Skip to Main Content
Centers for Medicare & Medicaid Services

View Public Comments for Draft Guidance for the Public, Industry, and CMS Staff Coverage with Evidence Development in the context of coverage decisions

Mertz, Alan

January 25, 2013

VIA electronic delivery

Marilyn Tavenner
Acting Administrator
Centers for Medicare & Medicaid Services
7500 Security Boulevard
Baltimore, MD 21244-1850

RE: Draft Guidance - Coverage with Evidence Development

Dear Ms. Tavenner,

Thank you for the opportunity to submit comments on the 2013 Draft Guidance for the Public, Industry, and Centers for Medicare & Medicaid Services ("CMS") Staff - Coverage with Evidence Development ("CED") in the Context of Coverage Decisions ("the Draft Guidance"). This draft guidance has the potential to guide stakeholders regarding CMS' plans for why, when, and how CED will occur. Our comments highlight issues CMS should clarify in final guidance.

The American Clinical Laboratory Association (ACLA) is the leading national association representing the interests of clinical laboratories. ACLA members include national and regional commercial laboratories, academic institutions, healthcare systems, and companies that are leading innovators in the emerging field of personalized medicine, including but not limited to molecular diagnostics. A complete list of ACLA members can be found on our website,

ACLA previously commented to CMS in response to its November 7, 2011, request for stakeholder comments on experiences with CED. ACLA also commented on the 2006 Draft Guidance implementing the CED policy. The primary issue of concern to ACLA members in 2006 - the potential for CED to delay development and clinical use of new and innovative clinical laboratory tests - persists today.

Indeed, recent actions by CMS and local contractors on coverage and payment for new clinical laboratory tests have amplified those concerns. Given the pace of new discoveries in medicine - which has only accelerated since 2006 due to advances in understanding of genetics - it is critical for innovators and funders of research in clinical laboratory medicine to know what evidence is required for Medicare coverage and payment and to have a well-defined process with identified end points.

Indeed, the Draft Guidance cites the National Bioeconomy Blueprint in describing the goals of CED. The agency's goals include, "...better defining the parameters and guidance for CED so it can be used more widely and effectively as a driver of innovation... and create predictable incentives for innovation..." ACLA firmly supports the notion that CMS' coverage and payment policies have the potential to drive innovation and create incentives. To that end, we offer recommendations on how CED should work in the field of clinical laboratory services.

Recommendations for Draft Guidance

  1. CMS should continue to adhere to the principles and guidelines included in the 2006 CED guidance document.

In its prior CED guidance, CMS included a list of eight principles as well as other guidelines to guide the application of CED. These principles, as well as other guidelines in the 2006 CED guidance document, provided a useful benchmark for all stakeholders and helped ensure that CED would be applied consistently and when appropriate and would not unnecessarily limit access to the optimal diagnostic and therapeutic items and services for patients. In the Draft Guidance, CMS provides a different list of standards of scientific integrity and relevance to the Medicare population and deletes the principles governing the application of CED from the 2006 Guidance, saying that "some" of the 2006 principles are now moot. ACLA strongly disagrees. We believe that all the principles and guidelines of the 2006 Guidance continue to be relevant and urge CMS to continue them in the final 2013 CED guidance that it issues. ACLA believes a number of the 2006 principles are particularly important to include in the final CED guidance.

For example, one of the first principles of CED articulated in the 2006 guidance was that "CED will in general expand access to technologies and treatments for Medicare beneficiaries." Yet to date, CED has been used to restrict coverage of innovative drugs and genetic tests at the cutting edge of medical research (e.g. genetic tests for warfarin, off-label use of cancer drugs.) In the case of warfarin, Medicare's policy has led to private payers' deciding not to cover the genetic test until the Medicare trials are complete. The combined effect of those decisions has resulted in a situation where little data is being collected on the use of the test in "real world" patients. This inability to collect data is a significant barrier to innovation and to patient access.

ACLA also urges CMS to adopt certain principles described in the 2006 guidance document as attributes of Coverage with Appropriateness Determination (CAD) that, in our view, provide a useful approach for when and how to apply CED short of an AHRQ sponsored research study - which would be the only available option should the draft guidance be finalized as written. Those include:

  1. Establishing a process for "interim coverage with study participation" for new technologies that have strong evidence of analytical / clinical validity, and for which additional data can be collected through means other than clinical trials in human subjects (e.g. claims data, registries, and electronic health records; and
  2. Expanding the concept of "coverage with study participation" to include participation in studies that are funded by industry, non-profits such as PCORI, and government agencies such as NIH, rather than solely relying on studies commissioned and funded by AHRQ.

This approach would permit coverage with study participation under a broader range of circumstances than those for which CED has been used to date. This approach would also be well suited for collecting additional evidence for diagnostics, as described below.

  1. CMS' guidance should distinguish the evidence requirements for diagnostics vs. treatments, such as drugs and medical services.

Diagnostic tools are not treatments themselves, but offer the health care system ways to identify those who will benefit as well as to focus on more effective treatment, thereby reducing costs. Diagnostic services can be used to:

  • Improve patient outcomes and reduce the cost of care by helping physicians determine which patients do or do not require more costly, aggressive interventions, and evaluating which physicians are practicing in accordance with evidence-based best practices;

  • Predict the benefits or harms of taking specific medications or other treatments, moving drug treatment away from a "one-size-fits-all" approach to a "right drug for the right patient" or "right dose for the right patient" approach; and

  • Provide patients and physicians increased control over chronic conditions through personalized "real-time" treatment and disease management regimens - yielding rapid results tailored to a patient's unique circumstances.

Based on statute, Medicare pays for items and services that are "reasonable and necessary for the diagnosis OR treatment... [emphasis added.]" ACLA believes that the distinction between diagnosis and treatment is important, that the standards for what is "reasonable and necessary" can - and should - be different for diagnostics than for treatments, and, therefore, the standards for evidentiary requirements should reflect the very real distinctions between the two. Further, CED studies for diagnostics should not be duplicative of FDA or other methodologies.

Given the broad range of applications for clinical laboratory services, assessments of the evidence should take into account these varied applications in deciding what level of evidence is appropriate for any given test. For example, where a test is being used to confirm a diagnosis, less rigorous evidence may be sufficient to demonstrate whether the test is reasonable and necessary. Where significant therapeutic decisions may be made based on the test results, a higher level of evidence may be required. However, in no event should a clinical laboratory test be held to the same evidentiary standard as a treatment that the test may inform.

  1. CED should be used to evaluate clinical laboratory services only when significant evidence gaps exist; the absence of data on health outcomes does not constitute such a gap.

The draft guidance states that CMS has used CED in cases where "... we have had reasonable grounds, based on the available evidence, to question whether improved health outcomes reported in narrower settings would be realized by our beneficiary population..."

ACLA sees several problems with this application of CED. Those include 1) defining what "available evidence" is appropriate to guide the use of CED, particularly for diagnostics, 2) the value of analyzing subgroups (of Medicare beneficiaries, and other subgroups as well,) in an age of molecular medicine, and 3) the suitability of health outcomes as a reasonable standard for evaluating diagnostics. We discuss each of these issues in detail below.

ACLA believes that while requiring information on health outcomes may be appropriate for studying treatments, it is not appropriate in the case of diagnostics. In contrast to the typically direct relationship between a therapy and health outcomes, the relationship between a laboratory test and health outcomes is usually indirect. An example of an indirect outcome is when a diagnostic test is used to rule out a disease or condition.

For this reason, diagnostic test services should be reviewed using a level of evidentiary rigor that considers the role that diagnostics play in patient care. The perceived benefits gained from lengthy, comprehensive evidentiary methodologies must be balanced against the significant costs such methodologies would impose, including barriers or disincentives to medical innovation for costly prospective trials and delayed or denied access to diagnostics that otherwise could have provided valuable diagnostic information to help the clinician avoid negative health outcomes and their associated costs. Physicians want to know if the diagnostic works and is accurate to identify a condition which may have a clinical impact on the patient, which they will evaluate as an integral part of how they practice medicine. Erecting a health outcomes barrier for new diagnostics will have a devastating impact on the incorporation of innovation in diagnostics into clinical practice and severely impact the practice of medicine.

Furthermore, the field of outcomes research is far more advanced with respect to the study of therapeutics, when compared to the study of diagnostics. An international perspective helps to illustrate this point. The National Institute for Clinical Excellence (NICE), part of the British National Health Service, first identified the need for new methods to assess diagnostic tests in 2007. In 2009, the International Society for Pharmacoeconomic Outcomes Research (ISPOR) created a special interest group called the Health Technology Assessment of Medical Devices and Diagnostics Working Group. Interestingly, out of 11,300 ISPOR members worldwide, the Working Group has only 35 active members in 2012.

In summary, health outcome studies should not be a mandatory requirement for CED studies of diagnostics. Applicants should be permitted to demonstrate that a diagnostic is reasonable and necessary using adequate and appropriate surrogates for such studies, for example studies of diagnostics based on retrospective review of laboratory data and research that indicates the link between specific markers and specific disease. The diagnostic should be covered based on demonstrating the presence or absence of the condition. Physicians often want to know the presence of markers that demonstrate clinical variation to assist them in diagnosis, consideration of other clinical factors, and selecting among treatment modalities.

  1. CMS should evaluate a range of data sources, methodologies, and measures in determining whether CED is necessary.

Well accepted standards exist for evaluating most clinical diagnostic testing. For the most part, these standards focus on two necessary and complementary requirements (1) the analytical validity of the test - does it consistently and accurately measure the analyte for which it is testing - and (2) the clinical validity - does it consistently identify the clinical condition associated with the analyte in the patient population for whom the test is intended.

Beyond analytical and clinical validity, an assessment of the utility of a diagnostic test should be informed by evidence of the extent to which the results of the test can influence patient management. Because health outcome is dependent upon treatment and other diagnostic options, evidence requirements for new diagnostics should be tailored to the condition for which the test is designed.

Further, new evidence in the field of clinical laboratory science is being generated at an exponentially faster pace due to the use of high-throughput genetic sequencing and bioinformatics research. Analyzing data from registries and archived specimens are powerful new tools that are uniquely suited to the study of clinical laboratory diagnostics, as opposed to randomized clinical trials. We strongly believe that CED should allow for these surrogates for health outcomes when it comes to diagnostics.

  1. Medicare's use of CED should rely on appropriate methodologies for the topic being studied, and clinical trials in human subjects may not be necessary in the case of diagnostics.

There is a broad spectrum of evidence that could be considered in evaluating diagnostic services. It is rare for a laboratory to conduct prospective randomized clinical trials to show that a clinical laboratory test has clinical utility; this information usually is deduced from other available evidence. While RCTs may be the "gold" standard for some procedures and therapies, they may have significant limitations when applied to most diagnostic situations.

Double-blinded, placebo-controlled RCTs are often infeasible for new clinical laboratory tests, for several reasons.

  • Most personalized medicine tests have populations of potential patients that are too small to recruit enough patients for a statistically-valid trial.

  • Biomarker data are almost always discovered retrospectively, while analyzing results from trials that were not designed to test the relationship between the disease and a biomarker.

  • Running a prospective, placebo-controlled trial of a personalized medicine diagnostic test could be unethical, especially if the group receiving a placebo is known to have a gene that makes them a candidate for a potentially life-extending, life-saving treatment or helps them avoid adverse events.

Certain observational studies can provide useful evidence for clinical validity and clinical utility of laboratory services. Those include researching claims data, electronic medical records, laboratory data, and using patient registries to identify respondents in accordance with HIPAA requirements and appropriate research protocols.

Archived biospecimens can and often are used to perform and validate genetic tests. It is possible to use such samples, consistent with requirements for informed consent and appropriate treatment of patients and patient specimens, to determine whether an individual with a given genetic profile ultimately had a recurrence, or responded to a particular drug or therapy.

Such retrospective reviews of archived specimens and patient outcomes, in lieu of prospective clinical trials, can result in more rapid determination of the utility of a diagnostic procedure, without adversely affecting incentives to develop beneficial new test services.

The standards for conducting outcomes research in a prospective clinical trial involving human subjects are simply not comparable given this new paradigm. Indeed, the advancing field of personalized medicine may, in the long run, turn the everyday practice of medicine into a constant process of patient-centered outcomes research, with each patient being his or her own "N of 1" trial. Recent studies of the whole exome of tumors have shown that every tumor has its own unique biology, based on the patient's inherited genes and the tumor's mutational status.

Essentially, treating each patient constitutes an "N of 1" study, in which the treating physician and the laboratory work together to diagnose different parts of a tumor and determine the appropriate therapeutic choices for the patient. To complicate matters further, the constantly mutating status of cancer cells means that the diagnosis and choice of therapy are both constantly changing. Given these dynamics, establishing a clinical laboratory test's analytical and clinical validity and its influence on patient management should be sufficient to enable coverage of its clinical use.

  1. Medicare should recognize the rapid cycle of evidence generation for diagnostics, and not rely solely upon published peer reviewed studies.

In personalized medicine diagnostics, the pace of methodological innovation is so rapid that the peer-reviewed academic literature is years behind the "state of the science." Sustaining US leadership in personalized medicine diagnostics will require a regulatory and reimbursement environment that is evidence-based and patient-centered, but that also remains nimble and flexible to adapt as new scientific methods become feasible, particularly in areas such as bioinformatics and next-generation gene sequencing.

Clinical use of new clinical laboratory services could be delayed by years if regulators and payers consider only published studies from peer-reviewed journals of prospective randomized clinical trials, or, in the absence of such studies, refuse to accept other evidence of clinical validity and utility. For example, to determine whether a given test can predict recurrence of a condition, the only way to perform a RCT would be to track patients prospectively over a sufficiently long period of time to determine whether the condition returned prior to some set endpoint. It could be many years before the results of such a study would be known, thus delaying unnecessarily the availability of an important diagnostic tool and important information for physicians who might take that information into account in evaluating all the clinical evidence for their patients.

  1. CMS should apply CED within the NCD process only.

The 2006 guidance provided that CED would occur within the National Coverage Determination (NCD) process. CMS does not explicitly confirm this in the Draft Guidance, and statements included in its earlier solicitation for public comment on CED in fact included statements that suggested the agency intended to apply CED in other contexts, such as Local Coverage Decisions.

ACLA strongly urges CMS to adhere to this 2006 principle and to apply CED only within the national NCD process. In order for CED to be effective and achieve its goal of producing evidence that will enable the agency to make informed coverage decisions, CMS will need to rely on input from knowledgeable stakeholders with regard to both review of the available literature as well in design of the CED studies to produce the data the agency needs. The best way to accomplish this is to apply CED within the NCD process, which is well-known, transparent, and open to the public. Furthermore, it removes uncertainty and potential inconsistency regarding beneficiaries' access to a technology in one coverage area while it is being investigated in another.

  1. CMS should clarify the roles of Medicare Administrative Contractors (MACs) in making coverage and payment decisions.

The draft guidance states that, in describing the role of MACs for coverage of items and services, it does not include "an exhaustive list of potential MAC activities in this regard." The draft guidance does not clearly state whether, and if so, in what circumstances, a MAC can apply CED in the context of a local coverage decision. The draft guidance also states that "this policy does not withdraw Medicare coverage for items and services that may be covered according to Local Coverage Determinations (LCDs) or the regulations on category B investigational device exemptions (IDE)..." It also "introduces potential interactions between CMS and FDA in support of CED."

When taken together, ACLA finds these statements to be contradictory and confusing. Medicare coverage of a new technology through CED should be at the discretion of the local MAC unless there is a national policy of non-coverage. CED should be used as a part of the national coverage process, rather than being left to individual MACs to create different, potentially duplicative, and certainly costly requirements for clinical studies and which could result in inconsistent coverage policies. Finally, CED should not be used as a way to add new evidence requirements beyond those required for introducing a new technology onto the market. On the other hand, it is equally important that CMS adopt uniform guidelines and standards for local contractors engaged in making coverage determinations to avoid or minimize inconsistent coverage policies among contractors.

  1. CMS should ensure that review of evidence and decisions about evidence development are clearly explained and transparent.

In addition to applying CED through the NCD process, CMS should make extra efforts to ensure that the analyses of evidence and decisions about evidence development methods leading to the initiation or end of CED studies are clearly defined, explained and transparent before such CED studies are undertaken. The draft guidance takes steps in this direction by setting forth clearly when CMS expects CED to apply and its standards for scientific integrity. But CMS does not provide any guidance on how it plans to determine the evidentiary criteria for a particular application of CED or any basic contours of what such criteria might be. It also does not explain how it will review data generated from CED studies to make coverage determinations. Furthermore, initial research often leads to requests by agencies such as AHRQ for additional research on ever-more data points. Filling in the gaps in the draft guidance to ensure that at the outset all stakeholders know exactly the process CMS intends to follow when applying CED will minimize uncertainty. A clearly defined process will avoid unanticipated costs and delays for innovators, and promote patient access to the most appropriate treatments.

Moreover, CMS should make the entire CED process transparent - from the decision to invoke CED, to the development of the study design and evidentiary standards, to the decision to end a study, and finally to the process for using the new evidence to make a coverage determination. CMS discusses transparency in the draft guidance with regard to publication of the results of CED studies. Yet, typically, by the time clinical literature is published, the opportunity for CMS to benefit from the knowledge and experience of stakeholders will have passed and the research is already dated. The standards and the process must be transparent from the very beginning in order to ensure that CED is applied only for the most appropriate items and services, and in a manner that is fully transparent and will not impede beneficiary access and will generate the data the agency needs to make coverage determinations in a timely fashion.

  1. CMS should ensure that beneficiary access to clinical laboratory services is not interrupted pending the review of CED study data.

The suggestion in the Draft Guidance that coverage under CED could end during the interval between the end of the CED study and the completion of the agency's review of the study data is inconsistent with the whole notion of CED and should be rejected. If the item or service covered under CED was worthy of coverage at all before the study was even complete, there is no logical reason to terminate coverage after the end of the study unless or until the agency determines, after completion of its review of the study data, that the study data does not support continued coverage. If CED is appropriately applied, such instances should be extremely rare.

ACLA appreciates the opportunity to submit these comments. If you have any questions or require additional information, please contact Jen Madsen, Vice President of Policy and Regulatory Affairs, at 202-637-9466 or


Alan Mertz