|
HIV antiretroviral treatment: early versus later |
Mauskopf J, Kitahata M, Kauf T, Richter A, Tolson J |
|
|
Record Status This is a critical abstract of an economic evaluation that meets the criteria for inclusion on NHS EED. Each abstract contains a brief summary of the methods, the results and conclusions followed by a detailed critical assessment on the reliability of the study and the conclusions drawn. Health technology Highly active antiretroviral therapy (HAART) was assessed. Specifically, HAART commenced early (CD4+ T-cell count greater than 350 cells/microL) was compared with therapy started later (CD4+ T-cell count of >200 but <350 cells/microL). HAART consisted of first-line treatment with a protease inhibitor (PI) and two nucleoside reverse transcriptase inhibitors (NRTIs), second-line treatment with one nonnucleoside reverse transcriptase inhibitor (NNRTI) and two NRTIS, followed by two possible salvage therapies. One salvage therapy comprised one PI, one NNRTI and two NRTIs, the other comprised one PI, one NNRTI, two NRTIs and enfuvirtide (an HIV-1 fusion inhibitor).
Type of intervention Treatment for patients with the human immunodeficiency virus (HIV) and the secondary prevention of acquired immune deficiency syndrome (AIDS).
Study population The study population comprised a hypothetical cohort of people with HIV infection.
Setting The setting was tertiary care. The economic study was carried out in Baltimore, USA.
Dates to which data relate The effectiveness data were collected from studies published between 1997 and 2003. Average whole sale prices were collected for 2002 and the expected cost of enfuvirtide was collected from a study published in 2002. Resource use was determined by a decision model. The price year was 2002.
Source of effectiveness data The effectiveness data were derived from a review and synthesis of completed studies.
Modelling A Markov model was created to track the progression of HIV infection through six disease stages. These corresponded to CD4+ T-cell ranges of at least 500 cells/microL, 350 - 499 cells/microL, 200 - 349 cells/microL, 100 - 199 cells/microL, 0 - 100 cells/microL, and death. Six-monthly time cycles were used. Sequences of different first-line, second-line and salvage antiretroviral regimens (and their impact on transition probabilities) were modelled as treatment pathways. The time horizon of the model was 25 years.
Outcomes assessed in the review Six-monthly transition probabilities were estimated from the review. The authors do not appear to have carried out a systematic review of the literature. Instead, they appear to have selected studies that reported outcomes relevant to their model.
Study designs and other criteria for inclusion in the review Sources searched to identify primary studies Criteria used to ensure the validity of primary studies Methods used to judge relevance and validity, and for extracting data Number of primary studies included Six studies were included in the review.
Methods of combining primary studies The authors used single estimates to represent the majority of inputs for their model. The effectiveness data for disease progression given first-line treatment were taken from the Johns Hopkins HIV clinic; the untreated rate of disease progression was estimated from Mellors et al. (1997, see 'Other Publications of Related Interest' below for bibliographic details). The effectiveness of second-line treatment was estimated from a meta-analysis of studies with results placed into an equation devised by Hill et al. (1998) to estimate the impact of antiretroviral treatment (see 'Other Publications of Related Interest' below for bibliographic details).
Investigation of differences between primary studies Results of the review The 6-month transition probabilities without HAART were presented.
The probability of moving from 500 cells/microL to 350 - 499 cells/microL was 31.0% for viral load 3,000 to 10,000 copies/mL, 35.3% for viral load 10,001 to 30,000 copies/mL, and 42.7% for viral load greater than 30,000.
The probability of moving from 350 - 499 cells/microL to 200 - 349 cells/microL was 31.0% for viral load 3,000 to 10,000 copies/mL, 35.3% for viral load 10,001 to 30,000 copies/mL, and 42.7% for viral load greater than 30,000.
The probability of moving from 200 - 349 cells/microL to 100 - 199 cells/microL was 31.0% for viral load 3,000 to 10,000 copies/mL, 35.3% for viral load 10,001 to 30,000 copies/mL, and 42.7% for viral load greater than 30,000.
The probability of moving from 100 - 199 cells/microL to 0 - 100 cells/microL was 44.3% for viral load 3,000 to 10,000 copies/mL, 49.1% for viral load 10,001 to 30,000 copies/mL, and 58.3% for viral load greater than 30,000.
The probability of moving from 0 - 100 cells/microL to death was 44.3% for viral load 3,000 to 10,000 copies/mL, 49.1% for viral load 10,001 to 30,000 copies/mL, and 58.3% for viral load greater than 30,000.
The relative risk of transition between health states was 27% for first-line treatment, 26.53% for second-line treatment, 22.80% for salvage therapy, and 51.95% for optimised background.
Methods used to derive estimates of effectiveness The authors used some assumptions to supplement the information obtained from their review.
Estimates of effectiveness and key assumptions The authors assumed that the effectiveness of second-line and salvage therapy was less than the efficacy observed in clinical trials, because the effectiveness of first-line HAART in the Johns Hopkins trial was less than the clinical trial results observed in the review. Specifically, the effectiveness of second-line treatment was reduced by approximately 60%. The authors also assumed that the duration of first-line therapy was 4.5 years, that of second-line therapy was 4 years and that of salvage therapy was 3 years.
Measure of benefits used in the economic analysis The summary measures of health benefit used were life expectancy and quality-adjusted life expectancy. Utility weights to adjust life expectancy were taken from studies published by Freedberg et al and Tengs and Lin ((see 'Other Publications of Related Interest' below for bibliographic details). The benefits were discounted at a rate of 3%.
Direct costs A perspective for the cost analysis was not reported. The authors estimated 6-monthly costs for each antiretroviral regimen. HAART regimen costs were calculated using the recommended daily dosages and average whole sale prices reported in Drug Topics Red Book 2002. The expected cost of enfuvirtide was collected from a study published in 2002. Disease monitoring and other treatment costs for each CD4+ T-cell range were taken from an unpublished study. Resource use was determined using the decision model. The costs were discounted at a rate of 3%. The price year was 2002. The prices were reflated using the medical services component of the Consumer Price Index.
Statistical analysis of costs The costs were treated deterministically.
Indirect Costs The indirect costs were not estimated.
Sensitivity analysis Sensitivity analyses were carried out. Their aim was to explore the impact of different treatment pathways, the effectiveness and duration of each line of therapy, the components of salvage therapy, quality of life and resource cost.
Estimated benefits used in the economic analysis The life-years from a mean CD4+ T-cell count of 425 cells/microL were 9.73 when HAART was initiated at a CD4+T-cell count of greater than 350 cells/microL, 8.98 when initiated at between 200 and 250 cells/microL, and 7.71 when initiated at less than 200 cells/microL.
The quality-adjusted life-years (QALYs) from a mean CD4+ T-cell count of 425 cells/microL were 7.98 when HAART was initiated at a CD4+T-cell count of greater than 350/microL, 7.37 when initiated at between 200 and 250 microL, and 6.28 when initiated at less than 200/microL.
The incremental QALYs gained from starting HAART at a CD4+T-cell count of greater than 305 cells/microL, compared with between 200 and 350 cells/microL, were 0.75.
The incremental QALYs gained from starting HAART at a CD4+T-cell count of between 200 and 350 cells/microL, compared with less than 200 cells/microL, were 1.27.
Cost results The remaining lifetime costs from a mean CD4+ T-cell count of 425 cells/microL were $141,303 when HAART was initiated at a CD4+T-cell count of greater than 350 cells/microL, $122,229 when initiated at between 200 and 250 cells/microL, and $94,163 when initiated at less than 200 cells/microL.
The incremental lifetime costs gained from starting HAART at a CD4+T-cell count of greater than 305 cells/microL, compared with between 200 and 350 cells/microL, were $19,074.
The incremental lifetime costs gained from starting HAART at a CD4+T-cell count of between 200 and 350 cells/microL, compared with less than 200 cells/microL, were $28,066.
Synthesis of costs and benefits The incremental cost per life-year gained between starting HAART at a CD4+T-cell count of greater than 350 cells/microL, compared with between 200 and 350 cells/microL, was $25,567.
The incremental cost per life-year gained between starting HAART at a CD4+T-cell count of between 200 and 350 cells/microL, compared with less than 200 cells/microL, was $22,064.
The incremental cost per QALY gained between starting HAART at a CD4+T-cell count of greater than 350 cells/microL, compared with between 200 and 350 cells/microL, was $31,226.
The incremental cost per QALY gained between starting HAART at a CD4+T-cell count of between 200 and 350 cells/microL, compared with less than 200 cells/microL, was $25,806.
Sensitivity analyses showed that the incremental cost-effectiveness of early HAART compared with later HAART generally remained less than $50,000 per QALY.
The incremental cost-effectiveness was reported to be sensitive to changes in the magnitude of utility loss at each disease stage attributable to HAART.
Authors' conclusions Starting highly active antiretroviral therapy (HAART) earlier rather than later "increases total lifetime costs, increases life expectancy, and is a cost-effective strategy" when $50,000 per quality-adjusted life-year (QALY) gained is used as a threshold to define cost-effectiveness. This conclusion was shown to be robust in the sensitivity analysis.
CRD COMMENTARY - Selection of comparators The authors compared early versus later starting of HAART. This was justified by a discussion of the literature in this area and an explanation of remaining uncertainties at to the exact level of CD4+ T-cell count at which to commence therapy.
Validity of estimate of measure of effectiveness The authors did not carry out a systematic review of the literature, but used data from a specific clinic and published literature to populate their model. The authors clearly identified which sources of literature were used to inform different aspects of their model. Relatively few details of the methodology behind the review were reported. For example, the authors did not state how they chose papers for inclusion or ensured the validity of their estimates. The authors made the decision to reduce the effectiveness of second-line treatment by 60%, owing to a conflict between first-line results from their clinic and the meta-analysis of second-line results. This assumption concerning effectiveness was clearly not supported by the available literature and casts some doubts over the validity of the results presented. Although the ad hoc selection of studies to populate decision models is common place, it does not ensure that the best available evidence has been obtained and used.
Validity of estimate of measure of benefit The estimates of life expectancy and quality-adjusted life expectancy were modelled using utility values from the literature. These measures were appropriate to the authors' objective. They also provide estimates that enable the cost-effectiveness to be compared with a broad range of other technologies.
Validity of estimate of costs A perspective for the cost analysis was not reported. It was therefore not possible to assess whether all the relevant costs were incorporated. Nevertheless, the costs reported suggested that the perspective of the health care provider was adopted. The sources of the cost estimates were clearly documented. A statistical analysis of the costs would have enabled the reader to judge whether differences in total cost were statistically significant.
Other issues The authors were able to make some comparisons with the literature. They identified work with similar results as well as some with conflicting results. For the latter, the authors discussed the shorter follow-up as a potential reason for differences in the results. The issue of generalisability was not addressed and, with the use of effectiveness data from a clinic for which the population characteristics were not reported, it would be unwise to generalise the results of this study. The authors did not present the results selectively and their conclusions accurately reflected the results given.
Several limitations to the study were highlighted. For example, the authors assumed that the relative risk of disease progression with HAART and the duration of efficacy were similar across all viral load categories. The authors suggested that this might not be the case in practice and assessed the potential impact of this assumption not being true. In addition, the model did not take long-term side effects and their effect on life expectancy into consideration, although the authors argued that the impact of this would be minimal.
Implications of the study The authors did not make any recommendations for policy or practice further to their study, nor did they raise suggestions for further work. However, further work could consider identifying more accurately the optimal time to start therapy and continuing to answer the same questions in the light of newly emerging HIV therapies.
Bibliographic details Mauskopf J, Kitahata M, Kauf T, Richter A, Tolson J. HIV antiretroviral treatment: early versus later. Journal of Acquired Immune Deficiency Syndromes 2005; 39(5): 562-569 Other publications of related interest Freedberg KA, Losina E, Weinstein MC, et al. The cost effectiveness of combination antiretroviral therapy for HIV disease. N Engl J Med 2001;344:824-31.
Tengs TO, Lin TH. A meta-analysis of utility estimates for HIV/AIDS. Med Decis Making 2002;22:475-81.
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;126:946-54.
Hill AM, De Masi R, Dawson D. Meta-analysis of antiretroviral effectis on HIV-1 RNA, CD4 cell count and progression to AIDS or death. Antivir Ther 1998;3:139-45.
Indexing Status Subject indexing assigned by NLM MeSH Antiretroviral Therapy, Highly Active /economics /methods; CD4 Lymphocyte Count; Cost-Benefit Analysis; Drug Administration Schedule; HIV Infections /drug therapy; Humans; Longevity; Markov Chains; Models, Statistical; Quality of Life; Viral Load AccessionNumber 22005006457 Date bibliographic record published 30/04/2006 Date abstract record published 30/04/2006 |
|
|
|