On September 6th, 2023, the FDA released three draft guidances with the stated of goal of “…Part of Ongoing Efforts to Modernize the Premarket Notification [510(k)] Program”. I can not encourage industry and regulatory professionals enough to review these guidances and make comments through the Federal Register comment process that is available at each guidance tab before December 6th, 2023 (90 day cutoff of comment period). There is a lot to digest in these guidances, so this commentary is an effort to try and break down some major implications of these guidances as I can see them, starting with the draft FDA guidance titled “Best Practices for Selecting a Predicate Device to Support a Premarket Notification [510(k)] Submission“.
It is worth noting what the FDA states as their goals / intent with this guidance as noted in the introduction:
“FDA developed this document to provide guidance to industry and FDA staff about best practices in selecting a predicate device for premarket notification [510(k)] submissions. Specifically, this guidance recommends four (4) best practices to employ when selecting a predicate device used to support a 510(k) submission. The recommendations provided in this guidance are not intended to propose any changes to applicable statutory and regulatory standards, such as how FDA evaluates substantial equivalence, or the applicable requirements, including the requirement for valid scientific evidence. FDA developed this guidance to improve the predictability, consistency, and transparency of the 510(k) premarket review process. This guidance and associated recommendations are consistent with and are intended to be used in conjunction with the FDA guidance “The 510(k) Program: Evaluating Substantial Equivalence in Premarket Notifications [510(k)]” (hereinafter, 510(k) Program Guidance) and other relevant FDA guidances on 510(k) submissions.”
We will come back to elements of these key stated goals later, but these are worth calling out for reasons that will become clear in further review. From a historical standpoint, it is worth noting we’ve seen this type of effort from FDA before, with the April 2018 “Medical Device Safety Action Plan: Protecting Patients, Promoting Public Health“, which then evolved into a FDA “Safety Action Plan”, culminating in November 2018 when FDA announced “transformative new steps to modernize FDA’s 510(k) program to advance the review of the safety and effectiveness of medical devices“. Among these described “steps to modernize” the 510(k) program was a somewhat shocking recommendation, in that “the Agency considered making public on its website those cleared devices that demonstrated substantial equivalence to older predicate devices. FDA also considered focusing on predicates that were more than ten (10) years old as a starting point, so the public was made aware of those technologies. FDA’s goal in focusing on older predicates was to encourage manufacturers to continually offer patients devices with the latest improvements and advances.” The implication from this FDA proposal to “modernize” the 510(k) program was that there was some kind of age of 510(k) clearance aspect to legally and properly 510(k) cleared devices via the 510(k) Act (which dates back to 1976), suggesting that once 510(k) clearances were 10 years of older, that they may have some less favorable status to be used as a predicate in the process of a 510(k) clearance pursuance. This would obviously be a moving target as time moves on, so as an example in the year 2023 clearances from 2013 would be acceptable, but next year 2013 clearances would not.
Understandably, industry was not a fan of this suggestion by FDA to “modernize” things in such a fashion, which on its face seemed completely arbitrary. The comment period was opened, which saw some very lively exchanges from numerous impacted parties. Some of this commentary illuminated that this effort by FDA to “modernize” smacked of going beyond their legislatively granted authority via the 510(k) Act, but more importantly, it functionally made no real practical sense in the effort to “protect public health” or “promote safety”. The FDA, in their wisdom, accepted the commentary that pointed out that this age approach was not, as originally believed truly helpful as FDA stated:
“FDA reviewed all comments submitted to the docket and acknowledges that the initial proposal of focusing only on older predicates may not optimally promote safer and more effective devices. For example, if selecting a predicate for an implant, older devices may potentially have long-term safety and effectiveness data that establishes a history of safe and effective use. Conversely, when selecting a predicate for a device that includes software, a more recently cleared device could include modern safety features due to rapid technological advances that affect cybersecurity, interoperability, and modern software architectures.”
From a practical standpoint, I always scour to find the best possible predicates for my 510(k) clearance clients based on their desired indications for use language and technological factors, and if it so happens such clearances are more recent, as long as their aren’t other factors that make the newer predicate less desirable, I will advise a client to use that newer one (or ones where applicable). This is no different than my contemporary colleagues in this industry. But from a professional standpoint, I am deeply concerned whenever there is any arbitrary suggestion that a perfectly valid predicate, duly reviewed and cleared by FDA personnel, is somehow “too old” to be considered as a usable predicate. Some FDA concerns may hinge on what testing an older predicate required and/or performed, which is a valid discussion in the process of a 510(k) submission review, but I find this is best served in the actual discussion with FDA lead reviewer and their team members, as opposed to being told in a guidance in blanket manner what is required to select a predicate. We will get into this testing issue later in this discussion, but as a preview it appears FDA has not given up on this age concept, but is pursuing it by other means couched in a “best practices” effort, stating:
“After considering the docket comments, FDA believes that it may be more appropriate to modernize the 510(k) process with respect to the use of predicate devices by focusing on utilizing best practices when selecting a predicate device rather than just their age. Therefore, FDA is issuing this draft guidance to propose ways to encourage the use of best practices when selecting a predicate device. ”
We’ve examined FDA intent as stated and some historical context on previous efforts to “modernize” the 510(k) program, let’s look at the actual recommendations by FDA in this September 6, 2023 draft guidance. First off, a new step has been requested by FDA in “When considering the selection of predicate devices during 510(k) submission preparation, submitters should consider the list of legally marketed devices that they believe have the same intended use as the subject device and when any differences in technological characteristics do not raise different questions about safety and effectiveness, hereafter referred to as a “valid predicate device.”” Practically speaking, FDA is asking for the submitter to generate a list of identical predicates (if that is indeed possible), which has never been a requirement in prior 510(k) efforts, but this request for a list of identical predicates is stated by FDA later in guidance. My professional approach (and that of my colleagues) has always been to select the best predicate and then use that as basis of a 510(k) submission, not generated a list of matches, which seems burdensome and is counter to the least burdensome principle. The more concerning term in FDA’s recommendations, is that of the valid predicate device. In my professional experience, there are only more desirable and less desirable predicates for whatever 510(k) I am pursuing for a client, but I never say a duly 510(k) cleared device is not “valid”, because that suggests it is somehow fundamentally deficient. This is a dangerous concept for the FDA to operationalize unilaterally, particularly because that appears beyond the scope of the Congressional authority granted to FDA by the 510(k) Act. This use of valid predicate device will be repeated a lot, but I wanted to bring its implications to the fore, as the suggestion that there are “valid” predicates suggests the inverse, in that some duly 510(k) cleared devices are somehow “invalid” predicates.
The first of the four best practices described by FDA as “Predicate devices cleared using well established methods” is described as the following:
“FDA recommends selecting a valid predicate device that was cleared using well-established methods. These methods include those from a currently FDA-recognized voluntary consensus standard, an FDA guidance document, a qualified medical device development tool (MDDT), or a widely available and accepted method published in the public domain or scientific literature for the context of use, or found acceptable through the submitter’s own previous premarket submission. FDA recommends prioritizing predicate devices with methods developed within a consensus environment, and those subject to public comment or peer review.”
As I stated above, these are all features I would evaluate on behalf of a 510(k) submission client when looking at predicate. In most, but not all cases, a FDA product code has one or more identified consensus standards and/or associated guidances, but there could be situations where no such information exists for a product code. The MDDT tools are less prolific with only 14 published to date, so this applicability is limited. Trying to find a method “published in the public domain or scientific literature for the context of use” could prove exceedingly difficult depending on the product, particularly if a newer technology that is amenable to the 510(k) pathway by acceptable predicate usage. In a nutshell, this first stated “best practice” appears to be violating the least burdensome rule for the submitter of a 510(k), which should be an individualized process on its face that is addressed in the actual 510(k) review, and not part of a “pre-screening” process driven by the “valid predicate device” concept. At its core, this first stated “best practice” appears to also be penalizing the use of older 510(k) predicates, as their clearance may have pre-dated certain standards, guidances, MDDTs and/or methods published in public domain or scientific literature. This penalty on older predicate would appear to run counter to what FDA stated “they learned” in the public comment process in 2019, but “packaging” this age penalty in a different way.
The second of the four best practices described by FDA as “Predicate devices meet or exceed expected safety and performance” is described as the following:
“FDA considers it a best practice to select a valid predicate device that continues to perform safely and as intended by the manufacturer during use in its intended environment of use whenever possible. FDA recommends selecting a valid predicate device after considering how any reported medical device-related adverse events, malfunctions, or deaths may have a role in the safety and effectiveness of the device”
Further, the FDA states that once the submitter has prepared the aforementioned list of valid predicate devices, to search the MAUDE, MDR and Medsun databases searching for instances of reported medical device incidents related to the selected predicates, and ideally select a predicate that has no reported incidents at this databases wherever possible. At least the FDA does give some leeway on this in recognizing “If another valid predicate device is not available, FDA recommends that the submitter describe in the 510(k) submission how the subject device mitigates the known concerns with the predicate device used to support the 510(k) submission.” This gives an avenue for acknowledging these reports and trying to address the issues noted therein, but at a certain level this seems to ignore the critical aspects of risk management that is part of every device development, which at a high level explores and documents the risks that a product may present and methods as how best to mitigate them acceptably. I always counsel clients at the front end that risk management is a big part of FDA regulated device development, a 510(k) submission, and quite frankly, continuing obligations on the part of any device company marketing a device. Risk management documentation pursuant to FDA requirements on the part of a 510(k) submission are always present and are one of the most frequently discussed items by FDA review team to best address risk matters. I always am prepared, and appreciate, the FDA’s perspectives on these risk discussions and strive to find practical means to address FDA needs and the needs of the client.
But once again, with this second best practice, FDA is requiring additional burdensome steps on the part of the applicant, and encouraging at all costs the submitter selecting a device with no MAUDE, MDR or Medsun reports. Statistically (but not exclusively), these “no report identified” devices would likely come from newer cleared devices, because older cleared devices with a longer use history have higher odds of resulting in some kind of MAUDE, MDR or Medsun report. Mind you any such reports are not necessarily a sign of a fundamentally deficient device, there are a whole host of reasons for these reports to occur, and frankly there are varying quality of these reports, as it is widely available to public, which as a policy is a good thing. But a varying quality report example would be one with minimal detail, but all efforts by device company to contact individual(s) for additional information are not answered, leaving a mark on the device company with no ability to gather additional important context and details. This is the reality of the use of medical device products in the market, particularly with longer use history and widespread use patterns, in that MAUDE, MDR or Medsun reports can and do occur. What is more important are the resultant steps by a medical device company when any MAUDE, MDR or Medsun reports occur, because this shows their complaint handling and corrective / preventive action (CAPA) processes are working to rectify issues, as opposed to suggesting that any such incidence of reports is grounds for a predicate not meeting the “best practice” as a “valid predicate”. Once again, at its core, this second stated “best practice” appears to continue the effort to penalize, discourage or outright “not allow” the use of older 510(k) cleared devices from being used as predicates for new 510(k) submissions.
The third of the four best practices described by FDA as “Predicate devices without unmitigated userelated or design related safety issues” is described as the following:
“FDA recommends selecting a valid predicate device that does not have unmitigated use-related or design-related safety issues, including consideration of emerging signals or safety communications.”
The FDA goes on to recommend that Medical Device Safety and CBER Safety & Availability (Biologics) databases be queried to locate such use related and design related safety issues and exclude as “best practice” any devices as a “valid predicate” those that have any such reports “wherever possible”. Once again, FDA is requiring additional burdensome steps on the part of the applicant. This is a matter I professionally feel best suited in the robust risk management process noted above for any submission I work up, which doesn’t make a previously cleared device “less valid”, but it is incumbent on the submitter to show how their product can address and mitigate any risks related to such a predicate if it was indeed cited. As expected, these “no report identified” devices would likely come from newer cleared devices, because older cleared devices with a longer use history have higher odds of resulting in some kind of report. This is not excusing any device company from their risk management, complaint handling and CAPA process obligations, but this seems a return to revive the FDA 2018 predicate age approach.
The last of the four best practices described by FDA as “Predicate devices without an associated design related recall” is described as the following:
“FDA recommends selecting a valid predicate device that has not been subject to a design-related recall…Design-related recalls can indicate a fundamental flaw with the design of the device as cleared and commercially distributed…”
In this case, FDA is recommending a search of the Medical Device Recall database to query predicates for design related recalls, as “FDA considers it a best practice to select a valid predicate device that is not associated with a design-related recall whenever possible.” Professionally speaking, this has always been my process to try to select predicates with no recalls of any kind, including design related recalls. But this has been my effort to reduce the perception of selecting a “risky” predicate device, which is different than what appears to be heading towards an absolute FDA requirement in this draft guidance from FDA. This once again ignores the rigorous risk management methods that companies should and do employ in their development, 510(k) submission and ongoing obligations once devices are legally marketed. Not surprisingly, “no design recall” devices would likely come from newer cleared devices, because older cleared devices with a longer use history have higher odds of resulting in some kind of recall.
What are some of the practical implications of these recommendations? First off, as clearly suggested, the age of a predicate is once again implied based on all four of the “best practice” recommendations, which is a bit unsettling to say the least from a professional standpoint that the FDA would appear to be returning to an effort they stated in 2019 was not ideal, but taking this approach from different directions but the same similar goal. Second, what happens if a viable predicate is identified per this guidance parameters in good faith, but any one or a combination of the 4 “best practices” outlined in this guidance is triggered during a submitted 510(k) review process? While this seems unlikely, a 510(k) submission can take a significant period of time, particularly depending on additional information requests, so if a new consensus standard / FDA guidance / MDDT issues that FDA deems applicable to cited predicate, but wasn’t on radar at time of submission, what is the submitter to do? Per the FDA’s request to “make a list”, that list would almost certainly be from the same product code / regulation number for cited predicates, so these new consensus standard / FDA guidance / MDDT issues would apply across the board. Similarly, if any one or a combination of MAUDE / MDR / Medsun reports, Medical Device Safety / CBER Safety & Availability (Biologics) reports or a Medical Device Recall related to a design issue occurs tied to your cited predicate, what is one to do? Per the “make a list” request, any one of these events would like impact your entire list (if that is indeed practical to create). Does FDA have a means to address these kind of events, or does one have to start the whole process over? While one could say this is unlikely, when you have essentially 4 new FDA “requirements” that cover a broad spectrum, this is entirely possible that one or more of these could trigger. Returning to the least burdensome principle, these 4 best practice recommendations do level what could be a significant burden to the submitter, particularly if events trigger mid-submission process.
These are just some of the unintended consequences I’m not sure have been evaluated by FDA in producing this draft guidance. I would never advocate this for any client, but I could see some device companies interpreting this guidance to say the FDA are leveling significant penalties in submitting MAUDE / MDR / Medsun reports for any future 510(k) submissions, as well as penalties to initiating any kind of design related recall, so it would be best to not avoid reporting and/or recalls at all costs, particularly if the company planned future follow 510(k) submission based on their own clearance. Again, this is not what I would advise anyone to do, ever, but I think we all know unintended consequences are real risks that can stem from inappropriately constructed guidances. In a real sense, this attempt is running counter to goals of FDA to “protect public health” or “promote safety”, as the FDA is disincentivizing device makers from following and pursuing what should be primary goals and obligations as required by CFR 820 of device manufacturers.
At a certain level, I applaud the FDA in trying to communicate methods to best select predicates for a 510(k) submission to the public and industry. I think all of this came from a safety focused intent, which is always a laudable goal in any arena, but their practical impacts may likely run counter to what they are trying to achieve. Furthermore, these 4 best practice recommendations are too limiting, and run the real risk, if this guidance is implemented as a final guidance, of effectively removing what could be viable predicates from the 510(k) clearance corpus for future submitters use. I am not clear if this is consistent with current FDA authorities granted by Congress, as this in a sense would take duly 510(k) cleared devices and create a split class of “valid” predicates and “invalid” predicates. While some may say “but the instance of this would be small”, if you add up these 4 best practice recommendations the numbers could grow markedly. And I can’t help but worry once a predicate is deemed “invalid” by this guidance is accepted as final, would there be additional penalties that could be leveled at those duly cleared devices? Could those devices be removed from the market essentially with this guidance as a pretext? But most of all, what I find the most troubling is that this guidance is ignoring the processes that should be baked into any quality 510(k) submission in terms of risk management considerations by industry, and in some way diminish the professionals at FDA who work very hard to review and process 510(k) submissions towards clearance. It suggests that the FDA reviewers and their colleagues can’t adequately make clearance calls based on their own decision making processes while working with submitters. With all of this in mind, for all parties concerned, I hope that the FDA will limit any usage of this draft guidance by FDA reviewers, reconsider this guidance dramatically if it goes to final, or withdraw this guidance entirely.
