FDB: Thanks for speaking with us. Let’s start with your general thoughts about the UDI system, which has been with us now for five years. A code number that definitively identifies a specific instance of a medical device, we’ve heard the UDI colloquially dubbed the “NDC for devices.” From what you’ve seen, has it made a difference from the standpoint of the clinical user—and if so, how?
Dr. Drozda: I think the answer depends on whether a hospital has implemented point-of-care UDI capture. At Mercy, where UDI has been captured in the cath lab for years and is now occurring in the OR, UDIs have facilitated the provision of needed consumables at the point of care and even provided a last check on expiry dates. The clinical team also finds it much easier to assemble trays and manage inventory.
Dr. Tcheng: UDI has definitely made a difference. At Duke, we decided to introduce UDI first into our cardiovascular procedure areas from point of care to clinical documentation and inventory management. To do this, we involved numerous stakeholders, including supply chain, cath lab staff, the information technology team, physicians, and the heart analytics group. Several information technology aspects had to be coordinated, including bar code scanning, inventory management, clinical documentation systems and supply chain. To us, it has made a tremendous difference—just in time inventory, correct clinical documentation, and capture of the UDI in the implanted devices table of our EHR. It isn’t as simple as “just” substituting the UDI for another product identifier… instead, deployment requires a fair amount of leadership and governance.
Dr. Baskin: By September 2016, UDI labeling had become became a requirement for Class II and Class III devices, but there has not yet been a requirement that health systems actually use UDI in an effective way, I haven’t worked in a system that has deployed UDI for supply chain management, at the point of care, for documentation or billing, or in subsequent care of the patient treated with a device. Interestingly, when I was at the Hospital for Sick Children in Toronto, in 1999, we had a system in the Interventional Radiology suite that coordinated devices from supply chain management and just-in-time inventory control through device use or implantation to procedural documentation and related clinical research. This was, incidentally, before widespread use of EHRs and 8 years before U.S. law authorized UDI implementation. It is remarkable to me that 20 years later, my practice is further from the ideal rather than closer to it.
FDB: So, it sounds as if Duke and Mercy are making great strides, but that their sort of implementation is not yet widespread. Why is that the case? And why haven’t devices become more like drugs—which benefit from fully integrated, normalized data that’s accessible in all healthcare workflows? What’s holding devices back?
Dr. Baskin: On this and other health IT issues, my experience has been that there’s reluctance on the part of institutions to add features or complexity to electronic health systems that don’t have an immediate ROI or that aren’t required by law. Present systems for medical device purchasing focus primarily on cost per unit or bulk purchasing arrangements, and lack high-quality data on either patient or economic outcomes related to medical device use. Under these circumstances, this lack of evidence, along with considerations of non-interoperability, data blocking and the absence of regulatory standards, restrain progress.
Dr. Drozda: I’m not sure what is involved in workflows for NDC capture and use, but the major obstacles to implementing UDI capture at the point of care are all the processes and systems that must be changed to accommodate it. Supply chain, ERP, inventory management, and clinical information systems need to accommodate UDI, and that means more than just buying scanners. The supply chain team must use the data in inventory management, clinicians must begin using bar code scanning, and both will require training and support to make the change. All of this requires considerable investment by hospitals and expertise in change management.
Dr. Tcheng: I believe there are two primary impediments holding back the global use of the UDI: complexity and the absence of regulatory requirement. While the financial cost is relatively minimal, implementation takes a fair amount of effort across multiple stakeholders that must be tightly coordinated. The benefits of UDI are also not directly aligned; for example, the financial benefits over months may accrue to the hospital, but not to the individuals who actually have to use a scanner. Clinicians already include device information in procedure documentation, and having a computer insert that information can make document authoring harder, not easier. And the GUDID system doesn’t include much clinically relevant information about the devices themselves, so there is no incentive for clinicians to access the GUDID for data.
FDB: Would you say the need for data simply isn’t as great for devices as drugs?
Dr. Tcheng: To the contrary. Not infrequently, I need information at the point of care that is only on the package insert. But finding that information requires finding and reading a document while in the middle of a procedure—something that is quite difficult today. I would also like to make sure that the device I’m about to implant hasn’t expired. These are all things that a robust system with granular device data could achieve through the UDI.
FDB: That prompts me to ask what a practitioner would want to know or see or otherwise understand about medical devices and their attributes that he or she can’t access now?
Dr. Drozda: Immediate access to expiry date. Confirmation that the device chosen is the device the clinician wants. Recall information immediately prior to implant. Detail data that might be of use in implanting the device. Key characteristics of the device recorded at the point of care and incorporated into the chart that makes the information available to future clinicians—for example, an ED physician who ordered an MRI for a patient with an implant. Information like this might be available currently in an IFU, but that can be cumbersome to access in the midst of patient care. Drug information is similarly available in printed package inserts but with drugs, clinicians have easy ways to access that information electronically. Again, it’s not so much whether device information is available to the clinician at the point of care; it’s how readily accessible that information is.
Dr. Baskin: The need for data is compelling, but the belief that critical data—relevant to the individual circumstances of the patient and the proficiencies and preferences of the operator—cannot readily be accessed at the point of care frustrates current clinical practice and patient expectations.
The clinician likely wants to see device characteristics to promote optimal outcomes and minimize complications. The clinician also wants to know the indications for the device, its appropriate use, its known risks, and its limitations, both at the time of use or implantation and during its lifespan. It would also be helpful, similar to the case for drugs, to know how to recognize the device, its appropriate position and function, and the appropriate evaluation modalities for assessment.
FDB: In several medical conferences and white papers, we’ve seen reference to “AUDI data.” Can you talk more about AUDI data?
Dr. Tcheng: AUDI stands for “Augmented UDI Device Information”—a storehouse of granular, discrete, clinically relevant data about devices. The intent of AUDI is to provide a reference “single source of truth” about devices that could be used as a point-of-care resource and for clinical documentation, patient education, inventory management, comparative effectiveness, longitudinal assessment, device surveillance, clinical research, etc.
Dr. Baskin: AUDI data are those clinically and (arguably) economically relevant device characteristics that fulfill the needs outlined above, normalized by logical device categories or classes.
Dr. Drozda: Think of them as discrete data comprising key clinical attributes of a device that are not found in the GUDID but that can be accessed from a reference database via a UDI-DI link.
Dr. Drozda: For coronary stents, AUDI data would include dimensions such as length, diameter, strut thickness and structural material. Many of these hold for peripheral arterial stents as well.
Dr. Baskin: With central venous catheters, some are obvious: inside and outside diameter, length, material, lumen number. Some are not as obvious: taper, filled volume, coating and impregnation, MR compatibility, pressure rating, cross-sectional profile. And some are not direct physical characteristics: required vessel diameter, required sheath diameter, limitations of user modification, radio-opacity.
Dr. Tcheng: There are actually two projects demonstrating the use of AUDI data: the BUILD Project at Mercy focused on coronary stents and the RAPID project evaluating peripheral vascular interventional devices. Both projects rely on accurate “single source of truth” device parameter data to perform analyses that inform device objective performance criteria and longitudinal clinical outcomes.
FDB: How would that sort of AUDI information be used? Where, how, in what context?
Dr. Drozda: At the point of implantation, the clinician would have confirmation of key stent data for stent selection. All clinicians providing follow-up care would have key information; for example, for drug-eluting stents, there could be clinical considerations re: dual antiplatelet therapy discontinuation.
Dr. Baskin: In selection, it would be desirable to match characteristics to the needs, size, and vulnerability to complication of the individual patient. It would also be helpful to know the appropriate dimensions of the device vis-à-vis the target location, what the tolerances are for successful use or implantation, the appropriate sequence for deployment, etc.
FDB: Let me ask a different question, not about the use of data in the course of treating a patient, but in looking back at hundreds or thousands of patients to see how devices have worked. Would AUDI data be useful in that setting?
Dr. Drozda: Absolutely. It’s important to point out that the data would be used in analytics for research and by individual clinicians and Cath Lab leaders to assess device performance and safety. The fatal flaw or advantage of a stent may actually be an AUDI attribute shared across stent brands. Our article that ran in BMJ Surgery, Interventions, and Health Technologies includes a prototype analysis illustrating this point.
Dr. Tcheng: Definitely. Both comparative effectiveness and device surveillance are enabled by AUDI. For example, there are several different drugs released from drug-eluting stents that inhibit the stent restenosis process. A real-world application of AUDI would be to identify the universe of drug-eluting stents, and then classify the hundreds of stents by the drug being eluted. Yes, you could do this manually—subject to errors of transcription, etc. An electronic single source of truth would reduce the potential for error and would allow analysts to analyze the data—rather than spending time on data “janitorial” cleanup.
Dr. Baskin: Automated extraction of patient and AUDI data to relevant registries is the only practical way that timely and sensitive detection of safety and outcome signals is possible, both for devices overall and for individual device/operator/patient/salvage/outcome/cost data related to device characteristics. Such information may also be helpful to device manufacturers for market assessment, research and development, and cost reduction for post-market surveillance for approval and indications.
FDB: This all suggests that there are some significant opportunities for improvement related to medical device data, both in the provision of care and in the ability to evaluate devices in actual use. What must happen for the promise to become reality?
Dr. Baskin: On its own, UDI can achieve significant improvements only in strictly defined supply chain use cases, but its true value for clinicians lies in its ability to link granular data about devices to patients and outcomes over time and across venues. Further regulatory mandates would drive adoption, but we’ll also need a single source of truth of interoperable domain-specific data elements curated by a neutral party with a unified system of governance. There must be a master data management system and data warehouse for the database, with an ultimate aim to integrate UDI and AUDI into one system.
Dr. Drozda: Implementation of UDI by health systems bringing them to the point of care where they can be captured. Obviously, the technology must be in place to bring AUDI data to the point of care when a barcode is scanned, and reliable AUDI databases will have to be constructed and maintained.
Dr. Tcheng: A few thoughts:
- An app that can be used at the point of care
- UDI implementation roadmap “playbook” for enterprises to use
- Leadership and governance at each health enterprise advocating for the adoption and use of UDI
- Supply chain recognition of the potential of UDI
- “Plug and play” UDI and AUDI data interoperability—across inventory management systems, data integration engines, clinical documentation systems, supply chain master records, etc.
- Clinician advocacy
- Marketing/academic detailing
FDB: Any other comments?
Dr. Baskin: This is a public health issue. A federally funded demonstration project should be undertaken to show the value of AUDI in one or two use cases focused on the key drivers of adoption. If it can be shown that a granular meta-registry of clinically relevant device characteristics can improve the processes of device and patient selection, result in improved detection of critical safety signals, and reduce clinical, supply chain, and registry workflow burden, then an appropriate pathway to certification and compliance can be justified and defined.
Dr. Drozda: We have to create a burning platform or an ROI to motivate hospitals to implement UDI. That is the rub.
FDB: Then let’s hope this conversation provides some kindling to move this issue forward. Thanks to you all.
For more information on the FDB Prizm medical device database and how it can help your organization improve clinical, financial and operational outcomes, please visit https://www.fdbhealth.com/solutions/prizm-medical-device-database.