April D. Kimmel

The Cost Effectiveness of Combination Antiretroviral Therapy for HIV Disease

BACKGROUND Combination antiretroviral therapy with a combination of three or more drugs has become the standard of care for patients with human immunodeficiency virus (HIV) infection in the United States. We estimated the clinical benefits and cost effectiveness of three-drug antiretroviral regimens. METHODS We developed a mathematical simulation model of HIV disease, using the CD4 cell count and HIV RNA level as predictors of the progression of disease. Outcome measures included life expectancy, life expectancy adjusted for the quality of life, lifetime direct medical costs, and cost effectiveness in dollars per quality-adjusted year of life gained. Clinical data were derived from major clinical trials, including the AIDS Clinical Trials Group 320 Study. Data on costs were based on the national AIDS Cost and Services Utilization Survey, with drug costs obtained from the Red Book. RESULTS For patients similar to those in the AIDS Clinical Trials Group 320 Study (mean CD4 cell count, 87 per cubic millimeter), life expectancy adjusted for the quality of life increased from 1.53 to 2.91 years, and per-person lifetime costs increased from

Use of Genotypic Resistance Testing To Guide HIV Therapy: Clinical Impact and Cost-Effectiveness

45,460 to

Cost-effectiveness of enfuvirtide in treatment-experienced patients with advanced HIV disease

77,300 with three-drug therapy as compared with no therapy. The incremental cost per quality-adjusted year of life gained, as compared with no therapy, was

Treatment for primary HIV infection: projecting outcomes of immediate, interrupted, or delayed therapy.

23,000. On the basis of additional data from other major studies, the cost-effectiveness ratio for three-drug therapy ranged from

Preevaluation of Clinical Trial Data: The Case of Preemptive Cytomegalovirus Therapy in Patients with Human Immunodeficiency Virus

13,000 to

The relationship of preventable opportunistic infections, HIV-1 RNA, and CD4 cell counts to chronic mortality

23,000 per quality-adjusted year of life gained. The initial CD4 cell count and drug costs were the most important determinants of costs, clinical benefits, and cost effectiveness. CONCLUSIONS Treatment of HIV infection with a combination of three antiretroviral drugs is a cost-effective use of resources.

Prevention of Human Immunodeficiency Virus—Related Opportunistic Infections in France: A Cost-Effectiveness Analysis

Treatment with combinations of antiretroviral drugs that target HIV-1 protease and reverse transcriptase enzymes is effective in reducing HIV-1 replication (1). Use of these regimens, referred to as highly active antiretroviral therapy (HAART), has dramatically reduced HIV-related morbidity and mortality (1-3). Despite this success, however, treatment failure occurs frequently and is associated with the development of HIV-1 variants that have reduced susceptibility to antiretroviral drugs. It is not clear whether acquired resistance causes treatment failure or whether failure precipitates drug resistance (4, 5). Patients who experience failure of initial HAART, as diagnosed by increasing HIV RNA levels, are given second-line and salvage therapies, but these later regimens have substantially lower success rates than initial therapy, partly because of acquired drug resistance (1, 2, 6). Genotypic assays, also referred to as genotypic antiretroviral resistance testing (GART), can identify drug-resistant mutations. The detection of these mutations has been shown to predict virologic response in retrospective studies (7-9) and most recently in two randomized, controlled trials (6, 10). The Community Programs for Clinical Research on AIDS (CPCRA) trial 046 showed that among patients who experienced HAART failure, viral suppression (<500 copies/mL) at 12 weeks was more common among patients for whom GART and clinical judgment were used to guide the choice of a second-line or subsequent regimen (34%) than among those whose treatment was determined by clinical judgment alone (22%) (10). Results from the AntiretroVIRal ADAPTtation (VIRADAPT) randomized, controlled trial in France and Israel were similar, with 32% viral suppression (<200 copies/mL) in the GART group and 14% in the control group (6). These results suggest that routine use of GART to guide the choice of antiretroviral regimens after treatment failure may be beneficial, and this practice is now recommended by two recent sets of guidelines (1, 11). However, the results of these studies are relatively short-term, and the long-term benefits of routine use of GART are uncertain. Moreover, GART is expensive; the current cost is approximately

Lives Saved by Expanding HIV Treatment Availability in Resource-Limited Settings: The Example of Haiti

400 or more per test (Office of Payment, Boston Medical Center, Boston, Massachusetts. Unpublished data). Simulation modeling is one way to use available data to estimate the long-term health consequences and costs of clinical interventions, which may then be used to inform resource allocation (12, 13). The need to complement clinical trials with simulation modeling is particularly critical in HIV disease because trials increasingly rely on surrogate markers of outcome, such as HIV RNA levels and CD4 cell counts. Models may also be used to conduct exploratory analyses of potential health impact and economic value before data from trials are available, thus providing insight into the most important questions to address in future trials. Because of concerns that widespread use of potent combination antiretroviral drugs could increase transmission of drug-resistant virus, the potential use of primary resistance testing to guide initial antiretroviral therapy is also being considered (5, 14). Two recent studies reported disparate data on the prevalence of drug resistanceconferring mutations and phenotypic resistance to antiretroviral agents in recently infected patients (15, 16). However, the potential value of resistance testing to guide initial therapy will only increase as HAART-induced drug-resistant strains of virus become more widespread. Our study had two objectives: 1) to estimate the cost-effectiveness of GART to guide the choice of subsequent therapy after initial HAART failure and 2) to estimate the cost-effectiveness of primary resistance testing to guide the choice of initial therapy in antiretroviral-naive patients. Methods Overview The analyses for both secondary and primary resistance testing were performed by using a state-transition model (17, 18) of persons with HIV infection. In the model, persons could make transitions between health states at monthly intervals. We used the model in random (that is, first-order Monte Carlo) simulations of 1 million patients drawn from an initial probability distribution of HIV disease characteristics. These simulations produced estimates of numbers of opportunistic infections, quality-adjusted life expectancy, and lifetime costs. The performance of GART was expressed as its incremental cost-effectiveness ratio, defined as the incremental cost compared with no GART, divided by the incremental gain in quality-adjusted life expectancy compared with no GART. We adopted a societal perspective and discounted costs and clinical benefits at 3% per year. We followed the reference case recommendations of the Panel on Cost-Effectiveness in Health and Medicine to ensure that resulting cost-effectiveness ratios would be comparable to those from analyses of other interventions (12, 13). Model Structure Health states in the model were defined by a patients current and maximum HIV RNA level (or viral load), CD4 cell count, time receiving HAART, history of effective and ineffective antiretroviral therapy, and previous opportunistic infections. We divided HIV RNA level and CD4 cell counts into six strata (HIV RNA level: >100 000 copies/mL, 30 001 to 100 000 copies/mL, 10 001 to 30 000 copies/mL, 3001 to 10 000 copies/mL, 501 to 3000 copies/mL, and 0 to 500 copies/mL; CD4 count: >0.500 109 cells/L, 0.301 to 0.500 109 cells/L, 0.201 to 0.300 109 cells/L, 0.101 to 0.200 109 cells/L, 0.051 to 0.100 109 cells/L, and 0 to 0.050 109 cells/L). Progression of HIV disease was modeled as monthly transitions between health states (17-19). Patients could enter and exit temporary health states corresponding to acute episodes of Pneumocystis carinii pneumonia; toxoplasmosis; disseminated Mycobacterium avium complex; fungal infections; cytomegalovirus; and a residual category consisting of other AIDS-related complications, including bacterial infections, tuberculosis, and lymphoma (19). From these temporary health states, patients could die or survive, allowing a transition to a new chronic state. The CD4 cell count, which was used in the model as a surrogate marker of disease progression, was also used to predict the rates of opportunistic infections and HIV-related death. However, in the absence of antiretroviral therapy, the rate at which CD4 cell count decreases in any individual patient was determined by the patients HIV RNA level. If HAART resulted in at least partial viral suppression, the CD4 cell count increased rapidly at first and then more slowly (2, 20, 21). The duration of the CD4 cell count increase beyond the follow-up of clinical trials is uncertain; we assumed that the increase would continue in patients with sustained viral suppression (1, 22). The duration of viral suppression depended on adherence and the development of viral resistance. Virologic failure was defined as an increase in HIV RNA level for 2 consecutive months while receiving HAART. If treatment failed, we assumed that CD4 cell counts then began to decrease. However, in the sensitivity analysis, we allowed for a delay of 6 to 24 months before CD4 cell counts decreased (23). Before treatment, HIV RNA level was assumed to be at a set-point, or steady-state value (24). With successful first-line antiretroviral therapy, HIV RNA levels decreased to undetectable levels (<200 copies/mL or <500 copies/mL, depending on the assay). The simulations began with a target population of antiretroviral-naive patients who had a CD4 count of 0.250 109 cells/L. The distribution of HIV RNA levels among patients was based on the Multicenter AIDS Cohort Study, as follows: less than 500 copies/mL, 7.71% of patients; 501 to 3000 copies/mL, 16.33% of patients; 3001 to 10 000 copies/mL, 25.21% of patients; 10 001 to 30 000 copies/mL, 25.02% of patients; and 30 001 to 100 000 copies/mL, 25.73% of patients (25). All patients were assumed to receive standard first-line prophylaxis against P. carinii pneumonia, toxoplasmosis, and M. avium complex (26-29). All patients received indinavir, zidovudine or stavudine, and lamivudine as initial therapy, as in the three-drug arm of the AIDS Clinical Trials Group (ACTG) Protocol 320 (2). The alternative of a first-line therapy more effective than that used in ACTG 320 was also considered, using viral suppression rates from more recent studies (20, 21, 30). We assumed that HIV RNA level and CD4 cell count would be tested every 3 months in the absence of treatment failure. After an increase in HIV RNA level was observed, HIV RNA testing occurred monthly until diagnosis of failure. The focal choice in this analysis was whether to use GART to guide antiretroviral therapy, both after treatment failure and in the choice of initial therapy. In both of these decision contexts, we evaluated two strategies for choosing the HAART regimen: clinical judgment alone, and clinical judgment aided by GART. The relative reduction in the probability of HAART failure with the GART-guided regimen was estimated from the CPCRA 046 and VIRADAPT randomized clinical trials (6, 10). The possibility of later salvage therapy after failure of the subsequent HAART regimen (30) was considered in the sensitivity analysis, but we did not model the use of GART to guide third-line and subsequent salvage therapies. For each simulation, 1 million patient lives were simulated and followed until death. Cumulative quality-adjusted life-months and resource costs were recorded for each simulated patient and then averaged over the 1 million cases to estimate quality-adjusted life expectancy and expected cost associated with the strategies of GART and no GART. A sample size of 1 million patients was necessary to reduce sampling error to below the effect sizes of interest. Clinical Data Values for CD4 cell count and other variables used in the base-case an

Decision Maker Priorities for Providing Antiretroviral Therapy in HIV-Infected South Africans: A Qualitative Assessment

Objective:Enfuvirtide (ENF) has been shown to improve short-term virologic responses when given to highly treatment-experienced patients with advanced HIV disease. Because of the high cost of ENF compared with other antiretroviral agents, our objectives were to determine the potential long-term clinical impact and cost-effectiveness of ENF in these patients. Methods:We used a computer simulation model of HIV disease to project life expectancy, quality-adjusted life expectancy, cost, and cost-effectiveness of ENF in treatment-experienced patients. Input data were from the T-20 versus Optimized Regimen Only (TORO) 1 and 2 trials, 2 studies comparing ENF plus an optimized background regimen (OBR) with an OBR alone. Results:ENF plus an OBR increased projected discounted quality-adjusted life expectancy from 45.4 months with an OBR alone to 54.9 months, a difference of 9.5 quality-adjusted life-months. At the current annual ENF cost of

Diagnostic tests in HIV management: a review of clinical and laboratory strategies to monitor HIV-infected individuals in developing countries

18,500 per year (in 2001 US dollars), the incremental cost-effectiveness ratio for ENF plus an OBR was