for managing reported toxicities, including single-drug substitutions of another NRTI for patients who developed grade 3 or 4 toxicity (or after six months for stavudine) (Table 1) [11,12]. Those who experienced virologic failure switched to an appropriate second-line regimen (Appendix).
We conducted sensitivity analyses on all major parameters in the model, including the costs of first- and second-line ART; access to only one or three lines of ART; effectiveness of tenofovir compared to other agents; second-line ART stopping criteria; probability of and time to severe lactic acidosis and other ART toxicities; and initial mean CD4 count. Because of limited data on nephrotoxicity from tenofovir in resource-limited settings, we conducted two-way sensitivity analyses varying the incidence of nephrotoxicity in both first- and second-line ART from 1.6% (base case) to 16% and mortality from 1.0% to 90%. We also assigned utilities reflecting quality of life (QOL) to all health states and ART toxicities, and we projected quality-adjusted life expectancy. Utility weights ranging from 0.0 (death) to 1.0 (perfect health) were derived from studies conducted in the US or clinical judgment .
In a cohort with mean CD4 count of 318 cells/μL, co-trimoxazole treatment alone (no ART) led to a projected mean discounted life expectancy of 36.9 months (40.1 months undiscounted) with discounted lifetime cost of $610 (Table 2). Initial stavudine led to a discounted life expectancy of 115.5 months (145.3 months) at a cost of $5,560. Initial stavudine-to-zidovudine resulted in a mean discounted life expectancy of 115.6 months (145.5 months) at a cost of $5,570. Initial zidovudine led to a discounted life expectancy of 115.7 life months (145.6 months) at a cost of $5,720. Finally, initial tenofovir increased life expectancy to 125.9 months (162.2 months) with a total cost of $5,590 (Table 2).
Initial stavudine was associated with 280 deaths from ART toxicity per 100,000 people, compared to 130 deaths/100,000 when used with a six-month zidovudine substitution, 60/100,000 deaths from initial zidovudine, and 10/100,000 deaths from initial tenofovir (Table 2, right column).
Initial stavudine and initial stavudine-to-zidovudine had higher cost-effectiveness ratios than the next more costly strategies, representing an economically inefficient use of resources (“weakly dominated”) . Initial zidovudine resulted in lower life expectancy and higher cost than initial tenofovir, also representing an inefficient use of resources. Initial tenofovir had an incremental cost-effectiveness ratio of $670/YLS compared to no ART (co-trimoxazole alone).
With access to three lines of ART, discounted life expectancy for initial tenofovir increased to 132.0 months, and the incremental cost-effectiveness of initial tenofovir increased to $830/YLS compared to no ART. When we decreased the cost of the lopinavir/ritonavir component of second-line ART by 25–75%, the lifetime cost for all first-line alternatives was reduced. When these costs were decreased by 75%, initial tenofovir became even more cost-effective, at $390/YLS compared to no ART.
We assumed that all first-line regimens had equal 48-week effectiveness but varied the probability of long-term virologic failure for tenofovir compared to initial stavudine or zidovudine (Table 3) . When we reduced this advantage by 25–100% (ultimately rendering equal the efficacy of all first-line alternatives), tenofovir remained the only economically efficient strategy, with incremental cost-effectiveness ratios of $700-$740/YLS compared to no ART.
We then varied nephrotoxicity from tenofovir in both first- and second-line ART. When the frequency of nephrotoxicity was 1.6% in either first- or second-line regimens containing tenofovir, even a mortality rate from nephrotoxicity as high as90% had little impact on mean total life expectancy (Figure 1) or the cost-effectiveness results (Table 3).
We also varied the mean initial CD4 count between 50 and 500 cells/μL. A mean initial CD4 count of 500 cells/μL decreased the incremental cost-effectiveness ratio of initial tenofovir to $590/YLS, while a much lower initial CD4 count of 50 cells/μL increased the ratio to $920 cells/YLS.
In results adjusted to reflect quality of life, initial tenofovir resulted in a discounted life expectancy of 86.4 quality-adjusted life months (QALMs; 110.4 undiscounted) and a cost-effectiveness ratio of $1,000/QALY compared to no ART. All other strategies were economically inefficient. We performed the same adjustments for a scenario in which patients had access to only one line of ART, yielding similar results (Table 3).
We varied the cost of tenofovir-containing first-line ART from its base case of $14/month down to $7/month and up to $100/month, with commensurate changes in the tenofovir component of second-line ART for those who fail initial stavudine or initial zidovudine (Figure 2). When two lines of ART were available, initial tenofovir was associated with an incremental cost-effectiveness ratio less than three times per capita GDP in India. If only one line of ART was available, then the cost-effectiveness results were more sensitive to changes in tenofovir cost, but initial tenofovir remained a cost-effective option throughout a plausible range of costs.
Alternative ART stopping criteria for patients who failed second-line ART (stop immediately upon virologic failure or upon severe OI) did not substantially affect the cost-effectiveness results (Table 3). Other parameters with limited impact included the probability of lactic acidosis resulting from stavudine and its associated mortality, quality of life decrements from ART toxicity, time to ART toxicity, duration of neuropathy and lipodystrophy, and mean initial CD4 count.
Newer ART agents, including tenofovir, are both highly effective and safer than older NRTIs, but stavudine remains the backbone of first-line ART regimens in India and many other resource-limited countries due to its lower cost [7,22]. We conducted a cost-effectiveness analysis of first-line ART regimens in India, focusing on the role of tenofovir. We found that initial tenofovir increased life expectancy by about 10.4 months compared to stavudine or zidovudine. At the current price of tenofovir-containing ART ($14/person/month), the incremental cost-effectiveness ratio for initial tenofovir in India was $670/YLS compared to no ART. The alternative strategies represented a less efficient use of resources. The main determinants of these results were the cost of tenofovir, access to and cost of additional lines of ART, and mean initial CD4 count. Sensitivity analyses adjusted for quality of life increased the incremental cost-effectiveness ratio of the tenofovir-based strategy to $1,000/YLS.
Some studies suggest that tenofovir may be associated with nephrotoxicity and potentially renal failure . Moreover, renal disease is more prevalent in HIV-infected patients than the general population, regardless of ART exposure . Several US-based trials show no increase in nephrotoxicity from tenofovir with careful monitoring and minimal long-term consequences from any decline in renal function [24,25]. However, renal toxicity may pose higher risks in resource-limited settings, where experience with tenofovir and access to laboratory monitoring are limited, and where death from acute renal failure may approach 90% [26,27]. Sensitivity analyses on nephrotoxicity within a clinically plausible range did not offset the virologic advantages conferred by tenofovir, suggesting that the benefits would outweigh any potential harm.
In a cost-effectiveness analysis on the use of tenofovir for first-line ART in South Africa, Rosen et al. found that tenofovir reduced toxicity and had a high incremental cost-effectiveness ratio of about $9,000/QALY compared to stavudine using monthly costs of $17 for tenofovir and $3.39 for stavudine in US$2007 . There are several important differences between the Rosen and the current study including treatment efficacy, treatment cost, and analysis time frame. The current analysis used data reflecting improved effectiveness and smaller differences in cost between initial tenofovir and other first-line alternatives, and it evaluated clinical and cost outcomes over a lifetime rather than just 2 years.
We also examined a scenario in which patients had access to only one line of ART, reflecting current limitations in ART rollout capacity, as well as a scenario anticipating access to three lines of ART. We found that the increased efficacy of first-line tenofovir over the alternatives improved its cost-effectiveness when only one line of ART was available. We observed the opposite, though modest, effect when patients had access to a costly third-line regimen. The cost-effectiveness results were stable despite changes in the cost of initial tenofovir when two lines of ART were available, likely because the higher cost of initial tenofovir was offset by the alternative with other strategies, a higher-cost, tenofovir-containing second-line regimen. When only one line of ART was available, the cost of initial tenofovir was more important. These findings together emphasize the importance of policies enhancing access to second-line, and eventually third-line, treatment in optimizing the use of limited resources for HIV treatment. Patients with better adherence may be at greater risk for developing ART toxicities, since drug exposure is a risk factor for adverse events . Toxicities such as lipodystrophy and neuropathy may decrease both adherence and the durability of first-line ART, though this is not clear [30,31]. To address these uncertainties, we varied the effectiveness advantage of tenofovir [7,17]. Although some have shown that tenofovir and stavudine have similar virologic outcomes, we used data that likely reflect lower adherence over time for patients on stavudine due to more chronic toxicities, and we tested this assumption in sensitivity analyses .
A very small study suggested that tenofovir combined with nevirapine rather than efavirenz may reduce virologic suppression . If these early findings are substantiated, the selection of a non-nucleoside reverse transcriptase inhibitor (NNRTI) for use with tenofovir should take this into account .
The discounted total lifetime costs of three of the ART strategies (initial tenofovir, initial stavudine, and initial stavudine-to-zidovudine), were similar and all around $5,600. Patients who experienced virologic failure on initial tenofovir switched to a less expensive second-line regimen containing zidovudine. Because patients spend the remainder of their lives on the second-line regimen, total costs were about the same as for those who started on a less costly initial regimen. This suggests that second-line ART costs are more important contributors to total lifetime costs than first-line ART costs.
This analysis has several limitations. First, there are limited published data comparing the effectiveness of alternative ART regimens in resource-limited settings. For stavudine- and zidovudine-based first-line ART, we used data from an observational cohort in India, but for initial tenofovir we used US clinical trials, which may reflect better outcomes. We assumed that decreased long-term adherence is implicitly included in the higher long-term virologic failure rates for initial stavudine and zidovudine, and we examined this assumption in extensive sensitivity analyses. In a sensitivity analysis examining equal effectiveness (short-term and long-term) for all first-line ART regimens, we found that tenofovir resulted in equal discounted life expectancy as initial zidovudine, but tenofovir remained the only efficient first-line ART alternative, with an incremental cost-effectiveness ratio of $740/YLS compared to initial stavudine. We did not explicitly model loss to follow-up, but we did examine the consequence of reduced adherence on treatment outcomes using the long-term failure parameter and found that it would render tenofovir more attractive.
Second, there are limited data about ART toxicities in resource-limited settings. We used published data from India on adverse events related to stavudine and zidovudine but US-based estimates for nephrotoxicity and other toxicities. Sensitivity analyses (Figure 1) suggest that nephrotoxicity is unlikely to alter the finding that tenofovir represents an attractive component of first-line ART in resource-limited settings.
Third, this study did not examine outcomes for individuals co infected with HIV and hepatitis B. Given the burden of hepatitis B in India, this analysis likely underestimates the benefits of tenofovir-based therapy [34,35].
Finally, there are scant data on the quality of life impact of ART in India . We found that including plausible quality of life effects increased the cost-effectiveness ratio of tenofovir-based first-line ART to $1,000 compared to no ART from the base case ratio of $670/YLS, but this higher ratio still represents the most efficient use of resources and meets criteria for cost-effectiveness. We examined the quality of life impact for only selected health states in the model.
Determining the best use of limited resources for HIV care is challenging, because the additional expense of initial tenofovir could potentially mean that some HIV-infected patients may not receive needed treatment. Thus, while initial tenofovir may be deemed cost-effective by WHO criteria, it could challenge short-term budget constraints in some countries.
This study has several implications for guidelines on first-line ART in India. While stavudine and zidovudine are effective components of first-line ART, recent substantial price reductions in tenofovir make it a very viable treatment option. As the cost of tenofovir continues to decline, its use will become even more attractive. Incorporating tenofovir as part of first-line ART in India will improve survival, is cost-effective by international standards, and should be considered for HIV-infected patients in India.
We thank the Y.R. Gaitonde Centre for AIDS Research & Education, Chennai, India, and especially A.K. Ganesh and the Natural History Study team at YRG CARE. (Ms. Rasmi and Ms. Glory, research nurses; Mr. S. Anand, data manager; Mr. Guru and Mr. Siva, data entry personnel). We are also indebted to the Scientific Advisory Board for CEPAC-International.
We are grateful to all of the CEPAC-International investigators for their contributions to this project, including:
Y.R. Gaitonde Centre for AIDS Research & Education, Chennai, India: A.K. Ganesh. We thank the Natural History Study team at YRGCARE. (Ms. Rasmi and Ms. Glory, research nurses; Mr. S. Anand, data manager; Mr. Guru and Mr. Siva, data entry personnel).
Harvard School of Public Health, Boston, MA: Kara Cotich, Sue J. Goldie, April D. Kimmel, Marc Lipsitch, Chara Rydzak, George R. Seage III, Milton C. Weinstein, Service Universitaire des Maladies Infectieuses et du Voyageur, Centre Hospitalier de Tourcoing, Faculté de Médecine de Lille, France, Laboratoire de Recherches Economiques et Sociales, CNRS URA 362, Lille, France: Yazdan Yazdanpanah, Sylvie Burban INSERM U593, Bordeaux, France: Xavier Anglaret, Roger Salamon, Delphine Gabillard, Hapsatou Touré
Programme PAC-CI, Abidjan, Côte d’Ivoire: Xavier Anglaret, Thérèse N’Dri-Yoman, Siaka Touré, Catherine Seyler, Eugène Messou, Christine Danel, Eric Ouattara University of Cape Town, Cape Town, South Africa: Robin Wood, Catherine Orrell Perinatal HIV Research Unit, WITS Health Consortium, Johannesburg, South Africa: Glenda Gray, James McIntyre, Neil A. Martinson, Lerato Mohapi
Yale University, New Haven, Connecticut: A. David Paltiel
Massachusetts General Hospital, Harvard Medical School, Boston, MA: Wendy Aaronson, Ingrid Bassett, John Chiosi, Jennifer Chu, Andrea Ciaranello, Nomita Divi, Mariam O. Fofana, Heather E. Hsu, Benjamin Linas, Elena Losina, Bethany Morris, Brandon Morris, Anjali Saxena, Caroline Sloan, Lauren Uhler, Stacie Waldman, Lindsey L. Wolf.
We are also indebted to the Scientific Advisory Board for CEPAC-International, including: Richard Chaisson (Johns Hopkins Medical School, Baltimore, Maryland, USA); Victor De Gruttola (Harvard School of Public Health, Boston, Massachusetts, USA); Joseph Eron (University of North Carolina, Chapel Hill, North Carolina, USA); Raman R. Gangakhedkar (National AIDS Research Institute (NARI), Pune, India; Jonathan Kaplan (Centers for Disease Control and Prevention, Atlanta, Georgia, USA); Salim Karim (University of KwaZulu Natal, Durban, South Africa); Therese N’Dri Yoman (University of Cocody-Abidjan, Abidkan, Côte d’Ivoire); Douglas Owens (Stanford University, Palo Alto, California, USA); John Wong (Tufts-New England Medical Center, Boston, Massachusetts, USA).
Funding Source: Supported by the National Institute of Allergy and Infectious Diseases (K01 AI074495 and T32 AI007433 to M.A.B, K24AI062476 and R01 AI058736 to K.A.F.), Harvard University Center for AIDS Research (P30 AI060354 to M.A.B.), National Institute on Drug Abuse (K01 DA017179 to B.R.S.), Doris Duke Charitable Foundation Operations Research on AIDS Care and Treatment in Africa (ORACTA) Award (2007015 to K.A.F. and M.A.B.), Doris Duke Charitable Foundation Clinical Scientist Development Award (2005075 to R.P.W.), Harvard Medical School Eleanor and Miles Shore 50th Anniversary Fellowship Program for Scholars in Medicine Award (to M.A.B.), and the Lifespan/Tufts/Brown University Center for AIDS Research (P30AI042853 to T.F. and K.H.M.).
Details of the structure of the CEPAC-International model have been described elsewhere . Briefly, patients transition one by one through the state transition model, which stratifies patients by both CD4 count and HIV RNA [1,2]. These characteristics, as well as the transition probabilities between disease states in the absence of treatment, are based on published cohorts [3,4]. Patients may develop opportunistic infections (OI), and they transition between acute health states (OI, drug toxicity) and chronic health states until death. CD4-specific data on incidence of major and minor OIs in untreated HIV infection were derived from the YRG CARE cohort . ART initiation reduces HIV RNA levels and increases CD4 counts for those patients with a virologic response. This rise in CD4 count then decreases the probability of OIs and mortality [6,7]. We derived age- and sex-specific background mortality estimates from WHO life tables for India .
Those individuals meeting clinical criteria for virologic failure on stavudine- or zidovudine-based regimens switched to a second-line regimen containing tenofovir, emtricitabine, and boosted lopinavir, with a 48-week virologic suppression rate of 68%, a CD4 cell count increase of 121 cells/μL, and an estimated long-term probability of virologic failure of 2%/month [9,10]. Patients who failed initial tenofovir switched to zidovudine/lamivudine with boosted lopinavir; in the absence of long-term data on this combination, we used the same virologic and immunologic outcomes as for the other second-line regimen. ART failure was defined as a 50% fall in CD4 count from peak for first-line ART, a 90% fall from peak for second-line ART, or any severe OI after six months on ART(OI) [11,12]. CD4 counts we re measured every six months.
We used two lines of ART in the base case analysis, but we examined through sensitivity analyses the potential impact of three lines of ART. In these sensitivity analyses, we assumed that 24-week virologic suppression, CD4 count increase, and probability of long-term suppression were the same as for the second-line regimen, and we used a cost of $80/month.
For tenofovir-containing second-line ART, we used a monthly probability of nephrotoxicity of 0.04, a mean time to toxicity of six months, and for those who developed nephrotoxicity, we assumed a 1% probability of mortality [10,13]. For zidovudine-containing second-line ART, we used a monthly probability of anemia of 0.133 and a mean time to toxicity of four months . For both second-line regimens, we used a probability of grade 3/4 gastrointestinal toxicity of 0.032, and a mean time to toxicity of 11 months .
ART costs were converted from US$2008 to US$2005 using exchange rates of 42.38 (US$2008 to 2008 Indian rupee) and 0.02288 (2005 Indian rupee to US$2005) as well as the appropriate GDP deflators [16,17]. Health care utilization for each patient was multiplied by the daily costs for inpatient and outpatient care, and all costs were converted to US$2005 [12,18].
For severe lactic acidosis resulting from stavudine, we used a probability of acute mortality of 0.4 . For sensitivity analyses on quality of life, we used the best available data from the US, or clinical judgment (Table A1) .
Model runs simulate one million patients with each strategy to ensure stable model outcomes including OIs, toxicity, mean life expectancy, as well as total lifetime costs and cost-effectiveness. The model is programmed in the C++ programming language.
We varied the relative 24-week virologic effectiveness advantage of initial tenofovir over initial stavudine between 16% (the base case) and 0% (no advantage). Eliminating the effectiveness advantage – while maintaining the long-term (after 48-week) effectiveness compared to other first-line regimens – reduced discounted life expectancy from 125.9 life months to 121.3 life months and discounted lifetime costs from $5,590 to $5,510. This had minimal impact on the cost-effectiveness results: initial tenofovir remained the only efficient strategy, with an incremental cost-effectiveness ratio of $700/YLS compared to no ART.
Potential conflicts of interest. All authors: no conflicts.