Interview: FDA’s UDI Architect Jay Crowley on the Agency’s Approach to UDI and Pending Changes
Latest News | Posted: 2 May 2013

By Alexander Gaffney, RF News Editor

The US Food and Drug Administration (FDA) is, like many device regulatory bodies around the world, preparing to implement a huge overhaul of its medical device systems. Known as the Unique Device Identification (UDI) rule, the US regulator wants to build a system that allows for safer, more accountable medical devices.
But that process, which is global in scale, is not without significant challenges. In the wake of a flurry of UDI activity in recent weeks, we called up FDA’s Jay Crowley, senior adviser for patient safety at the Center for Devices and Radiological Health (CDRH) and the regulator responsible for FDA’s UDI effort, to talk about the philosophy and status of the UDI program, as well as what comes next for regulators and industry.
The following interview, which is lightly edited for clarity, was conducted on 22 April 2013.

Regulatory Focus: What is the promise of a well-built Unique Device Identification system?
Jay Crowley: The basic underlying issue is the lack of a consistent, standardized way to identify devices. When it comes to doing things like adverse event reporting or aggregating adverse events or other postmarket surveillance, we don’t currently have a way to unambiguously identify devices, which creates all sorts of problems for everyone.
In the pharmaceutical space, by contrast—at least in the US, anyways—we have national drugs code (NDC) numbers, which are used extensively to identify pharmaceutical products, creating a whole host of subsequent activities that can take advantage of NDCs like barcode administration systems, recalls and postmarket surveillance activities.
UDI provides a standardized, unamibiguous identification which supports all these sorts of activities for devices as well.
The global piece really represents two issues.
The first is the reality of the medical device space, where—unlike pharmaceuticals, which are often manufactured and labeled for a particular market—a medical device can be distributed in many, if not all regions of the world. The notion of a global UDI supports those types of global activities so we don’t need to have device manufacturers re-label products for different markets just to meet various identification needs.
The second is the reality of the economy today, which is the sort of global nature of trade. A globally harmonized approach to UDI would support traceability, which is going to become a lot more important in the coming years.
RF: A lot of what you just talked about addresses UDI from the regulators’ perspective. What benefits might industry derive from UDI?
JC: There are a host of benefits to almost everyone in the whole UDI system. One of the big drivers that isn’t really FDA’s bailiwick is the support that UDI will give to supply chain activities. There have been some studies done that talk about the cost savings that could accrue from the use of standardized identification, and there’s a tremendous benefit to nearly all of those involved, including manufacturers, hospitals, purchasers and distributors. Having UDI will, I think, really solve a lot of problems that people currently have with supply chains.
It provides some visibility, too, which is useful when you start talking about traceability and supply chain activities. There is a tremendous benefit from the supply chain end. Beyond that, people talk about having “rights”: Having the right product at the right place at the right time and for the right reason. The availability of the appropriate device at the right place is an obvious benefit and a big benefit that might come from UDI.
We do think that there are other, bigger-picture benefits to be had as well. We’ve started to do research on UDI along with the Brookings Institute that looks at the adoption and implementation of UDI, and is focused on three things in particular.
The first is on reimbursement activities. We see benefits to folks involved in those activities when you can start to conduct reimbursement analysis from UDI data. It could be used to support—and I’m always careful to use this term—comparative effectiveness kinds of activities, identifying which devices work best for each patient population.
The second piece we’re working on with Brookings is electronic health records (EHR) and the documentation of device use. Right now, patients and clinicians often don’t know which devices are implanted in them over the long-term—Which hip implant do I have? Has it been recalled?—and UDI can help these patients to remain informed.
Related to that, the third prong of the Brookings work is focused on connectivity and communication. So, that covers communication between clinical and patient. Let’s say you have a concern about a device—UDI can incorporate itself into the whole EHR space, opening up new ways to communicate information about a device from the manufacturer, to the clinician, to the patient.
RF: Shifting over to the IMDRF proposal and FDA’s UDI proposal. Both of those you’ve been very closely involved with. How do these two proposals work together, and how will they interact in the future?
JC: The goal of the Global Harmonization Task Force (GHTF) and now the International Medical Device Regulator’s Forum’s (IMDRF) work and guidance documents is to develop a common framework such that any regulator that chooses to develop a UDI system—if they follow that framework—we will still end up with an identical or similar approach to UDI such that we end up with a harmonized approach to UDI.
The documents of the IMDRF, which are based on GHTF documents, very much mirror what we are doing at FDA and in the EU. It all looks very much the same. I think our goal is to keep these proposals the same or in lockstep with one another so that we do continue to have a consistent approach to UDI. That was part of the IMDRF update to the GHTF guidance on UDI, and was in response to FDA’s July 2012 proposal on UDI.
We sat down with IMDRF and we looked at places where there were inconsistencies, and some things I went back to FDA and said we need to change this or that, and in the final rule we made some changes to the guidance that are reflected in the IMDRF draft, again taking into account our changes that we made to the UDI rule at FDA.
Our intent is to keep these proposals the same or similar—whatever the verbs are—such that FDA’s approach ends up being consistent with the IMDRF’s approach and the EU approach to UDI.
RF: Which countries are involved with this IMDRF effort?
JC: In addition to the US and EU, Canada, Australia, Brazil, South Korea, Japan, China and Taiwan. I think that’s all of them…
RF: So all of the major medical device markets, in other words.
JC: Yes. I mean, so not everyone is as involved at the same level, but all of these countries are somewhere in the process of developing their own UDI system. FDA is probably farthest ahead out of all the countries, but China has indicated to us that they are very interested in UDI and working toward a proposal of their own. The EU and Canada are obviously very interested in UDI as well. There’s a lot of interest in this.
I will also say that most regulators who are interested in UDI are really focused on traceability or traceability-related issues, so we seem to see those sorts of proposals emerging in many UDI proposals. We haven’t really addressed that in the FDA proposal. That’s probably only the significant difference—the incorporation of traceability requirements in the UDI activities of these other regulators.
RF: What timeline are FDA and IMDRF anticipating in getting this framework up and running?
JC: So, we intend to have the IMDRF guidance completed by the end of this calendar year so it’s out for comment until July 2013, and we’re going to meet in September to go over the comments and hope to have a draft ready for the management committee meeting in December 2013.
FDA is working to publish our final rule in June 2013. That was a FDASIA statutory requirement. And then we have a phased-in implementation over seven years that depends on the risk classification of the device.
RF: Looking at FDA’s docket on UDI, to say that there are a significant number of comments would be an understatement. Many of those comments are based on cost, while others seem to be more specific to a particular device or device category. What are some of the more common comments you’ve received from industry, and how is FDA addressing those concerns?
JC: There are approximately five different categories these comments fell into.
The first one which we really probably can’t do a whole lot about is the timeline issue. There were a number of manufacturers who wanted more time, and particularly Class III (high risk) devices, which are required to be marked within the first few years. And then there were user groups who wanted all of this done much, much quicker. Many people wanted all of this done within two or three years.
I think we’ll likely stick with the timeline that we suggested, because it’s a balance between the two issues.
There were also a lot of comments around our proposal for a standardized date format in the UDI rule. We received a lot of letters and complaints over the years from clinicians and others about the various ways dates, and especially expiration dates, are displayed on devices. So we suggested a standardized date format that was US-centric. So today would be April 22, 2013, while the EU date format would be 22 April 2013.
That really created quite a bit of havoc for device manufacturers who are selling their devices in Europe, which follows the ISO 8601 format and is an all-numeric, YYYY-MM-DD format. And we had a lot of comments, both from device manufacturers and a number of user groups that supported that ISO standard format.
We’ll likely move in that direction, or very close to it. Our goal was to come to a standardized date format, but I don’t think we really cared which one it was so long as it could be understood by everyone.
There were also a number of comments on something we hadn’t really anticipated, and really no one had brought up before which was kind of an existing inventory issue. There are devices that are manufactured and packaged and labeled and then sit in a facility for a while because it’s an odd size, or it’s not a hot-selling item, or whatever reason. So we’ll likely come up with some sort of exemption for existing inventory to avoid the marking requirements, just as FDA did for pharmaceuticals, so that these devices won’t need to be relabeled.
RF: Even if those devices are high-risk, Class III medical devices?
JC: Correct. We wouldn’t force them to relabel something that’s already been manufactured. We’re taking a prospective, not retrospective, approach. We didn’t really touch on this issue in the rule, but this was seen as a big issue by those in the implantable device space.
But we received a lot of comments, as you said. Many were very thoughtful and thorough—some were a little flaky—but most were very good, and many were about the direct marking of implants and the difficulties and the costs associated with doing that.
We’ve taken a look at that whole requirement and all of that and are in the process of rethinking our approach to that issue. Again, given what people reported to be very complex and very costly processes that would be needed to directly mark a device.
RF: And those concerns were based on the physical redesign of a device?
JC: Well, that, and just figuring out and validating how you would put a direct mark on an implant, particularly like orthopedic implants where there isn’t a whole lot of space where you can actually mark something. All the spaces on those types of devices have very specific purposes; either they’re intended to go into the bone or be articulating, so trying to figure out where you would put a mark on a lot of implants was just very complicated and costly, as is the technology redesign and validation. There were a lot of concerns raised about that process.
RF: What alternative approaches are FDA looking at to respond to those concerns?
JC: I think folks in general didn’t believe that the benefits of directly marking implants were going outweigh the costs in general. Perhaps the UDI could be on the label instead of the implant itself. We’re rethinking our approach to requiring direct marking on implants.
One final issue that was raised by industry was in regards to kits and final products, which we tried to address in our UDI rule. There are an enormous number and types of kits in the medical device field, and the field is used differently in different parts of the industry. What a “kit” means to the IVD space is different than what it means to the orthopedic space, which is different to what it means to the surgical convenience kit space… we just have tons and tons of different types of kits.
So, a lot of comments were concerned about how to approach the marking rule in regards to kits, and we’ve really tried to simplify our approach to kits and combination products to make it easier to implement for everyone. Trying to figure out how to approach it and how to carve out exceptions became really convoluted, so we’ve tried to improve it significantly.
So those were the major comments. As you said, many of the other comments were very technical, along the lines of, “How am I supposed to mark this ______ device.” We’ve tried to simplify and streamline the requirements so that people can understand more directly how these things apply to their devices. We also created an exception process that we expect people will use, but we didn’t want to create a process that would result in thousands of exceptions. We hope that the simplified process will prevent anyone from getting overwhelmed.
RF: In regards to managing the process, one of the things that caught my eye was the GUDID—the Global Unique Device Identification Database—that FDA was floating in its proposal. Could you explain this database a bit more and what you think it will be used for and how it will benefit the process?
JC: Sure. So, we actually built the GUDID. We actually have a draft user guide which is going through the guidance review process now. We hope to publish it in the next couple of weeks.
What the database does is provide meaning. We have these UDIs, device identifiers, the production identifiers, lot numbers, expiration dates—that’s all pretty straightforward. The device identifier is this unintelligible number or code that lacks any inherent meaning.
The GUDID—the “Good ID” database, as well call it—contains the device identifiers and metadata associated with that device identifier. And the device manufacturer is required to submit all this data to the database so that people are able to derive meaning from this UDI.
We expect that as certain activities—MDR reports, recalls, etc—are conducted, the GUDID database will be what provides meaning to the MDR report or recall. So if a particular device is recalled, I can look it up in the database and figure out what the device is.
We’ve been working on the database for many years, actually. We originally started it in about 2008, looking at what the database should look like, what data elements would be included, how it would operate, what data industry captures at present, how the database would capture that information, best practices, how we could make that data available, etc.
It’s already gone through pilot testing and user testing in the fall of 2012 with a subset of device manufacturers and other users. It’s intended to operate a lot like the NDC directory, allowing users to look up what a drug is and the information associated with it. GUDID will provide that for the medical device space, acting as a catalogue of all currently available devices. So if you need to go purchase a new device and you’re a hospital and you want a new IV catheter, you can go and actually find all currently available IV catheters that are being marketed. It will contain both historical records of what was distributed and currently marketed devices, and know the difference between the two.
We’ve also built a web service for the GUDID database, so if you’re a hospital and you scan your medical device in the operating room, that information will be populated automatically into the patients’ EHR and operating log, and we’ll send you back the record associated with that device identifier so you can populate whatever system the facility wants.
We’re working to build a downloadable database so hospitals can take this information and do with it what they will. It has a fairly robust web user interface as well, so you can go in and search, download, print records, and do whatever you want. It’s intended to facilitate the use of UDI by providing the meaning provided with the device, such as whether the device contains allergens, or was packaged sterile or not. We think this will act a lot like barcode administration databases on the pharmaceutical side. People can start to do the same thing with devices down the road, so to speak.
RF: You’ve referenced the NDC system quite a bit throughout this conversation. A former FDA official, Susan Alpert, has talked about how the NDC code can act to legitimize the use of unapproved products. What is FDA doing to prevent a device that’s cleared in a foreign country with a compatible UDI system from being seen as approved or cleared in the US?
JC: Interesting question. The database is set up to only accept Class II and Class III devices that have a valid premarket submission number, and FDA only accepts devices with valid numbers to the database to make sure those devices and cleared or approved. We also require a listing number to be validated with FDA, so that is cross-checked with the FDA Unified Registration and Listing Systems (FURLS) database. We’ve tried to take the steps that we can at this point to assure that whatever information we have in the UDI database represents devices that are legally allowed to be on the US market.
But it’s an interesting point: How would someone differentiate a device, without going to the database, from identifying a device that has a UDI from one that shouldn’t be in this country? And I don’t know. That’s a question that no one has actually raised before.
Hopefully one of the purposes of the database will be for people to know that there is a source of truth for information on approved devices in the US. So if you go and look for a device in the GUDID database, and it isn’t there, that means the device isn’t legally supposed to be in the US. Though we do have some exempt devices, like Class I devices, which means that could potentially get a bit flaky…
RF: Is the GUDID database going to support photos, similar to the MedWatch database, so that you don’t have counterfeit devices just trying to copy a manufacturer’s UDI?
JC: We’ve talked about including photos. There have been a number of requests from several sectors for that. We haven’t built that functionality yet, but it is something that we are considering.
I think eventually we’ll get to the point where all this will converge. It’s something we would like to incorporate into future iterations of the database.
RF: Is there something about UDI that hasn’t been covered to your satisfaction?
JC: I think what’s going to be interesting as we move down this path toward UDI is that basically everyone nods appropriately when we talk about global harmonization of UDI—”Yeah, that makes sense, and we should do that.” But the devil is in the details on this one.
I think it’s going to be very interesting to try to keep this notion of harmonization front and center. You know how regulators are—they get into something that they want to do, and they want it done in a way that supports whatever efforts they’re in the process of doing. Keeping everyone focused on maintaining this is going to be a challenge.
I remain cautiously optimistic that we can do it, but I think that it’s going to take effort on everyone’s part not to diverge as we move down this path. That is, to me, going to be the trick to accomplishing all of this.
The other big issue that people are starting to try to wrap their heads around at IMDRF and elsewhere that hasn’t really gone anywhere is the interoperability of various global UDI databases, like FDA’s GUDID. Are there just going to be a bunch of databases around the world, or will there be a degree of connectivity between them?
RF: I imagine that would be useful in particular for conducting global recalls.
JC: Right. That said, these are not trivial issues to deal with. Those two issues, though, are really what we have to tackle next after we get the UDI rules done and start the implementation process. That’s where I think the next big effort is going to be.
RF: Thank you very much for your time, Jay.
JC: Thank you. I’m happy to talk about this, and thank you for the opportunity