| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ann Thorac Surg 2002;73:1403-1409
© 2002 The Society of Thoracic Surgeons
a London School of Hygiene and Tropical Medicine, London, United Kingdom
b Guy’s and St. Thomas’ Hospitals, London, United Kingdom
Accepted for publication December 20, 2001.
* Address reprint requests to Mr Jacklin, London School of Hygiene and Tropical Medicine, Health Services Research Unit, Keppel St, London WC1E 7HT, UK
e-mail: paul.jacklin{at}lshtm.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionMaterial and methodsResultsCommentReferences |
|---|
Methods. An economic model was developed to compare the cost-effectiveness of three management strategies: (1) coronary artery bypass grafting for all patients; (2) using positron emission tomography to select candidates for coronary artery bypass grafting, those without hibernation remaining on medical therapy; and (3) medical therapy for all patients. The model used data from our hospital and the published literature. A sensitivity analysis was also undertaken.
Results. Positron emission tomography was cost-effective in selecting patients for operation. In a hypothetical population of 1,000 patients, using positron emission tomography saved marginally more life-years and cost approximately £3 million less. Using positron emission tomography before coronary artery bypass grafting instead of all patients receiving medical treatment saved lives but was more expensive. The incremental cost per life-year saved was £77,000. The sensitivity analysis showed that the prevalence of hibernation and the survival rate of patients refused revascularization on the basis of the positron emission tomography scan were the areas most likely to influence cost-effectiveness.
Conclusions. Positron emission tomography may be cost-effective to select patients with poor left ventricular function for coronary artery bypass grafting.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentReferences |
|---|
Patients with ischemic heart disease and severe left ventricular (LV) dysfunction have a poor prognosis [1]. Prognosis can be improved, if myocardium with impaired contraction but significant amounts of viable tissue (hibernating myocardium) can be differentiated from necrosis and blood flow improved by revascularization [2–4]. Performing an operation in patients with LV dysfunction, however, has increased risks [5], and in the absence of hibernating myocardium the patient is unlikely to derive significant benefit [4, 6]. Patients with impaired ventricular function are also often more expensive to treat because of a greater utilization of intensive care resources.
Positron emission tomography (PET) is an imaging method using short-lived isotopes of naturally occurring elements, which decay by positron emission, to examine physiologic variables such as perfusion and glucose metabolism. Positron emission tomography can be used to establish the metabolic integrity of dysfunctional myocardium and is regarded the most accurate technique to detect hibernating myocardium [7]. If PET is used to select patients with severe LV dysfunction for operation, our hypothesis is that this would allow a more cost-effective use of resources.
One of the difficulties in performing an economic evaluation of PET is lack of data. The prevalence of hibernation in patients with LV dysfunction referred for operation is not established, and although uncontrolled studies suggest there are significant short-term to medium-term benefits, the role of revascularization versus medical treatment has not been examined in a long-term randomized prospective trial [8]. The variation in the cost of the same technology or treatment across different health-care systems and different populations poses further problems.
The aim of this research therefore was to develop a model using available data to test the hypothesis that PET would be cost-effective in selecting patients with poor LV function for revascularization. The design of the model would allow changes of input data to reflect differences in key variables such as cost, prevalence of disease, and survival of patients over time and locality, and to take account of new research, different health-care systems, and different characteristics in referral populations.
|
Material and methods |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentReferences |
|---|
A decision analysis model was developed using Microsoft Excel (Microsoft Corp, Redmond, WA), in which mathematical formulae were entered into a spreadsheet, which allowed the user to estimate the potential cost of treating a set number of patients according to a particular management strategy and what the outcome will be. The input data, eg, the number of patients treated, the cost of procedure, and the survival of patients treated with CABG, can be altered by the user to reflect his or her own local environment. The outcome measure in the model was the number of life-years generated. If 1,000 patients were treated and there was a 100% survival rate, 1,000 life-years would be generated by running the model. We made no attempt to measure quality of life. We assumed all patients referred for consideration of operation had technically operable disease and that appropriate risk assessment was carried out.
We estimated the cost and outcome of treating 1,000 hypothetical patients using the model for three different strategies (Fig 1):
|
More details about how we chose the values for the default data to enter into the model are given below. It should be remembered, though, that the input variables, eg, prevalence of hibernation, can be changed by the user to reflect local practice.
Data from hospital sources
The overall cost from admission to discharge of a patient undergoing CABG of £5,785 was taken from the prices charged by Guy’s and St. Thomas’ Hospital Trust to other parts of the U.K. National Health Service. The full price had been built up from the cost of staffing, consumables, equipment, and maintenance for the Cardiothoracic Directorate according to activity in the previous year. The total cost included separate costing for the use of intensive care facilities and ward facilities. These costings applied to all patients treated that year. We were specifically interested in patients with poor LV function, defined for these purposes as patients with ejection fractions of less than 30%. As these patients were likely to have different uses of resources such as the intensive care unit (ICU), which might affect their treatment costs, we estimated the cost of treating these patients surgically as follows.
We examined databases in Cardiothoracic Surgery, the ICU, and operating theaters to measure the resource use of all patients who had CABG during 1 financial year (Fig 2). The routine procedure in our hospital is that patients are admitted to the ward for preoperative assessment. After operation a patient is sent to OIR (overnight intensive recovery) or to the ICU. The ICU is used when it is anticipated that the patient will require more than 24 hours of intensive care, although bed capacity constraints may be a determining factor. The patient then returns to the ward before discharge. We measured how long the patients spent in each clinical area and what proportion were offered intensive care in the ICU and what proportion in the OIR unit. This enabled us to derive the average cost of treating a patient per day on the ICU, OIR, and the ward.
|
 |
 |
The cost of medical therapy was based on prices given to us by the hospital pharmacy for drug treatment within the National Health Service [14, 15] and did not include medical consultations or hospital admissions. A consultant cardiologist devised a typical recipe based on best clinical practice. The regimen used was simvastatin 20 mg once daily, aspirin 75 mg once daily, atenolol 50 mg once daily, furosemide 40 mg once daily, and lisinopril 30 mg once daily. It was assumed that patients continued on the same drugs after operation, which is probably current practice [3].
Data derived from literature review
The PET scan measures myocardial blood flow with ammonia N 13 and glucose metabolism with fluorodeoxyglucose F 18 (FDG). The pattern commonly associated with hibernating myocardium with PET is the preservation of FDG uptake in areas of regional reduction in blood flow. This pattern of uptake is generally referred to as mismatch. Where there is concordant reduction in blood flow and FDG uptake, the pattern is referred to as match and represents necrosis or scar. The annual survival of patients treated with medical therapy or revascularization varies according to the presence or absence of PET mismatch [4]. Retrospective analysis of 93 patients undergoing CABG revealed annual survival rates of 88% for patients with mismatch who underwent operation but only 50% for those who were treated with medical therapy. There was no significant difference in survival for patients with matched defects treated with operation (94%) compared with those who were treated medically (92%). All deaths in the surgically treated patients occurred in the perioperative period. The number of life-years generated in the model were based on our perioperative mortality rate of 9% for the patients who underwent CABG as it was comparable to that given in the study by DiCarli and coworkers [4]. We also assumed that that all surgical deaths occurred in the perioperative period. The number of life-years generated in the model for the medical patients was taken from the same study [4]. The sensitivity and specificity of PET to detect a mismatch pattern predictive of hibernation [10] and the nondiagnostic rate of the PET scan [11] were taken from the literature.
The prevalence of significant hibernating myocardium has been estimated at approximately 50%, and this was used as the default value [9]. The default values of the model variables are listed in Table 1.
|
|
Results |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentReferences |
|---|
|
|
These are thresholds for dominance rather than cost-effectiveness. They mark the point at which preoperative PET yields greater benefit in life-years and is less costly than routine revascularization. When this threshold is not reached, preoperative PET either produces less benefit or is more expensive. For example, at a PET sensitivity of 77.4%, the CABG (without PET) option generates more benefit but at a cost of almost £117 million per additional life-year.
|
Comment |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentReferences |
|---|
Our hypothesis was that the use of a PET scan in patients with severe LV dysfunction could improve the efficiency of cardiovascular care by limiting operations to those most likely to benefit. Some patients whose symptoms are predominantly those of angina rather than congestive heart failure may still warrant revascularization on symptomatic grounds. In the absence of viability, however, their prognosis is unlikely to be improved [2, 4] It could be argued therefore that a PET strategy might be more useful in patients with predominant symptoms of congestive heart failure.
We used a decision analysis model to collate the available data for economic evaluation. This is a relatively simple way to identify which variables are most important in evaluating cost-effectiveness and where further research should be focused when there is uncertainty about some of the data. It also allows flexibility to take account of new research and different health-care systems. It also has disadvantages; bias is the biggest potential problem [19], but we have tried to mitigate bias by being explicit about the assumptions and the data used.
Based on the default values in the model, the use of preoperative PET to select patients for revascularization is cost-effective. It is a cheaper strategy than performing CABG without PET because patients with myocardial necrosis (PET match) who cannot benefit from revascularization are refused CABG. The money saved by performing fewer operations outweighs the cost of introducing PET scans. Although the PET option was a lower cost strategy, it had minimal impact on the 1-year survival rates. Patients with myocardial necrosis, who avoid the risks of operation without potential for improvement in survival gain from having PET. Some patients with myocardial hibernation lose, however, when PET is introduced because the PET scan has a false-negative rate and will miss some patients with hibernation. These patients are refused an operation and lose the potential benefit. The net gains in survival made by patients with necrosis are balanced by the net losses experienced by patients with hibernation when the default values are used.
The sensitivity analysis confirmed that there was uncertainty surrounding the data available. The areas of uncertainty that might alter the cost-effectiveness conclusions were the prevalence of hibernation (PET mismatch) in the referral population and the survival of patients with myocardial necrosis (PET match) on medical therapy.
There is little information in the literature on the prevalence of hibernation. In the study of 93 patients with coronary artery disease and LV dysfunction, from which we have taken the survival data for the model, 43 patients (46%) had mismatch defects [4]. In another U.K. study the prevalence of significant mismatch in consecutive patients undergoing coronary angiography with ejection fractions of less than 30% imaged with PET was 52% [9]. The sensitivity analysis indicated that the PET versus no PET strategy remained dominant if the prevalence of hibernation did not exceed 53%. Another point to consider is how extensive the area of hibernation needs to be in an individual to affect their prognosis. The mass of hibernating tissue is directly related to improvement in function postoperatively. Approximately 20% to 50% of the myocardium probably needs to be hibernating to improve cardiac function [3, 20, 21] and may depend on the patient’s degree of LV dysfunction. Improvement in survival is probably also influenced by the mass of hibernating tissue present [22], but the minimum amount required to affect survival is likely to be lower than that required for functional improvement [4, 23, 24]. The study from which the 1-year survival figures were taken for the model showed a clear difference in survival for those with mismatch (>5% of the total of the left ventricle) who were treated medically compared with those treated surgically.
There are published survival data in only small numbers of patients with PET match and mismatch, all of which is retrospective [8]. One-year survival rates were used as the measure of effectiveness because this is what was available. If longer follow-up were used, hospital mortality rates would be of greater importance to the cost-effectiveness conclusions because mortality arising from operation would carry a greater potential opportunity cost (in terms of life-years lost). Studies suggest that using PET or thallium in the preoperative workup of patients to detect hibernating myocardium may result in lower perioperative mortality and complication rates [25, 26]. If this were true, it would add weight to the conclusion that using PET in the preoperative selection process was a cost-effective strategy. We did not attempt to measure quality of life, as the data available are again limited but do suggest potential advantages of treating patients with mismatch with revascularization [3].
The accuracy of PET might appear to affect cost-effectiveness conclusions. The threshold analysis suggested that the sensitivity of PET was more important than specificity. However, even if PET sensitivity was as low as 50%, the incremental cost of the CABG (without PET) option would be £181,101 per life-year saved (Table 3), which would probably not be considered cost-effective [27].
Medical therapy was the lowest cost strategy using the default values, but it was less effective than the two other strategies in most situations. Net gains in survival are only obtained when the prevalence of mismatch is less than 7.7% or if 1-year survival of mismatch patients on medical therapy is greater than 80%. The cost of medical therapy is underestimated, however, because we did not include the cost of medical consultations or admissions to hospital, as these data were not available.
The model examined only the use of PET in the preoperative workup of patients. We chose to examine PET as it is generally regarded as the most accurate imaging technique for hibernation [7]. It is, however, perceived as being the most costly, and the availability of PET in the United Kingdom is very limited. Other techniques can be used. Stress echocardiography detects contractile reserve using dobutamine. Contractile reserve influences survival in a similar way to the presence of mismatch on PET scanning [6]. Tracers such as thallium Tl 201 or technetium Tc 99m sestamibi using single-photon emission tomography rely on active cellular transport mechanisms for retention in viable tissue and can be used. All the techniques have similar sensitivities, but the single-photon emission tomography techniques have lower specificity [7, 10, 28 .] Echocardiography has been reported to have similar specificity to PET [28], but recent direct comparison of the two techniques in patients with poor LV function and a high number of akinetic segments has suggested that echocardiography may have a high false-negative rate [20]. The authors of this study suggested that it might be more cost-effective to perform echocardiography, which is a less costly test, first followed by PET for those patients with no evidence of viability on echocardiography. Others have suggested that the management and outcome of patients who are scanned using single-photon emission tomography rather than PET is unaffected despite the lower specificity [7]. We were unable to accurately cost these alternative options with the data available to us at our hospital, but the model could be adapted to do so. It might be that the assessment of viability using echocardiography or single-photon emission tomography or a combination of these techniques with PET might be more cost-effective.
It should be remembered that revascularization (with or without viability assessment) is only really cost-effective if the additional benefit gained in terms of survival is considered worth the additional resources that will be diverted from elsewhere within the health-care system. This economic analysis adds weight to the concept that a prospective trial of medical therapy versus surgical treatment is necessary [8, 18], although to randomize patients with hibernating myocardium to receive medical treatment is becoming more difficult to justify on ethical grounds.
Using costs obtained from a sample of patients undergoing CABG and PET in a large cardiothoracic unit, this study suggests that PET may be cost-effective in the selection of patients with poor LV function referred for CABG. The model we used appeared to be relatively robust but highlighted important areas of uncertainty. These included the percentage of patients with hibernating myocardium in the referral population and the long-term survival of patients without hibernation who do not undergo operation. A further issue is how the presence or absence of hibernating myocardium influences perioperative complications and mortality. These factors may have a significant bearing on the cost-effectiveness of PET and other methods of viability assessment and prevent definitive conclusions from being drawn. Nevertheless our model has indicated what the most profitable areas of research are likely to be.
|
References |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentReferences |
|---|
J., Frélich M., et al. Prognostic value of the amount of dysfunctional but viable myocardium in revascularized patients with coronary artery disease and left ventricular dysfunction. J Am Coll Cardiol 1998;32:912-920.This article has been cited by other articles:
|
|
 |
|
  M N Maisey Overview of clinical PET Br. J. Radiol., November 1, 2002; 75(90009): S1 - 5. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
