Cost-Effectiveness Studies: A Key to Innovation Development
Several speakers at the conference observed that outcome analysis/ cost-effectiveness studies are often key to innovation development. Such studies can enhance development/diffusion, as was the case for t-PA (tissue plasminogen activator) and implantable cardioverter defibrillators (ICDs). They can also impede development/diffusion, as was the case for intravascular ultrasound (IVUS).
Robert Young of the Fox Chase Cancer Center pointed to the reluctance to carry out cost-effectiveness studies resulting in new technology being used without adequate cost-effectiveness assessment. Richard Pazdur of the Food and Drug Administration (FDA) outlined some of the evaluation challenges facing metastatic melanoma. Fran Visco of the National Breast Cancer Coalition described how a patients’ advocacy group has fostered evidence-based medicine. Sean Tunis of the Center for Medicare and Medicaid Services (CMS) described the more formal approaches to coverage decisions now being adopted by CMS. Finally, Mark McClellan described a government initiative to develop a more seamless approval and coverage process.
TISSUE PLASMINOGEN ACTIVATOR (t-PA) CASE HISTORY
Dan Mark of Duke University began by saying that in terms of the economics of new therapies there are two key questions. The first question: is the new therapy good value for money? Expressed another way, is there an appropriate balance between the incremental health benefits that are
being produced and the incremental costs that are required to produce these benefits? If a therapy represents good value for the money, then the second question is whether there is money available to make the therapy generally available. Mark used the history of the development of t-PA (tissue plasminogen activator) to throw light on these two questions.
Mark outlined the development of thrombolytic therapy, streptokinase. In the 1980s, two very large RCTs were carried out in Europe to evaluate streptokinase. The ISIS2 trial was the second trial and was published in 1988. This trial showed that the combination of streptokinase therapy plus one aspirin taken in the emergency room lowered the mortality rate in heart attack patients from 13 percent to 8 percent. A paradigm shift in the treatment of heart attacks had been achieved. Instead of being a passive observer, cardiologists now had a therapy that could change the outcome for many heart attack patients.
Around this time Genentech developed recombinant t-PA. By 1985, the NIH had conducted the TIMI1 trial, which had shown that at 90 minutes after the onset of a heart attack the t-PA drug had twice as many open infarct arteries as the streptokinase drug. This was regarded as a major advance in the care of acute myocardial infarction patients. Treatment had gone from a low-tech bacterial enzyme (streptokinase) to a high-tech genetically engineered product (tissue plasminogen activator). In the United States, no trial was carried out to evaluate the mortality benefits of t-PA. Since the drug opened more arteries there seemed no question that patients would live longer.
In a more skeptical Europe, two large-scale trials, GC2 and ISIS3, were carried out. Both of these trials found that t-PA and streptokinase were equivalent in mortality rates. This was a shock to U.S cardiologists and it had a serious negative effect on the market share of t-PA. At the time of the GC2 and ISIS3 trials t-PA had a U.S market share of about 70 percent, with streptokinase having the remaining market share. t-PA’s market share dropped to 55 percent and at this point Genentech decided to fund the GUSTO-1 trial comparing streptokinase with t-PA. This 40,000 patient trial showed that the streptokinase patients at 30 days had a 7.3 percent mortality rate and the t-PA patients had a 6.3 percent mortality rate. At the time the streptokinase treatment cost about $300 while the t-PA treatment cost about $2,200, a seven-fold difference in price. Subsequent cost-effectiveness analysis showed that t-PA provided an extra life year at an estimated cost of $33,000 (Mark et al., 1995; Mark, in press). As a result of this analysis, t-PA was judged to be an economically attractive therapy. Recombinant thrombolytic agents (t-PA and others) now have 96 percent of the U.S. market with streptokinase having just 4 percent.
On a national level the economic cost of shifting from streptokinase to recombinant thrombolytic agents has been significant. Assuming there are
1.1 million myocardial infarctions per year in the United States and about 30 percent of these cases receive intravenous thrombolytics, shifting from streptokinase to t-PA (or a mutant of t-PA) adds about $627 million to the national health budget solely for this aspect of caring for myocardial infarction.
IMPLANTABLE CARDIOVERTER DEFIBRILLATORS (ICDs) CASE HISTORY
Mark Hlatky of Stanford University examined the early barriers to the development of ICDs. He said that the early ICDs were large, they could only give a shock, and they required open-heart surgery to be put in place. These initial technical limitations were at the beginning the major barrier to the use of ICDs. These limitations have now been largely overcome. The second early barrier was the high cost of both implanting and maintaining ICDs. This led to another major barrier, the lack of evidence of cost-effectiveness. The final early barrier was the need for specialized personnel to manage patients.
Regarding the determination of effectiveness, the RCT is regarded as the gold standard in the clinical community because it is the best way to ensure a fair comparison of two therapies. Further, the demonstration via an RCT that a therapy saves more lives is usually a tremendous marketing tool for the therapy.
Hlatky said that the ICD is an interesting case history for effectiveness studies. Ventricular fibrillation is fatal if not treated. So, it was argued that any time the ICD was activated a life had been saved. However, even though the ICDs were effective in converting ventricular fibrillation, many were skeptical that this meant the patients were actually living longer.
Eventually RCTs of ICDs were performed, but there were many problems. There was inadequate patient enrollment. Many physicians did not want to enroll patients into the studies because these physicians were convinced the device was beneficial. In addition, many patients did not want to be part of an experimental study. The second problem was what to compare the device against—no treatment or a drug treatment? The third problem was identifying who should be responsible for paying for the study— the National Institutes of Health, FDA, payers, or health care providers? The final problem was the generalizability of the trial findings. RCTs tend to be carried out in idealized circumstances in which patients are highly selected and the therapies are administered in the best clinical settings by the best personnel.
Hlatky said that there have been at least three major RCTs analyzing patients who had an episode of cardiac arrest or ventricular tachycardia. About 2,000 patients were randomized either to get an ICD or the best drug available. The results show that patients who were given an ICD had a 28
percent lower chance of dying of any cause and about a 50 percent lower chance of dying due to an arrhythmia (Connolly et al., 2000). Hlatky observed that this was considered convincing evidence that ICDs16 do enable people to live longer if they are chosen properly and have prior evidence of cardiac arrest.
INTRAVASCULAR ULTRASOUND (IVUS) CASE HISTORY
Elizabeth Nabel (NHLBI) considered the use of intravascular ultrasound (IVUS), a device inserted in blood vessels to acquire images using ultrasound technology. IVUS was first developed by academic cardiologists in the late 1980s. It was not FDA approved until the early 1990s. It has been used primarily as a diagnostic and research tool in many academic centers despite the fact that it has not been reimbursed for a number of years. Nabel argued IVUS illustrates that sometimes devices are developed but are not readily implemented, and their benefits may not be apparent for 10 to 15 years.
Nabel said that there are four different diagnostic uses for the device:
to characterize the nature of atheromatous plaque;
to detect plaque rupture;
to detect transplant arteriopathy; and
to ensure apposition in stent placements.
One of the major impediments to broad deployment of IVUS has been cost-effectiveness data. A study (Berry et al., 2000), carried out by the National Health Service R&D Health Technology Assessment Programme in Britain found that if IVUS were used at the time of angioplasty by traditional methods using balloon catheter or by stent deployment, the restenosis rate was approximately 16 percent. Without IVUS use, the restenosis rate was approximately 24 percent. The cost per restenosis event avoided was about £1,500 ($2,200). After extrapolation to long-term outcome, the calculated cost per quality adjusted life-year was approximately £6,500 ($10,000). The baseline quality gain was 0.03 years. These investigators argued that in terms of cost-effectiveness, the widespread use of IVUS was not worth the investment.
Nabel concluded that IVUS is a valuable clinical adjunct to angiography. It is estimated that 5 to 8 percent of all stents that are deployed use IVUS, and a number of clinical studies have shown that IVUS-derived
residual plaque burden is the most useful predictor of outcome following clinical interventions. Technical development has persisted despite the slow schedule of reimbursement. Major clinical applications were not anticipated at the time of initial development. Cost-effectiveness is unlikely to be demonstrated unless this diagnostic device is incorporated into a therapeutic device. In summing up, Nabel said that IVUS is an example where cost-effectiveness data showed that broad use in angioplasty was not justified, and, as a result, development of the technology has been slowed.
MEDICAL USAGE OFTEN PRECEDES EFFICACY ASSESSMENT
Robert Young of the Fox Chase Cancer Center pointed to the frequent occurrence of therapies that are widely used but poorly assessed for cost-effectiveness by referring to some of the tools for diagnosis of melanoma. He said that increasingly complex diagnostic technologies are emerging, ranging from photography, through digital imaging and epiluminescence microscopy, to qualitative image analysis. The last three are expensive and have been heavily promoted but poorly assessed for cost-effectiveness. Other widely used but poorly assessed therapies include maternal-fetal monitoring, bone marrow transplant in breast cancer, lung reduction therapy in emphysema, and Ca-125 and trans-vaginal ultrasound in ovarian cancer.
Young then outlined a number of reasons why this happens. First, payers (insurers, employers, HMOs, and CMS) have generally been passive about the need for objective assessment (via RCTs) of unproven technology. They have too often abandoned the design of clinical trials to whomever wants to do trials funded in whatever way they want to do it. As a result, the large amounts of incomplete but positive data generated result in the stimulation of public demand. Second, and perhaps equally important, physicians and patients in the United States are aggressive and interventionist about care. At the same time there is little congressional desire to control access to or utilization of new technology and courts tend not to side against the patient-doctor relationship.
Determining at the margin what medical innovation is worthwhile and what is not represents a major challenge. As Paul Citron of Medtronic pointed out, early cost-effectiveness studies for implanted devices are likely to present worst-case scenarios and could cause therapies to be abandoned prematurely. As a result there is a reluctance to pay for outcome analysis/ cost-effectiveness studies. Young expressed the view that payers ought to take a more active role in trial design and fund the key trials. They should pay for the care and experimental costs only in the context of properly designed clinical trials and they should fund several sites for confirmation and comparisons.
Sean Tunis of CMS agreed with Young that payers should take a
greater role in clinical research. He said that payers and purchasers of health care need to know the clinical effectiveness of new medical technologies and, as a consequence, these stakeholders have a responsibility to participate in the clinical research process. Tunis said that Medicare had taken an important step in this direction by paying for the routine costs of care in federally funded or approved clinical trials following a recommendation made by an Institute of Medicine committee. Medicare is currently paying for the cost of experimental interventions in two clinical trials. One trial is evaluating lung volume reduction surgery and the other carotid stenting.
EVALUATION CHALLENGES FACING METASTATIC MELANOMA AND OTHER CANCERS
As background to his presentation, Richard Pazdur of the FDA described the basis for New Drug Application (NDA) approval. It is necessary, first, to demonstrate efficacy with acceptable safety through adequate and well-controlled studies and, second, to generate product labeling that defines an appropriate patient population and provides adequate safety information. In recent years the FDA has developed initiatives (for example, accelerated approval, fast track, priority review) to bring drugs to the market earlier in the approval cycle.
Turning to the regulatory challenges posed by metastatic melanoma, Pazdur said that the disease is characterized by a high degree of biological heterogeneity. Survival may be influenced by prognostic factors, and therefore it is important for clinical trials to be well balanced with regard to these factors. The advent of biological therapies has led to the approval of Interleukin-2. The delivery of high dosage Interleukin-2 usually requires intensive medical support with stringent eligibility criteria for inclusion in trials, raising the question of selection bias.
With respect to future regulatory challenges for oncology in general, Pazdur said that targeted therapies may make drug regulation easier. Treatment effects on better-defined populations are likely to be greater, requiring smaller clinical trials. Targeted therapies may also provide opportunities to look at novel surrogate end-points. This challenges both the regulatory and scientific communities to demonstrate that improvements in surrogate measures translate into improved clinical outcomes.17
There are other challenges. With the explosion of new agents that are being developed for medical oncology, the industry must identify those agents that are really important since it will not be possible to carry out clinical trials for every promising compound that is discovered. Pazdur pointed out that it is estimated that less than 5 percent of eligible patients are actually enrolled in clinical trials. He concluded by saying that efforts have to be made to increase enrollment in clinical trials.
CONSUMER GROUPS CAN HELP FOSTER INNOVATION
Fran Visco of the National Breast Cancer Coalition said that the NBCC stresses the importance of educating consumers on what is quality care and how they should go about getting it. The organization has developed a quality care guide that gives consumers a set of core values and an understanding of evidence-based decision making. The guide does not attempt to tell consumers what choices to make.
The NBCC has also been active in fostering clinical trials. It was a strong advocate of the Rockefeller-Mack legislation to provide Medicare coverage of routine patient care costs in clinical trials and helped bring about President Clinton’s Executive Order mandating this coverage. The NBCC has also been active in getting the CMS to implement this legislation. At the time of the conference, however, Medicare had covered the routine patient costs of only two trials. One barrier for patients wanting to enroll in clinical trials is not knowing that they are taking place. As a result of initiatives by the NBCC and others, Section 113 of the FDA Modernization Act required the institution of a clinical trials data bank. The National Library of Medicine has set up the data bank but unfortunately the industry has not yet agreed to put their trials in the data bank. The Act does not provide any enforcement mechanisms.
The NBCC also favors direct consumer involvement in clinical trials to bring forward innovation, something recently endorsed by the Lancet (Lancet, 2001). An example of this is the Genentech-NBCC collaboration on the herceptin trial. NBCC participated in the trial steering committee, had representatives on the data safety monitoring board, and helped prepare the outreach materials. Genentech has publicly stated that as a result of the consumer collaboration herceptin was on the market two years sooner than otherwise would have been the case.
ROUTES FOR NEW TECHNOLOGY INTO THE MEDICARE PROGRAM
Sean Tunis explained that there are three main routes whereby new technology enters the Medicare program. One way is through the diagno-
sis-related group (DRG) for inpatient care and ambulatory payment classification (APC) system for outpatient care. If there exists a DRG/APC payment category for a particular condition new technologies can be added to the Medicare program by being billed under the existing DRG/APC for the condition. The second main route for new technology to enter the Medicare program is through the Local Medical Review Policy (LMRP) process. Individual insurance companies that pay medical claims at the local level have their own technology policies. These local insurance companies make most new technology coverage policy. The national level, the third main route, handles only a minority of coverage decisions, for example, where the new technology represents a significant medical advance or there are inconsistent local coverage policies.
Tunis said that over the last decade or two the national coverage process has drawn criticism for the length of time that it has taken to make coverage decisions and the lack of transparency of the process. In response to these criticisms CMS has made the following changes since 1999 (Tunis and Kang, 2001):
A Federal Register notice described how the coverage process works and the timelines for this process.
Every coverage decision is now accompanied by a memorandum posted on the world wide web explaining the rationale for the decision.
The establishment of the Medicare Coverage Advisory Committee created a public venue for dealing with both general and specific technology issues.
An evidence-based framework has been adopted for coverage decisions.
DEVELOPING A MORE SEAMLESS APPROVAL AND COVERAGE PROCESS
In his keynote address, Mark McClellan spoke about an initiative to create a more seamless approval and coverage process. The NCI, CMS, and FDA are jointly developing an integrated process which involves all parties getting together early in the process, for example, when a clinical trial is being designed. Through this process the manufacturer of the device/drug learns at an early stage the information requirements not just for the FDA’s safety and efficacy evaluation but also for the CMS and private insurers’ coverage decision process. This initiative currently includes technologies involved with imaging procedures in cancer and is organized through the Interagency Council on Biomedical Imaging in Oncology.
Berry, E., S. Kelly, J. Hutton, H. S. J. Lindsay, J. M. Blaxill, J. A. Evans, J. Connelly, J. Tisch, G. C. Walker, U. M. Sivananthan, M. A. Smith. 2000. Intravascular ultrasound-guided interventions in coronary artery disease: a systematic literature review, with decision analytic modeling, of outcomes and cost-effectiveness. Health Technology Vol. 4: No. 35.
Connolly, S. J., A. P. Hallstrom, R. Cappato, E. B. Schron, K.-H. Kuck, D. P. Zipes, H. L. Greene, S. Boczor, M. Domanski, D. Follmann, M. Gent, R. S. Roberts. 2000. Meta-analysis of the implantable cardioverter defibrillator secondary prevention trials. European Heart Journal 21:2071-2078.
Lancet editorial. 2001. How consumers can and should improve clinical trials. 357:1721.
Mark, Daniel B., Mark A. Hlatky, Robert M. Califf, C. David Naylor, Kerry L. Lee, Paul W. Armstrong, Gabriel Barbash, Harvey White, Maarten L. Simoons, Charlotte L. Nelson, Nancy Clapp-Channing, J. David Knight, Frank E. Harrell, John Simes, and Eric J. Topol. 1995. Cost Effectiveness of Thrombolytic Therapy with Tissue Plasminogen Activator as Compared with Streptokinase for Acute Myocardial Infarction. New England J. of Med. 332(21):1418-1424.
Mark, Daniel B. In press. Medical economics in cardiovascular medicine. In Topol E.J. (ed): Textbook of Cardiovascular Medicine, 2nd Edition. Philadelphia: Lippincott-Raven.
Tunis, Sean R., Jeffrey L. Kang. 2001. Improvements in Medicare Coverage of New Technology. Health Affairs 20(5):83-85.