How to Collect Data for FDA and CMS in the Same StudyHow to Collect Data for FDA and CMS in the Same Study

A recent guidance document, “Coverage with Evidence Development” (CED), from CMS marks the continuation of a trend toward increased requirements for clinical and economic evidence in support of national coverage decisions.¹ Under the Coverage with Study Participation (CSP) document, CMS can grant provisional Medicare coverage for a new procedure contingent on the development and capture of additional patient data to supplement standard claims data. The goal of this initiative is “to generate data on the utilization and impact of the item or service evaluated in the National Coverage Determination (NCD).”

The rationale for CED is clear: CMS believes that the pivotal clinical studies conducted to achieve FDA regulatory approval do not necessarily provide data that are sufficient as a basis for establishing coverage policy.

For medical device manufacturers, the need to present rigorous and extensive evidence on product performance and patient effect creates a strategic dilemma. Getting to market quickly and at minimal cost is a significant goal for device companies. Implementing trials that are larger and more complex than necessary for regulatory approval may delay product launch, which could decrease return on investment, and could create an opportunity for competitors to gain market share. From this perspective, the uncertain prospect of a future Medicare CSP decision is a small risk compared with the advantages of a streamlined trial that collects only data required for FDA submission. An OEM's thinking is that at least the product is on the market and that it is eligible for Medicare reimbursement.

It may be risky to take that approach, however. Once the CED study is complete, coverage reverts to the pre-NCD status, and there is no guarantee that a new and favorable coverage decision will be forthcoming.

Faced with this dilemma, some medical device companies are adopting a new approach to the design of pivotal clinical studies by structuring the program to meet both FDA and CMS evidentiary requirements. This article examines the FDA and CMS perspectives on evidence from clinical trials and offers suggestions for planning studies that will address the information needs of both agencies.

FDA and CMS: Different Missions, Different Perspectives

Planning a trial that can satisfy both FDA and CMS is challenging because the agencies have fundamentally different missions. FDA is a regulatory agency with the power to control the marketing of medical products. CMS is a purchaser of medical supplies and services with the ability to approve or deny payment. Both agencies derive their powers from deceptively simple statutory authorities.

FDA is required to determine the safety and effectiveness of a device in several ways, as follows:²

CMS operates under an even more concise mandate. The coverage determination process must assess whether an item or service is “reasonable and necessary for the diagnosis or treatment of illness or injury or to improve the functioning of a malformed body member.”³

Each agency has published extensive regulations, issued numerous guidance documents, and established detailed review procedures. As a result, each agency has its own perspective on the sources and quality of information that are necessary to achieve its objectives.

Both agencies rely on a hierarchy of evidence, which is an ordered list of types of evidence for clinical effectiveness. Higher-ranked sources are given more weight than those lower in the ranking. (See the sidebar, “Clinical Evidence Hierarchies.”)

Positioned at or near the top of both evidence hierarchies is the randomized clinical trial (RCT). RCT evidence is credible because the experimental design establishes internal validity (i.e., assurance that the outcome is a function of the variables that are manipulated). The validity of findings is enhanced by measures to minimize the chance of an erroneous finding caused by spurious relationships. Such measures include specifying inclusion and exclusion criteria, random assignments to treatment, and administration of treatment per protocol.

Although RCTs are often viewed as the gold standard for demonstrating the effect of a clinical intervention, relying on this particular study design can be problematic for both FDA and CMS, but for very different reasons.

FDA might prefer evidence from RCTs for ensuring that a device is safe and effective. However, the agency has another directive that can limit its ability to request RCTs. The least burdensome provisions of the FDA Modernization Act of 1997 mandate that

Any clinical data, including one or more well-controlled investigations…for demonstrating a reasonable assurance of device effectiveness shall be specified as a result of a determination…that such data are necessary to establish device effectiveness. (FDA)…shall consider, in consultation with the applicant, the least burdensome appropriate means of evaluating device effectiveness that would have a reasonable likelihood of resulting in approval.

As a result of this least-burdensome requirement, even a complex and invasive Class III device could gain approval based on relatively sparse data, such as those from a small (single-investigator) nonrandomized clinical study.⁴

From CMS's perspective, the main issue is the generalizability of data from the regulatory submission. The question that CMS must ask is: Do the studies conducted to secure FDA approval adequately demonstrate that the product conveys a medical benefit to beneficiaries? If the studies approved by FDA are not RCTs, the answer is usually no. Therefore, CMS must examine many other sources of data when formulating the NCD.⁵

The Devil in the Details: Design Perspectives for RCTs

Key to meeting FDA regulations and to gaining CMS coverage is the RCT. CMS is unlikely to give much weight to a lesser standard of evidence unless the procedure involves incremental improvement to a well-established existing device. And premarket applications submitted to FDA usually require an RCT also. Where the two agencies diverge is in presentation. Design of an RCT involves numerous decisions. These decisions range from determining the definition of the trial patient populations to choosing a selection of investigators.

Design features and study end points are often points of contention. Table I lists characteristics of RCTs and contrasts the FDA and CMS perspectives.

In addition, CMS can view poor design in pivotal trials as reason to give unfavorable NCDs (see Table II). A common mistake is using a comparator (i.e., the device that is used to compare efficacy with a product) that does not correspond to the current standard of care. Study designers should also take care to ensure that the study patient population is representative of Medicare beneficiaries.

Practical Considerations in Study Design

Table I illustrates that designing a trial to satisfy both FDA and CMS is not simple. However, the prospect of accelerating coverage decisions could make such design an attractive strategy for many medical devices. The following five tips can make a double-duty trial a success.

Organize Value Messages Early in the Process. Planning for the trial should begin early in the development cycle. The objective is to gain an understanding of how payers are likely to view the new procedure or device. What value messages will be communicated? Although CMS has stated that cost is not a consideration in its decision making, third-party payers are deeply interested in the potential economics of new therapies, including the following factors:

With an understanding of potential value messages, plans can be made to assemble the evidence needed to support a favorable coverage decision.

Focus on Critical Protocol Issues. Table I describes many potential differences in the CMS and FDA viewpoints on evidence from regulatory clinical trials. From the CMS perspective, two trial design elements are of particular concern. First is the comparator. CMS prefers a comparator that represents the current standard of care. However, product characteristics or FDA preference may dictate other comparators, such as historical controls or sham therapy. For example, FDA's approval of a total artificial heart was based on a clinical trial comparing the survival of patients receiving the device as a bridge to cardiac transplantation with survival among historical controls.⁶

Also, an FDA guidance for trials of percutaneous catheter ablation devices in the treatment of cardiac ablation accepts sham therapy as a comparator. The guidance states that it provides control for a placebo effect of intervention while avoiding potential risks of the ablation procedure.⁷

The second critical issue for CMS is the choice of end points. CMS is looking for sustainable improvement in patient outcomes (e.g., survival, functional status, quality of life) as opposed to end points that only capture clinical parameters (such as changes in blood flow).

Other trial design features that merit close attention are inclusion and exclusion criteria, as well as the duration of follow-up. CMS looks for evidence that the procedure is safe and effective in a Medicare population. Therefore, inclusion of the older age group is essential.

Unfortunately, the number or proportion of older patients that need to be included is unclear. Follow-up studies should extend over a sufficient period of time to demonstrate the device's durability.

Leverage Existing Trial Designs. Data elements that are routinely collected in the case report form, such as medical clinical events that involve a medical resource (e.g., hospital admissions), can be used for both FDA and CMS applications. These events can be assigned to standard billing codes. Common billing codes include those for diagnosis-related groups (DRGs), current procedural terminology (CPT), and ambulatory payment classification (APC). Once these events have been coded, unit costs are assigned based on Medicare national average payment rates or other fee schedules.

The report forms that capture important types of resource utilization include the following details:

The serious adverse-event report is possibly the most important because it documents all hospital admissions during the follow-up period. Hospitalization is typically the major cost driver in a course of treatment. Serious adverse-event reports typically include enough information to assign a DRG code with reasonable accuracy.

Careful attention to the design of these reports can facilitate resource coding. For example, descriptions of procedures should ideally correspond to CPT. Minor additions to an existing report page might improve the precision of later analysis without increasing investigator burden. For example, a form documenting hospital admissions could include a field for the number of ICU or other high-intensity care days. Such days are associated with a substantial increase in cost of the hospital stay.

Include Patient-Reported Outcomes. One of the benefits of innovative medical technology is increased survival. Implantable cardioverter-defibrillators are a case in point. However, many innovative devices have no direct effect on survival but instead offer the patient an improvement in quality of life. Examples include hip and knee replacements that restore mobility, implantable pumps that alleviate pain, neurostimulators that enable patients with Parkinson's disease to perform activities of daily living, and cochlear implants that restore hearing to the profoundly deaf.

Fully documenting the value of these types of technology necessitates measurement of patient-reported outcomes (PROs). FDA recognizes PROs as valid end points and applies increasingly rigorous review criteria to both protocol designs and submitted data.⁸ CMS is concerned about the effect of novel treatments on beneficiary functional status, activities of daily living, and quality of life.

Defining the optimal approach to documenting PROs can be complex. Device makers need to anticipate the effect of a therapy on the patient's symptom, function, and psychological well-being. While many device trials routinely include the MOS Short Form 36 (SF-36) quality-of-life survey, there is no one-size-fits-all patient outcomes metric. The PRO survey forms included in a pivotal trial must be relevant and validated for the particular condition.

Engage CMS and FDA in Dialogue. Once a preliminary trial design is in place, the process of negotiation begins. Manufacturers expect to meet with FDA at least once, and sometimes several times, to negotiate an acceptable protocol. Bringing this early-stage protocol to the CMS coverage policy group for review may not be common practice, but this is an essential step in planning a trial designed for double duty.

CMS may suggest changes in key protocol elements such as comparators or end points. If this happens, the manufacturer must go back to FDA for further discussion. Ideally, FDA and CMS representatives would be brought to the same table, but this is unlikely. For that reason, shuttle diplomacy may be required to establish a protocol acceptable to both agencies.

Trial designers should address whether including CMS concerns in an FDA submission could muddle the data FDA uses to make the basic regulatory approval. In other words, information that is extraneous could complicate FDA regulatory approval. For example, a therapy may be very good at increasing the blood levels of a certain analyte, but when it comes to proving increased survivability, there may be an unclear picture. So, if a single trial includes CMS-type data, FDA could develop concerns.

It's entirely possible that back-and-forth dialogue will reveal that designing a trial that fully meets the evidence needs of both agencies is not feasible. If this happens, device makers should remember the priorities. Achieving regulatory approval is the primary objective, and CMS will not consider coverage for a product that fails to secure FDA approval. In this case, planning must begin for additional trials or a registry that will assemble data required for CMS.

Sidebar: Clinical Evidence Hierarchies

Both FDA and CMS rely on hierarchies for assigning weight to clinical data. For example, in 21 CFR 860, FDA lists the following as its preferred clinical documents in descending order of importance:

CMS employs a similar evidence hierarchy. An example of this is its decision memorandum for NCD’s implantable cardiovertor-defibrillators, issued in June 2003. The NCD offers “a representative list of study designs ranked from most to least methodologically rigorous in their potential ability to minimize systematic bias” as follows:

Note that the randomized controlled trial (or what FDA calls a well-controlled investigation), is at the top of both lists.

Conclusion

Virtually every pivotal study can provide information that will help Medicare and other third-party payers understand the economic and patient outcomes implications of a new procedure. Trials with random assignment to treatment and active standard-of-care controls are most likely to address issues important to Medicare. However, a thorough understanding of the payer perspective and careful planning are prerequisites for designing a study that will address coverage issues while not impeding the path to regulatory approval.

Because clinical trials are inherently conducted in a somewhat artificial environment, there are few pivotal studies that fully satisfy Medicare coverage information needs. Therefore, planning for coverage approval should include supporting studies based on literature synthesis, patient survey research, or economic modeling.

Acknowledgments

Suggestions and comments provided by Barbara Calvert at Abbott Laboratories, Robin Strongin at Polidias Associates, and Bryan Luce of UBC.

Gregory de Lissovoy is a senior research scientist in the Health Care Analytics Group of United BioSource Corporation (UBC).

References

1. “National Coverage Determinations with Data Collection as a Condition of Coverage: Coverage with Evidence Development” [online] (Baltimore, MD, CMS, 12 July 2006 [cited 29 December 2006]); available from Internet: www.cms.hhs.gov/CoverageGenInfo/03_CED.asp.

2. Food Drug & Cosmetic Act: Section 513 [online] (Rockville, MD, FDA, 2000 [cited 29 December 2006]); available from Internet: www.fda.gov/ohrms/dockets/dockets/00n0088/bkg0001.pdf.

3. Social Security Act 1862, “Exclusions from Coverage and Medicare as Secondary Payer Section (a)(1)(A)” [online] (Baltimore, MD, Social Security Administration, 7 December 2006 [cited 29 December 2006]); available from Internet: www.ssa.gov/OP_Home/ssact/title18/1862.htm.

4. “Guidance for Industry and Staff: Premarket Approval Application Filing Review” [online] (Rockville, MD, CDRH, 1 May 2003 [cited 29 December 2006]); available from Internet: www.fda.gov/cdrh/ode/guidance/297.pdf.

5. “Medicare Program: Revised Process for Making Medicare National Coverage Determinations” [online] (Baltimore ,MD, CMS, 26 September 2006 [cited 29 December 2006]); available from Internet: www.cms.hhs.gov/DeterminationProcess/Downloads/FR09262003.pdf.

6. JG Copeland et al., “Cardiac Replacement with a Total Artificial Heart as a Bridge to Transplantation,” New England Journal of Medicine 351, no. 9 (2004): 859–867.

7. “Guidance for Industry and FDA Staff: Clinical Study Designs for Percutaneous Catheter Ablation for Treatment of Atrial Fibrillation” (Rockville, MD, FDA, January 9, 2004).

8. “Guidance for Industry Patient-Reported Outcome Measures: Use in Medical Product Development to Support Labeling Claims, Draft Guidance” [online] (Rockville, MD, FDA, February 2006 [cited 29 December 2006]); available from Internet: www.fda.gov/cber/gdlns/prolbl.pdf.

Copyright ©2007 Medical Device & Diagnostic Industry