Picking a predicate that survives review

A 510(k) can satisfy substantial equivalence on paper and still draw a reviewer's skepticism from the first page, because the comparison only holds up as well as the predicate underneath it. FDA's guidance on evaluating substantial equivalence, current since 2014, tests the comparison a sponsor makes once a predicate is already in hand. It says almost nothing about how that predicate should have been chosen in the first place — a gap FDA's draft guidance, Best Practices for Selecting a Predicate Device to Support a Premarket Notification [510(k)] Submission, issued September 6, 2023, was written to close. The draft has sat unfinalized for nearly three years, which would normally mean shelving it. Sponsors report the opposite: reviewers are already asking questions shaped by its four criteria, guidance status notwithstanding.

A comparison test that assumes the predicate is already right

FDA's 2014 guidance, The 510(k) Program: Evaluating Substantial Equivalence in Premarket Notifications, walks a reviewer through the decision tree Section 513(i) of the FD&C Act sets up: same intended use, then either the same technological characteristics as the predicate or characteristics that don't raise different questions of safety and effectiveness. Every step in that tree takes the predicate as a given and tests the comparison against it. Nothing in the 2014 guidance tells a sponsor how to choose among several eligible predicates in the first place — a device type accumulates dozens of cleared entries over the years, and the SE guidance is silent on which of them is the right one to build a new submission against.

Four criteria, still in draft

FDA's September 2023 draft names four things a well-chosen predicate should be able to show: it was cleared based on well-established methods — FDA guidance, FDA-recognized consensus standards, or a qualified Medical Device Development Tool; it meets or exceeds the safety and performance a reviewer would currently expect for that device type; it carries no unmitigated use-related or design-related safety issue; and it hasn't been the subject of a design-related recall. None of the four is a new regulatory standard — the draft is explicit that it isn't creating one — and none of them has been finalized as of this writing. What's changed in practice is which predicates draw follow-up questions during review, well ahead of any finalization giving the criteria formal weight.

“Well-established methods” tests the predicate's own file, not just its outcome

The first criterion is the one most likely to eliminate an otherwise attractive predicate. A device cleared fifteen years ago, before the consensus standard your own device will be tested against existed, wasn't cleared on a well-established method by today's reading — it was cleared on whatever the state of the art was then. Building a new submission's technological-characteristics argument against that older device inherits its dated basis, even though the new device itself might test cleanly against current methods. The criterion asks a sponsor to check the predicate's own testing basis, not just confirm the predicate exists and was cleared.

Chains of predicates move further from the original than a single clearance suggests

A predicate is often itself downstream of an earlier predicate, which was downstream of one before that — a chain industry commentary sometimes calls predicate creep, where each individual comparison is defensible but the chain as a whole has drifted years or decades from the device it traces back to. The 2023 draft's safety-and-performance and recall criteria are aimed partly at this drift: a predicate several generations removed from the original design is worth checking on its own current merits, not credited automatically because something upstream of it once cleared.

Where the choice gets pressure-tested, and where it doesn't

FDA's Refuse-to-Accept review checks that a submission names a predicate and includes the comparison the SE guidance requires — the same presence test that governs every other section of the acceptance checklist. It does not evaluate whether that predicate was the right one to choose. The mechanism built for that question sits upstream of filing entirely: a Pre-Submission built around a specific, answerable question can put a candidate predicate in front of the reviewing division before a sponsor commits a full submission to it. A predicate rationale worked out after acceptance, in response to an additional-information request, is solving a problem a Pre-Sub question could have closed months earlier.

A predicate-selection worksheet built around the four draft criteria — scored against the predicate's own testing basis, not just its clearance date — is previewed in the launch catalog. If FDA finalizes the guidance with different criteria, or your program already weighs a fifth one, the shelf takes that correction directly.