Image by Gerd Altmann from Pixabay
FDA’s 510(k) process requires the comparison of a candidate device to a similar device that is already legally marketable. The objective of this comparison is to show that the proposed device is at least as safe and effective as the already cleared device. This assumes that the predicate is itself safe and effective. Predicate comparison is generally done without a clinical trial, a situation that sometimes generates heated controversy, especially after a device proves to be problematic.
In a crowded field the question arises of which of several already cleared devices one should choose for comparison. If we believe in constant progress, then the answer should be that the comparison should be made to the best currently available device, however best might be defined. For some attributes, there might be a relatively straightforward parameter that reflects a best performance. For example, fatigue resistance might be a critical parameter that can be measured and compared, assuming in vitro results reflect actual in vivo performance. If there were several devices on the market, they could be studied and ranked in order with respect to their relative fatigue resistance. It would then be nice if a new design were shown to be at least as good as the best. But what if the new device isn’t that good, but it is better than the worst? You could then choose the worst as your predicate or any other that your device is better than.
It might be the case that the worst available device is in fact good enough with respect to real-world performance. In the fatigue example, this would be the case if the worst device did not exhibit actual fatigue failures. This could be the case if the fatigue testing is not predictive of actual use conditions. But then why are we measuring a parameter that doesn’t actually matter? On the other hand, if devices pass the test and undergo field failures anyway, then the test is not helpful.
The existence of a worst device does not, according to FDA, mean that such a device is not safe and effective. This is based on the circular argument that because all legally marketed devices must be safe and effective, any device that is already legally marketed must be safe and effective. This logic ignores our experience with recalls and market removals of once legally marketed devices that are now off the market.
FDA has advanced two ideas for addressing the “at least as good as the worst” approach. One1, suggested in November 2018 by the now departed commissioner Dr. Scott Gottlieb, suggests that newer predicates are to be preferred over older ones. This is based on the apparent assumption that there is always progress in design such that each successively cleared device represents an improvement. This is a nice thought, but there is little evidence that it is always true, and examples where it clearly is false. This proposal included the use of shaming in which the use of a predicate older than 10 years would result in a public listing.
The second FDA initiative, as reflected in the February 2019 Final Guidance Document on a Safety and Performance Based Pathway2, is to replace the comparison-to-a-predicate basis with optional performance standards for selected devices. Based on this guidance, FDA intends to maintain a list of device types appropriate for this methodology, accompanied by guidance documents that identify the performance criteria for each device type, as well as the associated testing methods. This is similar to procedures allowed for some Abbreviated 510(K)s. Devices under this program and their performance criteria have not yet been named.
These future performance requirements may include the use of consensus standards. However, the creation of a performance standard raises questions similar to which predicate should be used. Should the performance requirement be based on the best available existing device, the worst, or something in between? Or should the standard be set to be better than the current best in order to force progress? While FDA has said the performance standard approach would be less burdensome, it is not obvious that this would always be the case, since it is possible that required (or recommended) testing could be more burdensome that what manufacturers might otherwise get away with.
The goal of continuous improvement in medical devices, and associated patient safety, would seem to suggest that new devices should be at least as good as the current best, if not better. Unfortunately, the current system does not ensure that this is the case.
- "Statement from FDA Commissioner Scott Gottlieb, M.D., on how modern predicates can promote innovation and advance safety and effectiveness of medical devices that use 510(k) pathway" https://www.fda.gov/news-events/press-announcements/statement-fda-commissioner-scott-gottlieb-md-how-modern-predicates-can-promote-innovation-and
- "Safety and Performance Based PathwayGuidance for Industry and Food and Drug Administration" https://www.fda.gov/regulatory-information/search-fda-guidance-documents/safety-and-performance-based-pathway?