At Lab InfoTech Summit 2007 in March, Anand Dighe, MD, PhD, gave a talk titled “Radiofrequency Identification for Tracking Assets and Reducing Errors.” Dr. Dighe is assistant professor of pathology at Harvard Medical School and director of the core laboratory and director of information management in the Department of Pathology at Massachusetts General Hospital, Boston. An abridged version of his remarks follows.
RFID, or radiofrequency identification, is a technology by which a physical object can be automatically identified, very much in the same vein as bar codes. This technology is all around us. I personally use RFID to pay for my gas with a Mobil Speedpass and to get through doors at the hospital with my hospital ID.
The RFID market in 2006 was about $4 billion globally. Health care in the United States was only about $90 million of that market. However, if you look at what hospitals say they’re going to do, about half of them have RFID spending planned in the upcoming year. And that’s up from just 10 percent two years ago. So there’s this big upswing in RFID interest in health care. People have found very useful applications for equipment tracking, patient tracking, and process and capacity management.
The two main types of RFID are passive and active. They each have different strengths and weaknesses. Probably 90 percent of the RFID tags produced globally are passive RFID tags. They’re durable, they’re cheap, and they can be made very small. There are three components to an RFID system: the tag, the reader, and the middleware. The RFID tag in many systems is a little bigger than a postage stamp. It has coils on it, and it has a microprocessor. It’s called a passive tag because it’s passively powered. There’s no battery on the tag; it gets its power from the RFID reader. The memory capacity of a tag generally is hundreds to thousands of characters.
The first step in a passive RFID system is energizing the tag. (Related article: Graphics) The passive RFID tag gets its energy from the reader, which creates an electrical current in the tag, and that current can then do useful things. It will access the data that’s on the microprocessor and send radio waves encoding that data back to the reader. The reader’s job is to pass the tag data to the server and middleware, which are then going to interact with you—display something, access an application, or open a door.
It’s important to compare and contrast RFID and bar codes. Bar codes have had a pretty dismal adoption rate in health care. In the retail sector, before they started using bar codes, they created a standard—the universal product code, or UPC. Further, in the retail sector they could measure success in terms of reduced labor costs and increased customer and revenue growth. Health care is a fragmented industry. We don’t have robust standards. As a result, we have bar codes on many things. If you go into a hospital, you’ll see bar codes all over the place, but for the most part no one’s actually scanning them or using them. When we do try to use bar codes in health care we often do it in a way that introduces a new and unfamiliar workflow. If we scan the wrong bar code or if we misprint the bar code, serious new errors can be introduced. Successful use of bar codes in health care can be difficult to measure as it is often hard to quantitate outcomes such as better patient care or patient safety.
Why is scanning a patient wristband bar code so difficult? First, you’ve got to find the wristband. There’s probably an IV line that you have to get out of the way. You have to contact the patient, with some nosocomial infection risk, and rotate the wristband to find the bar code. It’s not going to scan well if it’s on a curved surface. And soiled or wrinkled wristbands aren’t going to work at all.
Compare that to scanning an RFID wristband: You point the reader within three inches of the wristband, and it scans with a near-100 percent success rate. The big differentiator is the marginal cost. A bar code is essentially a font, so its cost is solely the cost of the label. There’s real engineering that has to be done to make a passive RFID tag. The passive RFID tags most commonly used in health care are in the $1 range, so we have a pretty expensive proposition.
Passive RFID tags can be made in many forms. There are smart labels for blood bags and patient ID wristbands with embedded RFID tags. There’s even an FDA-approved human-implantable RFID tag that can go into your deltoid or forearm. It has a unique number in it and can be linked to your medical record. At a bar in Spain, they’ll plant a passive RFID chip in your arm and link it to your credit card, so the next time you come in you can pay for your drinks with your arm.
Passive RFID tags are useful for many things other than paying for your martini in Spain. Passive RFID chips are being used extensively in distribution centers, manufacturing plants, and warehouses. A forklift with cases of RFID-tagged products goes through the warehouse exit door and the tags are automatically read by a passive RFID reader positioned next to the door, giving the company instant inventory and supply chain visibility. Another application is a shelf with a built-in passive RFID reader. When you put RFID-tagged packages onto this “smart shelf,” you have instant inventory. Target, Wal-Mart, and the U.S. Department of Defense are all implementing RFID due to its promise to improve their supply chain and reduce costs.
Passive RFID is used in situations where you want to know what something is, and perhaps where it was last. If you want to know where something is right now, you’re going to want to use active RFID. (Related article: Active RFID: What is it, where is it now?.) Let’s say an active RFID tag has Jim Smith and Jim Smith’s medical record number on it. This active tag is going to broadcast, at regular intervals, radio waves out into space looking for a reader. If you have a network of RFID readers, you can then locate that tagged object in space.
There are two types of active RFID reader networks. One is based on standard Wi-Fi connectivity, which uses your existing wireless infrastructure. The problem is that you may need more access points to locate tagged objects than you may have in your current Wi-Fi network. The other type of reader network is a proprietary network that you have to set up separately from your current wireless infrastructure, and that can be costly.
Another way of classifying RFID tags is by the frequency with which the reader and the tag interact. Certain versions of passive RFID tags are classified as being “low” or “high” frequency tags. These are attractive in health care because they penetrate water and tissue very well, although metallic objects can interfere with their signals.
Other types of RFID tag systems use ultrahigh frequency (UHF) and microwave frequency tags. These types may be either active or passive tags. Probably 90 percent of the RFID tags currently in use are UHF passive RFID tags. Those are the ones used in global supply chain management applications. Many of the active RFID technologies are in the UHF range as well.
The third classification of tags is memory, that is, is it a read-write tag, or is it a read-only tag? We’ve been talking so far about read-write tags—tags that have Jim Smith and Jim Smith’s medical record number on them. In reality, these tags are uncommonly used in large RFID implementations. It sounds like a great idea to put the patient’s medical record on the wristband, but there are reasons not to do that. You have to program all that information and worry about it being up-to-date, and you have to worry about data corruption. The more typical use of RFID is read-only. You use a read-only unique integer programmed onto the tag by the factory. You associate this unique number with Jim Smith and then whenever you scan this number, the middleware retrieves whatever component of Jim Smith’s information is needed.
There’s a lot more to RFID, though, than just the reader and the tag. The cost of the tag receives a lot of the focus, but we also have to consider the cost of interfacing. We’re not going to be able to simply link an RFID system to our existing enterprise applications. We’re going to need a layer of middleware and perhaps a proprietary application to view and make use of the data.
Some of the new tags being produced combine RFID with sensor technologies. You can have sensors for virtually anything, including light, temperature, vibration, or moisture. One example currently in use is an RFID tag that has a sensor for changes in acceleration. If you have an active RFID tag that broadcasts itself only when it moves, that’s going to save battery life. In the case of an RFID-tagged ultrasound machine that’s not moving, you don’t want the active RFID tag to wear down its battery life by constantly broadcasting its current location.
We’ll now shift gears and talk about the experience with RFID at Massachusetts General Hospital. I’ll talk about two projects: blood transfusion safety and asset tracking in the operating rooms.
A blood transfusion, although much, much safer than it has been in the past, is still a hazardous treatment to receive. One in 38,000 transfusions is with ABO-incompatible blood. Mistransfusions are most often errors caused by failure to pay attention at a critical moment. That moment in blood transfusion is typically the bedside clerical check. Blood in the operating room is often given at a time when the patient is crashing. It’s not a great time to calmly compare the identification numbers on the patient wristband and blood bag. Luckily, these types of errors are amenable to technology. Dr. Walter Dzik and collaborators at Mass General have been using passive RFID tags on wristbands, and they’ve been putting a second passive RFID tag on the blood bag. There must be a match for the patient to get the blood. You walk the blood into the operating room and scan it with a passive RFID reader. If the identification number on that blood is wrong, it’s going to display the message, “Stop. No match,” and an audible alarm will sound, so it’s very clear to the user that they made a mistake.
The other example uses active RFID. Operating rooms generate about 40 percent of a hospital’s revenue, and cases rarely start on time. The Mass General Operating Room of the Future (ORF) is both a laboratory and a working OR. The ORF is a test site for looking at new operative and perioperative technologies, including RFID. The question asked by anesthesiologist Dr. Warren Sandberg [co-leader of the ORF project] and his collaborators was: “Can RFID be used to improve operational efficiencies by revealing new relationships between items?” The desire was to look at operating room assets and see how they interact.
High-value assets—patients, surgical staff, cleaning crew, mobile equipment—are all tagged with active RFID. Just being able to see where assets are on a map is useful, but the real utility here is to have a rule in your middleware that says, “After the patient’s and surgeon’s RFID tags have left OR 38, automatically text-page the cleaning staff to clean that operating room.” Another rule could be, “After the cleaning crew has left, alert the staff to start the next case.”
The results have been encouraging such that Mass General has spread the technology to all the operating rooms now, from just the few that were used for the pilot. The early results showed that time between cases is faster, with additional cases able to be performed in the same amount of time, just due to knowing the status of the operating rooms and where the assets are.
One caveat for RFID in health care is “Just because you can doesn’t mean you should.” As highlighted by JCAHO, wrong-sided surgery is a serious patient safety concern. One solution is a passive RFID tag that gets programmed with whatever surgery you’re going to have. For instance, “Remove left kidney.” The adhesive tag gets applied to the patient’s arm, and when the patient arrives in the operating room, someone on the surgical staff has a handheld computer with an RFID reader. They scan the patient’s RFID tag, and hopefully they see the operation they expect, “Remove left kidney.” The startup costs for such systems are about $100,000, and the tags cost about $2.50 each, so if you’re doing 40,000 surgeries a year, you’re talking about another $100,000. Also, the system has to get the operative information from somewhere, so it needs to interface with the surgical information system, and you’ve got to maintain it and train people to use it.
The approach taken at many other hospitals is decidedly lower-tech. They’ve used a Sharpie magic marker to comply with this JCAHO initiative by writing cues for the correct side of surgery on the patient’s skin. The key is the right technology for the right role, whether it’s a Sharpie, a bar code, passive RFID, or active RFID. You’ve really got to demonstrate operational effectiveness, not just efficacy.
One thing we’ve learned is that it’s important to understand the public perceptions of RFID. This is a quote from a California state senator: “How would you like it if, for instance, one day you realized that your underwear was reporting on your whereabouts?” This sounds way out of the mainstream, but it’s really not. There have been boycotts of big companies due to their plans to put an RFID chip on every product in their store.
It is possible that in the future every item you buy with a passive RFID tag could be directly traceable to you. The unique tag ID would be stored with your name and credit card number at checkout. So the idea is, you’re walking down the street and someone can scan you with an RFID reader and find out who you are, what books you’re carrying, and, perhaps, what underwear you’ve got on.
Potentially, any unauthorized “rogue” RFID reader could read a tag. The data could be encrypted but that adds complexity and cost. The tag could also be detached or deactivated at checkout, but operationally that would be difficult. So the bottom line for us in health care is that this is a real risk and a real public perception that’s out there. We have to address this on any RFID project we consider.
There is a significant opportunity to use RFID to improve the practice of anatomic pathology. In anatomic pathology, we have a large number of high-value assets, and they all need to be uniquely identified. We also have a lot of points in our process where patient ID is essential and places where we make mistakes—accessioning specimens, making cassettes, labeling slides. Another complication of pathology not found in many other industries is that our high-value assets are often immersed in caustic chemicals or stains, so we have to find technology that will work in that setting.
Step one is not to think about RFID. It’s to understand your process. You’ve got to map your current process, identify steps where you think auto-ID might make a difference, and decide what your goals are. It’s really important to assess whether bar codes can accomplish the same thing. Another big consideration is that any new technology must interface with existing systems. We want to have the technology help drive and improve our existing workflow.
In the histopathology lab, after implementing bar codes, we have people scanning things all the time. We would like to make the scanning of the cassette and slide an automatic part of the process, and this can be done with RFID. Build RFID tags into the cassettes and the slides and build readers into the workstations people use, and you can make the process seamless and much safer.
ProPath is a high-volume AP practice in Dallas that uses Cerner CoPath Plus. What they’ve done in pilot mode is to incorporate a passive RFID tagging system. They have a glass-encapsulated passive RFID tag placed in each specimen cassette. The tags come pre-encoded from the factory with unique serial numbers. They’re not writing “colon biopsy” or any information about the specimen on the tag. All they’re doing is associating that unique tag number with the specimen. The tag is then scanned at numerous points throughout the processing cycle to improve patient ID, drive workflow, and create an audit trail. You can’t just plug these RFID readers right into Cerner CoPath Plus. You need a layer of middleware that is going to permit detailed tracking information and interaction with enterprise applications.
What’s the value proposition for RFID in pathology? It’s still unclear. Hard return-on-investment data are difficult to obtain. You really have to measure things and get a sense of what the costs of your errors and inefficiencies are and how you can improve things with RFID. There’s a lot of soft ROI, of course, like employee satisfaction—people don’t like to scan those bar codes all the time—and, potentially, patient safety. All of that may be hard to quantitate, and it may be hard to justify the expense of putting passive RFID tags on all your assets in pathology.
A JCAHO national patient safety goal that’s proposed for 2008 says: “The organization investigates and initiates the planning for the use of technology for the assist with patient identification.” They’re mainly talking about bar coding, but it’s clear they really want us to start identifying our assets better.
It is tough to figure out where this is going in health care. Health care is fragmented and there are few robust standards. As networks get more complex, the need for standards grows. Our network in health care is getting more and more complex, yet we don’t have a good set of standards like the consumer products industry or the manufacturing industry. Lastly, we’re going to be collecting a lot of data with RFID. The data analytics are just as important as the data gathering. If you can’t do anything with the data, you might as well not collect it. It is essential to tie RFID systems into our routine workflows and put in the hard work of interfacing RFID with our existing systems. When that is accomplished, RFID can become a valuable tool to improve the practice of pathology.