Early access registrations are open for Headstart, a predictable, fixed-price program for becoming compliant.

Updated October 11, 2020

Is Our Software a Medical Device (and Do We Need to Certify It)?

Dr. Oliver Eidel
Getting Started

Question

We’re a startup which is currently developing healthcare software. We’re wondering whether this software would be a medical device. That’s important because then we would need to get it certified before we’d be allowed to publish it.

Short Answer

It depends. Think about what your software does, consider writing this down (it’s called an intended use). Then, read the definition of a medical device according to the MDR. Do you fulfil that definition?

If that didn’t help, read the long answer.

Long Answer

It really depends. On what? On how your software is supposed to be used (the intended use). Let’s look at an example: Slack. Everyone who’s ever worked in a hip startup knows Slack. In case you don’t: It’s a chat software which persists messages.

Is Slack a Medical Device?

Clearly, Slack is not a medical device. It’s not certified as one. That’s right. Because of its intended use. What is Slack’s intended use? I don’t know. But after briefly browsing their website, I’d come up with this copy-pasted version of stuff they mention:

Work more easily with everyone, stay on the same page and make decisions faster by bringing all of your work communication into one place. You can use messages in channels with attachments, voice and video calls.

So it’s somewhat of a glorified communications software. Fair enough. Now, is this a medical device? After glancing quickly at the definition of a medical device, it becomes clear that this is not a medical device. Because it’s not doing diagnosis, prevention, monitoring (etc.) of a disease, injury or disability. And all the other things stated in the definition.

Great, Slack is not a medical device. But could it be? Yes! Let’s say Slack’s intended use would be something like this:

Communications tool for physicians for sharing health care data (medical reports, vital signs, etc.) of patients and recording of treatment decisions.

So physicians would be using it as a messenger. They would be sending around medical reports of patients and vital signs. Something like “check out this patient on my ward, he has diabetes type 2 and his blood pressure just dropped to 100/60” and then the other physician dude replying “yo thanks man I’ll check it out”.

And that makes it arguably a medical device, because now you’re using it to monitor disease, at least in my opinion. This isn’t the greatest example because it’s somewhat of a gray area (“it’s not monitoring diseases! A blood pressure machine does that, but Slack is just a messenger!”) but I hope the general idea is clear.

If you’d do a quick risk analysis on this (“what happens if it goes down?”), you could also argue that it’s a medical device. If the entire hospital only runs on this one tool for monitoring vitals and communicating treatment decisions, this tool becoming unavailable due e.g. due to a software problem would be a huge risk which would directly endanger patients.

At the latest, if you’d add an app to Slack which suggests suitable diagnoses based on a patient’s history, then you’d definitely be a medical device. Because you’re doing diagnosis of disease (see the medical device definition).

Takeaways: It Depends on How You Use It

Whether something is a medical device or not doesn’t depend your technology. It doesn’t even depend on your product. It depends on how it is used, the context of use. That’s what the intended use is about.

When you start your product / software development, you should be very clear on what your intended use will be. This isn’t regulatory overhead (it’s a matter of writing 1-2 pages) and will help you greatly in understanding what you actually wanted to develop. In product development, this sometimes is referred to as a high-level vision document.

See this article on how to write an intended use. But at this initial stage, don’t get bogged down in regulatory templates. You should have an idea of what your software is about and in which context it is used.

Examples of Medical vs. Non-Medical Devices

When talking to medical startup founders, they always found it enlightening to hear some surprising examples of software which would be classified as a medical device - or not. So here’s a small collection.

Hospital Management System (HMS)

Not a medical device.

Software which does the administrative stuff inside a hospital, e.g. assigning patients to wards, billing, etc. This is somewhat surprising, but becomes more clear when looking at the medical device definition. It doesn’t do any diagnosis, treatment, etc. of patients.

However, there’s an interesting distinction here: Once such a software records medication per patient, then it’s a medical device as it’s involved in the treatment of patients.

A quick risk analysis of this also makes this clear: Imagine information being stored incorrectly and leading to wrong medication doses for patients. Sounds dangerous.

If physicians can also write down regular observations into this software (e.g. from bedside rounds), it becomes somewhat of an gray area, at least to me. Now you’re recording notes which are essentially monitoring disease. But you could also argue it’s just a note-taking tool. I don’t know.

Picture Archiving and Communication System (PACS)

Not a medical device.

Wait, what?

A PACS is the “backend” of any radiology department. It stores the images and makes them accessible to radiology workstations. The viewer (i.e., the frontend) which physicians use to view images is separate software, often from another vendor (discussed below as separate example).

It’s intended use could be phrased as “receive and store images, serve them to workstations”. Doesn’t sound like a medical device. And that’s right: Right now, PACS viewers are not classified as medical devices.

Things however change if they’d do some sort of advanced image processing (e.g. highlighting suspicious areas) or suggest diagnoses (e.g. “hey this looks like cancer”). Then it’s clearly a medical device.

Unintuitive, but that’s the way it is.

By the way, what happens if a PACS system goes down? Then the hospital is screwed. I witnessed this a few times during my time in hospitals. Sounds more like a medical device, actually..

PACS Viewer

A medical device.

It’s the frontend to the PACS (see above) which physicians interact with. It displays the images and includes some tools for simple image manipulation, e.g. brightness / contrast and drawing onto circles or arrows onto images. Because medical diagnoses are made based on the way images are displayed, this is a medical device.

An App for Tracking Blood Sugar Values for Diabetics

A medical device.

You’re doing monitoring of disease by tracking blood sugar values of diabetic patients.

There are some interesting (well, at least from a regulatory perspective) aspects to this: If you’d only offer this to healthy patients, would it be a medical device? Tracking blood sugar for those people doesn’t mean you’re monitoring disease as they don’t have any.

But if you would offer a feature of early detection of diabetes to those people, then it would be a medical device again - you’d be doing diagnosis of disease.

An App to Cure Back Pain

A medical device.

You’re treating (or at least attempting to) a medical condition, back pain.

An interesting angle to this: If you’d market it as “an app to improve back muscle soreness” you could get away with it not being a medical device, because “back muscle soreness” is not a real definition of a disease. “Back pain” however is.

A Sleep-Tracking Device

It depends. What’s your goal? If it’s marketed as “a sleep tracking device for monitoring sleep apnea”, then it’s a medical device. You’re monitoring or even diagnosing a disease (sleep apnea).

If however it’s intended “just for fun” to monitor sleep of healthy users, then it’s not a medical device. Just like smartwatches.

Case in point: Emfit. They sell both of those devices. A device to detect and monitor sleep apnea. And another device for healthy people to monitor their sleep “just for fun”.

Check out the intended use of the “just for fun” product:

[…] intended to be used for tracking the impact of free-time activities on sleep quality and progression of recovery on a healthy subject. […]

Makes sense! A great product, by the way. I own one and held a talk on some analysis of its data.

A Symptom-Diagnosis Chatbot App Like Ada

A medical device.

There are multiple apps out there which offer a chatbot-like interface for users to input their symptoms and receive a diagnosis and treatment advice (e.g. “go to the hospital soon”), the most prominent one being Ada here in Germany.

Thankfully, this is a rather clear case. You’re diagnosing disease based on user input, just as in the definition.

(I’ve worked as a consultant for Ada Health.)

Congratulations! You read this far.

Get notified when I post something new.

Sign up for my free newsletter.

I work as a regulatory consultant for Healthcare software startups. I try to publish all my knowledge here so that startups can certify their medical devices themselves in the future.

If you're still lost and have further questions, just send me an email and I'll be happy to answer them for free. More about me

No Cookie For You Privacy Policy Imprint
No QMS on this planet will save you from creating crappy software.