Figures
Abstract
Background
Genotypic antiretroviral resistance testing (GRT) in HIV infection with drug resistant virus is recommended to optimize antiretroviral therapy, in particular in patients with virological failure. We estimated the clinical effect, cost and cost-effectiveness of using GRT as compared to expert opinion in patients with antiretroviral treatment failure.
Methods
We developed a mathematical model of HIV disease to describe disease progression in HIV-infected patients with treatment failure and compared the incremental impact of GRT versus expert opinion to guide antiretroviral therapy. The analysis was conducted from the health care (discount rate 4%) and societal (discount rate 2%) perspective. Outcome measures included life-expectancy, quality-adjusted life-expectancy, health care costs, productivity costs and cost-effectiveness in US Dollars per quality-adjusted life-year (QALY) gained. Clinical and economic data were extracted from the large Swiss HIV Cohort Study and clinical trials.
Results
Patients whose treatment was optimized with GRT versus expert opinion had an increase in discounted life-expectancy and quality-adjusted life-expectancy of three and two weeks, respectively. Health care costs with and without GRT were $US 421,000 and $US 419,000, leading to an incremental cost-effectiveness ratio of $US 35,000 per QALY gained. In the analysis from the societal perspective, GRT versus expert opinion led to an increase in discounted life-expectancy and quality-adjusted life-expectancy of three and four weeks, respectively. Health care costs with and without GRT were $US 551,000 and $US 549,000, respectively. When productivity changes were included in the analysis, GRT was cost-saving.
Citation: Sendi P, Günthard HF, Simcock M, Ledergerber B, Schüpbach J, Battegay M, et al. (2007) Cost-Effectiveness of Genotypic Antiretroviral Resistance Testing in HIV-Infected Patients with Treatment Failure. PLoS ONE 2(1): e173. https://doi.org/10.1371/journal.pone.0000173
Academic Editor: Bruce Schackman, Cornell University, United States of America
Received: November 29, 2006; Accepted: December 22, 2006; Published: January 24, 2007
Copyright: © 2007 Sendi et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This cost-effectiveness analysis was financially supported by the Swiss National Science Foundation (SNF project # 3345CO-103560/1), the Stiftung Forschung Infektionskrankheiten Basel and the Swiss HIV Cohort Study (SHCS project # 400), supported by the Swiss National Science Foundation.
Competing interests: The authors have declared that no competing interests exist.
Introduction
The advent of potent antiretroviral therapies (ART) a decade ago has led to a substantial decline of morbidity and mortality in HIV infected patients [1]–[3]. Since then many new compounds and drug classes for the treatment of HIV infection have been developed that have substantially increased the complexity of HIV patient care [4]. The emergence of resistant mutations to antiretroviral drug compounds and classes, in addition, may jeopardize the success of HIV treatment and accelerate disease progression [4]–[6]. Genotypic antiretroviral resistance testing (GRT) helps to distinguish between antiretroviral drugs to which HIV has become resistant and compounds that effectively suppress viral replication [7]–[9].
It has been documented in randomized controlled trials that GRT based treatment optimization leads to a higher viral load reduction than standard care alone in treatment-experienced patients [8], [10], [11]. Published clinical guidelines now recommend GRT in patients with antiretroviral treatment failure or in recently infected patients who may have acquired drug-resistant virus [12]–[14]. The cost of GRT guided therapy, however, is substantial (e.g., $US 625 per test in Switzerland and $US 400–500 in the USA) [15]–[17], which has prompted a debate about the appropriate use and financing of antiretroviral resistance testing in Europe and the USA [15], [17]. Health insurance companies may still be reluctant to finance GRT although it's cost-effectiveness has been documented in a few countries.
Early and more recent cost-effectiveness studies suggested that antiretroviral treatment optimization using GRT is cost-effective [17]–[19]. However, these studies did not specifically investigate job productivity changes, did not include the results of more recent long-term studies of antiretroviral resistance testing, and did not make use of the same large homogenous database on HIV disease to describe both clinical and economic outcomes. We developed a comprehensive model of HIV disease using the Swiss HIV Cohort Study (SHCS) database to describe disease progression [20]. We included resistance testing data from long-term studies and several clinical trials, and used data from patients enrolled in the SHCS to estimate the clinical effect, cost and cost-effectiveness of treatment optimization using GRT as compared to expert opinion in patients with antiretroviral treatment failure.
Methods
Study design
We developed a mathematical model of HIV disease to assess the incremental impact of antiretroviral therapy guided by genotypic antiretroviral resistance testing versus expert opinion alone in patients presenting with treatment failure defined as i) a viral log reduction of <1 log HIV RNA copies/ml during the first six months after initiation of antiretroviral therapy (ART), ii) a viral load increase of >1 log HIV RNA copies/ml within two months of ART, or iii) two consecutive viral load assessments >200 copies/ml after reaching undetectable plasma HIV RNA levels (<50 copies/ml) [21]. Outcome measures included life-years, quality-adjusted life-years (QALYs), health care costs and productivity costs over a patient's lifetime. Model parameters were mainly derived from patients enrolled in the Swiss HIV Cohort Study (SHCS), one of the largest cohort studies on HIV disease with over 14'000 patients enrolled to date [20].The analyses were conducted from the health care as well as societal perspective and costs and effects were discounted at an annual rate of 4% and 2%, respectively. The discount rate from the societal perspective reflects the current interest rate in Swiss government bonds whereas the discount rate from the health care perspective reflects the rate typically used by Swiss social health insurers [22], [23].
Disease model
A state transition model of HIV disease with mutually exclusive health states was developed to describe the course of HIV disease after antiretroviral treatment failure, the starting state of patients entering the model [24], [25]. The failing treatment schedule is then either maintained or replaced by another regimen, based on expert opinion alone or with information available from genotypic antiretroviral resistance testing. Reasons for maintaining the failing treatment regimen may include the lack of more effective treatment options, patients refusing the required number of pills per day, or stable CD4 cell count despite virological failure [21]. Patients may reach virological suppression (<50 copies/ml) or have a detectable viral load during the subsequent two years with the likelihood of virological suppression being modeled as a function of the level of resistance of the HIV to the prescribed drug regimen [21]. To model the long-term impact of treatment optimization following genotypic antiretroviral resistance testing health states in patients without AIDS in the third year and thereafter were further stratified according to CD4 cell count (strata: 0–200 cells/mm3, 201–500 cells/mm3, >500 cells/mm3) and viral load (<1000 copies/ml, ≥1000 copies/ml), and in patients who experienced an AIDS-indicator disease by CD4 cell count only (strata: 0–200 cells/mm3, 201–500 cells/mm3, >500 cells/mm3). Patients are always at risk of dying due to an HIV-related or unrelated cause and were modeled until death. The disease model was populated with data extracted from the SHCS; the starting age of patients is 33 years, the average age of patients enrolled in the SHCS, and 80% of patients are male [20], [21].
Antiretroviral resistance testing
The probability of maintaining the failing antiretroviral regimen or switching to a new regimen with or without information available from GRT was derived from Haupts et al. [21], a study within the SHCS that assessed the impact of GRT on the selection of salvage regimens in patients presenting with treatment failure (Table 1). The prescribed drug regimen may then either contain no drugs to which a resistant mutation is reported or may contain one or more drugs to which the virus is resistant (Table 1). The probability of achieving viral suppression was then conditioned on the level of viral resistance of the virus to the final chosen drug regimen as defined in Haupts et al. [21]. Patients were repeatedly tested for the presence of HIV drug resistance mutations during the two-year follow up period, if necessary, to improve the likelihood of viral suppression [21]; patient management, adherence and virological outcomes therefore reflect a real-world setting. The distribution of patients across CD4 cell counts and viral load at the end of the two-year follow-up period was used to populate the model for projecting long-term costs and clinical outcomes over a patient's lifetime. In this subsequent model, the relative risk of experiencing treatment failure when expert opinion alone was used versus GRT in virologically suppressed patients was derived from published randomized controlled trials [8], [10], [11], [26], [27].
HIV disease progression
HIV disease progression was modeled by means of a transition probability matrix derived from the SHCS using the dataset from the period 1996–2004 before GRT became widely available. The probability of transitions to a health state with a viral load above or below 1000 copies/ml, a different CD4 cell stratum, the risk of developing an AIDS-indicator disease and dying was extracted from the SHCS database by pooling observations from patients on highly active antiretroviral therapy over consecutive six-month periods [28]–[30], which reflects the average period between patient visits in the SHCS (Table S1). The risk of dying from causes not related to HIV disease was derived from Swiss life tables (www.statistik.admin.ch).
Quality of life
HIV disease may not only affect survival but may also substantially affect the patient's wellbeing [31]. In order to calculate quality-adjusted life-years (QALYs) the length of time a patient spent in a specific health state was adjusted for the quality (i.e., utility) of that state on a scale ranging from zero (death) to one (best possible health state). The utility values used in our study are derived from a study on quality of life in patients enrolled in SHCS [32] by transforming health state values assessed on a visual analogue scale into standard gamble utilities using methods as described in detail elsewhere [33]. We used regression analysis methods to derive the utilities for the different health states shown in Table 2. The disutility of experiencing an AIDS-indicator disease was derived from a meta-analysis of utility estimates in HIV-infected patients (Table 2) [31]. Since a proportion of patients may partially include the effects of ill-health on income in health state valuation [34]–[36], utilities were adjusted by increasing their value by 4.5%, as derived from Sendi and Brouwer [35]. This approach was chosen to exclude any income effects of HIV disease in QALY estimation since productivity costs are already included in monetary terms (Table 2).
Costs
In the analysis from the health care perspective we only included costs due to health care resource consumption (Table 2). We used microcosting [23] as reported in detail previously to assess the costs associated with antiretroviral therapy, drugs for the prevention of opportunistic diseases and other drugs such as antihypertensive, lipid-lowering or antidiabetic agents [15], [29]. Ambulatory costs included the costs associated with a doctor's visit, CD4 cell count and viral load measurements, blood chemistry, blood count and other diagnostic procedures such as radiological, cardiovascular and endoscopic examinations [15], [29]. The cost of GRT was $US 625 [15]. To approximate in-patient costs we used charges to Swiss health insurers for patient who stayed a minimum of one night in a hospital [15]. All costs were expressed in 2005 $US using the Swiss consumer price index for health care and the exchange rate of July 1st, 2005 ($US 100 = CHF 128, www.oanada.com).
With the advent of potent antiretroviral therapy HIV infection has become a chronic disease, predominantly in young patients in the working age [3]. It is therefore important to include in the analysis from the societal perspective the impact of GRT-based antiretroviral treatment optimization on the patients' ability to work and hence productivity [37]. The number of hours a patient worked was recorded during each 6-month visit where SHCS enrollees are regularly examined [29]. We therefore extracted the number of hours a patient worked in each health state as defined above from the SHCS database and attached an average Swiss wage rate of $US 20 per hour (www.statistik.admin.ch) to estimate productivity changes over a patient's simulated lifetime (Table 2). The difference in productivity between patients whose treatment were optimized with GRT versus expert opinion only was then subtracted from the health care costs associated with GRT [38], [39].
Sensitivity analysis
We used probabilistic sensitivity analysis to assess the uncertainty around the cost and effect estimates as recommended by most recent guidelines [23], [40], [41].
Hereby a distribution is ascribed to each model input parameter using Bayesian methodology and noninfomative prior distributions [42]–[44]. We used normal distributions for cost and quality of life estimates with the corresponding standard error to model uncertainty associated with the mean input parameters, and we used a log-normal distribution for modeling the uncertainty associated with the relative risk of a treatment failure without GRT [44]. The uncertainty around count variables were modeled by means of a Dirichlet distribution, the conjugate distribution of the multinomial distribution for modeling the probability of events [45]. We used Monte Carlo simulation to estimate the uncertainty around the point estimate of incremental costs and effects of GRT versus expert opinion by sampling 5000 times from all input distributions and then recalculating the model using specialized software (TreeAge Pro 2005, TreeAge Software Inc., Williamstown, USA). The joint distribution of the resulting incremental costs and effects in our model were then summarized in terms of cost-effectiveness acceptability curves that describe the probability that the intervention is cost-effective as a function of the maximum willingness to pay per QALY gained [46], [47].
Results
Health care perspective
The main results of our analysis using the base case input parameters are shown in Table 3. Patients whose treatment was optimized with information available from GRT versus expert opinion had an increase in undiscounted life-expectancy and quality-adjusted life-expectancy of 2 and 6 weeks, respectively. When a 4% discount rate was used these figures were 3 and 2 weeks (Table 3). Undiscounted health care costs with and without GRT were $US 763'000 and $US 761'000, respectively, and $US 421'000 and 419'000 when a 4% discount rate was used. This corresponds to an expected cost-effectiveness ratio of $US 35'000 per QALY gained (Table 4).
The point estimate and 95% credible intervals of incremental costs and effects are shown in Figure 1A. The probability that GRT will lead to a better health outcome is 89% and there is a 95% probability that incremental quality-adjusted life-months (QALMs) lie between −0.4 and 1.6. The probability that GRT will increase health care costs is 100% with the additional costs lying between $US 800 and $US 2900 with 95% probability. Summarizing the uncertainty around cost and effect estimates in terms of cost-effectiveness acceptability curves, i.e. the probability that the intervention is cost-effective for all possible willingness to pay per QALY gained, we see that at a willingness to pay of $US 35'000 or more per QALY gained GRT is the preferred treatment option (Figure 2A).
Bars indicate the 95% credible intervals for incremental costs and effects. One quality-adjusted life-year corresponds to 12 quality-adjusted life-months. The bars cross each other at the median. The mean point estimate of the bivariate distribution of incremental costs and effects is indicated as a dot. Dotted horizontal and vertical lines at zero indicate no difference between costs and effects of the two strategies. The broken lines indicates a threshold (i.e., maximum willingness to pay) of $US 50'000 per QALY gained. The area to the right of the threshold line (i.e., the point estimate and the respective part of the distribution) is considered as cost-effective if the decision-maker's maximum willingness to pay per QALY is $US 50'000.
Vertical lines indicate the threshold where genotypic antiretroviral resistance testing and treatment optimization based on expert opinion have the same probability of being cost-effective. At a higher willingness to pay per QALY gained (e.g. at $US 50000/QALY) genotypic antiretroviral resistance testing has a higher probability of being cost-effective than expert opinion alone.
Societal perspective
The main results of our analysis from the societal perspective, i.e. including the effects of ill-health on income in monetary terms, are shown in Table 3. Patients whose treatment was optimized with information available from GRT versus expert opinion had an increase in life-expectancy and quality-adjusted life-expectancy of 3 and 4 weeks when a 2% discount rate was used (Table 3). Discounted health care costs with and without GRT were $US 551'000 and $US 549'000, respectively (Table 3). Patients who did not receive GRT, however, had an expected discounted income of $US 399'000 versus $US 401'000 when treatment was optimized with help of GRT (Table 3). This gain in productivity more than offsets the additional health care costs due to GRT (Table 3 and 4) and GRT is therefore a dominant strategy from the societal perspective.
The point estimate and 95% credible intervals of incremental costs and effects are shown in Figure 1B. The probability that GRT will lead to a better health outcome is 91% and there is a 95% probability that incremental QALMs lie between −0.4 and 2.3. The probability that GRT will increase health care costs is 52% with the incremental costs lying between $US −3800 and $US 1900 with 95% probability. Since the societal costs including productivity changes are skewed towards negative values, the point estimate of incremental costs leads to cost savings (Table 3) whereas the probability of achieving cost savings with GRT compared to expert opinion is slightly lower than 50%. Summarizing the uncertainty around cost and effect estimates in terms of cost-effectiveness acceptability curves, i.e. the probability that the intervention is cost-effective for all possible willingness to pay per QALY gained, we see that at a willingness to pay of $US 850 or more per QALY gained GRT is the preferred treatment option (Figure 2B). Using the widely mentioned societal threshold of $US 50'000 per QALY, GRT is cost-effective with a probability of 88%.
Discussion
We found that GRT increases projected life-expectancy and quality-adjusted life-expectancy by 2 and 6 weeks, respectively. This increase in quality-adjusted survival is clinically meaningful and comparable to the benefit of elective surgery in patients with symptomatic gallstones (gain in life-expectancy 1.7 months) or the benefit of Hepatitis B vaccination in newborn babies whose mothers have Hepatitis B (gain in life-expectancy 2 weeks) [48]. GRT is a dominant strategy from the societal perspective as the additional health care costs incurred by adding GRT to the treatment plan are more than offset by the increase in job productivity in HIV infected patients. When we included the uncertainty with respect to the input parameters of the model in the analysis, GRT has an 88% probability of being cost-effective when we used $US 50'000 per QALY gained as a threshold to determine whether an intervention represents value for money.
However, when only health care costs are considered, the cost-effectiveness ratio of GRT is $US 35'000 per QALY gained, which is similar to the cost-effectiveness ratio of implantable cardioverter-defibrillators ($US 35'000 per QALY gained) [49] and more cost-effective than other HIV treatment efforts such as antibiotic prophylaxis against M. avium complex infection in patients with AIDS ($US 80'000 per QALY gained) [42] and non-HIV interventions such as the use of drug-eluting stents in patients with coronary stenosis ($US 80'000 per QALY gained) [50].
In a previous study we have shown that GRT reduces health care costs over a two-year follow-up period due to the reduction of ambulatory and in-patient costs and a reduced consumption of non-HIV medication [15]. When compared to treatment optimization based on expert opinion, GRT increases overall health care costs by $US 1800 only (Table 4). The inclusion of productivity costs in our analysis, however, has a major impact on the results because with newer antiretroviral therapy HIV infection has become a chronic disease and patients, mostly adults in their working age, retain their ability to work [29], [37]. This is particularly true in countries with a low unemployment rate such as Switzerland.
In the present paper we also considered the most recent methodological developments with respect to the inclusion of productivity costs in economic evaluation to avoid a biased cost-effectiveness ratio [34], [35]. Since current evidence suggests that respondents or patients do not consistently include the effects of ill-health on income in health state valuation, omitting productivity costs in monetary terms from the analysis may lead to an underestimation of job productivity changes [35]. On the other hand, including productivity costs without adjusting utility weights in cost-effectiveness analysis can lead to an overestimation of job productivity changes [35]. We therefore included productivity costs in monetary terms and used adjusted utility weights to avoid double-counting of productivity changes.
Our study has several limitations. We used a mathematical model to approximate the real-world and project long-term costs and outcomes of GRT. However, we used best available evidence from one of the largest cohort studies on HIV disease to reflect a real-world setting [20]. The gain in (quality-adjusted) life-expectancy is slightly more conservative than those reported by other groups, which range from 1.3 months to 1.4 years [17], [18]. These differences could be due to differences in modeling natural disease history. However, our estimate is in line with the conservative scenario of Corzillius et al. [18] and supports the modeling process validity of both research groups [51]. Furthermore, we did not explicitly model patient compliance, which has been shown to substantially influence results [52]. Patient compliance, however, is already incorporated in the estimation of transition probabilities derived from the SHCS. In addition, resistance accumulation reduces viral fitness and may maintain CD4 cell counts over prolonged periods associated with clinical non-progression for some time. We explicitly modeled the possibility of maintaining the failing antiretroviral regimen, which reduces the short-term chance of viral suppression but may prevent the development of multi-resistant virus and hence preserves future drug options [53].
Our study has been conducted in Switzerland where health insurance is compulsory by law for all patients. However, our results are equally relevant for other countries such as the USA where health insurance is often funded by employers. The inclusion of GRT in health insurance plans not only leads to a benefit to the patient but also to the employer by increasing job productivity. Public health insurance coverage in the USA, on the other hand, is tied to disability status. Patients with disability status on public health insurance may also benefit from GRT based treatment optimization, as a better health state may increase the likelihood to return to work. Our results suggest that GRT represents high value for money and should be offered to all patients who can benefit from it.
Supporting Information
Table S1.
Transition probability matrix for HIV disease progression generated by pooling consecutive six-month observations from patients on HAART enrolled in the SHCS between 1996–2004. Numbers indicate observations, the likelihood is shown in brackets. To calculate the posterior probability with a non-informative prior, the number of observations in each cell with an allowed transition is increased by one.
https://doi.org/10.1371/journal.pone.0000173.s001
(0.05 MB DOC)
Acknowledgments
The members of the Swiss HIV Cohort Study are M. Battegay, E. Bernasconi, J. Böni, H. Bucher, Ph. Bürgisser, S. Cattacin, M. Cavassini, R. Dubs, M. Egger, L. Elzi, P. Erb, M. Fischer, M. Flepp, A. Fontana, P. Francioli (President of the SHCS, Centre Hospitalier Universitaire Vaudois, CH-1011- Lausanne), H. Furrer (Chairman of the Clinical and Laboratory Committee), M. Gorgievski, H. Günthard, H. Hirsch, B. Hirschel, I. Hösli, Ch. Kahlert, L. Kaiser, U. Karrer, O. Keiser, C. Kind, Th. Klimkait, B. Ledergerber, G. Martinetti, B. Martinez, N. Müller, D. Nadal, M. Opravil, F. Paccaud, G. Pantaleo, L. Perrin, M. Rickenbach (Head of Data Center), C. Rudin (Chairman of the Mother & Child Substudy), P. Schmid, D. Schultze, J. Schüpbach, R. Speck, P. Taffé, P. Tarr, A. Telenti, A. Trkola, P. Vernazza (Chairman of the Scientific Board), R. Weber, S. Yerly.
Author Contributions
Other: Participated in the design of the study and interpretation of the results, and contributed to manuscript preparation: HG. Responsible for the design and statistical analysis of the cost-effectiveness analysis, interpretation of the results and wrote the first draft of the manuscript: PS. Participated in the statistical analysis of the study and interpretation of the results: BL MS. Contributed to the interpretation of the results and manuscript drafting: JS. Was involved in study design, interpretation of the results and writing of the manuscript: MB.
References
- 1. Palella FJ Jr, Delaney KM, Moorman AC, et al. (1998) Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection. HIV Outpatient Study Investigators. N Engl J Med 338(13): 853–860.
- 2. Egger M, Hirschel B, Francioli P, et al. (1997) Impact of new antiretroviral combination therapies in HIV infected patients in Switzerland: prospective multicentre study. BMJ 315(7117): 1194–1199.
- 3. Sterne JA, Hernan MA, Ledergerber B, et al. (2005) Long-term effectiveness of potent antiretroviral therapy in preventing AIDS and death: a prospective cohort study. Lancet 366(9483): 378–384.
- 4. Yeni PG, Hammer SM, Hirsch MS, et al. (2004) Treatment for adult HIV infection: 2004 recommendations of the International AIDS Society-USA Panel. JAMA 292(2): 251–265.
- 5. Grant RM, Hecht FM, Warmerdam M, et al. (2002) Time trends in primary HIV-1 drug resistance among recently infected persons. JAMA 288(2): 181–188.
- 6. Hermankova M, Ray SC, Ruff C, et al. (2001) HIV-1 drug resistance profiles in children and adults with viral load of <50 copies/ml receiving combination therapy. JAMA 286(2): 196–207.
- 7. Ochoa DE, Arnedo M, Xercavins M, et al. (2005) Genotypic and phenotypic resistance patterns at virological failure in a simplification trial with nevirapine, efavirenz or abacavir. AIDS 19(13): 1385–1391.
- 8. Ena J, Ruiz de Apodaca RF, Amador C, Benito C, Pasquau F (2006) Net benefits of resistance testing directed therapy compared with standard of care in HIV-infected patients with virological failure: A meta-analysis. Enferm Infecc Microbiol Clin 24(4): 232–237.
- 9. Hirsch HH, Drechsler H, Holbro A, et al. (2005) Genotypic and phenotypic resistance testing of HIV-1 in routine clinical care. Eur J Clin Microbiol Infect Dis 24(11): 733–738.
- 10. Torre D, Tambini R (2002) Antiretroviral drug resistance testing in patients with HIV-1 infection: a meta-analysis study. HIV Clin Trials 3(1): 1–8.
- 11. Clevenbergh P, Durant J, Halfon P, et al. (2000) Persisting long-term benefit of genotype-guided treatment for HIV-infected patients failing HAART. The Viradapt Study: week 48 follow-up. Antivir Ther 5(1): 65–70.
- 12. Vandamme AM, Sonnerborg A, Ait-Khaled M, et al. (2004) Updated European recommendations for the clinical use of HIV drug resistance testing. Antivir Ther 9(6): 829–848.
- 13. Hirsch MS, Brun-Vezinet F, Clotet B, et al. (2003) Antiretroviral drug resistance testing in adults infected with human immunodeficiency virus type 1: 2003 recommendations of an International AIDS Society-USA Panel. Clin Infect Dis 37(1): 113–128.
- 14. Hirsch MS, Brun-Vezinet F, Clotet B, et al. (2003) Questions to and answers from the International AIDS Society-USA Resistance Testing Guidelines Panel. Top HIV Med 11(4): 150–154.
- 15. Simcock M, Sendi P, Ledergerber B, et al. (2006) A longitudinal analysis of health care costs after treatment optimisation following genotypic antiretroviral resistance testing: does resistance testing pay off? Antivir Ther 11: 305–314.
- 16. Sax PE, Islam R, Walensky RP, et al. (2005) Should resistance testing be performed for treatment-naive HIV-infected patients? A cost-effectiveness analysis. Clin Infect Dis 41(9): 1316–1323.
- 17. Weinstein MC, Goldie SJ, Losina E, et al. (2001) Use of genotypic resistance testing to guide hiv therapy: clinical impact and cost-effectiveness. Ann Intern Med 134(6): 440–450.
- 18. Corzillius M, Muhlberger N, Sroczynski G, Jaeger H, Wasem J, Siebert U (2004) Cost effectiveness analysis of routine use of genotypic antiretroviral resistance testing after failure of antiretroviral treatment for HIV. Antivir Ther 9(1): 27–36.
- 19. Chaix C, Grenier-Sennelier C, Clevenbergh P, et al. (2000) Economic evaluation of drug resistance genotyping for the adaptation of treatment in HIV-infected patients in the VIRADAPT study. J Acquir Immune Defic Syndr 24(3): 227–231.
- 20. Ledergerber B, von Overbeck J, Egger M, Luthy R (1994) The Swiss HIV Cohort Study: rationale, organization and selected baseline characteristics. Soz Praventivmed 39(6): 387–394.
- 21. Haupts S, Ledergerber B, Boni J, et al. (2003) Impact of genotypic resistance testing on selection of salvage regimen in clinical practice. Antivir Ther 8(5): 443–454.
- 22. Krahn M, Gafni A (1993) Discounting in the economic evaluation of health care interventions. Med Care 31(5): 403–418.
- 23.
Drummond MF, O'Brien BJ, Stoddart GL, Torrance GW (1997) Methods for the economic evaluation of health care programmes. Oxford: Oxford University Press. 2nd edition ed.
- 24. Briggs A, Sculpher M (1998) An introduction to Markov modelling for economic evaluation. Pharmacoeconomics 13(4): 397–409.
- 25. Sonnenberg FA, Beck JR (1993) Markov models in medical decision making: a practical guide. Med Decis Making 13(4): 322–338.
- 26. Durant J, Clevenbergh P, Halfon P, et al. (1999) Drug-resistance genotyping in HIV-1 therapy: the VIRADAPT randomised controlled trial. Lancet 353(9171): 2195–2199.
- 27. Panidou ET, Trikalinos TA, Ioannidis JP (2004) Limited benefit of antiretroviral resistance testing in treatment-experienced patients: a meta-analysis. AIDS 18(16): 2153–2161.
- 28. McCarthy BD, Wong JB, Munoz A, Sonnenberg FA (1993) Who should be screened for HIV infection? A cost-effectiveness analysis. Arch Intern Med 153(9): 1107–1116.
- 29. Sendi PP, Bucher HC, Harr T, et al. (1999) Cost effectiveness of highly active antiretroviral therapy in HIV-infected patients. AIDS 13(9): 1115–1122.
- 30. Sendi PP, Bucher HC, Craig BA, Pfluger D, Battegay M (1999) Estimating AIDS-free survival in a severely immunosuppressed asymptomatic HIV-infected population in the era of antiretroviral triple combination therapy. Swiss HIV Cohort Study. J Acquir Immune Defic Syndr Hum Retrovirol 20(4): 376–381.
- 31. Tengs TO, Lin TH (2002) A meta-analysis of utility estimates for HIV/AIDS. Med Decis Making 22(6): 475–481.
- 32. Zinkernagel C, Ledergerber B, Battegay M, et al. (1999) Quality of life in asymptomatic patients with early HIV infection initiating antiretroviral therapy. AIDS 13(12): 1587–1589.
- 33. Mrus JM, Yi MS, Freedberg KA, et al. (2003) Utilities derived from visual analog scale scores in patients with HIV/AIDS. Med Decis Making 23(5): 414–421.
- 34. Krol M, Brouwer W, Sendi P (2006) Productivity costs in health-state valuations: does explicit instruction matter? Pharmacoeconomics 24(4): 401–414.
- 35. Sendi P, Brouwer WB (2005) Is silence golden? A test of the incorporation of the effects of ill-health on income and leisure in health state valuations. Health Econ 14(6): 643–647.
- 36. Meltzer D, Johannesson M (1999) Inconsistencies in the "societal perspective" on costs of the Panel on Cost-Effectiveness in Health and Medicine. Med Decis Making 19(4): 371–377.
- 37. Sendi P, Schellenberg F, Ungsedhapand C, et al. (2004) Productivity costs and determinants of productivity in HIV-infected patients. Clin Ther 26(5): 791–800.
- 38. Johannesson M, Meltzer D (1998) Some reflections on cost-effectiveness analysis. Health Econ 7(1): 1–7.
- 39. Johannesson M, Meltzer D, O'Conor RM (1997) Incorporating future costs in medical cost-effectiveness analysis: implications for the cost-effectiveness of the treatment of hypertension. Med Decis Making 17(4): 382–389.
- 40. Claxton K, Sculpher M, McCabe C, et al. (2005) Probabilistic sensitivity analysis for NICE technology assessment: not an optional extra. Health Econ 14(4): 339–347.
- 41. Doubilet P, Begg CB, Weinstein MC, Braun P, McNeil BJ (1985) Probabilistic sensitivity analysis using Monte Carlo simulation. A practical approach. Med Decis Making 5(2): 157–177.
- 42. Sendi PP, Craig BA, Meier G, et al. (1999) Cost-effectiveness of azithromycin for preventing Mycobacterium avium complex infection in HIV-positive patients in the era of highly active antiretroviral therapy.The Swiss HIV Cohort Study. J Antimicrob Chemother 44(6): 811–817.
- 43. Briggs AH (1999) A Bayesian approach to stochastic cost-effectiveness analysis. Health Econ 8(3): 257–261.
- 44. Briggs AH, Goeree R, Blackhouse G, O'Brien BJ (2002) Probabilistic analysis of cost-effectiveness models: choosing between treatment strategies for gastroesophageal reflux disease. Med Decis Making 22(4): 290–308.
- 45. Briggs AH, Ades AE, Price MJ (2003) Probabilistic sensitivity analysis for decision trees with multiple branches: use of the Dirichlet distribution in a Bayesian framework. Med Decis Making 23(4): 341–350.
- 46. van Hout BA, Al MJ, Gordon GS, Rutten FF (1994) Costs, effects and C/E-ratios alongside a clinical trial. Health Econ 3(5): 309–319.
- 47. Hunink MG, Bult JR, de Vries J, Weinstein MC (1998) Uncertainty in decision models analyzing cost-effectiveness: the joint distribution of incremental costs and effectiveness evaluated with a nonparametric bootstrap method. Med Decis Making 18(3): 337–346.
- 48. Wright JC, Weinstein MC (1998) Gains in life expectancy from medical interventions–standardizing data on outcomes. N Engl J Med 339(6): 380–386.
- 49. Sanders GD, Hlatky MA, Owens DK (2005) Cost-effectiveness of implantable cardioverter-defibrillators. N Engl J Med 353(14): 1471–1480.
- 50. Kaiser C, Brunner-La Rocca HP, Buser PT, et al. (2005) Incremental cost-effectiveness of drug-eluting stents compared with a third-generation bare-metal stent in a real-world setting: randomised Basel Stent Kosten Effektivitats Trial (BASKET). Lancet 366(9489): 921–929.
- 51. Sendi PP, Craig BA, Pfluger D, Gafni A, Bucher HC (1999) Systematic validation of disease models for pharmacoeconomic evaluations. Swiss HIV Cohort Study. J Eval Clin Pract 5(3): 283–295.
- 52. Wood E, Hogg RS, Yip B, Harrigan PR, O'Shaughnessy MV, Montaner JS (2003) Effect of medication adherence on survival of HIV-infected adults who start highly active antiretroviral therapy when the CD4+ cell count is 0.200 to 0.350×10(9) cells/L. Ann Intern Med 139(10): 810–816.
- 53. Jiang H, Deeks SG, Kuritzkes DR, et al. (2003) Assessing resistance costs of antiretroviral therapies via measures of future drug options. J Infect Dis 188(7): 1001–1008.