The use of unique device identification (UDI) labels for medical devices will soon begin, along with a global UDI database (dubbed GUDID) hosted by the U.S. Food and Drug Administration. Here, “soon” has a multiyear horizon, starting in September 2014 for Class III devices (and a few others), and extending through 2020. This then is either an opportunity for substantial procrastination or serious planning with respect to putting UDIs to good use. It is possible that some manufacturers will provide UDIs prior to the mandated deadlines. Purchasers may be able to use demand to speed such adoption. As UDIs become more available, and eventually universal, the long-running multiyear creation of the UDI system will be more-or-less complete.
The next key question is how to use the UDI system to achieve any or all of its touted values. Looking up a device in the GUDID if you have the number in hand is relatively straightforward, although it is likely to require typing in the number. Why exactly someone would want to do this is not clear to me since the scenario implies that you know the UDI, but do not have other documentation about the device. Perhaps the UDI might come from a medical record, but the idea that the medical record has the UDI but not other descriptors would be a highly undesirable use of the system.
Most other uses of UDI require systems involving several common elements. First, the UDI must be captured and made part of a permanent record. This record might be the hospital inventory, its supply chain, or it might be the patient’s medical record (i.e., the electronic health or medical record). To be practical, this will almost certainly require scanned entries, especially given the length of the code. It has also been suggested that UDIs be used as part of insurance billing records, presumably only when a device is being charged for separately. For individual patient records, what gets captured will have to be worked out. Most obvious is implants where the UDI, along with other information, would certainly be desirable. At the other extreme are large volume consumables where it is unlikely that the users will want to capture the UDI for every product used. In between are medical devices such as infusion pumps and ventilators, where you may want to know exactly which of several similar devices was used on a patient, especially after the fact. This could call for a great deal of scanning. Alternatively, and ideally, a device could automatically communicate its UDI to the appropriate record in an unambiguous way.
Once in a suitable record, the UDI must be searchable electronically to be most useful. Without that function, the use of such records to find a specific device would not be practical. It would also be helpful to be able to search or screen for only part of the UDI. For example, if there was a device recall covering a range of serial numbers, one wouldn’t want to have to search each recalled serial number individually. Instead, it would be better to first search using the device identifier portion of the code. Or perhaps a range of serial numbers or productions dates would want to be searched. The practicality of doing this depends in part on how the UDIs of recalled devices are communicated and how the patient data is structured. If the capture is by printed alphanumeric characters, then these would have to be typed into the appropriate search function. Better would be scannable codes. Ideal would be direct electronic transmission into the search function of the records to be searched. This requires a degree of standardization across multiple platforms, something that is certainly not at hand and I have not seen any specific plans to make this happen.
For public health purposes, recorded UDIs would have to be uploaded into a suitable higher level system, while retaining linkages to the original records and yet not compromising patient privacy. However, we are still far from having interoperable record systems that can communicate seamlessly either with each other or with a data integrator. Moreover, in many cases, that data will be scattered as one provider implants a device, others deal with the consequences of the device, and yet another takes the device out—all of whom at best maintain their own EHRs. Further, the UDI will have to be passed along from provider to provider since in most cases there will not be a single electronic record. Reporting device adverse events has a similar problem of getting the UDI into the reporting system in a way that it can be subsequently used to collate reports. Perhaps it can be expected that providing the UDI will become a required part of medical device reports or MDRs.
In these uses, it should be noted that there is a big difference between capturing the UDI and being able to process electronically the UDI containing records based on a full or partial search. For hospital equipment records, this capability will hinge on the equipment database software, whether it is an in-house creation or vendor provided. One advantage of using UDIs in equipment inventories is eliminating the many alternative ways in which the same device might be listed, for example, CT scanner; Scanner, CT; BrandX CT; etc.
On a somewhat peripheral note, the UDI rule uses the terminology “life-sustaining” and “life-supporting.” These terms have not been a part of the FDA medical device clearance process previously. and thus they have now been defined by the FDA using product codes. Such product codes have found relatively little other use outside of FDA functions. For the UDI system, eventually these definitions won’t matter because all devices will eventually be required to have UDIs. However these FDA designations may in turn play a role when other organizations such as The Joint Commission (TJC) or the Centers for Medicare & Medicaid Services (CMS) use the same terms. This perhaps not-always-welcome clarification is different from what has occurred previously. For example, there was at one time a good deal of discussion about whether a defibrillator was life sustaining. TJC’s George Mills ended this discussion by saying yes, they were. Similarly, the much discussed CMS maintenance clarification uses the terminology “critical equipment,” which is defined as that “for which there is a risk of serious injury or death to a patient or staff person should the equipment fail.” Instead of puzzling this out on a case-by-case basis in each hospital, wouldn’t it be handy if you could just look the device up and determine if it met this ambiguous definition? Or maybe not if too many things were thus categorized.
UDI is coming slowly, after a long period of discussion and development. Will we put it to good use? Will we get the tools necessary to put it to good use?
William Hyman, ScD, is professor emeritus of biomedical engineering at Texas A&M University. He now lives in New York where he is adjunct professor of biomedical engineering at The Cooper Union.