| |
Liver transplantation is effective, but is it cost-effective?
|
| |
| |
Editorial
Liver Transplanation, December 2003, Volume 9, Number 12
Kiran Bambha
W. Ray Kim
From the Division of Gastroenterology and Hepatology, Mayo Clinic and Foundation, Rochester, MN.
In 1983, The National Institutes of Health released a consensus conference statement that endorsed liver transplantation as an effective nonexperimental treatment for end-stage liver disease. Now, two decades later, liver transplantation has become a widely accepted therapy for improving survival and quality of life for patients with either acute or chronic end-stage liver disease. The outcome of liver transplantation has improved steadily as donor and recipient selection criteria have been refined, new surgical techniques have been developed, and newer immunosuppressive regimens have been introduced. Current 1- and 5-year post-liver transplantation survival rates in the United States are approximately 85% and 75%, respectively.
However, for logistic and ethical reasons, there have been no randomized controlled trials directly comparing liver transplantation with conservative therapy. The success of liver transplantation has been measured by comparing observed posttransplantation survival with the predicted natural history by using validated disease-specific prognostic models. Primary biliary cirrhosis (PBC), primary sclerosing cholangitis (PSC), and alcoholic liver disease (ALD) are three chronic hepatic conditions for which there exist well-validated natural history models. Studies comparing posttransplantation survival with predicted survival for these three liver diseases have shown 1- and 5-year survival rates of 83% and 78% for PBC, 94% and 86% for PSC, and 84% and 72% for ALD, respectively. Data from these and similar studies showed that liver transplantation is an effective therapy for prolonging survival in patients with end-stage liver disease.
Although the effectiveness of liver transplantation in improving survival of patients with end-stage liver disease is evidenced in the medical literature, much fewer data are available on the cost-effectiveness of the procedure. The underlying health economic issue that analyses of cost-effectiveness are designed to help address is, "What allocation of resources will achieve the greatest gain in health attainable without exceeding the allotted budget?" One approach to this issue is to calculate the cost-effectiveness ratio, which, with regard to liver transplantation, is costs associated with liver transplantation (expressed in monetary units) divided by the effectiveness of liver transplantation (measured, for example, in years of survival). However, for results of a cost-effectiveness analysis to be truly informative, a medical intervention should always be evaluated in light of the alternative interventions with which it is competing. Therefore, the incremental cost-effectiveness ratio may be calculated as the difference in costs between the two competing interventions divided by the difference in effectiveness between the two.
As shown by Longworth et al in their report, "Midterm cost-effectiveness of the liver transplantation program of England and Wales for three disease groups," in this issue of Liver Transplantation, an economic comparison is made between performing liver transplantation for patients with end-stage liver disease and taking a conservative medical approach toward treatment of these patients. The investigators use their directly observed data for costs and outcomes for a group of patients with PBC, PSC, and ALD who underwent liver transplantation. Because there have been no randomized controlled trials directly comparing liver transplantation with conservative management for end-stage liver disease, costs and outcomes data for the comparator group (i.e., absence of liver transplantation) were estimated by using validated disease-specific models for PBC, PSC, and ALD that predict the natural history of disease in the absence of intervention. Thus, for each patient included in this cost-effectiveness analysis of liver transplantation, data reflecting a patient's observed transplantation costs and outcomes were included, along with estimations for each patient of what their costs and outcomes would have been in the absence of liver transplantation (shadow costs and shadow outcomes).
There are many factors that must be included in a cost-effectiveness analysis. Accurate assessment of costs of the medical intervention being considered is critical. Costs are derived from multiple sources and may be direct (i.e., hospital facilities, personnel, laboratory, medication, imaging, and procedural costs) or indirect (costs associated with lost wages or lost productivity). Although the concept of costs needed to produce these health services is relatively straightforward, determining costs accurately can be difficult in real life. For example, many economic analyses estimate costs from published charges or payments for hospital care, physician services, laboratory testing, medications, and other services, which often are not representative of true economic costs.16 In that regard, Longworth et al are commended for having undertaken a resource-based estimation of costs associated with liver transplantation, including inpatient and outpatient care, hospital overhead costs, medications, testing, length of transplantation operation, and staff costs. A similar method has been used by Showstack et al17 in the United States. In the Longworth study, information on transplant center resource utilization was collected prospectively for each patient beginning at the time of initial assessment for liver transplantation. Additionally, despite the paucity of published data, the investigators also attempted to include costs associated with the process of organ procurement in their analysis.
With regard to indirect costs, methodologic issues arise in trying to estimate costs associated with loss of productivity because of illness or premature death. Although measurement of these costs is difficult and may be inaccurate, their inclusion in a cost-effectiveness analysis generally is recommended. In the report by Longworth et al, indirect costs associated with liver transplantation and conservative management were not explicitly considered. This was caused in part by the perspective of this analysis, namely, that of the payer (the British National Health Service). However, exclusion of indirect costs has the net effect of underestimating the cost-effectiveness of liver transplantation because indirect costs are likely to be greater in the conservative management strategy compared with liver transplantation.
The other critical component of a cost-effectiveness analysis is measurement of the effectiveness of the medical interventions under consideration. A variety of measurements of effectiveness may be used in a cost-effectiveness analysis. These include years of life gained by the prevention or delay of death because of the medical intervention, years of life lost because of side effects or complications of the intervention, or quality-adjusted life-years (QALYs) gained by the prevention or delay of morbid events because of the intervention. QALYs consider the quality, as well as quantity, of life affected by the medical intervention and currently are the preferred unit of effectiveness in cost-effectiveness analyses. Although in theory, use of quality-adjusted measurements is desirable, there are difficulties and limitations associated with their practical application, including how and in what population quality of life should be measured.
Longworth et al expressed outcomes in terms of QALYs for patients who underwent liver transplantation and for their derived comparator group. Health-related quality of life was assessed by using the well-known validated EuroQol EQ-5D questionnaire, which was sent to patients at multiple times pretransplantation and posttransplantation. Health-related quality of life scores were valued relative to scores for the UK general population. However, it is worthwhile to note that only 48% to 76% of patients had usable EuroQol EQ-5D values at the main times in this study, and missing scores for some patients were input. Although inputting missing values is an accepted means for dealing with missing data, this process introduces additional uncertainty into the analysis.
In addition to the two main components of a cost-effectiveness analysis, namely, costs and effectiveness of the medical interventions, several other factors need to be addressed in such an analysis. Three of these factors, discounting, patient population, and time horizon, are discussed here because they are particularly relevant to the report by Longworth et al.
Discounting must be factored into any cost-effectiveness analysis, and it is a basic principle of economics that the value of a dollar today is worth more than the value of a dollar in the future. This is a reflection of our time preference for money, meaning that the majority of people would prefer the same amount of money today rather than some time in the future. Therefore, costs occurring in the future should be valued less (or discounted). Discounting adjusts future costs and expresses them in terms of their present value. This gives greater weight to costs the earlier they occur. In a cost-effectiveness analysis, because outcomes of the medical intervention under consideration are being valued relative to costs and costs are being discounted, outcomes must also be discounted.
It is interesting to note that Longworth et al used different discounting rates for costs and outcomes (6% and 1.5%, respectively). The investigators report that these discount rates are in accordance with National Health Service guidelines 2001. Currently, for cost-effectiveness analyses conducted in the United States, it generally is agreed that discount rates used for both costs and outcomes should be the same to avoid the problem of artificially inflating or deflating the apparent cost-effectiveness of a medical intervention.18 In the investigation by Longworth et al, the greater discount rate applied to costs compared with effectiveness will have the effect of making liver transplantation appear more cost-effective than it may be. Overall, the short time horizon (27 months) makes the bias caused by differential discounting between costs and effectiveness rather small.
The two other factors that bear mentioning because they relate to this cost-effectiveness analysis of liver transplantation conducted by Longworth et al are issues surrounding the patient population and time horizon. Characteristics of the patient population used for a cost-effectiveness analysis will influence results of the analysis in terms of quantity and quality of costs and outcomes. It is important to keep in mind that there is an inherent selection bias underlying studies that focus primarily on patients listed for and/or undergoing liver transplantation. These patients are a select group of people who often have less comorbidity and are predicted to benefit from transplantation. Patients with significant comorbidities or who are otherwise deemed inappropriate liver transplantation candidates are not represented in these studies. However, the natural history models, such as those developed for PBC and PSC, are based on relatively unselected populations of patients with these liver diseases. Thus, although comparisons between survival predictions based on these models and observed survival in liver transplantation patient populations may suggest a greater survival advantage with liver transplantation, it is important to bear in mind that at least some of this difference in survival may be attributable to inherent differences that exist between the patient populations (i.e., liver transplantation patients versus historic cohorts of patients with liver disease).
With respect to the time horizon in this investigation by Longworth et al, patients were followed up during a 27-month period. This included an average of 3 months of waiting list time, with 2 years of posttransplantation follow-up. A long-term follow-up period would be ideal for such a therapy as liver transplantation because net costs and effects associated with the procedure will continue to change beyond the 2-year period. It is likely that the incremental cost-effectiveness ratio of liver transplantation would decrease with time, making liver transplantation appear cost-effective.
An important limitation of model-based (as opposed to trial-based) cost-effectiveness analyses is that these analyses entail many assumptions and estimations. The savvy reader needs to pay close attention to these assumptions and the evidence supporting them because, as we point out in this discussion of liver transplantation, certain assumptions may significantly influence the reported cost-effectiveness of a medical intervention. For example, in this report by Longworth et al, several potential biases may be present that work in opposing directions. There may be overinflation of the reported cost-effectiveness not only because of the differential discounting applied to costs and effectiveness in the study, but also because of the selection bias inherent in studies of transplantation patients, which influences observed patient outcomes. Conversely, truncation of the patient follow-up time at 27 months has the likely effect of making liver transplantation appear less cost-effective than it would had the study been conducted with a longer time horizon. Despite these uncertainties and difficulties, results of this analysis provide valuable information to continue to support liver transplantation and identify key determinants in resource allocation and health care policy decision making.
|
|
| |
| |
|
|
|
