|
Evidence-Based Coverage Policy
Many health plans apply evidence-based approaches to coverage decisions. The foundation of such approaches is the systematic review of information about the effectiveness of medical interventions. This paper discusses the principles underlying evidence-based coverage policy and how they are applied by two major programs: the Technology Evaluation Center of the Blue Cross Blue Shield Association and the Medicare Coverage Advisory Committee. Although such policies likely have limited effects on spending, they can help to direct medical resources toward effective care.
Coverage policy is the focal point for conflicts between the desire to speed the adoption of new medical technologies and the need to keep health insurance affordable. The National Institutes of Health (NIH), the National Science Foundation (NSF), and other federal agencies spend billions of dollars annually on biomedical research. This investment is intended to catalyze the invention and dissemination of new forms of medical care. Yet in its role as a purchaser of health services, the government must balance other goals and priorities. When they adopt new technologies, Medicare, Medicaid, and health programs of the Department of Veterans Affairs and the Department of Defense must pay for them. These agencies work within a budget that will not accommodate unrestrained growth in the use of expensive services and technologies. Thus, the federal government makes the development and dissemination of medical innovations a major policy goal and at the same time seeks ways to manage the financial consequences. Private payers balance similar concerns. Their interest is in reimbursing or providing effective medical care, and successful innovation is essential to their mission. There can be little doubt that declining illness and death from strokes and heart disease in developed nations, for example, are attributable in part to preventive and therapeutic health interventions developed and widely adopted over the past few decades.1 But whether innovations are dramatic and complex, like organ transplantation, or routine and simple, like antihypertensive therapy, they can drive spending higher. In recent years much of the increase in U.S. health spending has been attributed to increases in the use of expensive medical care, often in the form of new technologies.2 Pharmaceuticals illustrate the link between technology diffusion and expenditures. The rapid rate of introduction of effective new medications and their success in the market has been accompanied by a 17 percent increase in spending for prescription drugs in 1999 and nearly 19 percent in 2000.3 The anticipated introduction of new biotechnology products may push spending even higher. Other health spending components have also grown, leading to rises in insurance premiums and out-of-pocket costs. According to the U.S. Office of Personnel Management, the average premium for plans in the Federal Employees Health Benefits Program (FEHBP) rose by about 11 percent between 2000 and 2001.4 Employers reported premium increases of 8.3 percent in 2000 and an estimated 11 percent in 2001.5 Furthermore, many private insurance plans shifted more costs onto beneficiaries in the form of higher deductibles, increased copayments for office visits, tiered copays for prescription drugs, and, in some cases, increased copays for hospital admissions. This recent trend contrasts with the period 1990–1997, when workers’ out-of-pocket costs fell.6 Employees and their dependents were not the only Americans to feel the effects of rising costs. Elderly Americans, largely shielded from increases in Medicare costs, saw sharp increases in Medigap premiums among plans that cover prescription drugs. A recent report stated that the premiums for Medigap plans that did not offer a prescription drug benefit rose 15.5 percent between 1998 and 2000 while premiums for plans that covered drugs rose 37 percent.7 When the diffusion of technology makes insurance too expensive, fewer people are able to pay the premiums for even the least costly health plans, and many either remain uninsured or drop insurance coverage. Patients may want the most effective or convenient new drug or the most promising new surgical procedure when they fall ill or are injured, but they lack enthusiasm for the increases in premiums and out-of-pocket payments needed to pay for such technologies.
Health insurance coverage policy is a focal point for these conflicts because coverage is a gateway for new medical technologies. With out insurance reimbursement, many patients would be unwilling and unable to pay for many medical technologies. Thus, manufacturers depend upon insurance reimbursement to create a favorable market for their new products, and providers depend upon insurance to be able to incorporate the new technologies into their practices. Coverage policy, in its broadest sense, is intended to promote value in medical care by using reimbursement to favor the use of effective care and avoid payment for ineffective care. However, coverage policy involves other exclusions that may or may not be related to effectiveness. Specific parameters of coverage are set first by the broad categories of medical products and services that insurance contracts state will be eligible for reimbursement. The legislation creating Medicare, for example, names fifty-five benefit categories. Products and services that fall outside these categories are not eligible for reimbursement: typically, cosmetic services and unproven therapies, outpatient drugs, and most mental health services. Reasonable and necessary. Coverage policy typically focuses on the reimbursable goods and services within the contractually defined categories. Most insurance contracts list covered goods and services in groups rather than individually, because excessive detail can prove unworkable. The rapid introduction of new drugs, devices, and procedures means that any such list will become obsolete if it is not updated frequently. Instead, coverage policy usually consists of a set of procedures for deciding which items are covered, based on broad language about the scope of coverage. For example, a brief statement in Medicare’s authorizing legislation, Title XVIII of the Social Security Act, is the basis for all Medicare coverage policy: "No payment may be made for any expenses incurred for items and services that are not reasonable and necessary for the diagnosis or treatment of illness or injury." This broad language excludes nonmedical goods and services, and what it includes depends in part upon the definition of "illness or injury." A treatment that improves memory would presumably be covered when used for a patient with memory loss that is considered "pathological" but not for one who seeks to boost her "normal" memory. Although subsequent legislation has modified the scope of Medicare coverage, its central principle has never been altered: Covered items and services must be both "reasonable" and "necessary." Although there are many similarities, the criteria for regulatory approval to market a drug or device are not the same. Food and Drug Administration (FDA) approval requires demonstrated safety and efficacy, conditions that must be met before the Centers for Medicare and Medicaid Services (CMS, formerly the Health Care Financing Administration) or commercial insurers would consider a regulated drug or device to be eligible for unrestricted coverage. But an item that is safe and efficacious may not be considered either "reasonable" or "necessary" (perhaps because it offers no advantage over, or is much less attractive than, covered products and services that are used for the same indication). Furthermore, to obtain FDA approval, a manufacturer only needs to establish efficacy for a specific indication, which is often narrowly defined; there may be inadequate data to determine whether coverage is appropriate for an "off-label" indication or for a population that differs from the ones included in preapproval studies. Thus, although the FDA approval process often provides crucial information, it does not obviate the need for coverage decision making. Medical necessity. Lists of services covered by private health insurers have varied levels of detail, but they also tend to rely on broad principles to guide coverage determination. The most popular health plan offered to federal employees, the Blue Cross Blue Shield Service Benefit Plan, adds the following qualification to its list of broadly defined categories of covered and excluded services: "Although we may list a specific service as a benefit, we will not cover it unless we determine it is medically necessary [italics added] to prevent, diagnose, or treat your illness, disease, injury, or condition." Most private plans have a similar medical necessity provision. Usually insurers have no contractual obligation to provide reimbursement when the medical good or service has not been proven to confer a health benefit, either because the data do not support its use or because the absence of data makes it ineligible for coverage under an experimental exclusion. When is there enough evidence about effectiveness to conclude that a medical procedure is no longer experimental and is "medically necessary"? There has never been a uniform definition of medical necessity, but during the postwar period, when the term first entered insurance contracts, it was interpreted to mean medical care that most doctors believed to be appropriate. In the wake of many studies documenting practice variation and high rates of inappropriate care, along with a growing appreciation that many accepted practices were ineffective and even harmful, this interpretation has been supplanted by a general consensus that only interventions that are of demonstrated effectiveness can truly be considered necessary.8 Thus, determining whether an intervention has proven effective for a specific clinical use is now the key to coverage determination. Evidence-based medicine. This is also the central task of evidence-based medicine, a movement that promotes the adoption of medical practices whose effectiveness has been demonstrated in a convincing body of well-designed studies. An important feature of evidence-based medicine is the attempt to rate the quality of evidence and to give greater weight to those studies that have a strong design and are well executed. Formal ratings of the quality of evidence of effectiveness were established by the Canadian Task Force on the Periodic Health Examination and subsequently the U.S. Preventive Services Task Force (USPSTF).9 The U.S. Agency for Healthcare Research and Quality (AHRQ) has established Evidence-Based Practice Centers (EPCs) to promote the evaluation of evidence of clinical effectiveness in areas of substantial interest to professional societies, individual clinicians, payers, and government agencies.10 Many professional societies, such as the American College of Physicians–American Society for Internal Medicine (ACP-ASIM), the American College of Cardiology/American Heart Association, and the American Urological Association, have adopted evidence-based approaches to practice guidelines. The popularity of evidence-based medicine rests in part upon the belief that an evidence-based approach will help improve quality and reduce inappropriate variation in medical care. Evidence-based coverage policy draws upon the same belief in an effort to use health insurance to help promote effective medical care.
Comprehensive assessment of evidence of effectiveness is the central activity of evidence-based coverage policy. It draws upon formal methods for summarizing and integrating information, such as meta-analysis and decision analysis.11 It compiles and analyzes the evidence, to determine whether an intervention improves outcomes that are meaningful to patients, such as freedom from pain, disability, and death. Intermediate measures of health, such as improvements in laboratory values or physiological parameters that a patient does not experience directly, have a limited role. They are considered to be links in the chain between the intervention and changes in survival and perceived health, whose importance depends upon the evidence that an improvement in the intermediate health measure causes an improvement in a final health outcome. Adequacy of evidence. The most crucial—and controversial—question for evidence-based coverage policy concerns the adequacy of evidence. What standards must the body of evidence meet to support a decision to offer coverage? The standard of evidence establishes the burden of proof. One policy option is to cover an intervention unless there is compelling evidence that it is more harmful than beneficial. This amounts to a presumption of efficacy, requiring minimal or no evidence to support a decision to cover and a high evidence standard to reject coverage. Another standard is the "best guess": cover if the preponderance of evidence, whether extensive or meager, suggests that the technology is beneficial. This corresponds to a physician’s approach to choosing a treatment when none of the available options has been studied well. The "best guess" demands little evidence; opinion, as expressed by a panel of physicians or scientific experts, might be enough. In the past, many consensus and guidelines panels used this method, but leading professional societies now use a much more structured and rigorous approach when they draft guidelines or consensus statements regarding the use of medical technologies. Typically they impose the burden of proof on the proponents of a new intervention. This is also the approach used in evidence-based coverage policy: The new technology is not considered efficacious until there is adequate evidence to demonstrate efficacy. Although they recognize and frequently apply evidence-rating schemes such as those of the USPSTF, few organizations that produce guidelines or coverage recommendations have attempted to set down detailed rules about what constitutes "adequate evidence." Judging the adequacy of evidence is often subjective, yet there is little disagreement at the two extremes: when available studies are small and of poor quality, and when very well designed studies exist. Well-designed studies are not always available because they are usually time-consuming and can cost tens of millions of dollars. The incentives to sponsor such studies are strong for manufacturers of drugs and some devices, when the studies are required for regulatory approval for marketing. The large clinical studies submitted in support of applications for FDA approval are responsible for much of the cost of introducing a new medication to the U.S. market.12 According to a 1993 study frequently cited by the pharmaceutical industry, the cost of bringing a new drug to market in 1990 averaged $500 million.13 Devices are also subject to FDA regulation, although their approval processes are less uniform than for drugs and do not always represent a comparable level of scrutiny. Unlike drugs and devices, most medical and surgical procedures are not subject to direct government regulation. For these procedures, coverage policy is the most important incentive to conduct studies of efficacy, since insurance reimbursement is nearly as important for the success of a medical intervention as FDA approval is for a regulated drug or device. Effects of low standards. Adopting a low standard of evidence can reduce the costs of introducing a medical product or service and can shorten the time required to plan, finance, and conduct high-quality studies. Interventions then can be introduced and disseminated more rapidly than with a rigorous standard. But choosing a relatively low standard of evidence would have three major drawbacks. First, ineffective treatments and potentially misleading diagnostic procedures would be disseminated more readily. Belief, rather than proof of effectiveness, would be the basis for reimbursement. Second, having a low standard would limit the incentives to conduct or sponsor studies of efficacy. Useless treatments might never be unmasked, and patients would continue to receive them for many years. Sometimes ineffective treatments will become obsolescent as their lack of effectiveness is discovered or as more attractive alternatives are introduced. However, dissemination of an intervention into routine clinical practice is unlikely to reveal serendipitously that a treatment is ineffective. For example, radical mastectomy was the standard of care for breast cancer for decades. Although it was disfiguring and caused considerable morbidity, surgeons, other physicians, and many patients strongly believed that it was superior to more limited operations. Only after millions of women had been treated in this way did a randomized trial reveal that radical mastectomy was no better at protecting women from cancer than more conservative techniques were. If not for the trial, which led to recommendations that other approaches should be used to treat most early-stage breast cancer, radical mastectomy might still be the standard of care.14 The third drawback of a low standard of evidence is that it is unlikely to direct medical care toward diagnostic procedures, treatments, and other forms of medical care that work, simply because it would lower the barrier for adoption of tests and treatments with little scientific support. Thus, evidence-based coverage policies represent an important component of any strategy to ensure that resources are targeted toward effective medical care.
Implementation of these principles requires answering a series of questions. The first concerns the definition of the specific product or service to be evaluated. It is not difficult to define a specific drug or device. Complex surgical and medical procedures, however, can consist of so many individual components, including factors such as skill that are unique to the treating physician, that the services provided in different settings are not comparable. For such procedures, the data used to support a coverage decision may not be relevant to the services that are provided in other settings. The most important questions arise in evaluating the data. How well designed must a study be to be worth considering? If many studies of varied quality are available, how much weight should be placed on each? The answers to these questions can determine the outcome of a coverage determination process. If well-designed studies are more likely to give results in agreement with one another than with poor studies, a coverage determination process that imposes a low evidence standard can seem to make arbitrary and unreliable decisions. Yet setting a uniformly high evidence standard—for example, considering only well-designed, large, randomized controlled clinical trials—would exclude from coverage broad areas of medical care for which well-designed studies are unavailable. Although some might argue that only the highest-quality evidence—that is, well-designed randomized controlled clinical trials—is sufficient to establish effectiveness, evidence-based coverage processes are usually flexible. They balance the need for rigor against the limitations of the current state of medical knowledge. They recognize that the relevant question is whether there is convincing evidence of effectiveness, rather than whether studies of a particular design are available. Flexibility has characterized two of the most important examples of evidence-based coverage processes, a program led by the Blue Cross Blue Shield Association (BCBSA) and the Medicare Coverage Advisory Committee (MCAC). BCBSA’s Technology Evaluation Center. The Technology Evaluation Center (TEC), a BCBSA program administered in partnership with Kaiser Permanente, is one of the best-established processes for evaluating scientific evidence for medical technologies.15 This program "examines and synthesizes the best existing scientific evidence to determine the safety and efficacy of new medical technologies. The goal is to determine whether these technologies improve health outcomes of the patient, such as length of life, quality of life and functional ability."16 TEC develops detailed evidence reports on about thirty technologies each year. After undergoing review and modification or ratification by the TEC program’s Medical Advisory Panel (MAP), the reports are distributed to subscribing health plans and provider groups. The MAP includes physicians and researchers from academe; appointees of the ACP-ASIM, the American Academy of Family Physicians, the American Academy of Pediatrics, and the American College of Surgeons; and physicians associated with health plans. Although neither the TEC program nor the MAP makes coverage determinations or recommendations regarding coverage, payers use the information TEC provides as part of their coverage deliberations. TEC’s relevance to coverage extends in large part to its application of five criteria "to assess whether a technology improves health outcomes." If all of these criteria are met, the technology is considered effective, based upon the totality of scientific evidence. (1) "The technology must have final approval from the appropriate governmental regulatory bodies." This criterion amounts to a simple determination that the technology has passed the necessary regulatory hurdles to be marketed generally. Care that does not require formal approval by a government agency, such as surgical procedures, automatically passes this hurdle. (2) "The scientific evidence must permit conclusions concerning the effect of the technology on health outcomes." This asks about the adequacy of evidence: Is the scientific evidence base sufficient to determine whether the technology is effective? Effectiveness is interpreted broadly, to mean that the intervention improves any of several health outcomes, such as ability to function, freedom from pain, quality of life, and survival. The next two criteria are only meaningful when criterion 2 is met: (3) "The technology must improve the net health outcome." (4) "The technology must be as beneficial as any established alternatives." Criterion 3 asks whether the technology improves health outcomes, when compared with doing nothing. Criterion 4 asks whether the technology is at least as good as established alternatives. Notably, these criteria mention health outcomes, but not cost. (5) "The improvement must be attainable outside the investigational settings." This criterion addresses potential differences between the settings in which the technology was formally studied and the "real world" of clinical medicine. Usually the first published studies of a new intervention report its use in specialized facilities with highly trained personnel. Other institutions may be unable to match the outcomes of these pioneers, at least early on, and they may never do so when the treated condition is rare. Even for common diseases, only institutions that apply a treatment hundreds of times each year may be able to achieve results equal to those reported in the literature. The TEC criteria have been tested in the course of evaluating hundreds of technologies, and similar criteria are used widely throughout medicine. Many other evaluations of evidence for clinical practice, such as guideline development efforts (for example, the USPSTF), similarly assess both the quality of evidence and the size of the health improvement from the intervention. Rating the quality of evidence is essential, since design shortcomings in studies of medical technologies can lead to spurious findings of effectiveness. Studies with small sample sizes can be vulnerable to both Type I (falsely concluding that an ineffective intervention is effective) and Type II (concluding that an effective technology is ineffective) errors. Criterion 2 does not require that studies meet a specific quality standard, but it would only be met if at least some of the available studies were sufficiently well designed and large to make it unlikely that the results are due to chance alone. In addition, the design would have to be sufficiently free of important biases to ensure that improvements in health outcomes were not falsely attributed to a diagnostic test or treatment that had no effect. Specifically, the evidence should consist of well designed and well conducted investigations published in peer-reviewed journals. The quality of the body of studies and the consistency of the results are considered in evaluating the evidence...The evidence should demonstrate that the technology can measure or alter the physiological changes related to a disease, injury, illness or condition. In addition, there should be evidence or a convincing argument based on established medical facts that such measurement or alteration affects the health outcomes.17 Because the base of scientific evidence is fluid, a technology will often pass that second criterion after having failed it on an initial review. But new studies may not confirm that a promising technology is effective. TEC’s MAP determined in 1996 that there was insufficient evidence to conclude that autologous chondrocyte transplantation, a procedure to treat severe osteoarthritis of the knee, was effective. 18 At that time the principal evidence consisted of a single published report of the outcomes of the treatment in twenty-three patients. Further studies were expected to become available shortly thereafter as part of an FDA agreement with the company responsible for the technology. When the MAP revisited the topic late in the following year, several more studies had become available, but they consisted solely of poorly controlled studies and case series. 19 Fluctuationsin symptoms over time, the difficulty in assuring that patients included in different studies were comparable, and features of the procedure itself led the MAP to conclude that effectiveness could not be assessed by making before-after comparisons of patient outcomes or comparisons of outcomes of different treatments reported in different studies. Thus, the MAP concluded for a second time that the evidence was not adequate. In 1995 the TEC program evaluated lung volume reduction surgery, a treatment for severe emphysema.20 At the time, the supporting studies consisted of reports of overlapping series of patients with severe emphysema treated at a single institution. Their conditions were expected to worsen inexorably with conventional treatment, yet most of them improved after surgery. The MAP concluded that the treatment met the TEC criteria, "when patient selection criteria and perioperative support systems used in the available studies are applied." This conclusion was largely based upon short-term (three-to-twelve-month) follow-up of patient outcomes, and most of the outcomes evaluated were intermediate outcomes, such as pulmonary function tests. Matching the patient selection criteria was critical to the MAP’s conclusion, since there was much uncertainty about whether the reported results would apply to other patients whose emphysema was either more or less severe and who had a different set of comorbid diseases. In 1999, when the TEC program updated its assessment, results from randomized controlled clinical trials of the procedure were not yet available, but new studies reported outcomes in many more patients. They revealed that the procedure was more risky than the early literature suggested and that the short-term improvements in pulmonary function tests and other intermediate outcomes were not reliable predictors of improvements in overall quality of life or mortality. Furthermore, the outcomes of the procedure appeared to vary greatly. It was unclear whether the relatively favorable outcomes observed in case series could be attributed to the selection of patients with better prognoses, or if they were attributable to the treatment. Thus, the MAP concluded on its second review that the evidence was not adequate to draw conclusions about effectiveness.21 Medicare Coverage Advisory Committee. The Health Care Financing Administration (HCFA, now the CMS) established MCAC in 1999 to improve and clarify processes for determining which medical products and services are covered by Medicare. From the outset, this has been a public process. MCAC meetings are public, notices of meetings and copies of meeting materials are made available to all interested parties, and there are many opportunities for public commentary before, during, and after MCAC meetings. MCAC brings together a diverse range of expertise and interests to ensure that evidence evaluations will be based upon a comprehensive assessment of the clinical benefits that result from adoption of the interventions under consideration. MCAC is composed of an executive committee, consisting of a consumer and industry representative and the chairs and cochairs of six panels that evaluate the technologies. The panels are Medical and Surgical Procedures; Drugs, Biologics, and Therapeutics; Laboratory and Diagnostic Services; Diagnostic Imaging; Durable Medical Equipment; and Medical Devices and Prosthetics. The executive committee is responsible for providing guidance to the panels and facilitating coordination among them. The committee’s original charter also gave it the responsibility for reviewing and ratifying panel reports and submitting them to HCFA. When HCFA convened MCAC, it did not give the panels detailed guidance about the procedures they should use to develop their recommendations. Two panels met before the first meeting of the executive committee. The Drugs, Biologics, and Therapeutics Panel evaluated autologous stem-cell transplantation for multiple myeloma in August 1999, and the Laboratory and Diagnostic Services Panel evaluated human tumor assay systems in November 1999. When the executive committee first met in December 1999, it heard the reports of these panels. Members of the executive committee expressed concerns about the lack of a consistent process for determining whether the evidence was sufficient to support a conclusion that the technologies were beneficial. The executive committee declined to ratify the conclusions of the panels, opting instead to develop interim recommendations to guide the panels. These guidelines were intended to ensure that panel deliberations would be relevant and useful to the coverage process, and that the processes would be broadly consistent and predictable. 22 The executive committee believed that recommendations that emerged from such an approach would be fair to all parties, open, and supported by an impartial and complete review of the relevant evidence. The interim guidelines suggest that panels evaluate evidence as part of a two-step process. For the first step, "the Panels must determine whether the scientific evidence is adequate to draw conclusions about the effectiveness of the intervention in routine clinical use in the population of Medicare beneficiaries." This criterion is a standard evaluation of the adequacy of the scientific evidence base, similar in intent to the second criterion used by TEC, but specific to Medicare beneficiaries. The second step assesses the direction and magnitude of benefit. The interim guidelines also state that "evidence from well-designed studies ... must establish how the effectiveness of the new intervention compares to the effectiveness of established services and medical items." Thus, if the evidence is adequate, the panels need to assess whether an intervention is less effective, more effective, or about equally effective when compared with standard medical services. In the year since the executive committee first approved a draft of the interim guidelines on 1 March 2000, MCAC panels held meetings devoted to a variety of topics: treatment of urinary incontinence (12–13 April 2000); electrical stimulation for the treatment of wounds (17 October 2000); sacral nerve stimulation for treatment of urinary incontinence (18 October 2000); and ambulatory blood pressure monitoring (21 February 2001). The executive committee also discussed FDG positron emission tomography on 7 November 2000, in response to requests from Congress and the Department of Health and Human Services (HHS). These deliberations were centered upon evidence reports, which are structured reviews of the literature, and other written materials evaluating the technologies under consideration. In the first application of the interim guidelines, the Medical and Surgical Procedures Panel evaluated biofeedback as a treatment for urinary incontinence. It was asked to answer the following question: "Is the scientific evidence adequate to draw conclusions about the effectiveness of biofeedback as an adjunct to pelvic muscle exercises (PME) in routine clinical use in the Medicare populations for the following three indications: 1) stress incontinence, 2) urge incontinence, and 3) post-prostatectomy incontinence?" The panel unanimously responded that the evidence was inadequate for the second and third indications, and all but two panelists agreed that it was inadequate for the first. Despite the panel’s near unanimity, the process was criticized: Advocates for biofeedback, along with some panel members, stated that the question was ill posed, believing that biofeedback with PME should have been compared with nontreatment, rather than with PME alone. One stated that biofeedback should be covered even if the evidence base was inadequate. The early experience with the interim guidelines made it possible to identify potential problem areas, such as the completeness of the literature distributed to the panels, communication with the public, and the framing of the questions. The MCAC process was refined as panels gained more experience, and the executive committee continued to revise and clarify the interim guidelines. The MCAC process remains relatively new, and MCAC has evaluated less than half as many technologies in its history as TEC considers in a year. Nevertheless, it appears to have overcome many of the problems that it confronted as a new program grappling with contentious issues in a public arena. Choosing technologies to evaluate. Only a subset of candidates for coverage are formally evaluated through either the TEC or MCAC procedures. Although the Blue Cross Blue Shield plans, Kaiser Permanente, other commercial health plans, and Medicare use results of these evidence-based processes, they do not rely on them for all decisions. For example, many treatments for Medicare beneficiaries are administered as part of a hospitalization. Since Medicare reimburses hospitals for the admission rather than for the service provided, such procedures bypass coverage determination. Most Medicare coverage decisions are made by local Medicare carriers, and only about a third of the national coverage decisions that HCFA has issued since MCAC was established have gone through MCAC. These formal processes are used selectively because they require substantial time, resources, and logistical support. Established interventions that have never undergone evaluation are infrequently reviewed in a formal process, and many new interventions receive coverage (or denial of coverage) based on much more limited reviews. The 1999HCFA notice of procedures for making national coverage decisions stated that the MCAC process would generally be used, except when our review requires little or no outside input. Issues that fall into this category are usually those for which the medical and scientific information submitted by the requestor (as well as any additional information available to us) is overwhelmingly in favor of, or against, coverage.23 The small number of technologies evaluated as part of MCAC implies that most technologies considered for national coverage require little outside input. A review of the technologies that TEC and MCAC have chosen to study, however, suggests that additional principles, although they may not be explicit, also influence the selection of technologies for the formal coverage process. The first of these is the potential impact of the technology. An intervention that has large potential effects on health, and is used widely or is expensive, is likely to be chosen over a treatment that appears to have little potential effect. The second principle is uncertainty about the impact of the intervention. As the guidelines imply, a formal process of evidence evaluation adds little when there is either an overwhelming amount of high-quality, consistent evidence or little evidence. Between these extremes, the evidence is often unclear—perhaps the outcome measures are indirect, the number of patients studied is small, the studies are not randomized or have conflicting results, the study population is not the population of interest, or the follow-up is too short. Then structured deliberations informed by a detailed review of the evidence are likely to be most helpful. The third principle is interest in using the intervention. If physicians have strong interest in using a new medical technology, payers need to decide whether to offer reimbursement. Manufacturers, patient advocates, and political figures can also be strong advocates for rapid evaluation of a new technology. The fourth and final principle is novelty. Evidence-based processes tend to evaluate new technologies rather than ones that have been in widespread use for many years. On occasion, older technologies are subject to evaluation. Some of the treatments of urinary incontinence that MCAC’s Medical and Surgical Procedures Panel evaluated, for example, had been used for many years but had not been adopted as part of a national coverage decision. Such instances may be exceptions, in part because coverage policy is likely to be more successful at limiting the diffusion of an ineffective new intervention than reversing the use of an established one. Although the effectiveness of many widely used diagnostic tests and treatments has never been scrutinized closely, withdrawing coverage from them would be difficult to justify without strong evidence that a better alternative was available. Separating evidence evaluation from coverage decisions. Both TEC and MCAC have deliberately separated evidence evaluation and coverage decisions. The TEC process established this precedent in part to ensure that the evaluation of the TEC criteria would consider only the effectiveness of the intervention. TEC and MCAC deliberations are an important input into the decisions of coverage authorities, such as health plans’ medical policy committees. Such authorities are able to consider effects on overall utilization, needs in specific patient populations, the precise clinical situations involved, and the availability of other diagnostic procedures and treatments in local settings. Panels that consist largely of outside experts, such as those of TEC and MCAC, are less suitable for working out the details of coverage than for evaluating evidence. Consistency of coverage decision making. All evidence-based coverage policies are designed to promote consistency in coverage decision making. A uniform standard of evidence is one way to ensure consistency and predictability: If exactly the same kinds of studies are required, and the statistical criteria are the same for all technologies evaluated, there is little doubt about whether an evidence standard will be met. Yet uniform evidence standard can have different consequences in different clinical specialties and for different medical technologies. The lack of rigorous efficacy studies for services that are unregulated or loosely regulated poses a challenge for coverage decision making. Well-designed clinical trials may not be performed, even when the incentives are strong, because in some areas they are difficult to carry out. Double-blinding, ensuring that neither the physician nor the patient knows whether the patient is receiving the intervention being studied, is very difficult for some procedures. Blinding the patient to surgical treatment usually means performing a sham operation, which itself can cause substantial morbidity. Compounding such difficulties is a public goods problem: Who other than the NIH has the incentive to fund a study of an intervention like an operation that cannot be patented, be copyrighted, or otherwise assure that those who fund studies will receive compensation? Thus, many medical procedures never undergo the rigorous scrutiny of a randomized trial. In some clinical areas few innovations would meet a high standard of evidence. Bodies making coverage decisions must grapple with the possibility that few treatments or diagnostic procedures would be available for entire classes of disease. TEC has dealt with this problem by basing its decisions on the availability of convincing evidence, rather than the availability of studies of a particular design. When TEC evaluated multivisceral organ transplants and fetal surgery for prenatally diagnosed sacrococcygeal teratoma, the sample sizes were very small (fewer than ten), and there were no randomized trials. However, the literature suggested that this condition was uniformly fatal in early childhood. A high percentage of fetuses who underwent the operation survived well beyond birth and were apparently cured. The MAP concluded that the survival of surgically treated children who had a disease with a devastating natural history was adequate evidence of efficacy. Costs and coverage decision making. Neither TEC nor MCAC explicitly considers costs. Expensive interventions can be covered under both sets of criteria as long as there is adequate evidence of health benefit, however small. The "Notice of Intent" to draft coverage criteria that HCFA circulated for comment in 2000 assigned a narrow role to costs. Costs would be considered only when the agency decides to cover a new intervention whose health effects are nearly identical to those of another intervention already covered by Medicare. If the technology under consideration did not greatly increase costs to the Medicare program, it would be covered. In all other cases, in which the technology provided at least some added or unique benefit, costs would not be considered. Even this circumscribed role for costs was controversial, and as of this writing the CMS has not announced that it will consider costs in any form in the Medicare coverage determination process. Private plans have not explicitly adopted cost-effectiveness as a criterion either; although TEC has provided cost-effectiveness information on a few occasions, it has stated that such information is separate from the TEC evaluation of effectiveness, and health plans have not asked that such information be made available routinely. This approach is consistent with TEC’s efforts to ensure that its evaluations are focused solely on effectiveness; if it also considered costs, the program might be perceived as a component of cost containment efforts rather than as a vehicle to improve care. Indeed, when the coverage process ignores costs, it is unclear whether or how evidence-based coverage policy affects spending. An evidence-based process raises the bar for new technologies by requiring studies that demonstrate effectiveness. However, slowing the adoption of a new technology does not necessarily lower costs to health plans. The products and services used in lieu of a technology for which evidence is lacking can be more costly. The cost to the payer that results from a coverage decision is simply the reimbursement that it pays. Medicare and most commercial health plans have separate processes for deciding what to cover and what the reimbursement level will be. Medicare reimbursement rules are complex. For inpatient care, Medicare pays professional fees on a fee-for-service basis, while hospital fees are set by a diagnosis-related group (DRG)–based prospective payment for the entire admission. Unless a new DRG is established for the intervention, hospitals providing it will not receive unique reimbursement. If the intervention costs the hospital $20,000 but the DRG-based reimbursement for the condition is $8,000, the hospital, not the CMS, bears the $12,000 difference. Reimbursement may increase in future years if the technology is widely adopted despite the loss to the hospital, since DRG payments are updated to reflect changes in the costs that hospitals report for caring for patients in that DRG, but hospitals have strong disincentives to provide a costly technology to Medicare beneficiaries. Setting reimbursement levels in the outpatient setting has been more complex, but it will become similar to the inpatient setting when Medicare completes its planned transition to prospective payment for outpatient procedures. Although reimbursement levels can be set so that they correspond to the benefits produced by a good or service, many other factors—such as costs of producing the good or service and general market pricing—are principally used when setting reimbursement. Consequently, the abilities to apply standards for coverage and to set reimbursement can influence payers’ costs but are far from being comprehensive tools for cost containment.
Nearly all health insurance plans have a formal coverage determination process. Yet their approaches vary, and little is known about the consequences of alternative approaches to coverage determination. The speed of adoption of new technologies, the use of medical services, and the quality and outcomes of care provided to enrollees could all be affected by coverage policy. Speed of adoption. An evidence-based process can slow adoption of a technology in several ways. The lengthiest delay is from the time that it takes to conduct high-quality studies to establish effectiveness. The evidence assessment process itself takes varied amounts of time; a cursory process of reviewing evidence need not be time-consuming. However, most evidence-based coverage processes seek to be comprehensive and rigorous in their evidence reviews. Reports that summarize the evidence, particularly in areas in which there are many potentially relevant studies, take many researcher-hours to complete and usually take several weeks, even months, to prepare. Giving interested parties opportunities to participate in an evidence review process prolongs the process further. MCAC’s deliberations are public and require adequate notice of meetings and involvement of numerous interested parties. A small committee could scan a handful of studies and reach a decision about evidence adequacy in an afternoon, but such a process would likely omit important studies, could fail to assure the representation of diverse viewpoints, and could lack credibility and consistency. There are inherent limits to the speed of a process that includes a high-quality evidence review. The CMS, under pressure to reach national coverage decisions quickly, is striving to balance speed and the integrity of the coverage determination process. Quality of care. Evidence-based coverage policy can help to improve the quality of care by promoting better studies of the effectiveness of diagnostic procedures and treatments and by using reimbursement to favor effective care. Although evidence-based coverage decisions delay the adoption of treatments and diagnostic procedures until adequate scientific evidence is available, they also make it more likely that good studies will be conducted. Revenues to a firm producing a medical device or instrument will be much greater if it is covered, so they will have incentives to sponsor or carry out the needed studies. Incentives to study a medical intervention or, for the patients, to participate in a trial are much weaker if the intervention will be covered in the absence of a trial. Patient recruitment into randomized controlled clinical trials of high-dose chemotherapy for breast cancer, for example, became far more difficult after several health plans decided to cover the procedure. When the trials were completed, they did not demonstrate improved survival, a finding much less positive than earlier, less rigorously designed studies yielded. Such studies not only establish whether an intervention is effective; they can help establish which groups of patients are most likely to benefit, improve our understanding of how they should be combined with other interventions, and quantify benefits more precisely.
Rapidly evolving technology places a heavy burden on any systematic approach to evaluating medical interventions. Many technologies need to be evaluated, and reevaluation will often be necessary, since new evidence accumulates even for stable technologies. Expectations will rise as dramatically new technologies are introduced, many with claims of unique effectiveness. Recent advances in genomics and proteomics, to give two prominent examples, fuel such hopes today. Scientific advances may well turn into medical breakthroughs. If they do, the delays inherent in an evidence-based coverage process may seem to be little more than frustrating obstacles to improved health care. Arguably, however, rapid innovation in medical care heightens the importance of processes that can determine which interventions work, in which patient populations. With rapid innovation, the potential for overuse of services increases. Health insurance inherently encourages overuse, or "moral hazard," since enrollees do not bear the full cost of the services they consume.24 A fully rational and well-informed patient whose insurance had a 20 percent co payment would seek to consume health services up to the point at which every additional dollar spent provided only $0.20 of benefit. Coverage policy is one of many instruments that insurers use to help limit moral hazard. It is complementary to provider incentives (under managed care) and increased patient cost sharing. However, as long as insurance pays a substantial part of the costs of care, cost sharing will not be enough to eliminate moral hazard. Thus, coverage policy is likely to remain a crucial component of any health care financing system. Cost containment is not the most important reason to adopt evidence-based coverage policy, however. The promotion of effective care is. Evidence-based coverage policy sends a clear signal about which innovations are considered effective, a form of validation that is particularly important for interventions that are not subject to regulatory review. Although legislators are likely to be wary of any health care reform legislation that includes comprehensive and detailed lists of covered services, processes to determine which services are covered remain central to current health reform proposals, and it is likely that the success of any health care financing system will depend upon its processes for determining coverage. Coverage policies that are evidence-based are likely to be more credible and gain broader acceptance than existing alternatives. Unless manifestly better policies are developed, or far greater cost sharing renders coverage policy less important, evidence-based coverage policy will continue to be used. Evidence-based coverage decision making may evolve, and debates about the adequacy of evidence will continue, yet the basic principles of evidence evaluation for coverage decision making are likely to be as relevant in the future as they are now. In the face of rising expenditures, cost containment is likely to return as one of the greatest challenges to the U.S. health care system, and coverage policy will undoubtedly be part of the response. It is likely to find its greatest value, however, as a vehicle for the improvement of the quality of care. We can only expect that if the fruits of biotechnology are as plentiful and attractive as many hope, they will readily meet a high standard of proof of effectiveness, and that evidence-based coverage decision making will help ensure that they are used as effectively as possible.
Alan Garber is a staff physician with the Department of Veterans Affairs, and is professor and director of both the Center for Health Policy and the Center for Primary Care and Outcomes Research at Stanford University in Palo Alto, California. He is a member of the Blue Cross Blue Shield Association’s Medical Advisory Panel and chair of the Medical and Surgical Procedures Panel of the Medicare Coverage Advisory Committee. This research was supported in part by the Homer Laughlin Fund and by Robert Wood Johnson Foundation Grant no. 039396. Naomi Aronson, Linda Bergthold, Sean Tunis, Victor Fuchs, and Steven Sheingold made helpful comments. The views expressed here are those of the author and may not be shared by these individuals, the Blue Cross Blue Shield Association, the Centers for Medicare and Medicaid Services, or the Department of Veterans Affairs.
This article has been cited by other articles:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
The Role Of Coverage...
Principles Of Evidence-Based...
