The biggest mistake innovators make with medical device classification is assuming that just because a product looks like yours is already on the market as a Class I device, yours will be too. It does not work that way in Europe or the UK. Unlike the US system which often relies on comparing your new gadget to an old one, the EU MDR 2017/745 and UK MDR 2002 operate on a strict rule-based system found in Annex VIII. Your classification is determined entirely by your specific Intended Purpose and how you describe what the device does, not by what your competitors are doing. If you get this wrong, especially with the new Rule 11 for software, you might find yourself unexpectedly needing a Notified Body, facing massive delays, or having to rewrite your entire technical file because you accidentally claimed your wellness app treats anxiety.
The Predicate Trap Is Real
I have lost count of how many times I have sat across from a founder who is absolutely convinced they are right. They slide a piece of paper across the table, or share a screen because who meets in person anymore, and show me a competitor’s product. “Look,” they say. “This device does exactly what ours does. They are Class I. So we are Class I.”
It is heartbreaking to tell them they are wrong.
In the United States, the FDA uses a predicate system. You find a device that is substantially equivalent, and you ride its coattails. It makes sense there. But here? In the UK and the EU? That logic will get your application thrown out.
The European system ignores what is already on the shelves. It does not care. It cares about the rules. Specifically, the classification rules in Annex VIII of the MDR. You have to run your device through these rules as if it is the only object in the universe.
If you build a thermometer, you don’t look at Braun. You look at the rules for active devices used for diagnosis. If you build software, you look at Rule 11. I think people struggle with this because it feels lonely. You want the comfort of knowing someone else has done it. But relying on a competitor’s classification is dangerous because you don’t know when they were approved. They might be sitting on an old MDD certificate that is about to expire, while you are trying to enter under the stricter MDR. You cannot copy their homework.
Software Rule 11 Changed Everything
If you are building software or an app, you need to sit down for this. Maybe grab a drink.
Under the old directive, almost all software was Class I. It was the wild west. You could self-certify, slap a CE mark on it, and go sell. Those days are dead. They are gone.
The Medical Device Regulation introduced Rule 11. This rule was specifically designed to catch Software as a Medical Device (SaMD) and drag it up the risk ladder. It is aggressive. Basically, if your software provides information that is used to make decisions with diagnosis or therapeutic purposes, you are likely looking at Class IIa or higher.
Class IIa means you need a Notified Body. You cannot just sign a declaration of conformity yourself. You need an external auditor to come in and check your work. And if you haven’t noticed, there is a massive shortage of Notified Bodies right now. The waiting lists are long. Expensive too.
I saw a team recently that built a symptom checker. They thought it was simple triage. Class I. But because the algorithm suggested potential conditions and advised on urgency, it fell squarely into Rule 11. They went from a planned three-month launch to an eighteen-month certification nightmare. They burned through half their runway just waiting for an audit slot.
It is brutal. But it is the reality. If your code does anything more than store data or simple communication, you are probably not Class I anymore.
Intended Purpose Is Your Only Shield
So how do you survive this?
It comes down to words. Specifically, your Intended Purpose statement. This is the holy grail of Medical Device Classification. The regulations don’t classify the hardware or the code; they classify what you say it does.
You can have the exact same physical object classified in two different ways depending on the label. A scalpel used for surgery is a medical device. A scalpel used for cutting clay in an art studio is… well, it’s just a knife. It’s a tool.
The problem is that marketing teams love to use big, bold words. They want to say the device “cures,” “treats,” “diagnoses,” or “monitors vital signs.” Every time they use one of those words, the risk class goes up. I often see a disconnect where the regulatory person writes a conservative technical file, but the website claims the device is a miracle worker. That is a compliance disaster waiting to happen.
You have to lock the marketing team in a room. Seriously.
If you want to stay in a lower class, you might need to de-scope your claims. Instead of “diagnosing arrhythmia,” maybe the software “tracks heart rate patterns for wellness.” The difference is subtle to a user, but to a regulator, it is the difference between a Class IIb device and a non-medical gadget.
Be precise. Be boring. Boring is good in regulation. Boring gets approved.
The Wellness Borderline Zone
This brings us to the grey area. The “borderline” products. This is where things get really messy.
There is a massive industry of health apps, fitness trackers, and sleep monitors that are trying desperately not to be medical devices. They want to be “general wellness” products. And that is fine. The regulators don’t want to police every step counter in the world.
But the line is thinner than you think.
If your app claims to help with “depression,” that is a medical claim. Depression is a disease. If it claims to help with “sadness” or “mood,” you might be okay. See the difference? It feels silly, but legal distinctions often are.
There is a guidance document called MDCG 2019-11 regarding software that is worth reading. It helps define qualification and classification. But even with guidance, it is subjective. I’ve seen companies get a nasty letter because a blog post on their site implied their wellness tracker could help manage diabetes. Boom. You just claimed a medical purpose. Now you are an illegal medical device.
It seems unfair that a blog post can trigger a regulatory enforcement action, but that is the world we live in. Your website is considered part of your labelling. If you talk like a doctor, you get regulated like a hospital.
UKCA and CE Are Not Friends
Let’s talk about the Brexit elephant in the room.
It used to be so easy. One CE mark covered the UK and Europe. You did the work once, and you sold everywhere from London to Lisbon. I miss those days.
Now, Great Britain has its own rules. We have the UK MDR 2002. And here is the funny part: the UK MDR 2002 is currently based on the old European Directive (MDD), not the new Regulation (MDR). So, you have the EU operating on the new, super-strict rules, and the UK operating on the old rules (for now), but requiring a separate mark called UKCA.
You cannot assume that your EU classification holds water in the UK, or vice versa. While the classification rules are currently similar because the UK hasn’t fully diverged yet, the paperwork is separate. You need a UK Responsible Person. You need a separate technical file structure in some cases.
I know startups that got their CE mark under MDR and thought, “Great, we can just apply for UKCA.” Not exactly. And it works the other way too. If you are a UK company, your UKCA mark is worthless in France. You need both.
It is double the admin. Double the fees. And double the headache. You have to plan for this divergence in your classification strategy. Just because you are Class I in the UK under the old rules doesn’t mean you won’t be Class IIa in Germany under the new rules. In fact, that is a very common scenario for software right now.
Why Startups Fail Audits
When the auditor comes, they are going to look at your technical documentation. This is where the bodies are buried.
A common pitfall is a lack of justification. You state “This is a Class I device.” Okay. Why? “Because it is.” That is not an answer. You need to write a justification report that goes through Annex VIII rule by rule. “Rule 1 does not apply because…” “Rule 2 does not apply because…” “Rule 10 applies, but…”
You have to show your working out. It is like math class. The answer matters less than the method.
Another huge failure point is clinical evaluation. Under the MDR, even Class I devices need a Clinical Evaluation Report (CER). You can’t just say the device is safe. You have to prove it with data. And here is where the “equivalence” trap comes back. Under MDR, claiming equivalence to another device to use their data is extremely hard. You need access to their technical file. Do you have access to your competitor’s trade secrets? Probably not.
So you end up with a classification that might be correct, but you have no clinical data to support it because you can’t use the equivalence route. This forces you to generate your own data. That means clinical investigations. That means money.
I sometimes see companies try to fudge this. They write a literature review that is just a list of Google Scholar articles. The auditor will tear that apart. It needs to be systematic. It needs to be methodical.
Also, spelling mistakes in your technical file make you look sloppy. I once saw a file where “sterile” was spelled “sterrile” throughout. It makes the auditor wonder what else you didn’t check. If you can’t spell, can you code? Can you manufacture? It is a trust thing.
Don’t Trust Similar Devices
I want to circle back to this because it is the most dangerous habit.
You go on Amazon. You buy a pulse oximeter. It says it is FDA approved and CE marked. You think, “I can make this cheaper.” You assume their regulatory path is your regulatory path.
But you don’t know the history of that device. Maybe they are “grandfathered” in under old rules. Maybe they are flying under the radar and haven’t been caught yet. There are thousands of products on the market that are technically misclassified, just waiting for a Competent Authority to notice. If you copy them, you are just copying their liability.
And things change. The guidance changes. The MDCG (Medical Device Coordination Group) releases new documents all the time. What was true in 2019 might not be true in 2024. You have to stay current. You have to read the boring PDFs.
I strongly suggest you check the official MDCG guidance documents yourself. Don’t take my word for it. Go to the source.
It is tempting to look for shortcuts. We all want to move fast. But Medical Device Classification is the foundation of your entire business. If the foundation is cracked, the house falls down. It doesn’t matter how good your technology is if you are legally barred from selling it.
There is also the issue of accessories. Did you know an accessory to a medical device is classified in its own right? If you make a special cover for a probe, that cover might be a device. It gets complicated fast.
Sometimes it is necessary to bring in an expert early. Not when you are ready to submit, but when you are sketching on the whiteboard. An external eye can spot a Rule 11 trap from a mile away.
The Bottom Line
Classification isn’t just a box to tick. It is the strategy that defines your budget, your timeline, and your survival. I have seen great products die because the founders didn’t respect the rules of the game. They assumed logic would prevail. But regulation isn’t about logic; it’s about the law.
Take the time to define your intended purpose with surgical precision. Don’t let your marketing team write checks your regulatory team can’t cash. Understand that software is scrutinized more heavily than ever before. And please, for the love of all that is holy, do not just copy what you see on the market. Do the work. Read the rules. It is tedious, unglamorous work, but it is the only way to build something that actually lasts.
