Awareness is growing about the challenges of developing and maintaining safe and effective wireless medical devices. What with IEC80001 moving forward (due to be finalized next year) and the recent series of wireless medical device workshops, people in hospitals and among vendors are asking more of the hard questions about wireless. Amongst the turmoil, participants are jostling for position. This post looks at common problems with Wi-Fi, a report from U.K. alliance ERBI, and some alternatives to Wi-Fi.
Problems with Wireless
Those of us who are old enough, think back to the golden age of wireless medical devices — channelized analog telemetry. These systems were so basic and limited in scope (a couple dozen transmitters typically covering just a single 30 bed unit) that they had few problems and required little maintenance. Today, larger hospitals are pushing the envelope with a few hundred patient monitors and a thousand or more wireless infusion pumps. These wireless devices are using sophisticated client radio/access point (AP) communications protocols to maximize capacity, whether using Wi-Fi or WMTS. We’ve since left the golden age far in the past.
Radio frequency (RF) spectrum is a shared resource. There’s no getting around that fact, even with “dedicated” spectrum. The ether in which wireless signals move is like gases in the atmosphere or chemicals in water. There are no ways to practically segregate RF signals to specific areas, except for a Faraday cage. In a health care facility, some shielded rooms in Radiology qualify as Faraday cages, but little else. Much of the rest of a health care facility consists of objects and structures that seem to perversely confound and obstruct RF communications in ways like partially blocking and attenuating signals, creating multipath interference, and radiating both intentional and unintentional interference. Intentional interference is where two or more users of a portion of wireless spectrum get in each others way, disrupting or degrading the communications of one or both parties. When there are problems with two or more wireless devices using the same spectrum, this is intentional interference, often referred to as coexistence problems. Unintentional interference comes from electromechanical devices that accidentally spew RF signals as a consequence of some degradation or failure. Common sources of unintentional interference are florescent light balasts, blow dryers, paper shredders, elevator motors, or faulty microwaves. You can see a bunch of examples of RF interference on a spectrum analyzer (which everyone doing wireless medical devices should have, and know how to use) here.Read More
The archetypal wireless medical device is the telemetry monitor for measuring electrocardiographs . First introduced in the 1970s, cardiac telemetry systems were pretty straight forward. Analog signals were transmitted with each telemetry transmitter/receiver using its own dedicated channel. Medical device vendors placed ceiling mounted antennas connected with coaxial cable back to central radio frequency (RF) transmitter/receivers in a wiring closet. There were no other wireless medical devices. Nor were there any wireless LANs – or even wired local area networks, for that matter.
A lot has changed in almost 30 years – I mean besides feeling older.
The nirvana that was the 1970s came to an abrupt end on February 27, 1998 at 2:17 pm, when, “WFAA-TV channel 8 television began broadcasting on digital TV channel 9 and continued until 10:35 p.m., shutting down transmission a few times to allow a tower crew to work on the antenna.” This and subsequent tests of digital television broadcasts by the Dallas broadcaster, knocked Baylor University Medical Center’s (BUMC) telemetry off the air. Fallout from this intentional (and completely legal) interference resulted in the creation of the new WMTS frequencies for use by telemetry monitors. Between that fateful day in 1998 and 2006, BUMC has spent $6.6 million shifting frequency and upgrading the telemetry systems at their hospitals. (You can read about BUMC’s ordeal reprinted from the AAMI publication Biomedical Instrumentaiton and Technology Journal story on this FDA web page.)Read More
I received the following blog post from Stephen Olsen, Principal at Integra Systems. Steve has spent more than 20 years in the wireless industry in engineering, sales and business development. Steve’s wireless experience extends beyond health care to include public safety, cellular and 802.11.
In the past I’ve extended an invitation to a few select industry experts and thought leaders to post their writing. Steve is the first to take me up on my offer. Enjoy:
Over the last few years, MobileAccess and InnerWireless have generated considerable interest in broadband Distributed Antenna Systems (DAS) for the healthcare market. These systems can support a wide range of applications (WiFi, cell phones, mobile radios, pagers, WMTS) and frequency ranges (400/800 MHz up to 6 GHz).
The appeal to providers is the idea that a broadband DAS will remove all wireless headaches: no more cell phone complaints, WiFi will work better, no more dead spots for mobile radios, no more tricky RF interference problems, etc. Disappointment ensues when the DAS does not live up to its promise.Read More
I was in hog heaven at this year’s AAMI meeting. Connectivity was a major theme, and during every time slot in the program there was at least one presentation dealing with connectivity. During my presentation Monday afternoon, there was one I really wanted to see that dealt with alarm notification.
Lots of discussion centered around the evolving role of biomeds and clinical engineers and the kinds of training they might need in the future. There were rumblings from some in the ACCE who wanted to hold their annual meeting at HIMSS next year rather than AAMI. There certainly is a life-critical systems role that needs to be filled, and clinical engineers could fill that role. To this observer, it seems that clinical engineers will slowly become marginalized if they do not move in the “systems” direction. Even biomed techs will need IT skills to manage and support increasingly complex and pervasive medical device systems.
During the GE sponsored breakfast, there was a session on managing RF in your hospital. Reportedly the perennial “WMTS versus ISM” debate reared its tired ugly head. For many reasons mentioned here in the past (just google “WMTS” in the search box on the left colum). The WMTS bands will never have the bandwidth or (more importantly) the management tools to support more than a small portion of the wireless medical devices in a hospital. Only the usual suspects can even afford to develop the prorpietary radios required for WMTS, which is why 802.11 has seen so much uptake with device vendors.
But the inherent limitations of WMTS do not make 802.11 a slam-dunk. In fact, recent experience has highlighted the need for more rigorous RF engineering, wireless LAN design, and ongoing RF and network monitoring to ensure a reliable network. Hospitals are perhaps the most hostile environment for wireless networking. When it comes to networks, hospitals are faced with both selecting a hardware vendor that best meets their needs and a VAR (value added reseller – the indirect reps used by IT vendors to sell their products) who really knows what they’re doing. Only the best VARs can design and install a reliable network that supports all the big apps: data, wireless VoIP, positioning, and medical devices.
In a nod to presidential politics, “It’s the workflow, stupid.” To most, connectivity is about extracting data and moving it some place else. The real objective is to automate workflow – and how connectivity is implemented has a huge impact on what workflows it supports, and ultimately the usability of the system. A fundamental piece of this workflow is patient context, the association between a patient, their medical devices, and the data that comes out of them. Patient context remains a concept that’s poorly understood by most users and vendors. Many still try to fudge patient context by associating the patient to a port number or bed location. Guess what? Patients move, and mobile devices especially, must establish patient context in the device itself to be safe and effective. I would love to see some of the fantasy-based risk analysis and mitigation documents done for certain connectivity features that I saw this week.
All of this gets to another big change reflected in this weeks conference. Stand alone embedded products are evolving into real systems that extend functionality way beyond the box itself. This “systemization” of medical devices requires some changes in thinking. No longer can you focus on building safe and effective boxes, and after the fact plugging them together with other stuff and be sure the result is still safe and effective. Nor can you manage and support interconnected devices simply by maintaining the device – the entire system must be configured and maintained as a whole.
One of the good things to come from the increased involvement of IT in device connectivity is their insistence on a test system to support the “production” system. They do this with all their software systems. An indicator that connectivity is an afterthought is the total absence of test fixtures for an integration lab. Another symptom is the scarcity of such labs in hospitals and the limited capabilities of most manufacturers’ verification labs. As systems grow and become more complex, hospitals will increasingly demand support for these labs – in the absence of test fixtures, that means customers with clout will insist on indefinite loaners so they can effectively maintain their systems.
During the ACCE Clinical Engineering Symposium Saturday morning, Bridget Moorman referred to medical device connectivity as “brittle.” I know more than one person had an epiphany upon hearing that term. Any change, no matter how small, along the chain from medical device to target computing device renders the device interface inoperable. Device firmware changes, pin-outs, cable connections, terminal server configurations, network configurations, and interface configurations – on either side of the interface – all result in failure. Planning for these interfaces (hopefully by the vendor before product development) must take this brittleness into account. At the very least, customers must be able to monitor their connectivity all the way to the device, not just a server or terminal server.
Finally we come to FDA regulatory issues. I met an FDA representative in the exhibits. She works on the Issues Management Staff, a tiger team that addresses patient safety related issues that reach a point where they must be dealt with. Can you guess one of the simmering issues that may soon become an Issue? That’s right, medical device connectivity. Much of the current regulatory framework (both vendors regulatory strategies and how the FDA manages the process) is based on standalone medical devices, and “oh, by the way, it gets plugged into all this other stuff to do… stuff.” We can expect to see regulatory perspectives shift increasingly to a systems view, especially when multiple vendors are involved.
The contortions many vendors go through to avoid FDA regulation is a symptom of this spreading systemization of medical devices. While the FDA has a responsibility to ensure safety and effectiveness, they are also responsible for accomplishing their mission in a way that doesn’t drive undeserving vendors out of business or stymie the development of innovative solutions that promise even better safety and effectiveness. Don’t expect them to accept the status quo for long. I ask everyone who’s skirting the regs if they are committed to building a quality product, and the answer is inevitably yes. All it usually takes to get a 510(k) is compliance with a basic quality system (the FDA’s Quality System regulation) and 60 days for the FDA to process your 510(k) paperwork. And yet the reticence to be regulated suggests that things like prototype code makes it into finished products all too often.Read More
I came across some interesting posts in the Biomed Listserv. A biomed from a 250 bed hospital is looking for feedback on GE and Philips telemetry systems. This 260 bed community hospital is going to buy a 12 channel system, and ramp up to about 150 of the devices over the next few years. All sorts of questions came to my mind, and came to this fellow as well:
Instead of what to purchase, please assure that you ask the question – from a clinical patient perspective – what are you attempting to accomplish? Telemetry is intended to allow patients to ambulate and improve their outcome and therefore leave the hospital sooner. It was developed specifically to assist post open heart patients get out of bed and let clinical staff still watch the heart. If you are installing hundreds of channels of telemetry to just add monitoring – what is the patient benefit?
Also the clinical operating concerns can not be underestimated. What nurse is responsible for patient care? How many nurses per floor? Is there a central station, monitor techs? Most telemetry systems in hospital are (my opinion) not properly set up technically or operationally.
Most patients on telemetry do not ambulate, so why not have bedside monitoring? Most nurses taking care of patients on telemetry can not even see the ECG being technically transmitted as only the monitor tech can. Most monitor techs are not licensed to manage patient care – RN’s are. Most telemetry units also lack physician criteria for admission and discharge causing the need to always have more telemetry. I think most adult hospitals could benefit from the lessons learned at pediatric hospitals. If the patient needs to be monitored and is not ambulatory, put a networked bedside monitor at the patient bedside. Some vendors – Spacelabs – even have modules that allow for telemetry to be at bedside for those patients that require same.
Great questions. Fitting the monitoring device to match the patient profile is important. And effective alarm notification is essential.
Every hospital has some unmonitored patients that could benefit from being monitored. Patients on pain medications, especially PCA pumps, should be monitored. According to the literature, almost half of all cardiopulmonary arrests in hospitals are unwitnessed. While the survival rate of witnessed codes is 22%, the survival rate for unwitnessed codes is just 1%. Increased monitoring can reduce adverse events, improve outcomes, reduce length of stay, and minimize legal liability.
For these reasons and more, the current standard of practice for patient monitoring is changing. A greater number of hospitals are using telemetry monitors to monitor previously unmonitored patients. Patients that would normally not qualify for the telemetry unit or some other high dependency unit are being monitored – and usually this means telemetry packs.
Telemetry is a good choice because these patients are usually ambulatory, and telemetry transmitters are light weight and small. At the same time, telemetry monitors are a bad choice for several reasons. Telemetry transmitters have no local alarms; you can be right next to a patient in arrest, and unless you see their lips turn blue (or some other physical sign) you won’t know there’s an alarm condition. Telemetry alarms traditionally annunciate at the central station, and some installations include additional audible alarms, message panels or flashing lights. A related weakness of telemetry transmitters is the lack of a display. To view the patient’s physiological parameters you must leave the patient and go to the central station.
Another telemetry system limitation mentioned above is the requirement to have someone actually watch the central station displays. Some hospitals have monitor techs man central stations in a central location – frequently called the “war room.” While this is a safe approach, it is also very expensive – an 800 bed hospital can spend $1 million per year on their war room. Other hospitals distribute their monitoring techs on individual nursing units. Depending on patient volumes on nursing units, this could be more expensive than the centralized approach. Finally, many hospitals put central stations in nursing stations and make the nurses on the unit responsible for surveillance. This is certainly the most cost effective approach, but also the most vulnerable to alarm fatigue, especially when you have one or two patients who constantly throw off false positive alarms.
The final challenge is cost. Telemetry monitors, as inadequate as they are, cost about $8,000 per channel – still too high for the many hospitals that have yet to adopt broader patient monitoring. A contributing factor to cost is the use of WMTS by GE and Philips. While both vendors justify their use of WMTS on the fact that it is a “protected” frequency, that provides protection only against intentional interference (people purposely using the same frequency). The major source of interference in hospitals is from unintentional interference – noisy hair dryers, florescent light ballasts, elevator motors, microwave ovens, the list is endless. While WMTS is a frequency band authorized by the FCC, there is no mandate to use it. In fact, there are no operating standards for WMTS to ensure coexistence between vendors. When they first switched their access points over to WMTS, GE and Philips interfered with each other’s systems. It was only a gentlemens agreement (and time) that facilitated technical adjustments to allow both vendor’s systems to operate in the same hospital. Systems using WMTS simply cost more money than systems that use your hospital’s infrastructure – and the wider you deploy their proprietary WMTS infrastructure the more it will cost you when you want to change vendors (and you always do, sooner or later).
While we can’t solve the cost issue today, there are alternatives to the traditional telemetry monitor. Our perfect solution would include a real patient worn monitor, meaning a device with a display and local alarms, and you really should have both ECG and SpO2. Such devices include the new Draeger Infinity telemetry system, and a couple of Welch Allyn monitors – the Micropaq and the Propaq LT. The Draeger monitors run on 802.11 b/g, and Welch Alllyn just announced support for 802.11a/b/g on their monitors at HIMSS. None of these products is perfect, but you get important clinical features and you’re not investing unecessarily in a proprietary infrastructure that only works with one vendor’s products – and in the case of GE, only supports telemetry – all their patient monitors run on WiFi.
Pictured above is the Draeger Infinity telememetry monitor, shown in bedside dock with trickle charge.
UPDATE: Reader Dan Davis MD suggests that most hospitals looking to broaden their patient monitoring are trying to reduce failure to rescue incidents (that’s patients who have an adverse event, go into cardiopulmonary arrest and usually die). Certainly saving lives is a primary motivator for increased patient monitoring. But I would argue that there is a greater need (and hopefully market demand) for electronic surveillance that alerts caregivers long before patients become obvious failure to rescue statistics. This requires the ability to identify patients who are pre-arrest, at a stage when clinical intervention is less expensive and much more successful. Patients who arrest in hospitals have low survival rates – 22% for witnessed arrests and a grim 1% for unwitnessed codes.
Dan also mentions Hoana Medical, and their passive non-invasive sensors that can be incorprated into patient beds as a possible solution. I’ve written about them a number of times (here, here and here), and they have a very interesting system – check them out. The only thing that I can see that keeps it from becoming a natural choice for hospitals is that the kind of patients who would benefit most (currently unmonitored patients) are also active – they are encouraged to get up and walk around. Hospitals need solutions that cover patients in bed and when ambulatory.Read More