|
Economic evaluation of the Randomized Aldactone Evaluation Study (RALES): treatment of patients with severe heart failure |
Glick H A, Orzol S M, Tooley J F, Remme W J, Sasayama S, Pitt B |
|
|
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 The treatment of advanced heart failure with spironolactone was studied.
Study population The hypothetical population used in the economic model was based on the study population in the Randomized Aldactone Evaluation Study (RALES). RALES was a randomised, double-blinded, placebo-controlled trial that enrolled 1,663 participants with severe heart failure and a left ventricular ejection fraction (LVEF) of no more than 35% who were receiving standard therapy, including an angiotensin-converting enzyme (ACE) inhibitor and a loop diuretic, with or without digoxin. Severe heart failure was defined as New York Heart Association (NYHA) Class III or IV with a history of Class IV within the 6 months before enrolment.
Setting The setting was unclear, but appears to have been primary and secondary care. The economic study was carried out in the USA.
Dates to which data relate The effectiveness data were based on the RALES trial. Randomisation began on 24 March 1995; recruitment was completed on 31 December 1996, with follow-up scheduled to continue through 31 December 1999. However, at the fifth planned interim analysis, the observed effect of spironolactone on the risk of death from all causes exceeded the pre-specified critical z value. Hence, the trial was stopped on 24 August 1998, after a mean follow-up of 24 months, on the recommendation of the data and safety monitoring board. The costs were expressed in 1999 US dollars.
Source of effectiveness data The effectiveness data were derived from a single study
Link between effectiveness and cost data The costing was undertaken on the same patient sample as that used in the effectiveness study. Resource use data were collected prospectively alongside the clinical trial. The costing was carried out retrospectively for the hypothetical study population in the model and appears to have been based on resource use data collected in the trial.
Study sample The study sample comprised 1,663 participants with severe heart failure (as defined already) and an LVEF of no more than 35% who were receiving standard therapy, including an ACE inhibitor and a loop diuretic, with or without digoxin. A total of 822 participants were randomly assigned to receive 25 mg of spironolactone daily and 841 were assigned to receive placebo. No power calculations were reported and the exclusion criteria were not provided. Further details can be obtained from Pitt et al. (see Other Publications of Related Interest).
Study design This was a randomised, double-blinded, placebo-controlled trial. Participants were enrolled in 195 centres in 16 countries. The trial was discontinued early, after a mean follow-up of 24 months, because an interim analysis determined that spironolactone was efficacious. No details of the loss to follow-up were provided in the current paper, but further details can be obtained from Pitt et al. (see Other Publications of Related Interest).
Analysis of effectiveness Details of the group used for the analysis, or the comparability of the groups at baseline, were not provided. The primary health outcome used in the analysis was the quality-adjusted life-years saved (QALYS). The QALYS were derived using quality-adjustment factors for each NYHA Class, which were derived from the responses of 1,601 participants in a study of left ventricular dysfunction to a visual analogue scale, the Ladder of Life questionnaire. This questionnaire was used to rate the current health of persons in the four NYHA classes as a fraction of healthy life, where 0 represented worst possible life and 1 best possible life. Further details are given in Pitt et al. (see Other Publications of Related Interest). The secondary health outcomes were years of survival according to NYHA functional class and years of life saved.
Effectiveness results Spironolactone therapy during the first 35 months' follow-up in RALES increased the QALYS (0.13 discounted QALYS, 95% confidence interval, CI: 0.07 - 0.18).
The average survival time during the first 35 months' follow-up in RALES for patients who received spironolactone was 2.28 years (95% CI: 2.01 - 2.34). This survival time was 0.22 years longer (95% CI: 0.11 - 0.3) than that for participants who received placebo (2.07 years, 95% CI: 1.86 - 2.13).
Spironolactone therapy also led to improved functional status. Of the 0.22-year increase in survival, 0.05 years were spent in NYHA Class I and 0.13 years in NYHA Class II.
Clinical conclusions The authors concluded that treatment with spironolactone therapy increased quality-adjusted survival time and improved functional status according to the NYHA classes.
Modelling A decision analytic model was used. This incorporated data from participants in RALES, as well as cost data from five countries that participated in the study.
Measure of benefits used in the economic analysis The primary health outcome was the QALYS. The secondary health outcomes were years of survival according to NYHA functional class, and years of life saved. The health outcomes were evaluated for the first 35 months of observation in RALES.
A bootstrap procedure was used to assess stochastic uncertainty in the analysis. The results from this were also used to estimate the probability that spironolactone has a cost-effectiveness ratio that falls below $20,000 per QALY.
Direct costs The costs were estimated for eight types of nonfatal and fatal hospitalisation, deaths outside of the hospital, ambulatory care, and spironolactone therapy. The daily costs of hospitalisation (by reason for admission) were derived from Belgium, Brazil, France, Spain and the UK. These five countries enrolled 70% (1165 of 1663) of the participants in the trial. The per-patient total hospital costs in these countries were estimated by multiplying days in hospital (by reason of admission) by the admission-specific daily cost estimates. To estimate hospitalisation costs for other developing countries (Mexico, South Africa and Venezuela), the days in hospital were multiplied by the daily cost estimates derived in Brazil. To estimate the costs for other developed countries (Australia, Canada, Germany, Japan, the Netherlands, New Zealand, Switzerland and the USA), the days in hospital were multiplied by the admission-specific average cost estimates from Belgium, France, Spain and the UK. The resulting costs from this method were not reported in the paper.
The cost of death outside of the hospital was estimated at $1,000. This included the costs of an ambulance, emergency services and emergency department care. A year of ambulatory care was estimated to cost $436. The cost of spironolactone therapy was based on the US average wholesale price ($43.80 per 100 25-mg tablets). A discount rate of 3% was used over the 35-month follow-up period.
Statistical analysis of costs Bootstrapping was used to assess stochastic uncertainty in the analysis. Discounted costs were included in this.
Indirect Costs The indirect costs were not included.
Currency The results were reported in 1999 US dollars ($). The costs were drawn from five countries (Belgium, Brazil, France, Spain, UK), but the conversion rates were not provided.
Sensitivity analysis The effects of five assumptions used in the models were tested in a one-way sensitivity analysis. The data varied were the survival benefit (+/- 33%), based on the 95% CI for the relative risk of death among the participants who received spironolactone, the costs of spironolactone (+/- 50%), the daily costs of hospitalisation (+/- 50%), ambulatory care costs (+/- 50%), and the discount rate (0% and 7%). No justification for these choices was given.
A sensitivity analysis was also undertaken to address the ways in which gynecomastia may affect quality of life. This was done by assuming that, first, gynecomastia had an additional quality-adjustment factor of -0.10 and, second, that patients who developed this side effect experienced it for the duration of the trial.
Best- and worst-case scenarios were reported by combining the values of variables that led to more optimistic results for spironolactone, and also those that led to more pessimistic results. The best case assumed an increased survival benefit (+33%), decreased spironolactone costs (-50%), decreased ambulatory care costs (-50%), and a lower discount rate (0%). The worst case assumed a reduced survival benefit (-33%), increased hospital costs (+50%), increased ambulatory care costs (+50%), a higher discount rate (7%), and a quality-adjustment factor for gynecomastia (-0.10).
For each analysis, the impact of the change in the assumption was reported on the point estimates of discounted costs, discounted years of life saved, discounted QALYS, and the ratio of discounted cost per discounted QALY. The impact of the change on the 95% CI for the ratio of cost per QALY was also reported. The CIs reflected both the uncertainty related to the specific measures being evaluated and stochastic uncertainty.
Estimated benefits used in the economic analysis The study reported estimates of survival time during the first 35 months of follow-up in RALES (overall and by NYHA class). The average survival time for patients who received spironolactone was 2.28 years (95% CI: 2.01 - 2.34). This was 0.22 years (95% CI: 0.11 - 0.30) longer than the 2.07 years (95% CI: 1.86 - 2.13s) of survival time for participants who received a placebo.
Of the 0.22-year increase in survival, 0.05 years was spent in NYHA Class I and 0.13 years was spent in NYHA Class II. There was also a small substitution of survival time between Classes III and IV. The results for discounted survival time were similar to those for undiscounted time (due to the short time horizon and the low discount rate).
Discounted quality-adjusted survival time was substantially less than discounted survival time (because of the relatively large diminutions in the quality-adjustment factors associated with survival time in the four NYHA classes). Patients who received spironolactone experienced 1.27 QALYS (95% CI: 1.12 - 1.30) during the 35 months' follow-up, whereas those receiving placebo experienced 1.14 QALYS (95% CI: 1.02 - 1.18). Thus, spironolactone therapy was associated with a gain of 0.13 discounted QALYS (95% CI: 0.07 - 0.18).
Cost results The reported discounted costs are for participants receiving either spironolactone or placebo for the first 35 months of follow-up in RALES.
The costs were $8,762 (95% CI: 7,612 - 9,860) for participants who received spironolactone and $9,475 (95% CI: 8,197 - 10,434) for those who received placebo. The apparent cost-saving of $713 (95% CI: $2,123 in savings to $783 cost) associated with spironolactone therapy was not statistically significant, (one-tailed, p=0.2).
These total costs comprised several elements. The discounted costs for nonfatal hospitalisation were $6,096 with spironolactone and $7,109 with placebo, a difference of -$1,012 (95% CI: -2,305 - 368) (the differences in these figures are due to rounding). For fatal hospitalisation, the discounted costs were $1,113 with spironolactone and $1,241 with placebo, a difference of -$128 (95% CI: -597 - 377). For deaths outside of the hospital, the discounted costs were $178 with spironolactone and $265 with placebo, a difference of -$86 (95% CI: -131 - -30). The discounted costs of ambulatory care were $948 with spironolactone and $861 with placebo, a difference of $87 (95% CI: 44 - 119). The discounted cost of spironolactone was $427, compared with $0 for placebo; a difference of $427 (95% CI: 377 - 441).
Synthesis of costs and benefits Spironolactone therapy either dominated placebo or had a ratio of cost per QALYS that was unlikely to exceed $20,300. These results were robust in both the one-way and multi-way sensitivity analyses.
One thousand bootstrap replicates of incremental costs and QALYS for those receiving spironolactone therapy and placebo during the first 35 months of RALES were determined. The point estimate indicated that spironolactone therapy dominated placebo (i.e. it saved $713 and lengthened discounted quality-adjusted survival time by 0.13 years). A total of 80.4% of the bootstrap replicates fell in the dominant south-east quadrant of the cost-effectiveness plane (i.e. where spironolactone therapy decreased costs and increased discounted QALYS). A further 19.6% of the replicates fell in the north-east quadrant (i.e. where spironolactone therapy increased costs and discounted QALYS).
The 95% CI for the comparison of costs and QALYS indicated that spironolactone may have dominated placebo (lower limit) or may have had a cost per QALY ratio as high as $6,650 (upper limit). The probability approached 100% that the cost per QALY ratio was less than $20,000.
Authors' conclusions Even after the implementation of current clinical guidelines, the addition of spironolactone therapy provides an opportunity to further reduce the large clinical and economic burden of patients with heart failure. Spironolactone therapy during the first 35 months of follow-up in RALES increased quality-adjusted survival time without increasing the costs. Spironolactone therapy either dominated placebo or had a ratio of cost per quality-adjusted life-year (QALY) that was unlikely to exceed $20,300. These results were robust in both the one-way and multi-way sensitivity analyses.
CRD COMMENTARY - Selection of comparators The comparator was standard therapy and placebo. The paper stated that, although current evidence-based clinical practice guidelines for the treatment of heart failure recommend an ACE inhibitor and diuretic, recent studies of beta-blockers and spironolactone have shown that these drugs may further reduce heart failure-related morbidity and mortality. However, the choice of placebo over beta-blockers was not justified. You should decide if this represents current practice in your own setting.
Validity of estimate of measure of effectiveness The model was populated using data from a randomised controlled trial (RCT), which was appropriate for the hypothesis. The hypothetical study population in the model was based on the RCT and appears to have been representative. Details of the patient groups and detailed analysis of the RCT results were not given in this paper, but are reported elsewhere (Pitt et al., see Other Publications of Related Interest).
Validity of estimate of measure of benefit The estimate of QALYS was modelled and the method used to derive the QALYs (visual analogue scale) was appropriate. However, these were derived by heart failure class and were not based on preference data collected directly in the trial. The authors acknowledged this was a limitation and carried out extensive sensitivity analyses.
Validity of estimate of costs The cost perspective was stated to have been societal. However, the costs included were those to the health care system and not those to society. Details of the resources and costs were not given separately for all resources, thus it is not possible to comment on them.
Other issues The authors stated that the study had a number of limitations such as missing data and the use of the average variable costs per day for hospitalisation to estimate the hospitalisation costs. Also, QALY results driven by heart failure class, and not by preference data collected directly in the trial, were used. However, the authors carried out extensive sensitivity analyses to account for these issues.
Implications of the study The authors stated that the addition of spironolactone therapy provides an opportunity to further reduce the large clinical and economic burden of patients with heart failure. However, they stated that a further study may be required to confirm these findings and evaluate the cost-effectiveness of spironolactone in practice.
Source of funding Funded by a sponsored research agreement.
Bibliographic details Glick H A, Orzol S M, Tooley J F, Remme W J, Sasayama S, Pitt B. Economic evaluation of the Randomized Aldactone Evaluation Study (RALES): treatment of patients with severe heart failure. Cardiovascular Drugs and Therapy 2002; 16(1): 53-59 Other publications of related interest Pitt B, Zannad F, Remme WJ, et al. The effect of spironolactone on morbidity and mortality in patients with severe heart failure. New England Journal of Medicine 1999;341:709-17.
Cook J, Glick H, Kinosian B, Heyse J. An assessment of "Ladder of Life" score as a value measure for health statue in congestive heart failure (abstract). Medical Decision Making 1993;13:386.
Indexing Status Subject indexing assigned by NLM MeSH Ambulatory Care /statistics & Cost-Benefit Analysis; Double-Blind Method; Economics, Pharmaceutical; Heart Failure /drug therapy /economics /mortality; Hospitalization /economics; Humans; Mineralocorticoid Receptor Antagonists /economics /therapeutic use; Quality-Adjusted Life Years; Randomized Controlled Trials as Topic; Spironolactone /economics /therapeutic use; numerical data AccessionNumber 22002000917 Date bibliographic record published 31/07/2005 Date abstract record published 31/07/2005 |
|
|
|