A predicate checklist nobody has to follow

Choosing a 510(k) predicate is usually read as a judgment call — find a cleared device close enough in intended use and technology to support a substantial-equivalence argument, and let the comparison tables do the rest. In September 2023, FDA wrote down, in public, what it actually wants that judgment to weigh. The document is still a draft. Three years on, it has not moved.

Four criteria, dated and specific

Best Practices for Selecting a Predicate Device to Support a Premarket Notification [510(k)] Submission recommends a predicate that was cleared using well-established, scientifically accepted methods; that meets or exceeds the safety and performance FDA already expects for its type; that carries no unmitigated use-related or design-related safety issue; and that has no associated design-related recall. FDA also asks for a short narrative in the submission itself, explaining how the sponsor weighed those four factors in picking the predicate it picked — a document FDA cannot yet require, because the guidance that would require it has not been finalized. Neither the eSTAR template nor the Refuse to Accept checklist has a dedicated field for that narrative today; a sponsor who writes one anyway has nowhere built-in to put it but the submission’s own rationale section, attached as supporting text rather than a scored form field.

Parked, not advancing

FDA’s own fiscal-year 2026 guidance agenda sorts every planned CDRH document into three lists — an A-list FDA intends to publish with priority, a B-list it will publish as resources allow, and an Under Construction list for documents it intends to keep developing with no publication date attached. The predicate best-practices guidance sits on the third list, the one with the least committed timeline of the three. Under FDA’s own good-guidance-practices rule, even a finalized version would create no legally enforceable obligation — a guidance states the agency’s current thinking, and a submitter may use a different approach provided it still satisfies the underlying statute. A draft sitting on the under-construction list asks even less of anyone than that, today or next year.

Nothing requires you to run this checklist today. That is exactly what makes today the cheap time to start running it.

The practice

Treat the four criteria as an audit you already have the data for, not a rule you’re waiting on. A proposed predicate’s own clearance record and decision summary speak to the first two factors. The device-recall and adverse-event record speak to the third and fourth — and they are worth checking more than once: a predicate clean at the moment you cite it can pick up a design-related recall or a MAUDE pattern afterward, while your own submission is still pending, or years later when the same predicate gets cited again for a new indication or a line extension on a different clearance entirely. A standing predicate list for your own cleared portfolio, re-checked on a schedule rather than only at the moment of filing, is the version of this habit that survives the guidance staying in draft indefinitely — and it costs nothing to start before FDA ever makes it mandatory. This is regulatory intelligence and method, never regulatory advice about whether a specific predicate you are relying on would hold up against FDA’s own four criteria.