CDRH Releases Preliminary Reports with Recommendations for 510(k) Program

By Carmelina G. Allis –

Since September 2009, the Center for Devices and Radiological Health (“CDRH”) has been reviewing the operation of the 510(k) program and the way CDRH uses science in the decision making process.  This week, the 510(k) Working Group and the Task Force on the Utilization of Science in Regulatory Decision Making each released a preliminary report with a series of recommendations. 

The preliminary reports are now open for public comment.  You can find them here.  Once CDRH has assessed public input, it will announce which recommendations it will adopt, along with projected timelines for doing so. 

It is clear that the proposals would have a significant impact on the process by which industry brings devices to market.  Dr. Shuren’s cover letter summarizing the reports restated the agency’s goals in this review process as looking at ways to foster medical device innovation, enhance regulatory predictability, and improve patient safety.  It is an open question whether all of the proposed changes would actually serve these goals.  Some of the proposals seem likely to increase industry’s regulatory burden without much improvement in patient safety, such as, for example, a requirement that manufacturers provide regular, periodic updates to CDRH listing any device modifications that were implemented without the submission of a new 510(k).

The reports do acknowledge the importance of enhancing innovation.  The twin goals of protecting patients from harm and promoting the development of safe and effective devices are given equal billing.  However, in our view, the majority of the recommendations will result, if implemented, in an increased burden on manufacturers, CDRH reviewers, and the 510(k) regulatory program.  It is not apparent that the recommendations, if implemented, will foster medical device innovation or enhance regulatory predictability.  If anything, these recommendations may add new scientific requirements, new regulatory hurdles, and additional uncertainty to a regulatory process that is already non-transparent and unpredictable.

The reports assert that the recommendations are preliminary and have not been implemented.  However, CDRH staff in recent months appear to already be following some of these recommendations.

Below we discuss the most significant recommendations issued by the 510(k) Working Group:

(1) “CDRH should clarify the meaning of ‘substantial equivalence’ through guidance and training for reviewers, managers, and industry.”

The 510(k) Working Group found that there is confusion both within CDRH and the public at large as to what constitutes the “same” versus a “new” “intended use,” and about when “different technological characteristics” raise “different questions of safety and effectiveness.”  As a remedy, the Working Group first recommends that the concepts of “indication for use” and “intended use” be consolidated into a single term, “intended use.”  (The dual terms “indications for use” and “intended use” have created confusion for years.)

Second, the Working Group recommends that CDRH better train reviewers and managers on how to determine “intended use” based on information available in the 510(k) submission.  The main intent of these proposals is to “reduce inconsistencies in [the] interpretation and application” of the term “intended use.”

Third, the Working Group recommends that CDRH explore the possibility of amending the Federal Food, Drug, and Cosmetic Act (“FDC Act”) to provide the agency authority to consider an off-label use of the proposed device when determining its “intended use.”  The declared intent of this proposed measure is to ensure that the manufacturer of the device does not seek clearance for a use that is not the actual use for which the device was intended to be marketed.

If FDA has sufficient evidence that the manufacturer is going to market the device for an intended use different from that proposed in its 510(k) submission, then the agency would likely deny 510(k) clearance or require that additional information be submitted in support of the off-label use.  This proposal would effectively revert to the days prior to the Food and Drug Administration Modernization Act of 1997 (“FDAMA”) when FDA would allege an “implied” intended use based upon off label capabilities of a device.  Since FDAMA, FDA has been required to accept the manufacturer’s characterization of its intended use, absent a finding by the CDRH Director that there is a reasonable likelihood that the device will be used for an intended use not identified in the proposed labeling for the device and that such use could cause harm.

It is not clear whether the agency would apply this change to a general class of devices or target specific device manufacturers based, for example, on their compliance history or publicly available information on their product development and marketing intentions.  It also raises the question as to whether FDA could unfairly subject a manufacturer to additional unnecessary pre-clearance scrutiny due to the bad acts of manufacturers of devices of the same type.

Fourth, the report alleges that FDA existing guidance does not “fully articulate a clear standard that may be applied consistently by reviewers and managers in determining which ‘technological characteristics’ to consider in their decision making, and how to determine whether such characteristics raise ‘different questions of safety and effectiveness.’”  The Working Group acknowledges that such lack of standard or criteria has resulted in inconsistencies in the 510(k) decision making process.  In fact, FDA’s determination whether technological differences raise different questions of safety or effectiveness has been notoriously subjective.

The report recommends revising existing FDA guidance to ensure conformance with language in the FDC Act regarding “different technological characteristics” and “different questions of safety and effectiveness.”  In addition, the Working Group proposes to revise existing FDA guidance to provide clear criteria for identifying different questions of safety and effectiveness, and to identify a core list of technological changes that generally raise different questions.  The goal of providing greater clarity is laudable, but achieving at will be challenging.  Unfortunately, it seems likely that, unless the revisions are device-specific, the proposed guidance, like existing guidance, will have to be written at a high level of abstraction given the heterogeneity of devices regulated as Class II.  It also seems unlikely that revised guidance could make this determination any less subjective and may result in ad hoc decision making.

(2) “CDRH should explore the development of guidance and regulation to provide greater assurance that any comparison of a new device to a predicate is valid and well-reasoned.”

Currently, all devices with 510(k) clearance are potentially available as predicate devices with no time limit.  The Working Group raised concerns about the continued availability of predicate devices that have either been withdrawn from the market for safety reasons or that have become obsolete due to technological advances.  The Working Group also recognized that the public 510(k) database often does not provide sufficient information to help submitters identify an adequate predicate device – and, according to the report, an adequate predicate device is one which raises no safety and/or effectiveness issues, does not have substandard performance, and has not been withdrawn from the market.  (This is not the definition in the FDC Act.)  The recommendations provided in the report include the development of a guidance document to address when a device should no longer be available for use as a predicate device.

The Working Group believes that rescission authority would be useful to eliminate inappropriate predicate devices.  Although the FDC Act does not state that the agency has 510(k) rescission authority, FDA believes that this authority is implicit in the statute.  The Working Group therefore recommends that CDRH consider issuing a regulation to define the scope, grounds, and procedures for exercising its authority to rescind a 510(k). 

FDA issued a proposed rule in 2003 allowing rescission of 510(k)s.  The proposed rule has languished ever since.  This proposal essentially revives the idea.  The problem is that the FDC Act did not really contemplate rescission of 510(k)s, and the withdrawal of a 510(k) creates a potentially cascading effect on other devices cleared in reliance upon that predicate, and also potentially compromises the regulatory status of devices remaining in the field.  What if the manufacturer has discontinued sale of a device with a rescinded 510(k), but still needs to ship accessories to the installed base?  Would that be lawful?  These kinds of practical issues have the potential to create a great deal of mischief.  It will be interesting to see if the agency can make rescission work without significant statutory changes.

The Working Group report also discusses the need to develop guidance on the use of more than one predicate device to establish equivalence, such as when the proposed device combines the intended use and technological characteristics of multiple devices.  One concern with the use of multiple predicate devices is that an analysis conducted for the report shows a greater mean rate of adverse event reports for those devices that cited more than five predicates in the 510(k) submission, something which the Working Group recommends that CDRH analyze further.  (The data used to support this conclusion are not described in the report.)  Recently, we have seen the agency require that manufacturers identify no more than two predicate devices, often times preferring that only one predicate be identified.

Moreover, the Working Group recommends that CDRH no longer allow the use of “split predicates,” which refers to those submissions where the manufacturer uses one predicate to claim intended use and another predicate to claim technological characteristics.  As stated in the report, “[c]oncerns have been raised that the use of a ‘split predicate’ may not allow for a valid comparison of safety and effectiveness because no such device exists . . . and therefore there is no real-world information about its risks and benefits.”  We believe that the agency has already started to implement this recommendation.  This approach may prove to be a formidable obstacle to technological innovation.  The report does not analyze the impact on innovation of a ban on “split predicates,” and, in the absence of “split predicates,” some devices may be inappropriately placed in Class III, requiring premarket approval.  On the other hand, if FDA improves the de novo process, making it less cumbersome and available more often, then a prohibition against “split predicates” might be more palatable.

(3) “CDRH should reform its implementation of the de novo classification process to provide a practical, risk-based option that affords an appropriate level of review and regulatory control for eligible devices.”

The Working Group recognized that CDRH’s implementation of the de novo classification process has been inefficient and has not been utilized optimally across the Center.  CDRH has averaged only about four de novos per year; the last de novo clearance was over a year ago.  One of the reasons is that the FDC Act requires a full 510(k) review prior to initiating the process, even when it is clear that there is no predicate for the proposed device.  Another challenge has been reviewers’ reluctance to agree to de novo review at the outset of the discussions.  Also, the agency develops device-specific guidance to serve as special controls for each device classified into Class II through the de novo classification process, which is a time-consuming process.

The Working Group recommends that CDRH revise existing guidance to streamline the de novo classification process, and proposes the use of pre-submission meetings to discuss the type of information that should be submitted to facilitate the de novo classification process.  The report also said that the Center should consider establishing a generic set of controls that could serve as baseline special controls for devices classified into Class II through the de novo classification process.

(4) “CDRH should take steps through guidance and regulation to facilitate the efficient submission of high-quality 510(k) device information, in part by better clarifying and more effectively communicating its evidentiary expectations through the creation, via guidance, of a new ‘class IIb’ device subset.”

One of the agency’s concerns is its inability to sometimes determine whether a predicate device identified in a submission is a modification of the original cleared device, therefore raising the question of whether it is a valid predicate.  To address this issue, the Working Group made several recommendations for the agency to have better control of unreported device modifications.

The Working Group recommends that CDRH revise existing guidance to clarify the types of changes to a device that do or do not warrant submission of a new 510(k).  Although this would be valuable, it will be impossible for the guidance document to capture all types of changes for all device types regulated via the 510(k) requirements.  Therefore, it is unclear that a revision to existing guidance will help streamline the 510(k) process, facilitate the submission of high-quality device information, or even enhance the agency’s control over unreported device modifications.  There is also a risk that CDRH will use the occasion to revise guidance to require more device modifications to undergo 510(k) clearance, which could slow iterative innovation and improvement in devices, to the detriment of patients.

Another recommendation suggests that CDRH require each manufacturer to provide regular, periodic updates to CDRH listing any device modifications that were implemented without the submission of a new 510(k).  It is unclear how this would be implemented or in what way submitting these tens of thousands of reports annually would foster medical device innovation, but it is certain to burden manufacturers and increase the workload on CDRH reviewers and managers.

The Working Group further recommends that FDA develop guidance to explain manufacturers how to make adequate, structured, and well-supported predicate comparisons in their 510(k)s.  Moreover, CDRH should explore the possibility of requiring each 510(k) submitter to provide as part of the 510(k), non-proprietary photographs and schematics of the device, as well as samples of the device so that CDRH staff could examine the device hands-on as part of the review of the device itself, or during future reviews in which the device in question is a predicate.  The prospect of release of proprietary information is of considerable concern to industry.

The Working Group also recommends providing additional guidance and training for submitters and reviewers on the appropriate use of consensus standards.  In addition, the Working Group recommends that CDRH consider revising Part 807 regulations to require, as is required for the more burdensome PMA applications, a list and brief description of all scientific information regarding the safety and/or effectiveness of the device reasonably known to the submitter.  While generating these summaries may sound like a minor task, for some devices the process will require substantial effort.

Another recommendation is the proposal to develop guidance to define a subset of Class II devices, called “Class IIb” devices, for which clinical information, manufacturing information, pre-clearance manufacturing site inspections, and/or additional post-market surveillance requirements would likely be required to support a substantial equivalence determination.  In addition to delineating the Class IIb types of devices, the Working Group recommends that the same guidance discuss what type of clinical data would be adequate to support clearance.  The Working Group also recommends that the agency continue its ongoing effort to implement a unique device identification system, and consider using “real world” clinical data, such as data from electronic health record systems, as a requirement for future 510(k) submissions.

These recommendations, if implemented, would not likely foster medical device innovation, as it would subject certain Class II devices to more stringent regulatory requirements.  It is unclear that these would be offsetting benefits in improved effectiveness or safety.  The purpose of the 510(k) process is to allow devices fast access to market.  Requiring the submission of clinical and manufacturing information for 510(k) clearance would defeat the purpose of the 510(k) program, making it into a lengthy and burdensome process, similar to that for PMAs, and seems better calculated to hinder, not foster, device innovation.  The review of manufacturing information currently is only required for a very small subset of devices; the intent here seems to apply this requirement more widely.  A small silver lining would be that companies would at least have greater certainty up front as to whether clinical data will be required.
 
(5) “CDRH should take steps to enhance its internal and public information systems and databases to provide easier access to more complete information about 510(k) devices and previous clearance decisions.”

Some of the most useful recommendations include: developing procedures and guidance for the development and assignment of product codes and developing a publicly searchable database for each cleared device that includes a 510(k) Summary, non-proprietary photographs/schematics of the cleared devices, and information identifying the predicate device and corresponding data in support of the clearance.  The Working Group also recommends developing a standardized electronic template for 510(k) Summaries to ensure they contain all information required by regulation.  Another recommendation involves requiring manufacturers to submit final device labeling post-clearance and periodic labeling updates.  The Working Group also recommends developing guidance and regulations to document 510(k) transfers of ownership, which is currently not required by the agency.

Implementing those changes would enhance a manufacturer’s ability to identify an adequate predicate device and provide accurate and up-to-date information on the intended use and technological characteristics of predicates.  All of these reforms are much overdue.  The current CDRH 510(k) database lacks meaningful data to help manufacturers identify adequate predicates.  For example, most 510(k) Summaries do not contain a detailed description of the cleared device, or a discussion of the data submitted in support of clearance.  Without that information, it is difficult and sometimes impossible to identify a predicate device or to determine the likely data support requirements.  And a lack of information on predicate devices introduces unpredictability and uncertainty to the decision-making process.

(6) “CDRH should enhance training, professional development, and knowledge-sharing among reviewers and managers, in order to support consistent, high-quality 510(k) reviews.”

The Working Group recommends that CDRH explore new avenues to enhance the professional development of its review staff and engage outside experts.  Most of these recommendations are reasonable reforms not likely to be controversial within industry.

A specific proposal is to establish a Center Science Council to serve as a cross-cutting oversight body to facilitate knowledge-sharing across offices.  The report also recommends developing a process for regularly evaluating the third-party program; for enhancing third-party reviewer training programs; and for sharing more information with third-party reviewers.  They currently are at an informational disadvantage compared to CDRH reviewers.

The Task Force also made several recommendations which, for the most part, are general in applicability to all of CDRH’s pre-market review programs.  For example, there is a proposal for CDRH to revise its 2002 “least burdensome” guidance to clarify the Center’s interpretation of the “least burdensome” statutory provisions, evaluate the success of the pre-IDE program and agency-industry interactions during the pre-IDE process, and create a mechanism whereby review offices can assemble an ad hoc team of reviewers from different divisions to accommodate unexpected surges in workload. 

In addition, the Task Force recommends that CDRH develop better data sources, methods, and tools for collecting and analyzing post-market information, and take steps to develop a model to help the agency respond to new scientific information, such as using “Level 1- Immediately in Effect” guidance documents or “Notice to Industry” to address a public health concern or a change in regulatory expectations regarding a particular type of device.  The “Notice to Industry” suggestion highlights one of the fundamental challenges for CDRH.  The concept is commendable.  The problem is that communicating “expectations” after they have been set is not very helpful.  Companies need advance notice of changes, and in at least some instances, the opportunity to comment on the changes.  Learning about “evolving expectations” that already have been set is too late for a company that has completed its 510(k) development program.