US regulators of medical devices, and indeed all regulators, must balance two competing needs when it comes to regulation: the desire for a process to be as simple as possible, and the desire for a process to be able to appropriately evaluate a product. If a regulatory process is overly complex, it will stifle the development of new products, harming consumers. Conversely, if the same process is too simplistic, certain products may not be evaluated appropriately, either stifling development or resulting in unfit products coming to market. With this in mind, a new paper from the law group Hyman, Phelps and McNamara (HPM) asks: What sort of balance in the 510(k) substantial equivalence program striking?
The 510(k) program is built on the premise that, in general, if a device has already been approved by FDA, a device that is "substantially equivalent" to the predicate device should, in theory, be safe for use as well. The devices that are cleared through the 510(k) process are still rigorously tested, but generally avoid the costly clinical trials that can be a burden on companies' bottom lines, as well as the substantial premarket application (PMA) fee of $248,000. In comparison, a 510(k) costs less than $5,000 to file.
The program has also been wildly successful, explains Jeffery Shapiro, director at HPM, with approximately 98% of all devices reaching the market via the 510(k) pathway. The pathway has also proven resilient and flexible, Shapiro adds in a paper. "Taken as a whole, the 510(k) program has proven adaptable to a wide variety of devices, and has allowed meaningful premarket review while fostering robust technological innovation. Under the 510(k) program, the device industry has flourished and allowed for increasingly safer and more beneficial device technology, thus providing a practical demonstration that the 510(k) program is working well."
But that doesn't mean the program is without its flaws. It has come under fire during several incidents in recent years involving devices that relied on either unsafe predicates, or deviated sufficiently enough from the predicate devices as to render them unsafe for human use. The most prominent example of all may well be Johnson & Johnson subsidiary DePuy's metal-on-metal hip implants, which relied on several changes, known as a "split predicate" device. Those changes, however, resulted in several unexpected mechanical failures, injuring patients and requiring numerous surgical revisions. In light of those problems, FDA has since required the devices to be approved under a PMA as a Class III (high-risk) medical device.
Are such failures the inevitable cost of an otherwise successful system? No, says Shapiro. "Certainly, there can be improvements," he writes in a position paper, but "the use of substantial equivalence as the standard of premarket review has strengths that have not been well articulated."
Leveraging the Strengths of the 510(k) Process
At present, observes Shapiro, FDA needs to answer six questions about a device before it can be cleared via the 510(k) pathway: Is the subsequent device comparable to the predicate(s); Does the device have the same use(s) as the predicate(s); Is the device technologically similar to the predicate(s); Are there any differences, and are they significant to the subsequent device's safety or efficacy; Are there acceptable methods for testing; and what do the data show?
The multiple aspects of the 510(k) program confer various advantages on the review process and subsequent outcome, Shapiro explains. For example, referencing predicate devices allows FDA's review staff to leverage its previous review experience with that device, both making it easier for regulators to target potential problems and subsequent device manufacturers to identify which features are acceptable to regulators. "FDA may require data to bridge the difference between [features], but the 510(k) program enables both FDA and industry to avoid continually re-inventing the […] wheel," Shapiro writes.
"In short, the open regulatory architecture of the 510(k) program enables FDA and industry to leverage the entire body of FDA's review experience," he continued. "Correspondingly, it enables specific sectors of the device industry to leapfrog through the regulatory process more rapidly than any one firm could on its own, each taking advantage of other firms' clearances as they steadily improve the technology."
This benefit can be leveraged in other ways, as well. Because FDA's prior findings basically show that the essential components and design of a device are safe, subsequent manufacturers can focus on the effects of modifications done to the device and not the device as a whole. This permits a disaggregation of components, and breeds a more efficient regulatory review process, Shapiro observes.
In addition, the process as practiced allows for greater predictability-something not always associated with either regulatory in general or the 510(k) process in particular. Shapiro argues that the 510(k) process is similar in many respects to our system of common law in that each subsequent legal decision builds on the precedence of a former one. As needed, higher courts-much like the directors of the Center for Devices and Radiological Health-may revisit a decision and overturn prior precedence in light of either new information or a new understanding of its impact. This also permits a sort of hierarchy of devices, Shapiro noted.
Potential for Improvement
With those thoughts in mind, as well as a reflection on the history and the problems associated with the 510(k) process, Shapiro presents several ideas for consideration.
The first need, he says, is for more responsibility. "A persistent industry criticism of FDA's implementation of 510(k) review is of its excessive unpredictability, particularly as to what testing data will be required to show substantial equivalence," he writes, adding that this is likely caused by a lack of transparency in 510(k) decisions. Because decisions evolve iteratively, Shapiro continues, if regulatory requirements or decisions are not well-publicized, regulatory predictability and the strength of the 510(k) program can be put at risk.
And the chief source of blame to Shapiro is the "lack of easily searchable information describing the basis for FDA's 510(k) decisions." Shapiro charitably describes FDA's 510(k) database as clunky and difficult to search, and the decision summaries contained within are often difficult to discern or draw conclusions from.
"This issue might seem like small potatoes," adds Shapiro, who goes on to propose a thought experiment in which judges have access to a complete and searchable database of decisions, but litigants and counsel do not, and that some of the decisions that can be found are partially or mostly omitted. "Would this situation not make legal research substantially more difficult? Would it not make judicial decisions wildly more difficult to predict?"
Shapiro answers this hypothetical emphatically in the affirmative, adding that were the 510(k) system actually the one relied upon by the judicial system, the "paucity of information would be considered intolerable."
Shapiro goes on to conclude that the system FDA already uses for in vitro diagnostic devices, under which it publishes full "decision memoranda" regarding its decision-making calculus, would be a strong step in the right direction. Other proposals, such as those to institute a standardized format, "could actually be harmful in eliminating the nuance and detail that pervades so much of substantial equivalence decision-making in the heterogeneous world of devices."
Finally, Shapiro notes that there "should be an up-to-date and publicly searchable 510(k) database consisting of all 510(k) files in full text searchable format (including the original submission, relevant correspondence, clearance letter, cleared indications for use, and FDA's decision memorandum)." While conceding that there are plenty of "obstacles to overcome"-trade secrets being the most obvious-he concludes that, "It is not beyond the capacity of the industry and FDA to jointly solve these problems."
"This more complete transparency would greatly improve the predictability of the 510(k) process, and also would reduce the costs associated with finding predicate devices," Shapiro concludes.
Read his paper here.