Medical Device Change Management Best Practices

Medical Device Change Management Best Practices

19 min reading time

Best Practices for Medical Device Change Management

Presented by Michael Drues and Jon Speer at 10x for Design and Manufacturing – San Diego 2018

Reading Time: 19 minutes

Michael Drues: I want to have a little bit of a discussion. And I want to invite all of you to participate as well because I genuinely believe this is a very important topic and a topic that, quite frankly, so many companies screw up.

Does anybody know what one of the most common reasons why companies get warning letters and 483’s from the FDA?

Response #1: Not following design change rules?

Michael Drues: Yes, not following design change rules, assuming the rules make sense. I’ve never made the assumption that rules make sense.

But, not doing what makes sense when it comes to changing a product. And we’re going to talk about that this afternoon.

For those of you in the audience, how many of you have been involved with bringing a medical device onto the market?

[Audience Signals]

Okay. How many of you have been involved, once your product has gotten onto the market, changing it in some way? Most of you.

Now, this is not a drug audience. But if I were doing this presentation at a drug audience and asked the same question, how many people do you think would raise their hands to that second question?


Should you do a letter-to-file? It might allow you to make a change without notifying the FDA.

But having “avoid FDA notifications” as a motivating factor is a bad idea, says regulatory expert Michael Drues.

He said, “Several medical device companies tell their R&D engineers, “If you’re going to change a device, only change it to the point we can handle that change internally, doing a letter-to-file. Don’t change it to the point we have to notify the FDA.”

Can you think of any better way to prevent or hamstring development than putting a limitation on your R&D engineers?

Medical Device Companies Are Paranoid About Changes to a Product on the Market

One of the many interesting differences between drugs and medical devices is medical devices, by their nature, are very iterative and companies make changes.

On the other hand, when it comes to drugs, drug development is not very iterative. Although, that’s going to change in the future.

But, more importantly, drug companies are absolutely paranoid to make a change to a product once it gets onto the market. To some extent, medical device companies can be paranoid as well.

I’ve worked with several medical device companies – including some of the largest medical device companies on Earth – that have said, as a matter of company policy, to their R&D engineers, “If you’re going to change a device, only change it to the point that we can handle that change internally, doing what we call a letter-to-file. Do not change it beyond a certain point where we have to notify the FDA via either a special 510(k) or a PMA supplement.”

Now, perhaps as a biomedical engineer, you understand that’s what makes my blood pressure just shoot through the roof.

Can you think of any better way to prevent or hamstring development than putting a limitation on your R&D engineers?

Change Management Is a Key Role for Any Engineer in the Medical Device Space

So, Jon, when we talk about change management in the context of medical device development, what does that mean to you?

Jon Speer: Well, first of all, I felt like I should have done the intro for the global medical device part.

Anyway, I started my career as a product development engineer. And to your point, changes happen.

The moment you launch a product, something happens. You’re going to change the material or a supplier. Manufacturing is going to find a better way or something you forgot.

And so, change management is really a key role for any engineer working in this space. So, you really have to analyze.

Those of you who know me or read anything I’ve written, I’m a huge design control nerd. And so, design control isn’t just a design and development activities. It’s not just something you do while something is prior to market launch. Design control is pervasive throughout.

So, anytime you make a change, you have to assess and evaluate – How does this impact the form, fit, and function? What are the V&V implications of that? What are the risk implications of that?

Design control is something that should be living throughout the entire product lifecycle as well.

Michael Drues: I could not agree more, Jon. Thank you for sharing that.

Why Many Companies Get in Trouble When They Change Products

And I want to come back to the question I asked a moment ago – why do so many companies get in trouble because of change management?”

And I don’t think it’s a matter of following the rules.

Scott Phillips: I think the post-market changes in my study of these numbers say that’s the big bucket, right? It’s a different team.

Michael Drues: So the reason why companies get in trouble is they’ve made changes to the product?

Scott Phillips: They make a post-market change; they don’t update the risk file; they don’t update their specifications; they don’t validate that the training was affected; it’s all these sorts of secondary things that are normally required.

And the original team would have done them, but the sustaining team doesn’t do anything.

Michael Drues: Okay. I think that’s part of the answer. That’s a good start. Anybody else want to add to what Scott said? Or, Jon, what would you add?

Jon Speer: I would just build on what Scott was offering.

In my experience, what happens during the design and development process is the product development team does, usually, a pretty decent job of documenting design controls and risk management activities. And the moment design transfer happens, that file gets archived and buried, and those are not the people who will continue to maintain that product once it’s in a manufacturing environment.

Now, you have a sustaining team or manufacturing team or some other group of people that did not design and develop the product whose responsibility it is to author these changes.

And they’re making changes blindly because that design history file, that risk management file oftentimes is buried somewhere in archive, and they don’t go back to that design history file and those design controls to see if the change they’re making has any impact upstream or downstream to the decisions that were made when that product was originally launched.

We have another opinion here.

Christine Zomorodian: Jon is going to probably smack me when I say this, but it depends. And what it depends on is a couple of things.

First of all, we talked about a situation with a contract manufacturer where the 510(k) holder is not doing the manufacturing. There tends to be some drift, in my experience, once you turn it over to the contract manufacturing organization. And, sometimes. management doesn’t want to even know what that drift looks like. So, that’s one reason.

The other reason is fear of questioning or the risk of questioning when they make a submission to FDA.

They don’t want to face that. They don’t want to take the risk of making a new submission and face questions about their product.

Michael Drues: So, I think all of the comments people have shared have been very good. They’ve all been part of the answer. But I think we’re still dancing around on the surface.

As engineers, we like to think in terms of root cause. But, oftentimes, when I hear people talk about root cause, they don’t get anywhere close to the actual root cause. They’re talking about the superficial manifestations of a much deeper problem.

And I think, quite frankly, the root cause to a lot of the problems we’re talking about here and a lot of the reasons why companies get in trouble when they change products is because of what’s up here [points at his head] or the lack thereof – the thinking, the analysis that goes into it.

What are the Administrative Options When Changing a Medical Device?

So let me, as my attorney friends like to say, lead the witness – if you’re going to change a medical device in an administrative way, you have two options to handle that.

What are the two options? Let me give you a hint –

  • you can either notify the FDA; and if we notify the FDA, how do we do that?
  • Or we cannot notify the FDA; and if we don’t notify the FDA, how do we do that?

So, with that big hint, what are those two options?

Jon Speer: So, if you’re going to notify the FDA, what are some forms that you might communicate a change you’re making to the FDA?

Response #4: Special 510(k).

Jon Speer: Special 510(k). Are there any other examples you can think of?

Michael Drues: How about if you’re a Class III device?

Response #4: PMA supplement.

Michael Drues: All right. So those are ways we can notify the FDA. What if we choose not to notify the FDA?

Response #5: Letter-to-file.

How Does a Company Decide Whether or Not to Notify the FDA?

Michael Drues: Letter-to-file. How does a company make that decision on whether or not they should notify the FDA?

Response #6: Follow the flow chart.

Michael Drues: Yes, you can follow it if you want to but – let’s be honest – does that happen in most companies? What’s the point of having a discussion if we’re not going to be candid?

So, at least in my experience, most people just don’t do that. And even if they do follow the flow chart – for those of you that heard my talk on substantial equivalence yesterday – give me any flowchart from FDA, I will make the result come out any way you want it – backwards, sideways, this way, that way.

So, talk about reality. How do you, in a company, make that decision as to whether or not you need to tell the FDA?

Scott Phillips: If it’s a new indication.

Michael Drues: If it’s a new indication, now the question becomes, what does “new” mean? How different from your original indication can your new indication [be]?

Scott Phillips: If your indication means now you’re [in the heart?], then it’s a new indication.

Michael Drues: But what if it’s not quite that extreme?

So, labeling changes, that’s one possibility. Yes, Sri?

Sri Punnamaraju: It automatically triggers an update to the risk management and the new risk is escalated. That is a quantitative way to say, “We assess the risk as high now, and we need to update that.”

Michael Drues: So I would agree some companies will get to that eventually, but that’s not usually where the conversation starts.

Jon Speer: Do you see what I have to deal with whenever we do a podcast?


Michael Drues: Because I talk about reality, and Jon talks about theory.

Jeff Gable: If they don’t notify, they think the FDA will say no.

Michael Drues: That’s an interesting spin, Jeff.

Of course, that could never happen in the real world, right? (sarcastically)

Jon Speer: In my experience, I think what, oftentimes, happens is we go through the FDA guidance document on deciding when to submit a 510(k). We follow the flowchart. We pick the path of least resistance.

We may or may not document that decision. But more times than not, we choose not to communicate changes to the FDA because we can justify it through whatever means.

And we may or may not document “the letter-to-file,” which is interesting to me because I see a lot of companies say, “Oh, that’s a letter-to-file,” and, then, they never document the decision. There is no letter in a file.

Michael Drues: There are, actually, some advantages to not documenting things down. When it comes to product liability, that’s a huge advantage. If we have time, we can come back to that.

The Industry-Wide Misconception About Letter-to-File

But again – let’s be honest here –

How many people have a friend who told us the perception in their organization is, if the company chooses to do a letter-to-file, that’s somehow less work than if they were to notify the FDA via special 510(k) or PMA supplement? How many people have heard of this?

I think I’m being facetious here. Most, if not all, medical device companies believe it is less work to do a letter-to-file. And this is where I believe the industry standard is flat out wrong.

Jon Speer: 100%. We go through the decision tree. If we choose to document it, we may print out that flow chart; we highlight or circle the items; we may write a paragraph or a memo and document it and check the box. And we’re done. That’s the most common practice.

Michael Drues: That’s a very common practice. And that is, in fact, the industry standard.

And the industry standard is 100% wrong.

Because, typically, the way that decision is made, we’re going to make this change, let’s decide first – letter-to-file or special 510(k) – and, then, do the other stuff that go into it.

I think that’s 100% backwards.

I work with companies all the time. I say,

“Whether you’re going to do a letter-to-file or a special 510(k), it makes absolutely no difference. The amount of work you have to do in terms of thinking, in terms of analysis, in terms of literature search, in terms of benchtop testing, maybe additional other kind of testing, is exactly the same.”

The only question is, “What do you do with that information?”

In some cases, you will take that information and literally put it into a file folder and put it in your three-drawer file cabinet. For those of you that don’t remember what a three-drawer file cabinet is, you can Google it. You’ll see a picture. That’s what we used in the olden days. That’s why we call it a letter-to-file.

Or you take that same information and you put it into a different package – we call it a special 510(k) or a PMA supplement – and we send it off to the FDA. The information is exactly the same. It’s just a matter of what we do with it.

And here’s another reason why I write a letter-to-files that way. If you make a change to a medical device and you choose, for whatever reasons, not to notify FDA and, in the future, some knock on your door comes – it’s the FDA – they say, “Hey, we’ve noticed you’ve made a change to this device. We don’t remember you ever coming and talking to us about it? What the heck is going on?”

I don’t want to be in a situation where I say, “Oh, gee, we forgot,” or worse, “Darn. You caught us.” I don’t want to say that at all.

I would say, “Oh, Mr. or Mrs. FDA reviewer, come on in. Sit down. Have a cup of coffee.”

Jon, are we allowed to give them coffee anymore?

Jon Speer: No.


Michael Drues: “Let me pull out my letter-to-file where we have completely documented the change we made, why we made it, all the testing we’ve done to support it. Oh, by the way, here is a list of other companies that have made similar changes and they didn’t notify you either.”

I work as a consultant for the FDA, so I see this from both sides. I want to make it painfully obvious to my friends at the FDA that we know what the heck we’re doing, that we’re not forgetting something, we’re not hiding something. We made a business decision that, based on the following reasons, we decided not to notify you.

The reason why I like that strategy is very simple. It is because – worst case scenario – if the FDA says, “Well, gee, thank you for sharing all this information with us. We think you should have let us know.”

“Fine. Not a problem,” I take all that info from my letter-to-file; I repackage it; I put special 510(k) on it, “You’ll have it next week.”

What do you think of that approach, Jon?

Jon Speer: Well, a lot of companies simply do this letter-to-file path to minimize the work.

And this is the key thing Mike is illustrating here – the letter-to-file is not a substitute for the work.

We still have to do the analysis. We have to make a decision as to – when we make a change – what is the impact of that. We may have to update design control activities. We may have to update risk activities. We may have to do additional testing and analysis.

“It’s just good, prudent engineering.” [To Drues] I’m stealing one of your phrases.

Michael Drues: They say if you’re going to steal, steal from the best. So, I’m flattered, Jon.

Jon Speer: But I think this is where companies get themselves in trouble because, “We’re in a rush; we have to hurry; we got to make this change. We got new components that came in from the supplier and they are a different color, but let’s just do a letter-to-file and let’s push on without doing the appropriate amount of work and effort that goes into that.”

So, that’s a problem. And if you do that on one change and, then, you make another change and, then, you make another change, and you handle all these things the same way, what does that look like?

What does this product – two or three changes down the road – look like when compared to the product you originally got clearance or approval for?

Michael Drues: Now, what Jon is talking about is what I call “change creep.” So, for those of you that are familiar with predicate creep, the idea here is exactly the same.

Change creep: You make a change to a device. That change is not significant enough to let FDA know. You make another change to the device. That change is not significant to let FDA know. You make a third change.

Each of those individual changes, considered one at a time, might not be significant. But over a long enough period of time, when your device is on the market, now those changes become significant.

At What Point Do We Notify the FDA? And How?

At what point do we notify the FDA? And, if so, how? Some of you are probably familiar with the phrase in the vernacular, the catch-up 510(k).

Officially, there is no catch up 510(k). There are only three types of 510(k)s. That’s not one of them. I’ve suggested to FDA many times we need to create one, but we don’t. At least, not yet.

But I refuse to use regulation as an excuse to hold me back. If the regulation does not have a way to notify FDA of these changes, we need a way.

What I will do is I will use a special 510(k) which has a significant change in it. But I will embed, into that special 510(k), the previous changes.

In other words, I will say, “As a matter of professional courtesy, I’m going to use this opportunity to notify you of all these other changes we’ve made to our device.” If FDA kicks that back on administrative reasons, I will fight them to the tooth because that defeats the entire purpose of being able to work together.

Jon Speer: So, question, Mike.

I do the analysis; I do the work; I do the testing; the effort to properly evaluate and analyze my change. And after I do that work and I’m stuck, I go through the decision tree and I’m not sure if I should do a letter-to-file or if I should do a special 510(k). Should I just go ahead and submit a special 510(k)?

Michael Drues: Some companies will take that approach. Some companies will be more conservative and, if they’re in that gray area, go ahead and file anyway. Other companies will be a little more – I’m not sure if aggressive is the right word – and decide to do the other way.

But most important to me, Jon – and I don’t want to put words in your mouth, but I think you agree – that decision can only be made once we’ve done all of that work first, right?

Jon Speer: Absolutely. So what about a pre-submission? I know you’re a big fan of pre-submissions. is this scenario where, if we’re on the fence, we might consider a pre-submission?

Michael Drues: You could. I am a huge fan of the pre-sub process. Like the catch-up 510(k), we don’t have an official pre-sub type to notify FDA of a change to an existing medical device. But there are ways.

Again, it’s a matter of perception here. I refuse to use regulation as an excuse to do prudent engineering. I want to be able to proactively notify FDA of my changes whether I do it in a submission, whether I do it in a pre-sub, however. I don’t care as long as we get it done.

Downsides to Making a Robust Letter-to-File

We made a joke, a moment ago, about documentation. And Jon was saying documentation, from a regulatory perspective, is a good thing. But there are some downsides to having documentation.

To illustrate the point we were trying to make here – I’m not going to ask for your own experience, your own company – how many people think or have heard of companies where, if they choose to do a letter-to-file, that letter-to-file is built out to the extent it would be if they were submitting something to the FDA? How many people would do that? How many people would not do that?

Some of you are not being honest. I’ve seen letter-to-files that are a paragraph long. A letter-to-file is a legitimate path, but it is not an excuse to not do what we should do as engineers. That’s what gets people into trouble.

So, what’s the downside of making a more robust letter-to-file?
I’ll give you a huge hint – product liability.

I spend a fair amount of my time working as an expert witness in medical device product liability cases. And one of the cool things I like about working with my attorney friends is, they don’t limit me to what FDA may or may not require.

They want to know from me, as a professional biomedical engineer, did the company do what they should have done? Who cares if it was required by FDA or not? That’s not an excuse.

Did the company do what they should have done, from an engineering or biological perspective? That’s the litmus test here.

Now, the problem with that, when I write letter-to-files – and I do these a lot – I want to write them from a regulatory perspective to cover my you-know-what, but I also want to write them from a product liability perspective to not expose my you-know-what.

There’s sort of a fine line. Maybe, Jon, you can describe it a little bit more?

Jon Speer: Well – all of us who have been in this space – sometimes we don’t even think about the potential litigation aspects. We have a healthy fear or maybe unhealthy fear of regulation i.e. the FDA.

And it seems like, a lot of times, what we’re doing is jumping through hoops to satisfy the regulator.

And, sometimes, that lends us to maybe going a little bit overboard and thinking of every single use case or mis-use case and documenting that in our risk assessments or documenting that in our testing.

And, from one perspective, you can say, “Well, I’m being thorough.” But, from the other perspective, if there were an adverse event or something happened to the product, that may come back to bite you.

So, this is totally gray. How do you know when it’s too much or not enough? Because you get it wrong in either scenario, you could have regulatory issues or you can have legal issues.

Michael Drues: Or, maybe, both.

So, to illustrate in a slightly different way, a lot of people I work with in companies, they tell me they fear the FDA and I say, “No, no. You should not fear the FDA. You should have a healthy respect for the FDA.”

After all, as one of my friends who used to be a senior reviewer was fond of saying,

“Physicians can kill patients one at a time, but an FDA reviewer can kill patients thousands at a time.”

And that’s something that, quite frankly, more people in our industry need to remember.

So, you should not fear the FDA. Whom should you fear?
You should fear product liability attorneys because they can impose a heck of a lot more damage than the FDA ever could.

And let me tell you – I’m being deposed in one of my cases in just a couple of weeks – the product liability attorneys are much better at finding documentation that the FDA ever will be.

I’m constantly amazed because – remember, I work as a consultant for the FDA. A lot of reviewers are personal friends of mine. Some of them go back to graduate school, before some of you in this room were probably born – they’re flabbergasted when I tell them about changes.

And FDA says, “Well, we don’t know. We weren’t notified.” Well, duh. How would you ever know about a letter-to-file unless something bad happens?

So, quite frankly, the agency has not a clue as to how many letter-to-files actually are done.

I don’t have a problem with the fact that they’re done. I have a problem with the way a lot of people do them. In other words, they use it as an excuse to not do what we should do.

Jon Speer: So I’ve gotten the signal there are just a few minutes left. What do you think about opening the floor to–?

Michael Drues: I was just going to say the same thing, Jon. Let’s do that.

Jon Speer: Who has questions about change management in medical device?

Hitesh Mehta: I’ve never worked on a drug side but, on the drug side, you said changes never happen?

Michael Drues: Let me qualify that.

Changes happen but you, typically, have to notify the FDA. I’m simplifying here, but there isn’t really an analogy to letter-to-file in the drug world.

Hitesh Mehta: So, there’s always a supplement which they have to submit?

Michael Drues: Of some kind, yes.

Hitesh Mehta: Okay. That makes it clear.

Michael Drues: Yes. Sorry about that. Thank you.

Jon Speer: Any other questions? All right.

Christine Zomorodian: Hi. I appreciate the discussion we’re having. But, for my money, this is really about risk management.

If you have an adequate and robust risk management system and you’ve identified, adequately, your product risks, when you have a change and you’re considering a change, you’re looking at your risk assessment and you’re making a determination if the changes you’re making are affecting requirements that are high risk.

Michael Drues: I could not agree with you more. You and I are singing the same song, just in a slightly different key.

But here’s my point – Do we need any guidance? Do we need FDA to tell us that?

To me, that’s basic engineering. To me, that’s what they used to teach in engineering school back in the day. I’m not sure they still do.

And to me – and I’m going to set the bar very high here purposely –

Anybody that changes a medical device without considering risk, as you just suggested, should not be in this business. To me, that is such common sense.

I’m sorry. I get frustrated with the amount of micromanagement that we have with regulation telling people to do things that, quite frankly, they should know to do anyway.

Jon Speer: And I’m just going to build on that. Back in the day, design control used to include risk management and human factors and usability. When did that go away?

It’s like we keep creating these other “disciplines.” Not that they’re wrong, it’s just that they were always there the entire time. When design control regulations were rolled out 20 years ago, and even before that, if you’re doing prudent engineering and good design and development work, risk is part of that process. It’s not a separate thing.

Michael Drues: And just to take that a tiny bit further, I agree 100% but we have to be careful about overgeneralizing.

I happen to be a subject matter expert for FDA in a few different areas, one of them being risk. So we can say “to take, into account, risk management,” but what the heck does that mean? What kind of risk and what detail and so on?

So, the devil’s in the details. And I would love nothing more than to be sitting in a court being able to say, “Well, this company followed the guidance.” The new change control guidance does specify risk in it now, which I think is a no-brainer for me. But if there were aspects of risk that they did not consider, you know where I’m going with that.

What’s the Typical Timetable for Approval of a Special 510(k) or PMA Supplement?

Jeff Gable: I’ve never dealt with either one of these special 510(k) or PMA supplement. What’s the typical timetable for clearance or approval of that additional information?

Michael Drues: Good question. We’ve got a lot of real-world evidence sitting in the room. Anybody want to share your own experiences? If you’ve submitted a special 510(k) or a PMA supplement, how long did that process take?

Hitesh Mehta: Correct me if I’m wrong, but the special 510(k), there is no notification back. You just send it and it goes into a black hole and it never comes out.


There’s no response back.

Michael Drues: Whew. I’m glad that it is 4:30 in the afternoon. We’ll continue that discussion at the pub, maybe. Anybody else want to give a little more quantitative number?

Response #10: … the special 510(k) does not require three months [inaudible 00:26:30]

Michael Drues: Now, we’re getting into some regulatory minutiae. But, bottom line, it depends on what product code you’re in, and it depends on your device.

We’re talking about a couple of months. We’re not talking about a huge amount of time here. But here’s the most important point, Jeff. In many ways, a special 510(k) is much easier than a traditional 510(k).

I don’t care what the contents or requirements are. I don’t care what the RTA Checklist is. The only important thing, to me, in a special 510(k) comes down to one and only one thing – to demonstrate that whatever change you’ve made, regardless of reasons, is not going to impact safety, efficacy, performance.

To me, that’s the only part of a special 510(k) worth reading. All the other stuff is pretty much copied and pasted from the traditional 510(k).

Jeff Gable: I’m a software engineer. And you can change software a lot faster than you can change hardware, which is one of the subjects of my talk tomorrow.

Say you can make software improvements, and you do the risk analysis; you do the testing; all that to make sure it’s safe, how fast can you feasibly push software updates and so forth?

Michael Drues: I’m not sure, Jeff, if you’ll like this answer or not.

Philosophically, to me, it makes no difference if you’re changing a hardware, if you’re changing a software. The mental process, the investigation isn’t exactly the same.

I’m not a programmer, but you have to go through that testing, Okay. You make this change in this part of your software. How do you make sure that’s not going to have impacts on other parts of your software or your hardware and so on?

It’s the law of unintended consequences, right. But there are zillion mechanical equivalents to that.

Let’s be honest, when the special 510(k) was created, we weren’t talking about software-based core products.

So, this is another example of how we’re taking newer technologies and trying to fit older regulatory models. I’ve never really thought about this before but, maybe, in the software world, we need a different mechanism to update FDA on those changes.

Regardless of what change you’re talking about; the most important thing is that we have that communication.

Joe Hage: Highly engaging. Highly interactive.

Michael Drues: And I’m sorry if I threw you a curveball.

Joe Hage: They don’t want to leave.

And so, you get to speak last from now on.


Dr.Michael Drues and… Jon.


email me