|Low doses vs. high doses of the angiotensin converting-enzyme inhibitor lisinopril in chronic heart failure: a cost-effectiveness analysis based on the Assessment of Treatment with Lisinopril and Survival (ATLAS) study
|Sculpher M J, Poole L, Cleland J, Drummond M, Armstrong P W, Horowitz J D, Massie B M, Poole-Wilson P A, Ryden L
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.
The use of lisinopril, an angiotensin-converting enzyme (ACE) inhibitor, aimed at reducing the risk of death and hospitalisation in chronic heart failure patients.
Economic study type
The study population comprised patients with New York Heart Association (NYHA) Class II to IV heart failure, and a left ventricular ejection fraction equal to or less than 30%. Patients were excluded if they had acute myocardial infarction, unstable angina, symptomatic ventricular tachycardia or unstable congestive heart failure. They were also excluded if they had undergone a revascularisation procedure in the preceding 2 months, or used various negatively or positively isotropic drugs.
The setting was the community. The economic study was carried out in the UK using multinational trial data.
Dates to which data relate
The period during which the effectiveness and resource use data were collected was not reported. The prices used were for 1997 to 1998.
Source of effectiveness data
The effectiveness evidence was derived from a single study, the results of which were published elsewhere (see Other Publications of Related Interest).
Link between effectiveness and cost data
The costing was undertaken prospectively on the same patient sample as that used in the effectiveness analysis.
Power calculations to determine the sample size, and the methods used to select the sample, were not reported. An overall sample of 3,164 patients was included in the analysis. There were 1,568 patients with a mean age of 63.6 (+/- 10.5) years in the high-dose group. The proportion of males was 79.8%, and 63.7% of the patients had coronary artery disease. There were 1,596 patients with a mean age of 63.6 (+/- 10.3) years in the low-dose group. The proportion of males was 79.3% male, and 64.9% of the patients had coronary artery disease.
The study was a randomised, double-blind clinical trial, carried out in 19 countries. The USA recruited 38% of the patients. The methods of randomisation and blinding were not reported. After an initial open-label period, which was intended to establish the tolerance of a daily dose of lisinopril of 12.5 to 15 mg, the patients were randomly allocated to the study groups. The patients were followed for 4 years (median follow-up: 46 months; range: 39 - 58). The loss to follow-up was not explicitly reported.
Analysis of effectiveness
The basis for the analysis of the clinical study was unclear. The primary health outcome used in the economic analyses was mean survival, although the risk of death or hospitalisation, the occurrence of adverse events, the proportion of patients stopping medication due to adverse events, and heart failure symptoms were presented. The study groups were comparable in terms of their demographics and medical characteristics.
After 4 years, there was an 8% lower risk of death in the high-dose group, (p=0.128), compared with the low-dose group. There was also a 12% lower risk of death or hospitalisation in the high-dose group, (p=0.002).
The patients in the high-dose group experienced more adverse events, which were easily managed by adjusting the therapy.
The proportion of patients stopping medication due to adverse events was 17% in the high-dose group and 18% in the low-dose group.
Heart failure symptoms were similar between the groups.
High-dose lisinopril appears to have been more effective than low-dose lisinopril in reducing the risk of death in patients with chronic heart failure.
Measure of benefits used in the economic analysis
The benefit measure used in the economic analysis was the life-years gained with lisinopril. These were derived from the clinical trial using Kaplan-Meier survival curves. A 2% discount rate was used.
A 6% discount rate was used since the costs were incurred over 4 years. The unit costs and the quantities of resources were reported. The cost/resource boundary adopted was that of the health service. The cost items included in the analysis were related to hospital stay and drug use. Hospital stay was computed using the average cost per inpatient day in English hospitals, which was derived from their financial returns in 1997 to 1998. These included the costs of diagnostic and therapeutic procedures. Drug use was derived on the basis of British National Formulary prices. It was assumed that patients in the high-dose group needed three extra visits to a general practitioner. Data on resource use were obtained from actual data derived from the clinical trial. The period during which the resources were collected was not reported. The prices used were for 1997 to 1998.
Statistical analysis of costs
Statistical analyses of the costs were carried out. Bootstrapping methods were used to determine the 95% confidence interval (CI) for the mean differences in costs between the study groups.
The indirect costs were not included.
Sensitivity analyses were carried out to take account of the uncertainty around the use of a specific discount rate. One- and two-way analyses were performed to test for the use of a 6% discount rate for both the costs and benefits, and for no discounting, respectively. In addition, a stochastic approach (bootstrapping method) was carried out to assess the probability that the high-dose therapy was more cost-effective than the low-dose therapy for given values of willingness to pay for an additional life-year (acceptability curve).
Estimated benefits used in the economic analysis
The estimated mean life-years per patient were 2.980 in the high-dose group and 2.896 in the low-dose group. The difference was 0.085 years (95% CI: -0.0074 - 0.1706).
The acquisition costs of lisinopril were higher in the high-dose group, but these were offset by lower hospital inpatient days.
The mean cost was 6,867 in the high-dose group and 7,264 in the low-dose group.
The mean cost difference was -397 in favour of high-dose lisinopril (95% CI of high-dose minus low-dose: -1,263 - 436).
Synthesis of costs and benefits
An incremental cost-effectiveness analysis was carried out to combine the costs and benefits of the two interventions. High-dose lisinopril dominated low-dose lisinopril, due to higher effectiveness and lower costs. Following the stochastic approach, the acceptability curve showed that high-dose lisinopril was more cost-effective than low-dose lisinopril with an 82% probability, if the decision-maker was unwilling to pay anything extra for an additional life-year. Also, there was a 92% probability of being cost-effective, if the decision-maker was willing to pay at least 3,600 per life-year gained. The acceptability curve and the overall conclusions of the study were not affected by variations in the discount rate.
High-dose lisinopril for the treatment of chronic heart failure has a high probability of being more cost-effective than low-dose therapy.
CRD COMMENTARY - Selection of comparators
The rationale for the choice of the comparator was clear. The authors wanted to determine the cost-effectiveness of high-dose in relation to low-dose therapy. In terms of whether a high dose is actually cost-effective, one would require information on the incremental cost-effectiveness ratio (ICER) of low-dose versus the next best technology. The next best technology might be no treatment since it was unclear whether low-dose lisinopril was established practice. You should assess which technologies would be appropriate comparators in your own setting.
Validity of estimate of measure of effectiveness
The analysis of the effectiveness was carried out using data derived from an international, randomised, double-blind clinical trial, which was appropriate for the study question. Thus, the internal validity of the analysis was relatively high. In addition, the time horizon of the study appears to have been sufficient to assess the medium-term effects of the therapies. The study sample was representative of the study population and the patient groups were comparable at baseline. However, clarity could have been improved in terms of reporting losses to follow-up, and thus, whether the basis of the analysis was treatment completers only or intention to treat.
Validity of estimate of measure of benefit
Life-years gained were selected as a benefit measure. These were derived from the effectiveness analysis using Kaplan-Meyer survival curves. It would have been interesting to have collected data on quality of life, and to have assessed the impact of the therapies using quality-adjusted life-years.
Validity of estimate of costs
All the categories of costs relevant to the perspective adopted appear to have been included in the analysis. The unit costs and the quantities of resources were reported separately. The estimated costs were treated stochastically using a bootstrapping method. The price year was reported and sensitivity analyses were carried out on the discount rate. The costs appear to be somewhat specific to the study setting. External validity could have been improved by more extensive sensitivity analyses.
The authors performed a bootstrapping exercise to express the uncertainty in the ICERs, leading to a cost-effectiveness acceptability curve. This permitted the expression of probability that the ICER between high- and low-dose lisinopril would be cost-effective, using appropriate decision rules. This avoids the problem that ICERs reflecting lower cost and lower effectiveness are interpreted identically to those reflecting higher costs and higher effectiveness due to them having the same sign. The more incidences (bootstrap estimates) of the latter that occur below the threshold, the better. However, the more incidences of the former that occur above the threshold, the worse the technology. This may not be such an important problem here since the plot of ICER estimates shows very few with lower costs and effectiveness. Indeed, 82% of the estimates showed dominance (higher effectiveness and lower cost).
The authors did not compare their findings with those from other studies. The issue of the generalisability of the study results to other settings was addressed, although sensitivity analyses were only carried out varying the discount rate. Consequently, the external validity of the analysis may have been reduced. The authors acknowledged that important variations could exist in terms of practice patterns and use of resources among the countries. However, this consideration is unlikely to affect the study results, due to the high probability that high-dose therapy is cost-effective. The effectiveness results were presented selectively, although they should be available in the other publication (see Other Publications of Related Interest). The authors' conclusions were in keeping with the scope of the study.
Implications of the study
The authors recommend the use of high-dose lisinopril in chronic heart failure patients. They justified this on the basis of effectiveness and cost-effectiveness, drawing attention to the high probability of dominance and the high probability of low ICER. The only caveat in this analysis was that the importance of the ICER between high and low dose depends on the size of the ICER between low dose and the next best technology (e.g. no treatment). For example, if the ICER was found to be very high (due to a very high increase in cost to gain little effectiveness), a low dose would not be the appropriate comparator for the high dose; it would be no treatment. This would make the probability of cost-effectiveness of high-dose lisinopril much lower.
Source of funding
Supported by AstraZeneca.
Sculpher M J, Poole L, Cleland J, Drummond M, Armstrong P W, Horowitz J D, Massie B M, Poole-Wilson P A, Ryden L. Low doses vs. high doses of the angiotensin converting-enzyme inhibitor lisinopril in chronic heart failure: a cost-effectiveness analysis based on the Assessment of Treatment with Lisinopril and Survival (ATLAS) study. European Journal of Heart Failure 2000; 2(4): 447-454
Other publications of related interest
Packer M, Poole-Wilson PA, Armstrong PW, et al. Comparative effects of low doses and high doses of the angiotensin converting-enzyme inhibitor, lisinopril, on morbidity and mortality in chronic heart failure. Circulation 1999;100:2312-8.
Subject indexing assigned by NLM
Angiotensin-Converting Enzyme Inhibitors /administration & Cost-Benefit Analysis; Double-Blind Method; Drug Costs /statistics & Female; Great Britain; Health Resources /utilization; Heart Failure /drug therapy /economics /mortality; Hospital Costs /statistics & Humans; Lisinopril /administration & Male; dosage /economics /therapeutic use; dosage /economics /therapeutic use; numerical data; numerical data
Date bibliographic record published
Date abstract record published