|
Evidence, Appropriateness, And Technology Assessment In Cardiology: A Case Study Of Computed Tomography
As the volume, complexity, and cost of new medical technology increase, the need for evaluating benefits and risks becomes increasingly important. Once the formal requirements for Food and Drug Administration approval and insurance coverage are met, however, few systematic criteria are applied to ensure patient benefit. A more systematic policy approach regarding new technologies is needed, with input from balanced groups reviewing evidence of clinical outcomes data to determine patient benefit. This paper examines cardiac computed tomography angiography as a case study; it proposes procedures designed to ensure that the benefits of new technologies justify their costs.
TECHNOLOGY IS A MAJOR DRIVER of total health care spending.1 Evidence-based technology assessment—that is, using study data showing which patients will have a net benefit from a drug or procedure—is a powerful way to ensure value for health care dollars. Evidence-based assessment’s importance is increasingly recognized, as health care spending exceeded 15 percent of U.S. gross domestic product (GDP) in 2004 ($1.9 trillion) and is expected to be as much as 20 percent of GDP by 2015.2 As the number and cost of medical procedures increase, along with expanding indications for use, health care expenses related to technology will accelerate even more rapidly. Unfortunately, evidence-based review of new technology is rarely applied properly, which often results in high spending for unproven procedures. This paper examines how the normal processes of adopting a new technology enable this to occur, looking at the role of technology in health care and how and what type of evidence is used by administrative agencies, insurers, and professional societies. In particular, I analyze the current evidence regarding cardiac computed tomography angiography (CTA) and how such evidence relates to current coverage policies. CTA presents a particularly useful case example for technology assessment because, like many complex, expensive new medical procedures, it is undergoing rapid adoption despite a relatively limited evidence base. The improvements in speed and processing with the CTA scanners allow higher-quality images to be obtained without the problem of motion artifacts. CTA offers the potential for accurate, noninvasive visualization of the coronary arteries. However, the clear benefits of CTA are not yet apparent.
Certain conditions are generally required before widespread adoption of a new medical technology can take place. First, of course, the Food and Drug Administration (FDA) must approve all new drugs and devices. Second, major insurers, including the Centers for Medicare and Medicaid Services (CMS), must agree to pay for the new technology. Finally, there must be physician and patient demand. FDA approval. Evidence evaluation formally begins during FDA review. New drugs and devices must be deemed "safe and effective" to gain FDA approval. Thus, clinical trials are designed to prove safety and efficacy but often do not show specific patient benefits in the clinical setting.3 Spurred in part by the Prescription Drug User Fee Act (PDUFA, first enacted in 1992 and revised in 1997 and 2002) and by pressure from industry and patient advocacy groups to speed approval of new drugs and devices, the FDA often encourages the use of biomarkers, or "surrogate outcomes," consistent with its Critical Path Initiative. In cardiology, the FDA’s encouragement is evidenced by its cosponsorship of the Fourth Annual Cardiovascular Biomarkers and Surrogate Endpoints Symposium in 2006. This symposium focused on biomarkers as tools to address the residual burden of cardiovascular risk and their role in drug development. Surrogate outcomes are appealing because endpoints can be reached with less time and expense than actual clinical outcomes data, which could require years of study and a large sample. However, reliance on surrogate outcomes can be misleading, with far-reaching consequences, if the surrogates have not been sufficiently validated.4 As it happens, in cardiology there are few reliable biomarkers and surrogate markers that have been clinically validated. Some intermediate outcomes, such as lower blood pressure and lower cholesterol, have been established as markers for clinical cardiovascular benefit. However, as Robert Temple of the FDA’s Center for Drug Evaluation and Research has stated, "In many cardiovascular diseases, surrogate and intermediate end points have not proved reliable predictors of outcome, the most striking examples being the failure of effects of antiarrhythmic drugs on VPB [ventricular premature beat] rates to improve survival and the similar failure of inotropic and vasodilator drugs with effects on exercise ability to improve heart failure outcome."5 Thus, to the extent that FDA approval relies on unproven surrogate markers, the process is inherently flawed. Insurance coverage. Following FDA approval, the next step for growth of a new technology is insurance coverage. Private insurers make their own coverage decisions but might be strongly influenced by Medicare coverage decisions. Medicare coverage is based on whether the treatment is "reasonable and necessary."6 No statutory definition of "reasonable and necessary" is provided, however, and what this phrase means, or should mean, has been frequently debated. In particular, this subjective standard is silent with regard to the need for evidence of improvement in quality or length of life. Quality of "evidence." Although "evidence" of benefit might include anecdotes, small case studies, and observational data, the widely accepted "gold standard" is a randomized controlled trial (RCT) using clinical outcomes. Medicare’s General Methodological Principles of Design, ranked from most to least methodologically rigorous in their potential to minimize systematic bias, include, in this order, RCTs, nonrandomized controlled trials, prospective cohort studies, retrospective case control studies, cross-sectional studies, surveillance studies, consecutive case series, and single case reports.7 Although RCTs might not be practical for every new drug or procedure, there have been unfortunate consequences of relying on something less. For example, prior to 1998, nearly fifty observational studies, epidemiological data, and laboratory data showed that hormone replacement therapy (HRT) had beneficial effects on the vasculature and lipids and was associated with reduced cardiovascular disease (CVD) in women.8 There had been a randomized trial using biomarker endpoints, as well as observational studies with clinical outcomes, supporting this view. Not until the first randomized trial with clinical outcomes, however, was it discovered that HRT did not decrease the risk of CVD in women.9 Evidence and professional guidelines. Cardiology’s leading professional societies, the American College of Cardiology (ACC) and American Heart Association (AHA), have codified their procedures for developing guidelines, performance measures, and clinical data standards. Guidelines generally are directed toward medical professionals and seek to define when and how to use relatively well established devices or procedures. The ACC and AHA grade the quality of the evidence leading to a particular guideline recommendation, using a simple ABC system: "A" for multiple RCTs, "B" for a single RCT or for observational data, and "C" for expert opinion. The ACC/AHA guidelines process provides a good framework for assessing medical technologies and for noting the quality of available evidence. Guidelines, however, are inherently limited; they are not designed to address larger policy issues such as cost/benefit analysis or appropriate use of a new technology. Moreover, guidelines are developed only for selected devices and procedures as the professional societies direct. Newer and less widespread procedures might not be the subject of guidelines until years after their introduction, if ever. Role of appropriateness criteria. The recent growth in cardiac imaging technology has led to renewed interest in a methodology to define appropriateness. Appropriateness seeks to provide recommendations on when the use of a technology or therapy is indicated by determining if benefits exceed risks. Ideally, criteria for appropriateness include cost-effectiveness and a risk-benefit analysis of available alternatives. The criteria should be simple, reliable, valid, and transparent.10 Appropriateness criteria can be thought of as a type of quality measure that considers under- and overuse of technologies. The appropriateness approach evaluates beneficial uses as well as risks and includes a middle-ground in which uncertainty of risks and benefits results in uncertainty on appropriate usage. Most importantly, appropriateness criteria, if implemented properly, improve health care delivery by ensuring a systematic review of evidence for a new technology’s benefit to patients. Moreover, if appropriateness criteria can be assessed early enough, they may be used to guide coverage decisions. The ACC recently established an Appropriateness Criteria Working Group, which has proposed a method to evaluate appropriateness "in the absence of ideal evidence for judging the appropriateness of interventions."11 The first few ACC appropriateness panels have addressed cardiac imaging, in light of the recent rapid growth in its use and the fact that heart disease is a leading U.S. cause of death. (The annual medical cost of CVD is an astonishing $403.1 billion.)12
Cardiac imaging of all types has grown 26 percent per year for the past ten years.13 By 2004, the total cost of imaging services either reimbursed by health insurers or paid for directly by patients came to nearly $100 billion.14 Some of the greatest growth has been in the use of computed tomography (CT) and magnetic resonance imaging (MRI), particularly for cardiac applications.15 The volume of all CT procedures (not counting head scans) per Medicare beneficiary increased 82 percent from 1999 to 2003.16 By 2002, on average, one of every four Medicare beneficiaries received at least one cardiac imaging test.17 Similar trends exist for privately insured patients. This explosive growth in imaging has been attributed to multiple factors, several of which are unrelated to improvements in image quality. In addition to technological advances, there has been a rise in entrepreneurial activity by physicians, increased testing attributed to defensive medicine, and increased patient demand in response to media attention and direct-to-consumer advertising.18 For example, Oprah Winfrey extolled the virtues of CTA on her popular television show, Matt Lauer touted CTA on the Today Show, and a Time magazine cover pictured a CTA image emblazoned with the headline, "How to stop a heart attack before it happens." (The Time headline was misleading; even CTA proponents do not suggest that there are data showing that CTA can prevent a heart attack.) CTA provides several potential advantages. The procedure provides high-resolution, dramatic color images of the coronary arteries, and it costs less and is less invasive than coronary angiography, the established procedure for coronary artery stenosis. However, it also poses major potential risks. Radiation exposure greatly exceeds that of coronary angiography, from 4 to 21.4 milliSieverts (mSv).19 In fact, a conservative estimate of 8 mSv for a typical multislice CT scan is 400 times the radiation dose of one chest x-ray. This radiation dose places CT scans at an intermediate (1–10 mSv) to moderate (10 mSv) level of risk under international guidelines—a risk level for which the corresponding benefit should be "moderate" to "substantial."20 Like coronary angiography, CTA also presents the risk of renal damage from the use of nephrotoxic contrast agents and of complications from the use of medicines to slow the heart rate to obtain a usable image. Despite the risks, the potential benefits described above, along with media attention and the revenue potential, have driven a rapid escalation in the use and overall cost of CTA. The most important question, though, is whether CTA leads to better patient care than the alternatives. On this question, despite its widespread adoption, the evidence is surprisingly paltry. Current evidence on CTA. In May 2006 the Duke Evidence-based Practice Center (EPC) reviewed the data on all CTA studies in a formal technology assessment for use by the Medicare Coverage Advisory Committee (MCAC).21 The Duke investigators noted that there were limited data consisting only of small, single-center studies. They also revealed that these studies suffered from selection bias. In particular, most of the studies excluded segments that could not be evaluated and patients with poor-quality images from analysis, inflating accuracy estimates. The investigators concluded that "studies conducted to date primarily fall into the ‘proof of concept’ category with study patients having a high pre-test probability of coronary artery disease. Future work will need to examine these tests in larger, less selected populations representing the clinical setting in which they are actually expected to be used."22 The first multicenter study of multislice CT (MSCT) was published in July 2006. This study found a high number of cases that could not be evaluated and a high false-positive rate, leading the authors to state that MSCT’s "routine implementation in clinical practice is not justified."23 CMS evidence review. The CMS convened the MCAC, an expert advisory panel, in May 2006 to review the current state of evidence on noninvasive imaging techniques (CTA and MRI) for diagnosing coronary artery disease (CAD) and to offer advice to the CMS. Curiously, in contrast to most MCAC meetings, no national coverage decision was under consideration. The discussion during the MCAC meeting focused mostly on CTA, with little discussion of cardiac MRI. (The author was a member of the MCAC panel.) At the public hearing, proponents of cardiac CTA testified, among other things, that CTA would decrease cardiac catheterization. They were enthusiastic about the newest sixty-four-slice scanners and presented some of the images the technology can produce. However, the proponents offered few data to support their claims. For example, no data were presented showing a reduction in cardiac catheterizations. Regarding the scanners, only six studies, involving only 397 patients, have been published. None of these studies included any data on patient benefit. Attendees and MCAC members expressed concerns about the lack of outcomes data for CTA, the lack of data on the appropriate patient population for CTA, the high incidence of false-positive results, and comparisons between sixteen-slice and sixty-four-slice technology. Overall, the MCAC panel found insufficient evidence showing benefits of using noninvasive imaging to diagnose CAD and, as a result, voted "a lack of confidence" in the evidence.24 As of this writing, the CMS has not acted upon the MCAC findings, and it is unclear how the CMS will use them. Thus, Medicare coverage of CTA has been left to local coverage decisions (LCDs) by the states’ various Medicare carriers. Local Medicare coverage of CTA. Eight Medicare carriers covering twenty-four states have finalized their LCDs, granting reimbursement for CTA. Five Medicare carriers in another thirteen states have drafted coverage determinations that soon will go into effect to reimburse for CTA. Noridian Administrative Services, a Medicare carrier for twelve more states, just released its coverage article for CTA and expects a draft LCD by the first quarter of 2007. Thus, as of this writing, forty-nine states cover (or soon will cover) CTA—less than a year after the Category III codes went into effect.25 In other words, despite the lack of any outcomes data, the limitations of studies to date, and the lack of confidence by the national MCAC, CTA soon will be reimbursed by Medicare in virtually all of the United States.26 Private coverage of CTA. Like MCAC, many private insurers have not found evidence of benefit of CTA in independent evidence reviews. For example, the BlueCross BlueShield Association (BCBSA) Technology Evaluation Center (TEC), Aetna, and WellPoint (UniCare) consider CTA to be "investigational" and not medically necessary, and they have denied coverage for its uses in evaluating CAD.27 These insurers have expressed several concerns regarding the methodology of the few studies that exist. The BCBSA evaluation center flatly called for more data: "Clinical trials comparing patients undergoing CTA as part of their diagnostic workup compared to patients not undergoing CTA may be required to demonstrate improved patient outcomes."28 WellPoint noted that there was, "most importantly, inability to translate diagnostic performance of CTA to expected effects on management and health outcomes."29 Most private insurers do not now cover CTA for diagnosis of CAD, in contrast to the more liberal local Medicare coverage for CTA. Current debate over CTA evidence and outcomes data. In cardiology, data regarding clinical outcomes—that is, nonfatal myocardial infarction (MI) or cardiovascular death—provide the most relevant evidence of benefit or lack thereof. However, there is a dispute over whether imaging tests should be supported by clinical outcomes data. An imaging test is only valuable, of course, to the extent that it leads to improved patient care. Proponents of CTA argue that if the image provides more-detailed information, it is inherently more valuable. Because of the lack of clinical outcomes data, these proponents essentially rely on the images themselves as the evidence of benefit, highlighting their superior resolution and detail. Moreover, they argue that awaiting clinical outcomes data (which can take years to develop) would irresponsibly deny patients the benefits of the improved CTA technology. The problem with these arguments is that there is no reason to believe that improved images alone lead to better patient care, in the absence of outcomes data. Moreover, insisting that the health care system adopt (and pay for) the new technology now merely fosters a continuation of a fundamental flaw in our health care delivery system. The "pay now, benefits might follow" approach helps explain why the United States, while spending more per capita on health care ($4,178) than any other nation, ranks only thirty-seventh in health system performance.30 This fact alone suggests that our health care delivery system has failed spectacularly to provide value for our medical spending. Under our present system, once a new technology is in use, particularly one such as CTA that requires major capital expenditure and specialized training, the financial and psychological investment itself promotes use. When a medical group buys a scanner, they must use it and bill for it. Moreover, widespread adoption and insurance coverage of a new technology is, as a practical matter, irreversible. The best way to ensure that we do not misallocate health care resources is to adhere to evidence-based review prior to, and even after, widespread adoption. It is important not to delay possibly life-saving interventions unnecessarily. However, this does not often apply, and the decision can be made case by case. CTA is not a life-or-death procedure; it is a diagnostic tool. There are alternative diagnostic methods available, and there has been no demonstration that CTA will lead to any benefit in outcomes.
Several entities seek to develop evidence of benefit regarding medical technologies, including private insurers, Medicare, professional societies, and independent groups. Unlike the European Union countries, the United States has no nationally coordinated policy on health technology assessment.31 This decentralized approach is partly attributable to the patchwork nature of the U.S. health care system and the long tradition of leaving patient care decisions solely in the hands of physicians. One tangible consequence of this decentralization is that the United States lacks widespread clinical electronic medical record (EMR) systems, also hindering data gathering for technology assessment. Experts suggest that a centralized process that produces a unified medical policy is best equipped to grapple with tensions among quality, innovation, and value.32 The broad principles of evidence-based medicine have become standard in health care. However, adequate data are not available for most everyday health care decisions. Thus, payers, purchasers, and policymakers have begun seeking new ways to create and use clinical research.33 Because it is difficult to demand evidence for technologies after they have been in use for a long time, such initiatives seek to provide an evidence base for newer technologies. One prime example of this effort is the CMS’s Coverage with Evidence Development (CED) initiative. CED has been described as a way to reduce the logjam between innovation and evidence-based coverage policy by permitting coverage of new technology that does not necessarily meet the "reasonable and necessary" criteria while accumulating data for the evidence base.34 This creative approach encourages (and, at least temporarily, rewards) innovation while also collecting data. CED has value, however, only to the extent that the CMS is willing to discontinue coverage when the accumulated evidence shows insufficient benefit. CED also addresses a major limitation on the data generally available to the CMS: The evidence usually is based on younger, healthier patients than those covered by Medicare. The CMS expects CED to allow the agency to get "into the business of medical-evidence development" by promoting the expansion of clinical trials of drugs and devices to the elderly. The CMS also seeks to include, through its CED program, a patient registry component, to provide a database of ongoing clinical use and results, which can be accessible by the research community and possibly others. In the case of cardiac imaging, data registries could help inform the choice of tests. There are multiple tests for cardiac imaging—CT, nuclear scan, MRI, and coronary angiography—all of which can be performed on one patient and separately paid for. Although there may be instances where such multiple imaging is helpful, it is not always in the best interest of the patient (because of the radiation, time, anxiety, or increased chance of a false-positive) or the system (expense). Properly used, data registries also can overcome economic barriers to more elaborate clinical trials, because they are much less costly than RCT. The CED approach is one strategy to establish a clinical research agenda oriented to decisionmakers and can channel public and professional demand for new technology into a mechanism to improve a weak evidence base. It also helps focus discussion on improving evidence with key stakeholders, including product developers, patient advocates, payers, and clinicians. However, the success of such a program depends entirely on follow-up and procedures to adequately assess the accumulated data, and to adjust approval and reimbursement policies accordingly. It remains to be seen whether the CMS can implement CED in this way. The U.S. patchwork system of evidence evaluation is inefficient at best and too often inadequate. Clinical outcomes showing benefit are not required before approval. Often, there is no follow-up, and the data used for approval and coverage decisions are insufficient and lack independence. To evolve toward a more systematic approach, we must first acknowledge the importance of evidence-based health policy and then incorporate into approval and coverage decisions an evidence-based approach that includes reevaluation over time. Rarely have decisions involving technology assessment been made by a balanced group with representation from all interested constituencies—physicians, industry, payers, government, and policymakers. Such a group could seek consensus on technology assessment and identify priorities for evidence gathering. To that end, the Center for Medical Technology Policy (http://www.cmtpnet.org) is working to provide a neutral forum where health care decisionmakers, stakeholders, and experts can work together to design real-world studies on emerging medical technologies. The most direct way to change the ad hoc approach of our current system would be to formally include evidence-based decision making in the coverage process. This change would ensure that use of new therapy and technology is tied to evidence of clinical benefit, resulting in a value-based health care system. Other approaches that do not consider the value of health care services in terms of quality or length of life can be very expensive without advancing the goal of improved health and health care. To avoid paying for new technologies with unproven benefit, the U.S. health care system must incorporate evidence-based procedures that are developed by balanced groups assessing risks and benefits.
Rita Redberg (redberg{at}medicine.ucsf.edu) is a professor of medicine in the Division of Cardiology at the University of California, San Francisco. The author thanks David Larson for his review of the manuscript. This research was supported by the Robert Wood Johnson Health Policy Fellowship and the Flight Attendant Medical Research Institute.
This article has been cited by other articles:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
The Regulatory Process And...
Cardiac Computed Tomography...
