Skip Navigation LinksHome > August 15, 2005 - Volume 39 - Issue 5 > HIV Antiretroviral Treatment: Early Versus Later
JAIDS Journal of Acquired Immune Deficiency Syndromes:
Epidemiology and Social Science

HIV Antiretroviral Treatment: Early Versus Later

Mauskopf, Josephine PhD*; Kitahata, Mari MPH, MD†; Kauf, Teresa PhD‡; Richter, Anke PhD*; Tolson, Jerry PhD§

Free Access
Article Outline
Collapse Box

Author Information

From *RTI Health Solutions, Research Triangle Park, NC; †Center for AIDS and STD Harborview Medical Center, University of Washington, Seattle, WA; ‡Duke Clinical Research Institute, Durham, NC; and §GlaxoSmithKline, Research Triangle Park, NC.

Received for publication September 4, 2003; accepted December 14, 2004.

Reprints: Josephine Mauskopf, Vice President, Health Economics, RTI Health Solutions, 3040 Cornwallis Road, PO Box 12194, Research Triangle Park, NC 27709 (e-mail: jmauskopf@rti.org).

Collapse Box

Abstract

Objectives: Cohort studies indicate that starting highly active antiretroviral therapy (HAART) when the CD4+ T-cell count is less than 200 cells/μL is associated with poor outcomes. These studies have been unable to address how early HAART should be initiated, however. This report uses a modeling approach to compare starting HAART at a mean CD4+ T-cell count greater than 350 cells/μL (early) versus less than 350 cells/μL but greater than 200 cells/μL (later).

Methods: A Markov model tracks people with HIV infection through 6 disease stages defined by CD4+ T-cell count ranges over a 25-year period. Transition probabilities between the disease stages for 6-month periods vary according to initial viral load. Sequences of different first-line, second-line, and “salvage” antiretroviral regimens are defined, and their impact on transition probabilities is estimated. HAART effectiveness is based on data from an urban hospital-based HIV clinic, supplemented by clinical trial data. The model computes the incremental cost-effectiveness of alternative treatment patterns and includes sensitivity analyses for a range of plausible alternative input values.

Results: Starting HAART earlier rather than later increases total lifetime costs by $19,074, increases years of life by 1.21 years, increases discounted quality-adjusted life-years by 0.61, and has an incremental cost-effectiveness ratio of $31,266 per quality-adjusted life-year. Early therapy is more cost-effective when the impact of HAART on well-being is smaller.

Conclusions: Initiation of HAART at a CD4+ T-cell count greater than 350 cells/μL may be cost-effective (less than $50,000 per quality-adjusted life-year) compared with initiating HAART at a CD4+ T-cell count less than 350 cells/μL but greater than 200 cells/μL and may result in longer quality-adjusted survival.

Over the past 15 years, many drugs that inhibit the replication of HIV have been developed and approved for use in individuals who are infected with this virus. In the United States, there are currently 7 nucleoside reverse transcriptase inhibitors (NRTIs), 3 nonnucleoside reverse transcriptase inhibitors (NNRTIs), 7 protease inhibitors (PIs), and 1 fusion inhibitor approved by the Food and Drug Administration (FDA)1 used in combinations of 3 or more. Because current drug therapies are not curative, treatment of HIV infection requires a lifetime sequence of treatment regimens, with the goal of maximizing life expectancy while maintaining as high a quality of life as is possible over that remaining life expectancy.2

Problems with side effects related to highly active antiretroviral therapy (HAART) as well as questions as to whether the drug combinations are equally efficacious when given earlier or later in disease progression have been debated in the literature. A recent study presented at the 14th International AIDS Conference3 indicates that HAART treatment may be just as effective if started when a patient's CD4+ T-cell count is between 349 cells/μL and 200 cells/μL as compared with starting when the person has a CD4+ T-cell count of 350 cells/μL or more. This study showed that HAART treatment is less effective if the start of treatment is delayed until after the CD4+ T-cell count has fallen to less than 200 cells/μL, however. The end point for this study was progression to AIDS. Several other studies have clearly demonstrated faster disease progression when treatment is delayed until the CD4+ T-cell count is less than 200 cells/μL,4-7 but these same studies have shown mixed results for those starting treatment when their CD4+ cell count is between 499 cells/μL and 350 cells/μL compared with between 349 cells/μL and 200 cells/μL. Currently, initiation of HAART is recommended when the CD4+ T-cell count has fallen to less than 350 cells/μL or when the initial viral load is greater than 55,000 copies/mL.8,9

Two recent Monte Carlo simulation modeling studies comparing the cost-effectiveness of early versus later initiation of therapy10,11 estimated that early initiation of therapy (>200 cells/μL) is more cost-effective than later initiation (<200 cells/μL) because the person is held for a longer time in a disease stage that is less severe. In these reports, initiation when the CD4+ T-cell count was 500 cells/μL or 350 cells/μL, respectively, was compared with initiation when the CD4+ T-cell count was 200 cells/μL, and efficacy data from clinical trials were used in the model. These studies did not compare the cost-effectiveness of starting therapy at 500 cells/μL with starting therapy at 350 cells/μL.

This report uses a Markov modeling approach for tracking HIV disease progression and focuses on a comparison of starting HAART when the CD4+ T-cell count is greater than 350 cells/μL (in the range 499 cells/μL and 350 cells/μL, mean value of 425 cells/μL) and starting HAART when the CD4+ T-cell count is less than 350 cells/μL (in the range of 349 cells/μL and 200 cells/μL, mean value of 275 cells/μL). This focus is important, because although there is now general agreement that HAART should be started before the CD4+ T-cell count falls to 200 cells/μL or less, there is still some uncertainty as to the value of starting HAART for people with CD4+ T-cell counts greater than 350 cells/μL rather than less than 350 cells/μL. In addition, recommendations to start treatment at a CD4+ T-cell count above 200 cells/μL were based on empiric studies with limited follow-up times.3-7 This report estimates the cost-effectiveness of early versus later HAART start using effectiveness data from a “real-world” urban hospital-based HIV clinic setting. An extensive sensitivity analysis is performed to estimate the impact on the cost-effectiveness of early versus later initiation of HAART on (1) use of an alternative treatment pathway, (2) duration of effectiveness of first- and second-line treatment regimens, (3) type and duration of “salvage” therapy regimens that are used, (4) initial viral load of the person with HIV infection, (5) impact of HAART and its side effects on quality of life, and (6) impact of HAART monitoring and side effects on annual cost of HIV infection in each CD4 cell count range.

Back to Top | Article Outline

METHODS

Model Overview

The viral load and CD4+ T-cell cost-effectiveness model is a Markov model that tracks the progression of HIV infection through 6 disease stages defined by CD4+ T-cell count ranges. Transition probabilities between the disease stages are estimated for 6-month periods and vary according to initial (set point) viral load. Sequences of different first-line, second-line, and salvage antiretroviral regimens (treatment pathways) starting at a specific disease stage are defined, and their impact on transition probabilities are entered into the model. The model tracks a cohort of people with HIV infection over a 25-year period and computes the incremental cost-effectiveness of alternative treatment pathways.

Effectiveness data from the Johns Hopkins HIV clinic were used to model the impact of the initial regimen on CD4+ T-cell count and viral load. The Johns Hopkins clinic provides longitudinal primary care to a large portion of HIV-infected patients in the Baltimore area. This urban HIV population has been well studied.12-14 The modeled effectiveness of the initial regimen was based on the 16-week response to initial HAART among previously drug-naive patients at the Johns Hopkins clinic during the year 2001. At this time, the first regimen for most patients consisted of 2 NRTIs and a PI, regardless of when HAART was initiated.

Effectiveness data from the Johns Hopkins clinic were supplemented with efficacy data from clinical trials to model the clinical effects of second-line HAART and salvage therapy. Because the effectiveness of first-line HAART among the Johns Hopkins cohort was less than clinical trial results for the same regimens among drug-naive patients, the impact of subsequent treatment regimens was assumed to be less than the efficacy observed in the clinical trials. Specific parameters for CD4+ T-cell counts and HIV viral load responses are discussed below.

Back to Top | Article Outline
Model Health States

The model specifies 6 health states based on the 5 CD4+ T-cell ranges: ≥500 cells/μL, 350 to 499 cells/μL (mean of 425 cells/μL), 200 to 349 cells/μL (mean of 275 cells/μL), 100 to 199 cells/μL (mean 150 cells/μL), 0 to 100 cells/μL (mean of 50 cells/μL), and death. Several previous cost-effectiveness models of new HIV treatments have used similar definitions of HIV disease stages.15-17 In any 6-month period, a person can stay in the same CD4+ T-cell range or progress to the next lower CD4+ T-cell range (Table 1).

Table 1
Table 1
Image Tools
Back to Top | Article Outline
Natural History Transition Probabilities

The natural (untreated) rate of disease progression through the 6 health states described previously was based on the experience of the Multicenter AIDS Cohort Study (MACS) population.18 The MACS is a longitudinal cohort study of the natural history of HIV-1 infection among homosexual men that was started in 1984/1985 at 4 study centers across the country. This study established the principle that the rate of CD4+ T-cell count decline is a function of the initial (set point) viral load as well as the viral load at any time point. Thus, clinical disease progression depends on changes in CD4+ T-cell count and viral load independently. Mellors et al18 reported the median and 95% confidence intervals for the annual rate of CD4+ T-cell change as a function of viral load for 4 viral load categories (<3000 copies/mL, 3001-10,000 copies/mL, 10,001-30,000 copies/mL, and >30,000 copies/mL as measured by sensitive branched deoxyribonucleic acid [bDNA] assay).

Using the MACS data, the mean waiting time (in years) for the 5 CD4+ T-cell ranges corresponding to the 4 initial viral load categories reported by Mellors et al18 was determined by Monte Carlo simulation. The annual health state transition probabilities representing disease progression in the absence of treatment were calculated as the reciprocal of the mean waiting times in each CD4+ T-cell range. Annual transition probabilities were converted to 6-month transition probabilities using a standard exponential adjustment (Pm = 1 − [1−Pn]m/n, where m is 6 months and n is 12 months) to allow for more flexibility in modeling treatment duration. Table 1 presents the estimated transition probabilities for each CD4+ T-cell range and initial viral load range combination. In each 6-month period, patients who did not transition to a lower CD4+ T-cell range were assumed to remain in their current disease state.

Back to Top | Article Outline
Impact of Treatment on Disease Progression

To estimate the impact of antiretroviral treatment on HIV disease progression, we used the work of Hill et al,19 who reported on the correlation between short-term clinical response and long-term progression. In the meta-analysis by Hill et al,19 16-week efficacy data were abstracted from 15 clinical trials and used to develop a logistic equation to estimate the relative risk of transition from one CD4+ T-cell range to the next based on the observed increase in CD4+ T cells and log viral load decrease at 16 weeks. The equation derived from the meta-analysis was as follows:

Equation 1
Equation 1
Image Tools

In the Markov model, treatment-induced changes in viral load and CD4+ T cells from the urban hospital, supplemented by clinical trial estimates, are input into Equation 1 to determine the relative risk of transition through the disease stages with treatment. Thus, the natural history transition probabilities for each of the viral load categories are weighted by the relative risk to derive the progression probabilities under treatment for each viral load category.

Back to Top | Article Outline
Initial Viral Load Distribution

To run the model, the distribution of the population of interest across the 4 viral load ranges must be determined. The choice of distribution is important, because it has a direct impact on the estimation of survival and cost via the disease progression process. We used the initial viral load distribution observed in the Johns Hopkins HIV clinic for individuals initiating antiretroviral therapy at a CD4+ T-cell count at or greater than 350 cells/μL (see Table 1). In the base case model, we only included people whose viral load was greater than 3000 copies/mL.

Back to Top | Article Outline
Choice of Treatment Pathways

In this study, we model a treatment pathway representing real-world experience in an urban setting. The treatment pathway included initiation of treatment with a PI and 2 NRTIs, followed after failure of the initial regimen by treatment with 1 NNRTI and 2 NRTIs, followed after failure of the second-line regimen by 2 possible salvage therapy options: (1) 1 PI, 1 NNRTI, and 2 NRTIs or (2) 1 PI, 1 NNRTI, 2 NRTIs, and enfuvirtide (an HIV-1 fusion inhibitor). This treatment pathway follows generally accepted guidelines for treatment of HIV infection that account for the development of resistance to antiretroviral therapy over time and propose switching to regimens that are most likely to be effective after developing resistance to previous regimens.20 Early and later initiation of antiretroviral therapy was modeled using the same treatment pathway. The impact of using a different treatment pathway was tested in a sensitivity analysis.

Back to Top | Article Outline
Treatment Effectiveness and Duration

Because of the development of drug resistance over time, the treatment-induced changes in viral load and CD4+ T-cell count required by Equation 1 (treatment effectiveness) and the duration of treatment effectiveness depend on whether the drug regimens were first-line, second-line, or third-line treatments, because each subsequent drug regimen may be less effective than the preceding one. Observational data from the Johns Hopkins urban clinic were used to estimate the treatment effectiveness for first-line treatment and reflect the use of a mix of PIs plus 2 NRTIs.21 The reduction in viral load with second-line treatment (assumed to be an NNRTI, efavirenz, and 2 different NRTIs) was derived from the Johns Hopkins data as well as from first-line data presented in a study by Staszewski et al.22 The relative efficacy of the efavirens regimen compared with the nelfinavir regimen in reducing viral load to less than 50 copies/mL in the Staszewski study was 1.75, and this was multiplied by the gain in viral load in the Johns Hopkins population and then reduced by approximately 60% because of second-line use rather than first-line use.23 The CD4+ T-cell count increase was assumed to be approximately 60% of that observed in the clinical trial. Data for the salvage therapy regimens were taken directly from the enfuvirtide clinical trial data.24 Table 1 presents the estimated changes in CD4+ T-cell count and the log reductions in viral load and their associated relative risks of progression based on Equation 1 for each drug regimen modeled.

The modeled durations for each treatment regimen in the base case analyses were based on a simulation developed by Richter et al25 and assumed that the duration of first-line therapy was 4.5 years, the duration of second-line therapy was 4 years, and the duration of salvage therapy was 3 years. The reduced duration of subsequent treatment regimens is assumed because of the likelihood of increasing resistance to antiviral therapy over time. Shorter durations before switching from each drug regimen were tested in a sensitivity analysis.

Back to Top | Article Outline
Cost Inputs

Cost parameters include 6-month costs for each antiretroviral regimen as well as disease monitoring and other treatment costs for each 6-month period by CD4+ T-cell range. HAART regimen costs were calculated from the recommended daily doses and the average wholesale prices reported in the Drug Topics Red Book 2002.26 The expected cost for enfuvirtide was taken from a Wall Street Journal article.27 Disease monitoring and treatment costs for a 6-month period in each CD4+ T-cell range were taken from an unpublished 1999 study (M. M. Kitahata et al, final report for RESA41066: utilization and costs of medical care after the introduction of potent combination antiretroviral therapy among HIV-infected persons in managed care setting, 1999) using data from the Group Health Cooperative of Puget Sound. All costs are reported in US dollars as of 2002 and are adjusted from prior years when necessary using the medical services component of the consumer price index (www.bls.gov/cpi/home.htm). The cost input parameter values are shown in Table 2.

Table 2
Table 2
Image Tools
Back to Top | Article Outline
Other Inputs

Other input parameters in the model include utility weights for each CD4+ T-cell range and discount rates for costs and benefits. The utility weights for the base case are taken from a report by Freedberg et al28 and shown in Table 2. Utility weights from Tengs and Lin29 show a greater decline in health-related utility as the disease progresses compared with the estimates by Freedberg et al28 and are used in a sensitivity analysis.

In the base case, a 10% reduction in utility was assumed for those taking HAART, but a sensitivity analysis tested the impact of a 20% reduction in utility while on HAART to give more weight to the side effects associated with treatment. A discount rate of 3% was used for the costs and health benefits in accordance with the recommendations of the Panel on Cost Effectiveness.30

Back to Top | Article Outline
Analysis

A base case analysis and sensitivity analyses were performed. The base case analysis uses the set of feasible input parameter estimates that were described previously. The base case analysis computes the remaining lifetime costs, health outcomes, life expectancy, and quality-adjusted life expectancy for people with HIV infection from the time that they are in the CD4+ T-cell count range of 499 to 350 cells/μL (mean value of 425 cells/μL). Costs and outcomes are compared for patients who start HAART early in their disease (when their CD4+ T-cell count is greater than 350 cells/μL) and patients who start HAART later in their disease (when their CD4+ T-cell count is less than 350 cells/μL but greater than 200 cells/μL). The incremental cost-effectiveness ratio of starting therapy when the patient's CD4+ T-cell count is greater than 350 cells/μL as compared with starting when the patient's CD4+ T-cell count is less than 350 cells/μL but greater than 200 cells/μL is computed.

A set of sensitivity analyses was performed with the goal of illustrating the impact of the effectiveness and duration of each line of therapy. In addition, the components of salvage therapy were varied to demonstrate the impact of the addition of enfuvirtide to optimized background therapy on cost and response for salvage patients. The quality of life and resource costs in each disease stage with and without HAART were also varied to determine the impact of these variables on outcomes. For example, adverse events associated with PIs might reduce quality of life and require treatment with cholesterol-lowering drugs, which would increase annual treatment costs. Finally, the impact of alternative viral load set points was determined.

Back to Top | Article Outline

RESULTS

Base Case Analysis

Results of the base case analysis are shown in Tables 3 and 4. These results indicate that in the base case, costs, survival, and quality-adjusted survival were all higher when HAART was initiated when the patient's CD4+ T-cell count was greater than 350 cells/μL as compared with when the patient's CD4+ T-cell count was less than 350 cells/μL. The incremental costs per life-years gained and the incremental costs per quality-adjusted life-years gained for starting HAART at greater than 350 cells/μL versus less than 350 cells/μL but greater than 200 cells/μL were $21,567 and $31,266, respectively. The incremental costs per life-years gained and incremental costs per quality-adjusted life-years gained were $22,064 and $25,806, respectively, when comparing starting HAART started at less than 350 cells/μL but greater than 200 cells/μL versus less than 200 cells/μL.

Table 3
Table 3
Image Tools
Table 4
Table 4
Image Tools
Back to Top | Article Outline
Sensitivity Analyses

The results of the sensitivity analyses (Table 5) showed that the incremental cost-effectiveness of early HAART (CD4+ T-cell count greater than 350 cells/μL) generally remains less than $50,000 per quality-adjusted life-year gained when compared with later HAART (CD4+ T-cell count less than 350 cells/μL but greater than 200 cells/μL) for tested variations in the parameter values. The incremental cost-effectiveness ratio for early HAART versus later HAART is a little higher when enfuvirtide is used for salvage therapy; however, the quality-adjusted life expectancy gain is lower when enfuvirtide is not used. The incremental cost-effectiveness ratio of early HAART versus later HAART is also higher when salvage therapy is continued indefinitely, but the quality-adjusted life expectancy gain is also higher.

Table 5
Table 5
Image Tools

A shorter duration of each treatment regimen from 4.5 years of first-line, 4.0 years of second-line, and 3.0 years of salvage treatment to 3.0 years of first-line, 2.5 years of second-line, and 1.5 years of salvage treatment results in a higher incremental cost-effectiveness ratio for early HAART and also reduces the quality-adjusted life expectancy gain from early HAART. The assumption of a completely different treatment pathway using efavirenz as a first-line treatment for 5 years and boosted PIs as a second-line treatment for a total of 7 years before salvage therapy results in a higher incremental cost-effectiveness ratio for early HAART compared with later HAART and a slightly lower gain in quality-adjusted life expectancy.

The incremental cost-effectiveness of early initiation compared with later initiation is sensitive to changes in the magnitude of the utility losses as the disease progresses as well as to changes in the magnitude of the utility loss (if any) at each disease stage attributable to HAART therapy. The incremental cost-effectiveness ratio for early HAART is higher for the base case, which has a lower utility loss with disease progression and a 10% utility loss associated with HAART therapy, compared with scenarios with a higher loss in utility with disease progression or a lesser impact of HAART on utility. An increased negative impact of HAART on utility is associated with a higher incremental cost-effectiveness ratio. An increased cost associated with HAART monitoring and side effects also slightly increases the incremental cost-effectiveness ratio for early therapy. Finally, our sensitivity analysis indicates that the incremental cost-effectiveness of early versus later therapy is higher for those with lower viral loads, because gains in life expectancy from early HAART are smaller in this population.

Back to Top | Article Outline

DISCUSSION

The results of the analyses using the Markov cost-effectiveness model and real-world data demonstrate that starting HAART earlier (at a CD4+ T-cell count greater than 350 cells/μL) rather than later (at a CD4+ T-cell count less than 350 cells/μL but greater than 200 cells/μL) increases total lifetime costs, increases life expectancy, and is a cost-effective strategy when this is defined as a strategy with a cost per life-year gained or cost per quality-adjusted life-year gained less than $50,000, a generally accepted benchmark in the United States. This is true under different assumptions about the treatment regimens used, duration of efficacy for first- and second-line therapy, set point viral load, and impact of HAART on disease treatment costs and individual utility. Early therapy is more cost-effective the smaller the impact of HAART on utility (or quality of life) is or the larger the negative impact on well-being is of more severe stages of HIV infection and for those with higher viral loads.

Our results also confirm the generally accepted view that starting HAART at a CD4+ T-cell count less than 350 cells/μL but greater than 200 cells/μL is cost-effective compared with starting at a CD4+ T-cell count less than 200 cells/μL. Our modeling exercise indicates that further life expectancy and quality-adjusted life expectancy gains can be expected when starting HAART at a CD4+ T-cell count greater than 350/μL and that this early treatment strategy is also likely to be cost-effective.

Unlike previous HIV Markov models,15-17 the Markov model used for this analysis allowed set point viral load to influence the rate of CD4+ T-cell count decline based on the study by Mellors et al.18 This model has several limitations, including the assumption that the relative risk of disease progression with HAART and the duration of efficacy are similar across all viral load categories and all CD4+ T-cell ranges. There is some evidence that this may not be the case, for example, that duration of effect may be longer when therapy is started earlier.31,32 Such differences would result in greater cost-effectiveness for starting HAART early (ie, lower incremental cost-effectiveness ratios than those presented in this report).

The model also did not compute possible reductions in life expectancy associated with long-term side effects of HAART (eg, dyslipidemia). The recent study by Schackman et al11 has shown that these effects are small and that less than 10% of the increased life expectancy attributable to HAART is likely to be lost as a result of the increased cardiovascular risk. Such a loss in life expectancy gain would not have a substantial impact on the cost-effectiveness ratios for early versus later HAART therapy.

Estimates of the impact of HAART on overall well-being in treated individuals are sparse,33 and the impact depends on the particular regimen being used; thus, we ran sensitivity analyses that assumed more or less loss in well-being than the 10% assumed in the base case. The results were sensitive to these changes, indicating the likely value of HAART regimens with more convenient dosing regimens and fewer side effects.

Finally, data on the number of different treatment regimens used and the duration of efficacy of each treatment regimen in real-world practice are limited; thus, the data used in the model are suggestive only. Our sensitivity analysis using a more effective set of regimens over a longer period indicated that the cost-effectiveness ratios are somewhat higher with the more effective and durable treatment pathways but that the ratio is still well below the $50,000 benchmark value. Total lifetime costs and life expectancy with these more effective regimens are considerably higher than with less effective and less durable pathways, as would be expected.

The loss of the ability to use certain drug classes because of the development of drug resistance is an important limiting factor in the duration of efficacy of HAART therapy. In this study, we chose to compare identical HAART treatment pathways started earlier or later in the disease course so as to focus on the question of when to start therapy while controlling for the different efficacies of different treatment pathways. We assumed that the person discontinues HAART after 3 different HAART regimens and experiences disease progression at the nontreated rate because of the development of drug resistance. This discontinuation occurs in our model earlier in the disease course for those starting HAART earlier than for those starting HAART later. We included a sensitivity analysis to demonstrate the impact of continuing salvage therapy and showed that although the total lifetime costs change, the cost-effectiveness of early versus later treatment changes little.

The results found in this study-that initiation of HAART at a CD4+ T-cell count greater than 350 cells/μL increases life expectancy and is cost-effective-are similar to those found in the modeling studies by Schackman et al10,11 but differ from those in a study presented at the 14th Annual International AIDS Conference.3 The latter study followed people who started with a CD4+ T-cell range of 350 to 499 cells/μL or 200 to 349 cells/μL for approximately 2 years. The primary end point in this study was progression to AIDS. When treating people in these CD4+ T-cell ranges with HAART, however, few people progress to AIDS within 2 years. Thus, the lack of a difference between early and later initiation of HAART in progression to AIDS may be a function of the relatively short period of observation rather than the starting points for treatment initiation. This is supported by a recent study by Palella et al7 using data from the HIV Outpatient Study (HOPS) study, with follow-up of approximately 4 years for those with a CD4+ T-cell count range between 350 cells/μL and 499 cells/μL, showing a trend toward higher mortality rates for those delaying HAART until their CD4+ T-cell count fell out of this range. Similar results have been shown by Opravil et al6 using data from the Swiss HIV Cohort Study. Thus, empiric studies with follow-up longer than 2 years are beginning to show health benefits associated with early versus later initiation of HAART. Both the Markov model used in this study and the Monte Carlo simulation model presented by Schackman et al10,11 follow each cohort for their remaining lifetime and are thus able to approximate the long-term benefits of early HAART that may not be apparent in shorter term empiric studies.

Our model differs from previous HIV cost-effectiveness models in using data from an urban hospital population rather than from a clinical trial population. The effectiveness of drug therapy in real-world practice is often lower than the efficacy seen in clinical trials. Our cost-effectiveness ratios are higher (less favorable) than those presented by Schackman et al,10,11 probably reflecting our use of the urban hospital population effectiveness values. Nevertheless, our study has shown that initiation of HAART at a CD4+ T-cell count greater than 350 cells/μL is likely to be cost-effective relative to later initiation, even in a real-world practice situation.

The results of this study indicate that physicians and patients making decisions about when to start HAART should balance the likely benefits on life expectancy with the possible negative impact on overall well-being associated with HAART. Clearly, drug regimens that have more convenient dosing regimens and fewer adverse effects can minimize the negative impact on overall well-being and thus would be likely to have greater value, assuming that their price was similar to the prices of drugs with less favorable side-effect profiles. For the time being, there seems to be little evidence to suggest that the initiation of HAART should be delayed on the basis of cost-effectiveness. This would be especially true if new drugs that have a more favorable side-effect profile become available.

Back to Top | Article Outline

ACKNOWLEDGMENTS

The authors thank Richard D. Moore from the Johns Hopkins Hospital for providing the effectiveness analysis for the urban hospital cohort used in this model and for providing comments on earlier drafts of the paper. The authors also thank Dr. Stephanie Earnshaw for assistance in developing the treatment pathways.

Back to Top | Article Outline

REFERENCES

1. Department of Health and Human Services (DHHS). AIDSinfo: overview of antiretroviral drugs. Available at: www.aidsinfo.nih.gov/drugs/. Accessed March 16, 2003.

2. Department of Health and Human Services (DHHS). Guidelines for the use of antiretroviral agents in HIV-infected adults and adolescents. February 4, 2002. Available at: www.aidsinfo.nih.gov. Accessed March 16, 2003.

3. Chene G, May M, Costagliola D, et al. Prognosis of HIV-1 infected drug naïve patients starting potent antiretroviral therapy. ART Cohort Collaboration [abstract TuOrB1140]. Presented at: XIVth International AIDS Conference; 2002; Barcelona, Spain.

4. Hogg RS, Yip B, Chan KJ, et al. Rates of disease progression by baseline CD4 cell count and viral load after initiating triple-drug therapy. JAMA. 2001;286:2568-2577.

5. Sterling TR, Chaisson RE, Moore RD. Initiation of highly active antiretroviral therapy at CD4+ T lymphocyte counts of >350 cells/mm3: disease progression, treatment durability, and drug toxicity. Clin Infect Dis. 2003;36:812-815.

6. Opravil M, Ledergerber B, Furrer H. Clinical efficacy of early initiation of HAART in patients with asymptomatic HIV infection and CD4 cell count >350 × 10(6)/L. AIDS. 2002;16:1371-1381.

7. Palella FJ, Deloria-Knoll M, Chmiel JS, et al. Survival benefit of initiating antiretroviral therapy in HIV-infected persons in different CD4+ cell strata. Ann Intern Med. 2003;138:620-626.

8. Sterling TR. When to start HAART, and what to start. Hopkins HIV Rep. 2003;15:5.

9. Dybul M, Fauci AS, Bartlett JG, et al. Panel on Clinical Practices for Treatment of HIV. Guidelines for using antiretroviral agents among HIV-infected adults and adolescents. Ann Intern Med. 2002;137:381-433.

10. Schackman BR, Goldie SJ, Weinstein MC, et al. Cost-effectiveness of earlier initiation of antiretroviral therapy for uninsured HIV-infected adults. Am J Public Health. 2001;91:1456-1463.

11. Schackman BR, Freedberg KA, Weinstein MC, et al. Cost-effectiveness implications of the timing of antiretroviral therapy in HIV-infected adults. Arch Intern Med. 2002;162:2478-2486.

12. Gebo KA, Diener-West M, Moore RD. Hospitalization rates differ by hepatitis C status in an urban HIV cohort. J Acquir Immune Defic Syndr. 2003;34:165-173.

13. Perez JL, Moore RD. Greater effect of highly active antiretroviral therapy on survival in people aged ≥50 years compared with younger people in an urban observational cohort. Clin Infect Dis. 2003;36:212-218.

14. Gebo KA, Diener-West M, Moore RD. Hospitalization rates in an urban cohort after the introduction of highly active antiretroviral therapy. J Acquir Immune Defic Syndr. 2001;27:143-152.

15. Moore RD, Bartlett JG. Combination antiretroviral therapy in HIV infection. Pharmacoeconomics. 1996;10:109-113.

16. Chancellor J, Hill A, Sabir C, et al. Modeling the cost effectiveness of lamivudine/zidovudine combination in HIV infection. Pharmacoeconomics. 1997;12:54-66.

17. Mauskopf J, Lacey L, Kempel A, et al. The cost effectiveness of treatment with lamivudine and zidovudine compared with zidovudine alone: a comparison of Markov model and trial data estimates. Am J Manag Care. 1998;4:1004-1012.

18. Mellors JW, Munoz A, Giorgi JV, et al. Plasma viral load and CD4+ lymphocytes as prognostic markers of HIV-1 infection. Ann Intern Med. 1997;126:946-954.

19. Hill AM, De Masi R, Dawson D. Meta-analysis of antiretroviral effects on HIV-1 RNA, CD4 cell count and progression to AIDS or death. Antivir Ther. 1998;3:139-145.

20. Yeni PG, Hammer SM, Carpenter CC, et al. Antiretroviral treatment for adult HIV infection in 2002: updated recommendations of the International AIDS Society-USA Panel. JAMA. 2002;288:222-235.

21. Moore R. Report to GlaxoSmithKline: additional analyses of HIV-1 RNA and CD4 response to HAART for use in the ViLoCD model. April 23, 2001.

22. Staszewski S, Morales-Ramirez J, Tashima KT, et al. Efavirenz plus zidovudine and lamivudine, efavirenz plus indinavir, and indinavir plus zidovudine and lamivudine in the treatment of HIV-1 infection in adults. N Engl J Med. 1999;341:1865-1873.

23. Baxter JD, Mayers DL, Wentworth DN, et al. A randomized study of antiretroviral management based on plasma genotypic antiretroviral resistance testing in patients failing therapy. AIDS. 2000;14(Suppl):F83-F93.

24. Lalezari JP, Henry K, O'Hearn M, et al. Enfuvirtide, an HIV-1 fusion inhibitor for drug-resistant HIV infection in North and South America. N Engl J Med. 2003;348:2175-2185.

25. Richter A, Hauber B, Simpson K, et al. A Monte Carlo simulation for modelling outcomes of AIDS treatment regimens. Pharmacoeconomics. 2002;20:215-224.

26. Drug Topics Red Book. Montvale, NJ: Medical Economics Company; 2002.

27. Fuhrmans V. New AIDS drug shows promise, but it may prove unaffordable. Wall Street Journal. July 8, 2002.

28. Freedberg KA, Losina E, Weinstein MC, et al. The cost effectiveness of combination antiretroviral therapy for HIV disease. N Engl J Med. 2001;344:824-831.

29. Tengs TO, Lin TH. A meta-analysis of utility estimates for HIV/AIDS. Med Decis Making. 2002;22:475-481.

30. Gold MR, Siegel JE, Russel LB, et al. Cost-Effectiveness in Health and Medicine. New York: Oxford University Press; 1996.

31. Brooks JT, Hanson DL, McNaghten AD, et al. Low CD4 count predicts less durable virologic response to highly active antiretroviral therapy (HAART) in naïve patients [abstract TuOrB1141]. Presented at: XIVth International AIDS Conference, 2002; Barcelona, Spain.

32. Chaisson RE, Keruly JC, Moore RD, et al. Association of initial CD4 cell count and viral load with response to highly active antiretroviral therapy. JAMA. 2000;284:3128-3129.

33. Gill CJ, Griffith JL, Jacobson D, et al. Relationship of HIV viral loads, CD4 counts, and HAART use to health-related quality of life. J Acquir Immune Defic Syndr. 2002;30:485-492.

Cited By:

This article has been cited 23 time(s).

AIDS Patient Care and Stds
To take or not to take: Decision-making about antiretroviral treatment in people living with HIV/AIDS
Kremer, H; Ironson, G; Schneiderman, N; Hautzinger, M
AIDS Patient Care and Stds, 20(5): 335-349.

Medical Decision Making
A systematic review of cost-utility analyses in HIV/AIDS: Implications for public policy
Hornberger, J; Holodniy, M; Robertus, K; Winnike, M; Gibson, E; Verhulst, E
Medical Decision Making, 27(6): 789-821.
10.1177/0272989X07306112
CrossRef
Revista Panamericana De Salud Publica-Pan American Journal of Public Health
HIV risk behaviors of Latin American and Caribbean men who have sex with men in Miami, Florida, USA
Akin, M; Fernandez, MI; Bowen, GS; Warren, JC
Revista Panamericana De Salud Publica-Pan American Journal of Public Health, 23(5): 341-348.

Clinical Infectious Diseases
When to Start Antiretroviral Therapy?
Wilkin, TJ; Gulick, RM
Clinical Infectious Diseases, 47(): 1580-1586.
10.1086/593311
CrossRef
Medical Decision Making
"It's my body": Does patient involvement an decision making reduce decisional conflict?
Kremer, H; Ironson, G; Schneiderman, N; Hautzinger, M
Medical Decision Making, 27(5): 522-532.
10.1177/0272989x07306782
CrossRef
AIDS Research and Human Retroviruses
Predicting CD4 lymphocyte count < 200 Cells/mm(3) in an HIV type 1-infected African population
Morpeth, SC; Crump, JA; Shao, HJ; Ramadhani, HO; Kisenge, PR; Moylan, CA; Naggie, S; Caram, LB; Landman, KZ; Sam, NE; Itemba, DK; Shao, JF; Bartlett, JA; Thielman, NM
AIDS Research and Human Retroviruses, 23(): 1230-1236.
10.1089/aid.2007.0053
CrossRef
Journal of Virology
Antiretroviral Therapy in the Clinic
Tsibris, AMN; Hirsch, MS
Journal of Virology, 84(): 5458-5464.
10.1128/JVI.02524-09
CrossRef
AIDS Patient Care and Stds
Patients' attitudes about rapid oral HIV screening in an urban, free dental clinic
Dietz, CA; Ablah, E; Reznik, D; Robbins, DK
AIDS Patient Care and Stds, 22(3): 205-212.
10.1089/apc.2007.0235
CrossRef
Hiv Clinical Trials
Efficacy, safety, and tolerability of long-term combination Antiretroviral therapy in asymptornatic treatment-naive adults with early HIV infection
McMahon, DK; DiNubile, MJ; Meibohm, AR; Marino, DR; Robertson, MN
Hiv Clinical Trials, 8(5): 269-281.
10.1310/hct0805-269
CrossRef
Jama-Journal of the American Medical Association
Antiretroviral treatment of adult HIV infection - 2008 recommendations of the International AIDS Society USA panel
Hammer, SM; Eron, JJ; Reiss, P; Schooley, RT; Thompson, MA; Walmsley, S; Cahn, P; Fischl, MA; Gatell, JM; Hirsch, MS; Jacobsen, DM; Montaner, JSG; Richman, DD; Yeni, PG; Volberding, PA
Jama-Journal of the American Medical Association, 300(5): 555-570.

Antiviral Therapy
When to initiate highly active antiretroviral therapy in low-resource settings: the Moroccan experience
Loubiere, S; el Filal, KM; Sodqi, M; Loundou, A; Luchini, S; Cleory, S; Moatti, JP; Himmich, H
Antiviral Therapy, 13(2): 241-251.

Revista Espanola De Salud Publica
Cost-effectiveness of initial adherence to antiretroviral therapy among HIV infected patients in Belo Horizonte, Brazil
Acurcio, FDA; Puig-Junoy, J; Bonolo, PD; Ceccato, MD; Guimaraes, MDC
Revista Espanola De Salud Publica, 80(1): 41-54.

Pharmacoeconomics
Optimal assignment of treatments to health states using a Markov decision model - An introduction to basic concepts
Bala, MV; Mauskopf, JA
Pharmacoeconomics, 24(4): 345-354.

Pharmacoeconomics
Economic models of antiretroviral therapy - Searching for the optimal strategy
Hellinger, FJ
Pharmacoeconomics, 24(7): 631-642.

Operations Research
The optimal time to initiate HIV therapy under ordered health states
Shechter, SM; Bailey, MD; Schaefer, AJ; Roberts, MS
Operations Research, 56(1): 20-33.

Hiv Medicine
Cost of medical care for HIV-infected patients within a regional population from 1997 to 2006
Krentz, HB; Gill, MJ
Hiv Medicine, 9(9): 721-730.
10.1111/j.1468-1293.2008.00613.x
CrossRef
Value in Health
A Predictive Model of Health State Utilities for HIV Patients in the Modern Era of Highly Active Antiretroviral Therapy
Kauf, TL; Roskell, N; Shearer, A; Gazzard, B; Mauskopf, J; Davis, EA; Nimsch, C
Value in Health, 11(7): 1144-1153.
10.1111/j.1524-4733.2008.00326.x
CrossRef
Bmc Medicine
When to start antiretroviral therapy: as soon as possible
Franco, RA; Saag, MS
Bmc Medicine, 11(): -.
ARTN 147
CrossRef
Plos One
Cost-Effectiveness Analysis of Option B plus for HIV Prevention and Treatment of Mothers and Children in Malawi
Fasawe, O; Avila, C; Shaffer, N; Schouten, E; Chimbwandira, F; Hoos, D; Nakakeeto, O; De Lay, P
Plos One, 8(3): -.
ARTN e57778
CrossRef
Quality of Life Research
The associations among coping, nadir CD4+T-cell count, and non-HIV-related variables with health-related quality of life among an ambulatory HIV-positive patient population
Armon, C; Lichtenstein, K
Quality of Life Research, 21(6): 993-1003.
10.1007/s11136-011-0017-2
CrossRef
JAIDS Journal of Acquired Immune Deficiency Syndromes
No Variability Across Centers in Adherence and Response to HAART in French Hospitals: Results From the ANRS-EN12-VESPA Study
Sitta, R; Lert, F; Gueguen, A; Spire, B; Dray-Spira, R; and the VESPA group,
JAIDS Journal of Acquired Immune Deficiency Syndromes, 52(5): 643-647.
10.1097/QAI.0b013e3181b26eb9
PDF (97) | CrossRef
Current Opinion in Infectious Diseases
Should HIV therapy be started at a CD4 cell count above 350 cells/μl in asymptomatic HIV-1-infected patients?
Sabin, CA; Phillips, AN
Current Opinion in Infectious Diseases, 22(2): 191-197.
10.1097/QCO.0b013e328326cd34
PDF (119) | CrossRef
AIDS
Cost-effectiveness of routine and low-cost CD4 T-cell count compared with WHO clinical staging of HIV to guide initiation of antiretroviral therapy in resource-limited settings
Athan, E; O'Brien, DP; Legood, R
AIDS, 24(12): 1887-1895.
10.1097/QAD.0b013e32833b25ed
PDF (214) | CrossRef
Back to Top | Article Outline
Keywords:

highly active antiretroviral therapy; early versus later; Markov model; costs; effectiveness

© 2005 Lippincott Williams & Wilkins, Inc.

Login

Search for Similar Articles
You may search for similar articles that contain these same keywords or you may modify the keyword list to augment your search.