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

AFTER 27 YEARS, the federal government's program to produce a fully implantable artificial heart is nearing one of its goals: a long-term device that can assist the ailing heart. This report is an evaluation of that program by the Institute of Medicine (IOM), at the request of the National Heart, Lung, and Blood Institute (NHLBI).

Heart disease is the leading cause of death in the United States. End-stage heart disease (heart failure) may be caused by coronary atherosclerosis, hypertension, viral or idiopathic cardiomyopathy, or congenital or acquired defects; it results in perhaps half of the nation's 700,000 heart disease deaths each year. Until now, transplantation has been the only effective treatment for most end-stage heart failure patients, but a severely limited supply of donor hearts has meant that only about 2,000 persons receive transplants each year.

NHLBI's predecessor initiated the artificial heart program in 1964; it has supported a number of academic and industry-based R&D teams since then. A primary goal has been the development of two types of mechanical circulatory support system (MCSS), ventricular assist devices and total artificial hearts, for both temporary and long-term use.

A ventricular assist device (VAD) aids the failing heart but does not replace it. Most current models of VADs are implanted in the abdomen, connected to tubes through which blood is drawn from one of the heart's ventricles and pumped into the circulatory system. A temporary VAD can support either ventricle; two VADs can support both. Such devices usually are employed to assist the left heart, the side primarily affected in most types of heart failure. The second category of MCSS, a total artificial

heart (TAH), is similar in concept to two VADs but replaces the diseased heart.

The long-term TAHs and VADs now being developed are powered by electricity that is transmitted across the skin surface in a transformer-like arrangement; the patient will carry rechargeable batteries on a belt or in an over-the-shoulder pouch. An implanted rechargeable battery provides temporary power for about 20 minutes, so the patient can change the external batteries or bathe.

MCSS use began in 1969 with the first temporary implant to support a patient awaiting a donor heart for transplantation. Temporary MCSSs are now also in routine—but still investigational—use to allow the heart muscle to recover functioning after open-heart surgery or after it has been damaged by an acute myocardial infarction (heart attack). In all, more than 1,300 TAHs and VADs have been used in these temporary applications, including some with the VAD itself placed outside the body. This report, however, focuses on devices intended for indefinite, long-term use to assist or replace the heart.

Several developers have made considerable progress toward both VADs and TAHs that are fully implantable. Technological advances have resolved some of the shortcomings observed in the mid-1980s, when Barney Clark, William Schroeder, Murray Haydon, and Jack Burcham received Jarvik-7 TAHs.

The first implantation of a long-term VAD is expected in 1992, in a clinical trial of a device manufactured by the Novacor Division of Baxter Healthcare Corporation. In this NHLBI-sponsored trial, 20 devices will be implanted over a two-year period by researchers at St. Louis University and the University of Pittsburgh. For the evaluation, each patient will be followed closely for two years. Full trial results will thus not be available until 1996 or 1997. Based on years of research including simulated bench testing and animal trials, the developers of this and other long-term MCSSs have confidence in the mechanical capabilities of their devices, but definitive findings about clinical efficacy and effectiveness will not be known for several years.

Estimating the cost of each device, once approved for routine use, is problematic at present. Based on information from developers, the committee expects that a VAD will cost about $50,000 and a TAH about $100,000, expressed in 1991 dollars. In addition, the hospital and physician care required to implant one will probably cost about $100,000, again based on 1991 costs. Costs for postimplantation care depend heavily on the frequency of such events as device problems and complications that require hospitalization.

This study's primary objective is to delineate important considerations in NHLBI 's funding decisions about the artificial heart program. Decisions are needed in 1992 about continuing to fund R&D with both types of MCSS. Another VAD developer besides Novacor is nearing the end of its contract for preclinical testing (the last step before clinical trials), and four contracts for earlier-phase R&D work on TAHs expire in September 1993.

Making R&D decisions about MCSSs is not simple. In 1988, NHLBI terminated work under the four TAH contracts so that it could devote additional funding to VAD development. This action generated considerable criticism from organizations interested in cardiac treatment research as well as from several U.S. senators. NHLBI subsequently reinstated the contracts.

In 1989, at NHLBI's request, the IOM began this evaluation of the artificial heart program. In the first phase, a committee was convened to plan a larger study. In its report, that committee formulated nine questions to become the focus of this full study by a second IOM committee; see The Artificial Heart Program of NHLBI: Plan for Evaluation (Washington, D.C.: National Academy Press, 1989). This summary reflects the nine questions and this committee's responses to them.

Because of the committee's charge to advise NHLBI about its future decisions in this area, the committee examined the decision-making processes involved in R&D funding. Peer review processes involving nongovernment personnel are used effectively in considering investigator-initiated applications for R&D grants and in certain priority-setting activities, but most other NHLBI decisions are based primarily on subjective professional judgment by the agency's executives and administrators. The committee believes that structured aids to decision making would be helpful to NHLBI in situations for which peer review is not appropriate; it recommends cost-effectiveness analysis (CEA) principles as an aid in making decisions about R&D funding allocations. The committee also identified and ranked as to importance 17 other decision-making criteria; it recommends that NHLBI consider using these criteria as a starting point in developing explicit criteria to aid in the exercise of judgment about R&D funding allocations.

Even before human trials of long-term MCSSs begin, predictions of their ultimate effectiveness can be made based on the results of temporary devices used for as long as a year. At least one fully implantable VAD is likely to be approved by the Food and Drug Administration (FDA) for general use by the late 1990s, although the clinical effectiveness of early

model VADs may be lower than that of heart transplantation (which now offers an 80 to 90 percent probability of two-year postoperative survival). As devices improve in reliability and longevity, their effectiveness is likely to approach today's transplantation outcomes. TAHs are being developed on a timetable 5 to 10 years behind VADs, with the first model not likely to be approved by FDA before 2005.

Epidemiological data about the natural course of end-stage heart disease are limited and incomplete. Estimating the number of patients in the United States who will ultimately become candidates to receive a long-term MCSS requires several considerations, among them:

• the effectiveness of the available devices, that is, whether quality of life and the probability of survival are likely to be better or worse than for medical treatment or heart transplantation;

• the number of patients suffering from end-stage heart disease, adjusted for the presence of severe comorbidities and for the proportion who experience sudden death before they would become candidates for an MCSS;

• the relative effectiveness of alternative forms of treatment, both those now in use and others likely to be developed, and thus their impact on the need for an MCSS; and

• the changing demographics and patterns of illness in the U.S. population.

The extent of third-party payments for implanting MCSSs will also influence the volume of use, but is virtually impossible to estimate this far in advance of FDA approval.

Based on current and projected mortality rates, the committee estimates that between 35,000 and 70,000 patients yearly are in the pool of potential candidates to receive a long-term MCSS, but practical limits (e.g., coverage restrictions by third-party payers, availability of qualified personnel) will hold the growth of this technology' s use below this range until about 2010. This estimate assumes that at least one VAD will be approved by FDA by 1999 or sooner and that coverage and payment decisions by policymakers and third-party payers will allow use to increase steadily between 1999 and 2010. The extremes of the potential pool of MCSS candidates correspond to typical upper ages (35,000 for age 75, 70,000 for 85) of patients who may be considered for these devices; however, other than to enable such estimates, no specific upper age limit for MCSS patients should be applied or recommended. Clinically determined criteria should be used in making individual decisions about MCSS use.

Estimating the proportion of all potential MCSS patients who need biventri-

cular support and thus a TAH is more difficult. VAD experience during the 1990s will determine, more precisely than is currently known, whether support of only the left ventricle will suffice for most patients or whether a substantial proportion of them will need a TAH. The committee's best estimate, assuming the availability of at least one FDA-approved TAH by early in the next century, is that 10,000 to 20,000 of the total number of potential MCSS recipients in 2010 will need a TAH; the other 25,000 to 60,000 will be candidates to receive a VAD.

The foregoing estimates are for the primary patient group, namely those who are in greatest need and who will receive most of the MCSSs implanted during the first 15 to 20 years of use. Thereafter, if devices are then at least as beneficial as is transplantation at present, others in what the committee has called the secondary group will also become potential candidates. The size of this group is even less certain. It could include as many as 200,000 persons per year after device availability becomes sufficiently widespread, assuming that resources are made available to finance care to such an extent, perhaps by the year 2020.

Two other factors may affect the need for MCSSs during the 2020s or possibly sooner. One is research that is beginning to yield understanding of the basic mechanism of heart failure. If NHLBI's continuing support of research on this topic results in translating this understanding into pharmaceutical advances, it may eventually be possible to slow or even halt the progression of heart disease or to postpone the point at which an MCSS will be needed. These advances are not likely to eliminate the need for MCSSs, but they may reduce the extent of need in both the primary and secondary groups of patients.

The second possible influence on the volume of future MCSS use is the development and implementation of interventions that are more effective in preventing the causes of end-stage heart disease than those currently employed. The committee's epidemiological projections take into account current trends in mortality rates as a result of such measures as hypertension control programs and dietary changes to reduce blood cholesterol levels, but it is possible that additional intervention methods will yield even greater reductions in the incidence of end-stage heart disease.

Health-related quality of life is important in assessing the effectiveness of MCSSs and other forms of heart disease treatment. Techniques exist to estimate a preference or utility value for each typical health state experienced by patients. Cost-effectiveness analyses are improved by incorporating these values, reflecting patients' quality of life.

Clinical trials of both drugs and medical devices are increasingly assess-

ing the quality-of-life aspect of patient outcomes. Among government agencies that fund research, NHLBI's requirement that health-related quality of life be assessed in clinical trials it supports is a pioneering step. All MCSS trials should use standard quality-of-life domains (assessment categories), so that study results can be compared.

MCSS patients and their families must learn to live with this technology for the rest of the patient's life. The committee recommends that all health care providers involved in use of MCSSs develop continuing support programs for patients and their families, to improve quality of life and allow them to learn from each other's experience.

Cost-effectiveness analysis (CEA) links the net benefit from using a technology to the cost of providing it. Employing CEA to examine MCSS use is particularly appropriate, because of the committee's wish to assure that the aggregate increases in health care costs resulting from MCSS use will be matched by improved patient outcomes.

Health-related quality of life can be reflected in a cost-effectiveness (C/E) ratio by using quality-of-life utilities that indicate a patient 's preferences for particular health states during and after treatment. Quality-adjusted life years (QALYs) are helpful in making comparisons among alternative forms of treatment; C/E ratios are usually expressed as the incremental cost per QALY gained from the particular treatment in comparison with another.

Cost-effectiveness ratios are best evaluated by comparing them with ratios for other forms of treatment for similar diseases. The committee estimates the incremental cost-effectiveness of using a TAH instead of medical treatment as $105,000 per added QALY (in 1991 dollars), a ratio of borderline acceptability. This C/E ratio is considerably less favorable than those for heart transplantation ($32,000 per QALY) and other generally accepted forms of heart disease treatment (e.g., $34,000 per QALY for coronary artery bypass surgery for two-vessel disease with severe angina; $7,000 to $13,000 for percutaneous transluminal coronary angioplasty for severe angina). The TAH C/E ratio is also considerably less favorable than ratios for other forms of treatment of catastrophic diseases, such as an average of about $50,000 per QALY for hemodialysis for end-stage renal disease.

Because the main focus of the report is TAHs, the committee's analysis did not include a specific C/E ratio for VAD use. The initial device cost will be lower than for a TAH, as will some of the detailed probabilities, but the benefit to patients will be very similar. The VAD C/E ratio thus is likely to be somewhat more favorable than the TAH's $105,000 per added QALY. Additionally, if VAD clinical trial results indicate that the commit-

tee's performance estimates can appropriately be revised, C/E ratios for both TAH and VAD use will improve; continuing technological improvements in the performance of the devices themselves may also have this result.

The committee believes that the currently estimated C/E ratio for TAH use is not so extreme that it affects the committee's basic conclusion — namely, that federal support for MCSS development should continue for an interim period. By making possible an earlier approval of TAHs for general use, continuing with TAH development for a two-to three-year period may benefit patients who would otherwise die if TAH development is delayed by a suspension of funding. This interim period will also see the availability of better information about the likely effectiveness of MCSSs in 2010 from the clinical trials and continuing temporary use of VADs. Offsetting these benefits is the possibility that, if the borderline C/E ratio is confirmed by forthcoming VAD trial results and leads to the suspension of TAH development at that point, the continued R&D expenditure will yield no return. The committee believes, however, that this possibility is outweighed by the potential future benefit to patients, particularly because a relatively small amount of R&D spending is involved ($5 to $10 million per year).

Estimating cost-effectiveness almost 20 years in the future is fraught with uncertainty, but the technique of sensitivity analysis allows examining the impact of such uncertainties. Varying the key parameters such as costs, complication probabilities, and discount rate shows that the C/E ratio applicable to TAH use is relatively stable.

Cost-effectiveness analysis can also be applied to the consideration of research and development options. The committee used CEA to examine alternative levels of R&D investment during the next phase of TAH development. Based on assumptions of the impact of increased funding on the length of time required to complete R&D and on the future selling price of TAHs, this CEA reveals that increasing the level of R&D funding may yield benefits greater than would otherwise occur, assuming that the underlying $105,000-per-QALY C/E ratio of TAH use is considered acceptable.

Among the many unusual aspects of MCSSs is the extent to which they have been scrutinized, decades in advance of their general availability. This study is the seventh originated by NHLBI and its predecessor institutes but the first to be conducted independently.

The committee is concerned that, despite this intense scrutiny, MCSSs may be used inappropriately, a concern similar to those expressed over the years about other new technologies. Inappropriate technology use is

inconsistent with the current emphases in health care on both improving the quality of care and eliminating unnecessary utilization and expenditures.

Given this concern, we note that never before has there been an opportunity such as the one this technology now presents to the health care community. Before VADs and TAHs go into general use, crucial activities can be undertaken to improve the likelihood that these devices will be used appropriately. The committee thus recommends that device developers, NHLBI, and other interested parties take advantage of the period of years during which MCSS development is being completed to participate jointly in the following activities intended to promote appropriate use:

• develop clinical practice guidelines or indications for use, at first on a provisional basis, with later revisions after clinical trial results are reported;

• commission technology assessments that include cost-effectiveness components, based on actual device use, in order to verify and update this committee's work;

• establish a registry of MCSS patients as a means of postmarketing surveillance and perform long-term follow-up studies of an adequate sample of all patients;

• develop guidelines for hospitals, third-party payers, and others to use in determining whether individual physicians have the training and experience necessary to implant MCSSs; and

• work with third-party payers to implement selective coverage programs, to avoid unnecessary and wasteful duplication and thus conserve expensive resources, as well as to ensure that the institutions at which MCSSs are implanted are suitably equipped and staffed to provide this care effectively.

Implantation of an MCSS is not a simple, time-limited treatment episode. Because of the patient's total dependence on the device and because problems can occur at any time, clinical trial subjects should be followed closely during the trials; they and other MCSS patients should be followed, through a registry, for the remainder of their lives. The committee recommends that trials be funded at a level sufficient to allow a full range of physiologic data (e.g., cardiovascular, renal, neurologic, hematologic) to be collected periodically, as well as health-related quality-of-life information and treatment cost data that are sufficient for cost-effectiveness analysis.

Maintaining a registry of MCSS recipients should be considered a routine aspect of this care. Its costs could possibly be supported through payments by hospitals at the time of each implant. Third-party payers would benefit from requiring surgeons and hospitals to agree to provide information to the registry and should provide reimbursement for registry

fees in their payment rates. Patients should also be requested to supply information to the registry on a continuing basis, such as during their periodic visits to check device performance.

For this type of device, long-term follow-up studies are important in order to detect problems that occur after the trial ends or that may not be revealed in the small number of cases included in clinical trials. The committee recommends that NHLBI and the Public Health Service's Agency for Health Care Policy and Research support long-term follow-up studies of an adequate sample of MCSS patients.

Because of the high cost of MCSSs both individually and in the aggregate, as well as their borderline C/E ratios, private third-party payers, Medicare, and state Medicaid programs are likely to resist approving coverage for MCSS implantations. Further, access to this technology by the 30 million or more citizens with inadequate or no insurance is of concern to the committee. The time is ripe for the United States to make clear decisions about access to health care, including costly new technologies.

As with clinical appropriateness, the committee recommends that the time until long-term MCSSs are approved by FDA be used by policy-makers at the national and state levels to decide whether at least MCSSs (if not other new technologies also) should be explicitly included or excluded in a package of basic health care services that is applicable to all forms of public and private health insurance. The committee recommends, as well, that policymakers begin to decide about equitable access to MCSSs and other technologies for those without personal resources or insurance, and what trade-offs in access to other services may be necessary to provide MCSS access.

Given the federal budget deficit and other constraints, early policy decisions on access are likely to be made at the state level, such as the efforts currently under way in Oregon and a few other states. For this type of effort to be most effective, the committee recommends that one or more organizations that have the respect of state governments be funded to aid states in considering these decisions.

Medicare, Medicaid programs, and other third-party payers sometimes approve coverage for a new technology but then establish for it a payment rate so inadequate that access to the technology is restricted. The committee believes this is an unacceptable way to ration access to care and recommends that any positive coverage decision include paying for the care at a rate that adequately reimburses the costs of the care, noting that fixing such rates will require accurate cost information.

An adequate informed consent process is an important component of any patient's decision to accept an MCSS implantation. Such a process should include a patient's advance directives concerning termination of treatment, as well as the legal designation of a proxy for such purposes. The committee recommends that protocols for all MCSS clinical trials include these requirements and other appropriate ethical safeguards; these requirements also apply to routine MCSS use and should be part of each institution's standard procedures in all long-term MCSS implantations.

Federal support for the development phase of health care R&D is rare; that phase of work is usually funded by industry. The development of MCSSs, however, presents an unusual situation. Because of technological, clinical, and regulatory uncertainties, adequate private-sector financing to develop long-term MCSSs has thus far not been available. Moreover, such support does not appear likely, at least until the first long-term VAD has been approved by FDA. The committee concludes that the potential benefits that patients will derive from these devices, once approved for general use, provide sufficient basis for NHLBI to continue support of R&D for both VADs and TAHs, as well as for other MCSS-related topics such as the development of alternative power sources. International-trade benefits may ultimately be an important outcome of this R&D, but the committee believes that potential benefits to patients should be the primary basis for these expenditures.

Collaborative research is particularly important to continuing R&D in this field in three respects: (1) interdisciplinary collaboration involving biomedical engineers, clinicians, and scientists; (2) collaborative research between academics and researchers in for-profit firms; and (3) formal R&D cost-sharing between government and private-sector developers. The procedures routinely used by NHLBI for the review and approval of proposals for investigator-initiated grants have, to date, also been used for contract awards by the artificial heart program. The committee recommends that peer review groups for MCSS development proposals include appropriate numbers of clinicians, biomedical engineers, and professionals with other relevant expertise (e.g., quality-of-life assessment, cost-effectiveness analysis), instead of the customary preponderance of life scientists. The committee also recommends that NHLBI consider adopting policies and structures governing the artificial heart program (and any similar ones that may exist) that allow for greater flexibility in arranging for collaborative R&D and cost-sharing.

Finally, the committee is concerned that traditional collegial communication among academic researchers may be adversely affected by the increasing support they are receiving from industry and other private-sector financing sources. The committee recommends that academic researchers avoid financial arrangements that restrict open communication and that universities establish policies and procedures that prohibit unreasonable constraints on communication.

Carrying MCSS R&D through to the development of a marketable device is affected by regulatory activities of FDA and of third-party payers including Medicare. In the past, technical criteria developed by NHLBI for its clinical trials and those applied by FDA in approving devices have not always been consistent. The two agencies are currently working together as the Novacor VAD trial protocol is developed; the committee recommends the continuation of this type of joint activity.

As third-party payers establish MCSS coverage and payment policies, the committee recommends that they recognize all aspects of the technology. Appropriate considerations include clinical benefit to patients— taking into account the information obtained from clinical trials —as well as patient need, access considerations, and the technology 's cost-effectiveness.

Coverage approvals may properly be narrow and restrictive initially but should be reconsidered periodically, based on additional clinical evidence of relative safety and clinical effectiveness and the impact of the new information on cost-effectiveness. As has occurred with other cardiovascular technologies (e.g., pacemakers, percutaneous transluminal angioplasty), indications for MCSS use can quite properly be expanded as experience is gained.

Although the precise number is uncertain, a substantial number of patients to be benefited by MCSSs are likely to need a long-term TAH because a VAD will not provide the biventricular support that they require. Continuing NHLBI R&D funding appears needed if a TAH is eventually to become available for use. Moving into the final preclinical stage of TAH development at this time is premature, however, because the results of early VAD trials will provide useful technological information as well as better information about TAH cost-effectiveness and the number of patients who will need one.

The committee is aware, in developing its main recommendation to NHLBI concerning TAH development, that the estimated cost-effectiveness

ratio for TAH use of $105,000 per added QALY is so unfavorable as to be a possible basis for suspending this R&D program. Further, the committee is aware how little can be said with certainty about the ultimate clinical effectiveness and cost of these devices, if development continues and they are approved for use in 2010 or sooner. In the short term, however, the cost of continuing to develop both TAHs and VADs can be held in the range of $5 to $10 million per year (in 1991 dollars), a level that is consistent with the artificial heart program's commitments to contracts in these areas in recent years.

The committee therefore recommends that NHLBI continue to support TAH development for an interim period, perhaps by extending the contracts of current TAH developers if their work, through peer review and use of criteria such as those applied by this committee, is deemed worthy of continuing federal investment. Such an interim funding period would allow an opportunity for NHLBI to consider the early results of the Novacor VAD trial and results of continuing temporary VAD use before deciding about the next five-year phase of TAH development. Doing so would also allow the developers to determine whether more funding than customary would enable earlier completion of R&D and, if focused on “manufacturability” (redesign to achieve greater quantity-production efficiency), whether it would yield a lower device selling price. Any changes in third-party payment policies or mechanisms also should be considered at that time. Under this recommendation, NHLBI's commitment to TAHs could be reexamined in 1994 or 1995, taking these factors into account in much the same manner as has the committee, but aided by another cost-effectiveness analysis that reflects updated estimates about projected TAH clinical effectiveness, complications, and costs.

Finally, both the committee's recommendations and any NHLBI action consistent with them should be understood by everyone involved notto imply a long-term commitment to TAH development. If clinical performance estimates do not improve as a result of experience during the interim period, NHLBI's proper course in 1994 or 1995 may well be to suspend all support for TAH development until further VAD experience has been gained.

The committee is concerned that, except for the Novacor clinical trial, NHLBI may end its contractual, targeted support for VAD development. Technological limitations or problems may develop with only one model of VAD moving into the clinical-trial phase of R&D; multiple approaches to VAD design may prove to be of significant clinical value. Lower-priced

devices may also result, if more than one manufacturer exists. The committee therefore recommends that NHLBI consider continuing its support for additional VAD development that is judged by peer review to have sufficient scientific merit, for an interim period ending in 1993 or 1994. At that time, further support could be considered in light of initial results from the Novacor trial.

Additional research is needed in two areas. First, any future R&D commitments to MCSSs should not detract from continuing NHLBI support of research aimed at an increased understanding of the mechanisms of heart failure and at developing new approaches to its treatment and prevention.

Second, as previously stated, relatively little is known about the epidemiological aspects of end-stage heart disease, although one current NHLBI-supported study will provide additional information concerning heart disease in patients aged 65 and over. The committee recommends that more epidemiological studies be performed of the natural course of heart failure in persons under age 65, with particular attention to including women and members of minority groups.

A wide range of persons within and outside the health care community will be able to apply some or all of this report's findings and recommendations. It is both a specific technology assessment that provides detailed guidance to NHLBI and one that can serve as a broader model. The process used in this evaluation and many of its general recommendations should be useful to policymakers at the state and federal levels, third-party payers, and others in the health care system.