Predicate devices and the 510(k): what makes a predicate appropriate, and how FDA actually decides

The 510(k) premarket notification is built on a single comparative idea: a new device can reach the market without a full premarket approval if it is "substantially equivalent" to a device already legally on the market. That already-marketed device is the predicate. Everything in a 510(k) review turns on whether the predicate is a legitimate benchmark and whether the new device measures up to it. Below is how FDA defines a predicate, the decision logic it applies, and how that logic has played out in real clearances and De Novo decisions.

What a predicate device is

FDA defines a predicate as a legally marketed device to which a new device is compared for a substantial equivalence (SE) determination ⁶. A device qualifies as legally marketed, and therefore usable as a predicate, if it falls into one of three categories ⁶:

1. a device legally marketed prior to May 28, 1976 (a "preamendments" device) for which a PMA is not required;
2. a device that has been classified or reclassified into Class I or Class II; or
3. a device that was itself found substantially equivalent through the 510(k) process ⁶¹⁵⁴.

A granted De Novo device also becomes usable as a predicate for future 510(k)s once it is classified into Class I or II ^136138140. Critically, a device that has been removed from the market at the initiative of the Commissioner, or determined by a judicial order to be misbranded or adulterated, cannot serve as a predicate ¹¹⁸.

The substantial equivalence decision framework

FDA reviews a 510(k) through a stepwise SE decision framework (often drawn as the SE flowchart) ²⁸¹¹²⁰:

1. Identify a valid predicate and compare the new device to it ⁸²⁰.
2. Same intended use? FDA can only find substantial equivalence if the new device has the same intended use as the predicate ^1381112.
3. Same technological characteristics? FDA asks whether there is any significant change in materials, design, energy source, or other features ¹³⁶¹².
4. Do different technological characteristics raise different questions of safety and effectiveness? A "different question" is one raised by the new characteristic that was not applicable to the predicate and that poses a significant safety or effectiveness concern ^2358912.
5. If not, do the data show the device is as safe and effective as the predicate? FDA then evaluates whether the submitted performance information demonstrates equivalence, which may include clinical or scientific data where needed ¹³⁶¹².
6. If the differences do raise different questions, the device is found not substantially equivalent (NSE) ³²².

Throughout, safety and effectiveness are the organizing lens for both the intended-use question and the technology question ¹³¹².

"Same intended use" and the indications-for-use trap

FDA determines intended use from the proposed labeling in the submission, including the indications-for-use statement ^84858690. Intended use is the general purpose or function of the device and it encompasses indications for use; indications for use describe the disease or condition the device diagnoses, treats, prevents, cures, or mitigates, plus the patient population ⁸⁴.

A change in indications does not automatically create a new intended use ¹⁸²¹. The difference becomes a new intended use, disqualifying the predicate, only when the change either raises a safety or effectiveness issue not raised by the predicate, or significantly increases a concern the predicate raised, such that a meaningful comparison to the predicate is no longer possible ⁸⁵. FDA's own examples include moving a device from external to internal use where infection risk changes, or expanding a cardiac ablation indication to treatment of atrial fibrillation where the predicate never established safety and effectiveness for that condition ⁸⁵. In those situations the predicate is not an adequate proxy and the device may need a PMA, a De Novo, or another pathway ⁸⁵.

Technological differences: "same questions" versus "different questions"

When technological characteristics differ, the pivotal judgment is whether the differences raise different questions of safety and effectiveness or merely the same questions that already applied to the predicate ^1745109. If the questions are the same, FDA can resolve them with performance data, drawing on bench, animal, and, where necessary, clinical testing depending on the device ^41019391. If nonclinical and analytical data are insufficient, clinical data may be requested ^9391109. If the differences raise genuinely new questions, the device is not appropriate for a performance-data comparison to that predicate and the review moves toward an NSE outcome or a different submission type ^1745322.

One predicate, many predicates, split predicates, and reference devices

FDA prefers a single predicate, because it simplifies the SE decision ⁸⁹⁹. Where more than one is used, the submitter should designate a primary predicate, the one with indications and technological characteristics most similar to the new device ⁸⁹⁹. Multiple predicates are appropriate only in defined situations: combining features from two or more predicates that share the same intended use, marketing a device with more than one intended use, or seeking more than one indication under the same intended use ^8911327. Each predicate must share the new device's intended use, and any technological differences must not raise different questions of safety and effectiveness ⁸⁹⁹.

A split predicate, using one device for intended use and a different device for technological characteristics, is prohibited as inconsistent with the 510(k) standard ⁸⁹⁹.

A reference device is not a predicate. It is a legally marketed device used only to supply scientific or technical information, such as a test method or standard reference values, and it may be invoked only after the early SE decision points are satisfied (for example at Decision Point 5a), with a scientific rationale ^911427117. It cannot be used to answer the intended-use or technological-characteristics questions.

What makes a predicate appropriate, and when FDA treats one as inappropriate

Beyond the legal-marketing threshold, FDA expects submitters to choose an appropriate predicate and, whenever possible, not to rely on a device that exhibits new or increased risks when an alternative predicate exists without that risk ⁶⁸⁸. Where a predicate is tied to recalls or postmarket safety concerns, FDA may seek additional data, sometimes including clinical data, for new submissions that rely on it ⁶⁸.

FDA's draft guidance, Best Practices for Selecting a Predicate Device, lays out four concrete factors ⁷¹³¹:

1. Use a predicate cleared using well-established methods, ideally methods drawn from an FDA-recognized consensus standard, FDA guidance, a qualified Medical Device Development Tool, or a widely accepted method in the public domain ⁷².
2. Prefer predicates whose safety and performance meet or exceed expectations for the subject device, considering the safety profile and frequency of adverse events ²⁴.
3. Select predicates without unmitigated use-related or design-related safety issues; if none is available, explain how any known concerns are mitigated in the new device through design and testing ²³⁴³².
4. Select predicates not associated with a design-related recall; if a recalled predicate must be used, describe the recall concerns and how the new device's testing addresses them in the 510(k) Summary ^2342432.

FDA also recommends screening candidate predicates through their publicly available 510(k) Summary and Indications for Use documents ⁷²³².

On the longer-term concern sometimes called "predicate creep," FDA frames the issue as a preference for modern predicates rather than age alone ³⁷. Older predicates may reflect outdated technology or superseded methods, may carry unmitigated design-related safety issues or design-related recalls, or may show high rates of adverse events or malfunctions suggesting fundamental design problems ^3731402328. FDA has said it wants to encourage "the evolution of safer and more effective medical devices" by using updated scientific methods where possible ²⁸, while acknowledging that some older implant predicates carry valuable long-term safety data ³⁷.

The Safety and Performance Based Pathway

For well-understood device types, FDA offers the Safety and Performance Based Pathway, which is still rooted in substantial equivalence but substitutes FDA-identified performance criteria for a device-by-device comparison against a specific predicate ¹⁷¹⁰². The submitter still identifies a predicate to confirm the same intended use and technological characteristics that do not raise different questions of safety and effectiveness ¹⁷. FDA maintains a list of eligible device types and sets criteria calibrated to the performance of legally marketed devices of that type; a device that cannot rely entirely on those criteria stays in the traditional predicate-based pathway ¹⁷⁴⁵.

The grounds for a "not substantially equivalent" finding

FDA finds a device NSE on several distinct bases ⁸⁵⁹:

• New intended use relative to the predicate ⁸⁴⁴⁵⁵²;
• Different technological characteristics that raise different questions of safety and effectiveness ^438441112;
• No identifiable valid predicate with the same intended use ⁸⁵⁹;
• Inadequate performance data to demonstrate substantial equivalence, even where the intended-use and predicate framework otherwise fits ^4464959;
• Failure to meet applicable special controls where they apply ⁴²⁷; and
• in a limitations-based SE determination, nonconformance with the labeling limitations FDA specifies ⁵¹.

New intended use and different technological characteristics are the core statutory grounds; inadequate data is a separate, review-stage basis for an NSE outcome ⁸⁵⁹.

When there is no appropriate predicate: De Novo

If no appropriate predicate exists, a low-to-moderate-risk device may use the De Novo classification process under section 513(f)(2) instead of a 510(k), provided general controls, or general plus special controls, give a reasonable assurance of safety and effectiveness ^14296147148. A granted De Novo classifies the device into Class I or II, allows immediate marketing, identifies special controls for Class II, and creates a device that can serve as a predicate for later 510(k)s ^136138140148. A submitter does not have to be found NSE under a 510(k) first before requesting De Novo ¹³⁷. FDA may decline a De Novo request if it identifies a predicate that could support an SE review, or if the device is not low-to-moderate risk or cannot be adequately controlled; a declined device remains Class III and typically needs PMA approval ^13813614052.

How FDA has actually decided predicate questions

De Novo decisions where FDA found no predicate. In Quantitation of Organophosphate Metabolites in Urine by LC/MS/MS (DEN130005), FDA recorded that "no predicate device exists" and marked the predicate and comparison sections "not applicable" ²³⁶. In the CDC DENV-1-4 Real-Time RT-PCR Assay (DEN120012), the predicate name and predicate comparison were both "not applicable," with a special-controls guidance to be promulgated ²⁴². In the Factor V Leiden Kit (DEN030005) and Mandometer III (DEN070014), FDA explained the device was automatically in Class III because it was not within a device type marketed before May 28, 1976 or later reclassified into Class I or II, which is the statutory reason no 510(k) predicate was available ²⁴⁰²³⁷.

510(k) clearances showing the SE argument in practice. These decision summaries illustrate the standard structure, same intended use plus performance data resolving technological differences:

• iFuse-TORQ Implant System (K213667): same intended use as the predicates, identical technological characteristics to the primary predicate, and ASTM F543 verification testing (axial pullout, torsional strength, insertion/removal torque) showing the differences did not raise new questions of safety and effectiveness ¹⁹¹.
• Single-use flexible ureteroscope (K240074): although the subject device lacked integrated biopsy functionality present in the predicate, the intended use was the same and the technological differences were argued not to raise new safety or effectiveness questions ¹⁹².
• Single-use video hysteroscope (K232003): the same lack-of-biopsy difference was addressed the same way, same intended use with differences not raising new questions ¹⁹³.
• Medical Diode Laser Systems (K211977): same intended use and similar technological characteristics, with minor differences resolved by compliance with IEC 60601-1, 60601-1-2, 60601-2-22, and 60825-1 ¹⁸⁹.
• SteadiSet Infusion Set (K242692): same intended use but different technological characteristics in design, materials, energy source, and features, which the submission concluded did not raise new questions of safety and effectiveness ¹⁹⁰.
• Eagle device (K230722): same intended use, with IFU wording differences argued not to change intended use, and both non-clinical and clinical performance testing relied on to support an acceptable safety and effectiveness profile despite differing technology ²⁰².
• OASIS MRI System (K192851): modified functions and added features were argued not to constitute a new intended use, with no significant change in technological characteristics, supported by performance comparison ¹⁹⁹.

The common thread is that a 510(k) succeeds when the submitter can (1) tie its device to a predicate with the same intended use, (2) characterize every technological difference, and (3) show with appropriate data that those differences raise the same questions the predicate already answered, not new ones. Where any of those links breaks, whether through a new intended use, a difference that raises genuinely new questions, an inappropriate or unavailable predicate, or inadequate data, the device is pushed toward an NSE determination and, ultimately, toward a De Novo request or PMA.