The Artificial Heart: Prototypes, Policies, and Patients (1991)

THE TRADITIONAL ROLE of cost-effectiveness analysis (CEA) is in helping to assess the overall value of specific medical technologies, but it can also be an important aid to decisions about research programs. This chapter considers the projected treatment cost-effectiveness of total artificial hearts (TAHs) and also illustrates a way in which CEA can assist in making decisions about the level of research on a particular technology.

To aid in the committee's work and to provide information about TAHs that will help others as they scrutinize this technology in the years ahead, the committee commissioned a cost-effectiveness study with two goals: (1) to project the cost-effectiveness of long-term TAH use, based on estimates of future device performance, clinical effectiveness, and costs, and (2) to develop information for the National Heart, Lung, and Blood Institute (NHLBI) to use in deciding future investment levels for TAH development. After providing general background, this chapter summarizes the CEA findings and discusses their implications for future TAH development; Appendix E provides the most important data elements and additional detail about the methods used in the analyses.

CEA and its counterpart cost-benefit analysis (CBA) are analytic techniques for comparing the positive and negative consequences from the use

of alternative technologies. Interest in health care CEA and CBA began to appear in the late 1970s, spurred largely by provider, payer, and consumer concern over increasing health care costs and government spending for health care services.

The uses of CEA and CBA for technology assessment can be categorized either by the type of technology (i.e., drug, medical device, or procedure) or by applications. CEAs and CBAs have been used for a variety of devices, instruments, and drugs. Such assessments require detailed analysis of the process, technical procedures, and personnel using the products; they may employ different analytic methods for diagnostic, therapeutic, or supporting applications.

The principal distinction between CBA and CEA is in the valuation of the effects and benefits of the alternatives. In measuring benefits, a CBA requires that all important effects and benefits be expressed in monetary (dollar) terms. CEA avoids this requirement by calculating the lives (or years of life) saved or lost per dollar expended. Further, assessments of the quality of life of the years saved may adjust or weight differences in health status, or “utilities.” These utilities usually range from a value of zero for the state of death up to 1.0 for a healthy state, but a state “worse than death” can receive a negative value; for example, the mean value assigned by the committee's assessment to a moribund end-stage heart disease patient in an intensive care unit or coronary care unit was −0.11, as discussed below.

In performing analyses, all future costs and benefits (including health consequences) should be discounted to their present value in order for them to be compared appropriately with one another. The discount rate attempts to adjust for the fact that a dollar not spent today would earn interest, which could then be made available for future health programs. For long-term projections, low discount rates tend to favor projects whose benefits accrue in the distant future or whose major costs occur in the near term. Accordingly, the selection of appropriate discount rates is often controversial and the use of such rates is usually subjected to sensitivity analysis.

Sensitivity analysis checks the importance of assumptions by testing a range of discount rates, varying the weights used to compute quality-adjusted life expectancy, and testing all important clinical and cost variables over a range, for example, from best to worst case. It is an important means to cope with problems of projections and uncertainties about the future.

A CEA or CBA should not serve as the sole determinant of a health care decision, but the process can improve decision making by considering not

only whether the technology is effective but also whether it is worth the cost. In general, a CEA is most useful for making a choice as to the lowest-cost technique to achieve a specified objective, benefit, or effect, or for setting priorities among technologies within a limited budget. A CBA is most useful for deciding whether to implement a technology when neither budget constraints nor alternative uses of resources are explicitly known.

Because CEA and CBA consider both economic and health outcomes, they offer promise in their ability to help with policy decisions that affect the quality of care under resource constraints. Social values, ethical considerations, and political realities may well, however, take precedence over analytical results. CEA and CBA techniques can be a useful tool for planning for the future, and prospective analytic simulation models can attempt to predict costs and effects or benefits of competing alternative programs.

The result of a CEA is expressed as a ratio, where the numerator is the total discounted net cost of the intervention for a defined group of patients and the denominator is the aggregate benefits those patients derive from the intervention, also discounted. The denominator is usually expressed in life years (or quality-adjusted life years) gained in relation to another form of (or no) treatment, yielding a cost-effectiveness (C/E) ratio of cost per added life year or quality-adjusted life year. Simply put, the greater the C/E ratio, the less favorable the intervention's effectiveness in relation to its cost.

The CEA performed for this committee provides a means of comparing the anticipated clinical benefits and costs of using a TAH with those of other forms of heart disease treatment as well as treatments for other diseases. The CEA examines this technology as of 2010, the earliest likely time when the TAH could be in wide use. It also examines the cost-effectiveness of the two main alternatives to mechanical circulatory support systems (MCSSs) for patients with end-stage heart disease: transplantation and conventional intensive medical treatment.

As described in more detail in Appendix E, the CEA was performed using a Markov simulation model that permits variations in assumptions about cost, clinical, and outcome variables. Using a 20-year time frame beginning in 2010 for the model, the typical patient moves through a sequence of potential “states” on a monthly, probabilistic basis. The committee and its consultants developed a number of estimates and assumptions for the CEA concerning the clinical effectiveness of TAH use, heart transplantation, and conventional medical treatment of end-stage heart disease. Costs were also projected, based on estimates such as anticipated lengths of hospital stays both for the initial procedure (implantation, transplantation, or regulation of medication dose) and for necessary device repairs or re-

placements as well as various types of complications that were determined to be likely. Costs were estimated in constant 1991 dollars, even though the scenarios were for the 2010-2030 period.

For MCSS use as well as for heart transplantation and conventional medical therapy, individual committee members participated in an assessment of quality-of-life “utility” or preference values. They assigned values to nine typical health states as shown in the appendix to this chapter, where 1.0 denoted full health and 0 denoted death. Because a number of committee members considered being hospitalized with end-stage heart disease as a state worse than death, one of the mean utility values is negative, as shown in Table 6.1. The mean values were incorporated into the cost-effectiveness analysis, yielding figures for estimates of the quality-adjusted life years (QALYs) gained from each type of treatment as well as simple increased life expectancy.

Table 6.2A and Table 6.2B summarize the results of the cost-effectiveness analysis for the primary or “base case,” applying the best estimates of the various assumptions discussed in Appendix E to the three alternative forms of treatment. The CEA shows that a patient will live an average of 4.4 years after TAH use, before adjusting for quality of life, in contrast with the six months expected if conventional medical treatment is received. Heart transplantation is calculated to provide an average 11.3-year life expectancy, because the committee's consultants estimated complications following transplantation at rates somewhat lower than current ones, in anticipation of improved outcomes between now and 2010. Adjustments for the quality of life reduce the life expectancies for all three forms of treatment. Total costs of TAH use over the average patient's lifetime are substantially higher than for a transplant, primarily because of the device's $100,000 cost.

The cost-utility data assessing the quality of life under the various health states are the group opinions of the committee, not of actual or prospective

patients, as would be ideal and eventually possible. Results of the CEA are thus presented both with and without incorporating these data, that is, both unadjusted and adjusted for the quality-of-life utility values.

Table 6.2B, comparing both TAH use and heart transplantation with conventional medical treatment, shows that a TAH yields an average increase of 2.85 years in quality-adjusted life expectancy at a net cost of $299,000, for a C/E ratio of $105,000 per QALY gained. Because of transplantation 's lower total cost and greater life expectancy, its C/E ratio is $32,000 per added QALY.

As discussed in detail in Appendix E, sensitivity analysis to examine the impact on the CEA's results of changing the assumptions for important variables has relatively little impact on the finding of primary interest, the C/E ratio of TAH use in contrast to conventional medical treatment. Increasing each of eight probabilities of key adverse events by 25 to 100 percent increases the marginal cost per QALY only 3 to 11 percent. If all costs are increased 25 percent in addition to all eight of the probabilities, the TAH's cost per added QALY becomes $165,000 instead of $105,000.

Reducing all eight of the variables in combination, as well as reducing costs 10 percent, yields an improvement in the TAH's cost per added QALY from $105,000 to $73,000. Discounting all of the costs and benefits at other than the 3 percent rate used in the base case has little impact; using a 10 percent discount rate yields a TAH cost of $117,000 per added QALY, only 11 percent greater than the $105,000 under the base case.

It is problematic to evaluate a single, isolated C/E ratio; these ratios must be interpreted in the context of C/E ratios for other uses of resources such as other forms of treatments. Table 6.3 reveals that the cost-effectiveness of a long-term TAH falls above the range of (and thus is considerably less favorable than) C/E ratios for other generally acceptable forms of heart disease treatment. Other applications of some of these interventions have even higher C/E ratios, for instance, coronary artery bypass surgery for very mild angina or care of a low-risk patient in a coronary care unit instead of an intermediate care unit. Such interventions with low-risk patients are generally deemed inappropriate, although they sometimes occur.

Further, the C/E ratio for TAH use is also considerably less favorable than those of other forms of treatment for catastrophic medical problems. The current average cost to the Medicare program for care of a patient with end-stage renal disease is about $30,000 per year, including hemodialysis either at home or in a facility, the latter being the most common form of treatment (IOM, 1991). The cost, and thus the cost-effectiveness, can vary dramatically, however, based on such factors as severity of illness, comorbidities, and whether patients regularly receive the costly drug erythropoietin for treatment of their anemia. Nonetheless, using an approximate utility value of 0.60² yields an average C/E ratio of roughly $50,000 per QALY (in

the case of hemodialysis, each year's cost matches with a year of life gained, because constant treatment is required). Further, Boyle et al. (1983) found a C/E ratio for neonatal intensive care of infants weighing from 500 to 999 grams to be $18,000 per added QALY in 1978 dollars, or $42,600 in 1991 dollars.

The primary charge of the committee was to make recommendations about NHLBI's future support for TAH development and the CEA thus focused on that device. CEA can, of course, be applied to a ventricular assist device (VAD) by substituting an estimate of $50,000 as the device 's cost instead of the TAH's $100,000 and revising other probabilities appropriately. For example, the impact of hard failure of a VAD is much less severe than with a TAH, as the patient's natural heart may be able to sustain the circulation long enough for the patient to reach a hospital and receive appropriate care such as repair or replacement of the device. While our CEA has not calculated a definitive number, the C/E ratio of VAD use when compared with conventional medical treatment is likely to be somewhat more favorable than the $105,000 for TAH use compared with medical treatment.

Eventually, if the clinical effectiveness of MCSSs approaches that of heart transplantation, these devices will be used with some patients (i.e., the secondary group described in Chapter 4) before their disease has progressed to end stage. In such a situation, MCSS use will likely halt or slow disease progression, at about the same total cost (including follow-up care) as for end-stage patients.

By definition, as a result of their less severe disease at the time of MCSS implantation, the average life expectancy of this type of patient—without receiving a transplant or MCSS—would be greater than for the moribund patients considered in the committee's CEA. (The impact of quality-of-life adjustments would likely not have a great effect on the CEA's outcome.) Consequently, the cost per added QALY of MCSS use with these earlier-stage patients would be even less favorable than the $105,000 C/E ratio applicable to moribund patients, because the cost would be divided by a smaller number of QALYs added. Only by incorporating into the CEA indirect costs and benefits, such as the added years of productivity gained by avoiding premature death, would intervention earlier in the disease course produce a more favorable result.

Economists have used various analytic tools in approaching decisions about R&D funding, in health care and many other fields. CEAs have been used in studies of a range of research topics, such as choosing from among alternative R&D proposals for cancer prevention (Weinstein, 1983) and vaccine development (IOM, 1985b), the appropriate federal role in R&D (Finneran, 1986), and the balance among R&D investments in various fields (Hartunian et al., 1981; Hatziandreu et al., 1988).

This portion of the committee's CEA has a more limited purpose, which is to assist NHLBI in determining the level of its continuing investment in TAH development during the 1990s. This analysis drew on the cost-effectiveness data generated for the first portion of the study in order to determine the relative cost-effectiveness of three scenarios with varying levels of R& D funding for TAH development after the four developers' current contracts expire in 1993. Under Scenario 1, the primary or base case, two developers would each receive $2 million per year for a five-year period of preclinical device readiness testing, and a 1999-2003 clinical trial of one of these devices would be funded at $2 million per year.

As explained in more detail in Appendix E, TAH developers advised the committee that increased funding levels of TAH development during the 1990s would likely yield long-term benefits, because the additional funding would allow earlier completion of R&D, making devices available for use sooner and thereby extending the lives of some patients who would otherwise die. The added funding would also perhaps reduce the selling price because of resulting design improvements and thus create long-term savings in health care costs, with the TAH selling price lowered from about $100,000 to $70,000 or $78,500, depending on the scenario. More details of Scenarios 2 and 3, describing the effects of the alternative levels of increased funding during the 1990s, are found in Appendix E; this CEA example calculates the benefits that will occur as TAHs come into general use in about 2010.

Under Scenario 2, treatment costs would be increased because an additional one-year cohort of patients would receive TAHs as a result of earlier R&D completion, but those costs and Scenario 2's additional R&D expense would be substantially offset by a lower TAH selling price. The C/E ratio for the patients receiving TAHs in 2008 and 2009 would be $40,000 per added QALY under Scenario 2 and $44,000 under Scenario 3, making it reasonable to undertake the additional R&D expense. Thus, if NHLBI believes that these benefits would flow from the added R&D spending andthat TAHs will in fact be used in substantial volumes despite their underly-

ing borderline cost-effectiveness, it may be beneficial to increase R&D funding over the base-case level of Scenario 1.

Sensitivity analyses (see Appendix E) show that the outcomes projected for the additional R&D funding are relatively stable when the assumptions are varied. Even reducing the TAH selling price to $50,000 only drops the C/E ratio to $85,000 per added QALY.

It should be understood that this CEA's results as well as the sensitivity analyses are dependent on a decision that TAHs will be used despite their borderline C/E ratio and that the projected benefits from the increased R&D spending are reasonably certain. If the commitment for continuing development of TAHs under the base case is notcertain to result in products that the health care system can afford to buy and use, then it would be folly to increase the R&D spending level. The same would be true if it is felt that more R&D funds might well not produce the gains assumed in Scenarios 2 and 3.

This CEA is useful as an example of how this technique can aid in deciding about funding levels withina single R&D program, in contrast with the discussion in Chapter 3 of using CEA as an aid in making funding allocations amongalternative research programs. It also provides a specific method that the NHLBI can use to help in deciding issues that it must face in the 1990s about continuing to support TAH development and, if so, at what dollar level.

The CEA performed for the committee reveals that the estimated benefits from using a long-term TAH compared with medical treatment yield a C/E ratio ($105,000 per QALY gained) that is considerably less favorable than ratios for both other generally acceptable forms of treatment for heart disease and other treatments for end-stage or catastrophic diseases. The estimates used in developing this CEA resulted in an average life expectancy (LE) of 53 months for TAH patients, substantially below the 11.3 years projected after heart transplantation. While the projected 11.3-year LE of a year-2010 transplant patient anticipates clinical gains over the next two decades, parallel technological improvements in TAHs—building, for in-

stance, on more than a decade's clinical experience with fully implantable VADs—may also yield a TAH by 2010 that is more effective than indicated by the performance estimates used in this CEA.

Currently, based on these estimates, the C/E ratio of TAH use is borderline; it remains to be seen what the results of VAD clinical trials will demonstrate about the estimated probabilities and other CEA assumptions. If probabilities of complications and similar events can be reduced as a result of VAD clinical-trial experience, doing so will yield an improved projection of the TAH's clinical effectiveness —and thus an improved cost-effectiveness—that can legitimately be used in future decision making.

NHLBI should, for now, recognize the borderline nature of the TAH 's estimated cost-effectiveness in deciding about future support of TAH development; this conclusion is discussed further in Chapter 10. Those developing clinical practice guidelines should also take the TAH's C/E ratio into account, as should third-party payers in their coverage and payment decisions.

Applying CEA to a narrowly defined question about a single R&D program allows the long-term effects of such options as alternative funding levels to be examined, in addition to CEA's broader R&D funding-allocation use discussed in Chapter 3. Subject to validation of the underlying assumptions, this portion of the committee's CEA shows that NHLBI may wish to increase its level of investment in the next full-scale phase of TAH development because of the benefits that may be derived, in life years gained, from earlier device availability. There may also be a potential for long-term savings as a result of a lower device cost to hospitals. It is not usually possible to take such considerations as these into account in making decisions about R&D programs, but the committee 's CEA example reveals the usefulness of this technique in providing precisely this type of information.

Beck, J. R., D. N. Salem, N. A. M. Estes, and S. G. Pauker. 1987. A computer-based Markov decision analysis of the management of symptomatic bifascicular block: The threshold probability for pacing . Journal of the American College of Cardiology 9:920-935.

Boyle, M. H., G. W. Torrance, J. C. Sinclair, and S. P. Horwood. 1983. Economic evaluation of neonatal intensive care of very-low-birth-weight infants. New England Journal of Medicine 308:1330-1337.

Finneran, K. 1986. The Federal Role in Research and Development: Report of a Workshop. (Committee on Science, Engineering, and Public Policy of the Na

tional Academies of Sciences and Engineering and the Institute of Medicine.) Washington, D.C.: National Academy Press.

Hartunian, N. S., C. N. Smart, and M. S. Thompson. 1981. The Incidence and Economic Costs of Major Health Impairments. Lexington, Mass.: D. C. Heath and Company.

Hatziandreu, E., J. D. Graham, and M. A. Stoto. 1988. AIDS and biomedical research funding: Comparative analysis. Reviews of Infectious Diseases 10:159-167.

IOM (Institute of Medicine). 1985a. Assessing Medical Technologies. Washington, D.C.: National Academy Press.

IOM. 1985b. New Vaccine Development: Establishing Priorities. Vol. 1. Diseases of Importance in the United States. Washington, D.C.: National Academy Press.

IOM. 1991. Kidney Failure and the Federal Government. R. A. Rettig and N. G. Levinsky, eds. Washington, D.C.: National Academy Press.

Torrance, G. W., and D. Feeny. 1989. Utilities and quality-adjusted life years. International Journal of Technology Assessment in Health Care 5:559-575.

Weinstein, M. C. 1983. Cost-effective priorities for cancer prevention. Science 221:17-23.

Wong, J. B., F. A. Sonnenberg, D. N. Salem, and S. G. Pauker. 1990. Myocardial revascularization for chronic stable angina: Analysis of the role of percutaneous transluminal coronary angioplasty based on data available in 1989 . Annals of Internal Medicine 113:852-871.