Mercy doctor advances potential for mining performance data on implanted devices

October 15, 2014


Millions of people worldwide live with an implanted medical device, whether a heart stent or an artificial hip. But currently there's no standard tracking system for implanted medical devices. If there's a problem with an implanted device, it's difficult to quickly identify and notify patients who have the devices in order to address or correct defects.

Mercy cardiologist Dr. Joseph Drozda Jr. leads a team of physicians and researchers working to change that. The team is part of the Healthcare Transformation Group, a collaboration between Mercy, Geisinger Health System, Intermountain Healthcare, Kaiser Permanente and the Mayo Clinic that seeks to improve health care safety and quality and contain costs by driving changes in the supply chain.


The current priority for the Healthcare Transformation Group's research and development team is to support the creation and adoption of a globally standard way to identify all medical devices. The U.S. Congress mandated the establishment of a unique device identification system for medical devices several years ago; last year the Food and Drug Administration published the requirements to implement the system. This system relies on device companies using unique identifiers that can be read by people and machines, and submitting certain information on these devices to an FDA-run database that will be made public. The unique device identifier system will be phased in from 2014 to 2020, starting with the highest risk devices, which are usually devices that are implanted, such as heart valves or pacemakers.

Health systems are required to report adverse events related to devices, and, in the future, they will be asked to include unique device identifiers in those reports.

Chesterfield, Mo.-based Mercy led a government-funded demonstration project to prove that the inventory tracking system, combined with the outcome information in the patients' records and descriptors about device attributes will make it possible for researchers to more rapidly assess the quality and safety of devices in large groups of patients.

Catholic Health World talked to Drozda about the health care systems' collaboration to develop a tracking system for implanted medical devices.

Was there a device recall or failure that prompted Congress to act?
The push for the unique device identifier began probably in the 2000-2002 timeframe, but it took full form in 2007 when the Congress put a requirement in FDA-enabling legislation that a unique device identifier system be established. There had been some high-profile device failures prior to this legislation, but the problems most often cited currently actually occurred after 2007. These included defective defibrillator wires that fractured causing inappropriate shocks leading to arrhythmias and deaths and failure of metal-on-metal hips.

What was the goal of the FDA-funded demonstration project?
The demonstration involved implementation of prototype unique device identifiers into the electronic health information systems at Mercy. We captured device identifiers in our supply chain database, our inventory software, our cardiac catheterization clinical software and our electronic health record. We then created a database with clinical data and unique device identifier-related device data. We are able to use that database, which we refresh on a regular basis, for ongoing monitoring of devices for safety, for performance and for research. We have produced some initial analyses out of the database that have shown the face validity of our data.

What sort of scientific learning might come from being able to track individual medical devices and integrate that data with other information?
Let me answer that by describing the database we created. We developed a system for capturing data on coronary stents as an initial effort to establish a generic methodology that could be used for all implanted devices.

This research database contains data extracted from the electronic health record at the time of implant and on a regular basis after that for follow-up. The database includes the unique device identifier, which is just a number. What we had to do was link the number to device attributes. So, for example, with coronary stents, these attributes include a stent's length and diameter, the material it's made of, and which drug is impregnated in it if it's a drug-eluting stent. Those kinds of attributes give you clinically meaningful information about the device.

We also included mortality data on our patients from the Social Security Death Master File. We were then able to do an analysis of coronary stents with respect to death rates following implantation. Baseline characteristics are in the database, so we can look at targeted populations, whether men, or women, or patients over 65, for instance. We can compare one drug-eluting stent against another, or stents of different diameters, or stents of different lengths with respect to their impact on mortality and other outcomes, for instance, the percentages of patients having heart attacks at 30, 60 or 90 days after stent implantation or the percentage of patients having repeat stenting procedures. We can refresh the database periodically with clinical data to monitor how patients are doing and capture these outcomes.

Ultimately, to realize its value, we need to be able to link our data set to identical data sets from other health systems. That's what we're going to try to do with the Healthcare Transformation Group. The four other health systems that are part of the group will create the same data sets that will be linked in a distributed data network, which will give us the power of a lot more patients enabling some sophisticated analyses. Also, large numbers of patients will allow us to identify issues sooner, and that's important when looking at safety.

Are there privacy concerns related to being able to identify people based on their implantable medical devices?
If we are able to create this distributed data network, the clinical data would continue to reside behind the firewalls of the individual health systems with the data available through queries for purposes of tracking and research. We can provide the (privacy) protections that are necessary. If you have a device in you that's endangering you in some way, I would think you'd want to know ASAP. If we don't have data systems that can find you and find that you've got the device, you are at some risk of avoidable harm.

If there is a problem with a device, will this system allow its use to be halted more quickly?
We can ultimately put in some proactive monitoring right on top of our data that can pick up signals indicating a possible problem that needs further investigation. I think a key issue is capturing the denominator when isolated incidents occur. For the FDA, its current way of monitoring devices depends on isolated reports: "We had about four of these devices that had problems." Well, is that four out of a thousand or four of a hundred thousand? They have a hard time getting the denominator. We will have that, we will be able to produce rates as we go and be able to say this is a signal, and be able to do it much more quickly.


Copyright © 2014 by the Catholic Health Association of the United States
For reprint permission, contact Betty Crosby or call (314) 253-3477.

Copyright © 2014 by the Catholic Health Association of the United States

For reprint permission, contact Betty Crosby or call (314) 253-3490.