Predicate device

Under §513(i) of the FD&C Act, a new device is "substantially equivalent" — and therefore clearable through 510(k) — when FDA finds that, compared to a legally marketed predicate, it has (i) the same intended use AND either (ii)(a) the same technological characteristics, or (ii)(b) different technological characteristics where (1) the differences do not raise different questions of safety and effectiveness, and (2) the information submitted demonstrates the device is as safe and effective as the predicate.

What counts as "legally marketed"? Four sources qualify: (1) cleared via 510(k); (2) reclassified down from Class III to Class II under §513(f)(3); (3) granted via De Novo under §513(f)(2); (4) grandfathered — in commercial distribution before May 28, 1976 and never required a PMA. A predicate must not have been recalled with a determination that it is not safe and effective.

FDA's 2023 draft guidance "Best Practices for Selecting a Predicate Device" sharply changed the expected predicate-selection rationale. Sponsors are now expected to choose predicates that:

• Were cleared without unresolved use-related or design-related safety issues (no associated recalls, no FDA safety communications, no high-rate adverse-event profile).
• Have well-characterised performance data — ideally including any consensus-standard testing that is now expected for the device type.
• Were cleared relatively recently — FDA strongly discourages cascading equivalence to very old predicates whose performance evidence is sparse by modern expectations.
• Are demonstrably comparable in intended use AND technological characteristics — not just nominally similar.

A single predicate is the strongest argument. A "reference device" (cited for a specific characteristic — e.g. a materials choice or a software feature — without claiming equivalence to it as a whole) is permitted; "split predicates" (claiming intended use from one device and technology from another) are not.

If the intended use differs from the predicate, the device cannot be 510(k)-cleared on the basis of that predicate — full stop. "Intended use" is the general purpose; "indications for use" is the specific patient population, anatomical site and clinical condition. The labelling must support the intended-use claim with performance evidence appropriate to the population and setting.

Common intended-use traps include: claims of long-term use vs short-term predicate; paediatric use vs adult predicate; implantable claim vs external predicate; sterile-field use vs general-clinical predicate. Any of those forces a different predicate or a De Novo / PMA pathway.

If technological characteristics are the same — materials, design, energy source, performance specifications, sterilisation method, etc. — the substantial-equivalence argument is straightforward and the performance data required is correspondingly modest.

If technological characteristics differ, the sponsor must demonstrate that the differences do not raise different questions of safety and effectiveness, AND must provide performance data showing the new device is at least as safe and effective as the predicate for each affected characteristic. Typical evidence includes:

• Bench performance testing against consensus standards (IEC 60601-1, ISO 11135 sterilisation, ASTM F1980 accelerated aging, etc.).
• Biocompatibility testing per ISO 10993 if materials or material processing differ.
• Electrical safety / EMC if electronic.
• Software documentation (IEC 62304 + FDA software-guidance content) if software-controlled.
• Animal data, simulated-use, or clinical data when bench testing alone cannot answer the residual question.
• Human-factors validation when use-related differences exist.

Reviewers follow FDA's published SE flow-chart for every 510(k):

1. Step 1 — Is the predicate legally marketed and not recalled for safety reasons? No → NSE.
2. Step 2 — Same intended use as predicate? No → NSE.
3. Step 3 — Same technological characteristics? Yes → go to Step 5. No → Step 4.
4. Step 4 — Do the differences raise different questions of safety and effectiveness? Yes → NSE. No → Step 5.
5. Step 5 — Are performance data adequate to demonstrate substantial equivalence? Yes → SE (clearance). No → NSE.

Not-Substantially-Equivalent (NSE) determinations route the device automatically to Class III by default. From there the sponsor either pursues PMA, requests Risk-Based Classification under §513(f)(2) (the De Novo pathway), or files a §513(g) request for classification information.

• Cascading equivalence to a 1995 predicate whose performance evidence is sparse — FDA pushes back.
• Predicate with an unresolved recall or safety communication — disqualifying.
• Split predicate (intended use from device A, technology from device B) — explicitly not permitted.
• Claiming "same intended use" when the indications for use are materially broader.
• Treating a reference device as a predicate — they serve different purposes.
• Omitting consensus-standard testing that has become expected since the predicate's clearance.
• No predicate-selection rationale — the 2023 draft guidance puts that rationale in the spotlight.

If after a careful search no predicate matches both intended use and core technological characteristics — common for genuinely novel devices, novel SaMD with new algorithmic claims, or novel combination products — pursue De Novo classification under §513(f)(2) instead of contorting a 510(k) around a weak predicate. De Novo grants come with FDA-defined special controls and immediately create a predicate-class for follow-on devices.