A 510(k) argues one thing: that your device is substantially equivalent to a predicate. Which means the predicate is not a citation in your file — it is your file. Part one of this course made the case that the comparison is the whole submission; this lesson is about the decision that sets the comparison. Choose well and the file almost writes itself. Choose badly and no amount of testing rescues the argument, because the testing obligations themselves were set by the choice.
What can legally serve
Four kinds of device can stand as a predicate: a device on the market before the Medical Device Amendments of May 28, 1976; a device cleared through the 510(k) process; a device granted marketing authorization through De Novo classification; and a device reclassified from Class III into I or II. One thing can never stand as a predicate: a device removed from the market at FDA's initiative, or found misbranded or adulterated in a way that led to its removal. Note what is absent from the list — PMA-approved devices cannot serve, because approval and clearance are different findings answering different questions.
The vocabulary of the comparison
- Primary predicate
- The single device your substantial equivalence argument leans on — the one with the same intended use and the closest technological characteristics. FDA's eSTAR template asks you to name it explicitly.
- Multiple predicates
- Permitted — with care. Legitimate uses include a device combining features of two cleared devices, or seeking more than one indication each supported by a different predicate. Every predicate cited must share the intended use you claim.
- Split predicate
- Citing one device for intended use and a different device for technological characteristics, when neither alone supports equivalence. FDA's 2014 substantial equivalence guidance calls this inconsistent with the statute. Files built on splits invite NSE.
- Reference device
- A device cited for its scientific methodology — a test method, an accepted performance benchmark — without serving as the predicate. Useful, legitimate, and no substitute for the comparison itself.
Finding it — and reading what you find
The search is unglamorous and public. Start in FDA's product classification database with your device type; it gives you the regulation number, the device class, and the three-letter product code. Then work the 510(k) database by that product code, newest clearances first. For each candidate, pull three documents: the 510(k) summary (the sponsor's own public account of the comparison that cleared it), the cleared indications for use, and — where one exists — the decision summary. You are reading for two things: an intended use you can honestly claim as your own, and technological characteristics close enough to yours that the differences can be either dismissed or tested.
- Same intended use as the predicate — read from the labeling, not from ambition. Fail this and the inquiry ends.
- Same technological characteristics — or different ones that raise no different questions of safety and effectiveness.
- Where characteristics differ: performance data — bench as a rule, clinical where the differences demand it — demonstrating the device is as safe and effective as the predicate.
Read the predicate's own history before you commit to it. Has it been recalled, and why? Has the manufacturer fielded warning letters about the device? Is it still marketed at all? A device removed at the agency's initiative is legally unavailable to you — and a predicate with a troubled record, while sometimes technically permissible, hands the reviewer questions your file then has to answer. The strongest predicate is usually recent, well documented, still on the market, and boring.
When the predicate is a set of criteria instead
For a growing list of well-understood device types — certain catheters, sutures, imaging accessories and more — FDA's Safety and Performance Based Pathway lets you demonstrate conformance to FDA-identified performance criteria instead of testing head-to-head against a specific predicate's performance. It is still a 510(k), and a legally marketed predicate still anchors the intended use; what changes is that the agency has pre-agreed what “as safe and effective” means for the type. If your device type is on the list and meets the published criteria, this is the cleanest comparison available — the argument is with the criteria, not with a competitor's spec sheet.
Where predicate choices go wrong
Choosing the market leader instead of the closest device
The best-selling device in the space is usually the most different from yours — newest features, broadest claims. Every unexplained technological difference becomes a testing obligation, and broad claims you do not share help nothing: the comparison runs on your intended use, not the predicate's market position. Choose proximity, not prestige.
Assembling an intended use no single predicate supports
Citing three predicates and claiming the union of their indications is a split predicate wearing a disguise. Each claimed indication must be supported by a predicate with that intended use — and the device as a whole must have the same intended use as its primary predicate. If your claim needs stitching, the claim is the problem.
Committing before reading the predicate's file
Teams choose a predicate from a spec sheet, design their testing around it, and only then order the 510(k) summary — discovering the cleared indications are narrower than the marketing, or that the device was recalled two years ago. The predicate's public record is a twenty-minute read. Do it before the choice hardens into a test plan.
Frequently asked questions
Can a predicate be decades old?
Legally, yes — the statute sets no age limit, and preamendments devices from before 1976 can still serve. Practically, an old predicate invites questions, because FDA expects performance testing against currently recognized standards regardless of what the predicate was tested to. The comparison may be legal and still leave you carrying a modern testing burden the predicate never faced.
Can I use a competitor's device as my predicate?
Yes, and companies do it every day. The comparison is built from public information — the predicate's 510(k) summary, its cleared indications for use, its labeling, and published specifications. You need no consent and no data from the predicate's manufacturer. What you cannot do is guess: every characteristic you compare against must be documented, not assumed.
What if no predicate exists?
Then substantial equivalence is the wrong question. A novel device of low to moderate risk belongs in the De Novo classification process, which creates a new classification — and the granted device then serves as a predicate for everyone who follows. Filing a 510(k) against a strained predicate usually ends in an NSE determination, which is the slowest possible route to the same De Novo.
Sources & further reading
- FDA Guidance — The 510(k) Program: Evaluating Substantial Equivalence in Premarket Notifications (2014) fda.gov
- FDA — How to Find and Effectively Use Predicate Devices fda.gov
- FDA 510(k) Premarket Notification Database — where every predicate's public record lives accessdata.fda.gov
- FDA — Safety and Performance Based Pathway: device types and performance criteria fda.gov
This lesson is provided for general educational purposes and reflects the regulatory landscape as of its publication date. It is not legal or regulatory advice.
