BBI Contributing Editor

SAN ANTONIO, Texas While the electronic medical record (EMR) more correctly, the computer-based patient record (CPR) has been a goal and a market about to emerge for more than a decade, things really seemed to come together at this year's Toward an Electronic Patient Record (TEPR) conference. More than 2,500 gathered here in mid-May at this year's meeting, along with a wide array of hopeful EMR and middleware vendors. TEPR is attended by companies offering practice management software, EMR and charting software, middleware, hardware, carts and wireless networking solutions. The market is rather robust at the moment and is being driven by a number of factors, some old and some new.

In addition to the recent Healthcare Information and Management Systems Society (HIMSS; Chicago, Illinois) call for a summit to discuss expediting adoption of the EMR, other factors are working together to drive computerized solutions into both the hospital and physicians' office market space. The Institute of Medicine (IOM; Washington) and Joint Commission for Accreditation of Healthcare Organizations (Oakbrook Terrace, Illinois) initiatives to reduce medical errors and to adopt computerized physician (drug) order entry, plus the imperatives of the Health Insurance Portability and Accountability Act on office automation, all are helping push the market. Add in a rapidly maturing wireless infrastructure and you have all the elements for impressive growth in this segment. The situation in the market is increasingly one of healthcare providers knowing they need to do something but finding themselves increasingly confused about what to do while adrift in a sea of fragmented products and partial solutions.

One message from TEPR that came through strongly, even more so than at the HIMSS meeting earlier in the year, was that wireless local area network technology is ready to become a foundational technology upon which to build the mobile platform for CPR, voice and practice management applications. New conferences are growing up around wireless in the medical space, such as the Mobile Healthcare & Electronic Order Entry Conference that will be held Sept. 8-10 in Minneapolis, Minnesota. This conference is sponsored jointly by TEPR and the Mobile Health Care Alliance (MoHCA). In addition, the Health Information Communications Technology marketplace will be convened in London on Dec. 2-3 and is open to European and U.S. technology vendors.

Several interesting presentations by providers who had successfully implemented secure wireless infrastructures were shared at the TEPR gathering, and the one thing they all agreed upon was that wireless technology was not yet mature enough to be safe and effective. Various familiar and new wireless LAN technology companies were attending or mentioned in presentations by wireless adopters. Symbol Technologies (Holtsville, New York) and Cisco Systems (San Jose, California) were prominently mentioned as having successful deployments, although the latter was not among the TEPR exhibitors. The greatest excitement seemed to be around Symbol's new Mobius product line.

Some newcomers that we saw at TEPR for the first time included ReefEdge (Ft. Lee, New Jersey), a new, venture-backed wireless networking company specializing in the ISM-band, 802.11 networking space. Newcomers like ReefEdge are not hampered by a large installed base of older, legacy networking systems and offer a cutting-edge set of features and products compared to companies such as Cisco or Lucent Technologies (Murray Hill, New Jersey). Other new providers, such as Airespace (San Jose, California) also were absent from the TEPR exhibit floor, but there was no lack of wireless and IT consultants at the conference. Microsoft (Redmond, Washington) and its partners consumed much of the convention floor space showing Office 2002 and Office 2003 beta-based XML-based solutions for the web-enabled EMR approach.

Advances on wireless front

Wireless technology convergence is happening. The 802.11 family of specifications can be confusing to both providers and some vendors alike. The standard has evolved now for several years and embraces a range of protocols and tools, from the earliest frequency hopping (FH) 2 Mbit/second networks, to the very well-known and popular 802.11b direct sequence (DS) 11 Mbit/second, the new, higher speed 802.11g OFDM (orthogonal frequency division multiplexing) protocol, and the lower-speed, Bluetooth 802.15.1 cable replacement protocols. On all of these are layered 802.11x security enhancements. In addition there is 802.11a, 54 Mbits/second protocol operating in the higher-frequency 5.6 GHz band. There were applications for and vendors showing all of these networking protocols at TEPR, as well as a few showing applications in the new, proprietary 608 MHz wireless medical telemetry spectrum (WMTS) band. This latter group was the definite minority, however.

Just as there were many network protocols, there also were many mobile hardware platforms that were shown, the smallest of which were the hand-held PDAs (personal digital assistants), including the popular Palm (Milpitas, California) and HP/Compaq (Palo Alto, California) I-paq models. Immediately above these were a variety of tablet PCs (that run Windows Tablet XP operating system) and above these were the laptops running Windows XP Professional or XP Home. These are the three levels upon which all mobile medical applications are being developed. The challenge of making applications run on screen real estate as limited as the smallest PDA on up to 19" screens is a daunting one that requires special tools and approaches, products such as those available from Counter Mind (Littleton, Colorado) and others. The trick in working with all of these platforms, particularly when developing web-based, thin-client applications, is to keep the data communications separate from the display formatting on the client (mobile) side of the application. Counter-Mind provides the toolset and middleware to accomplish just that. By slipping in its middleware product, the XML-based schemes and communications are standardized for all client application platforms, and then a custom display application is added that maps the tagged data to the specific display characteristics of the mobile platform being leveraged.

Counter Mind was simply one of many middleware products shown at TEPR. First Databank (San Bruno, California) and Cerner (Kansas City, Missouri) were there, as were SnoMed (from the American College of Pathologists; Northfield, Illinois), Apelon (Ridgefield, Connecticut) and Medicomp Systems' (Chantilly, Virginia) Medcin (structured language middleware), along with Trust Digital (Fairfax, Virginia), Vericept (Englewood, Colorado), Netilla (Somerset, New Jersey) and others who make secure, web-based applications and e-mail communications work.

An important underpinning of the mobile EMR is a secure and reliable, HIPAA-compliant ISM-band wireless infrastructure. This generally requires the ability to accommodate 802.11b DS WLAN inside the hospital or physician office and cellular mobile voice and data communications while traveling between the physician office and hospital/clinic. Numerous vendors, infrastructures and devices that provide just that were on display at TEPR.

In the wireless space, there have been two important and rapidly adopted developments. The first is the availability of dual-protocol access points (APs). APs that support the most popular 802.11b and the newer, higher-speed 802.11a networks are now available from Cisco via its acquisition of Linksys (Irvine, California) and others. Symbol's Mobius supports multiple access points and protocols as well. The other major change is the rapid adoption of 802.11g, the higher-speed ISM-band version of the similar technology used by 802.11a (at a higher frequency). Six months ago, amid concerns about incompatibility of 802.11b and 802.11g, it looked like the dominant two wireless technologies would be 802.11b in the ISM band and 802.11a at 5.6 GHz. Now that has all changed, with 802.11g being made to coexist well with 802.11b in the ISM band and offering a strong speed improvement over 802.11b. BBI's forecast, therefore, is that 802.11g will be widely adopted in the healthcare space and will quench much of the interest in 802.11a.

The advantages of the 802.11g solution is that it leverages the same wireless infrastructure (particularly with Symbol's Mobius products) as does 802.11b and has a better performance profile than 802.11a. 802.11g systems consume less power, have greater range, need to transmit less power and don't have as many absorption or multi-path issues as 802.11a systems, but deliver much of the potential speed of 802.11a networks. This is a powerful set of advantages in the mainly mobile healthcare environment, particularly in hospitals, where ripping out ceilings to install new AP infrastructure is a major disruption to providing care and also has infection-control implications.

Industry heavyweights like Intel (Santa Clara, California) could drive adoption even faster by upgrading its current mobile Centrino wireless bundle to include 802.11g technology. That company has announced that it will bring a combination 802.11b/ 802.11g bundle to the Centrino platform by the end of 2003, probably in time for the holiday technology buying season. The company had previously not planned to bring the 802.11g technology on line until mid-2004. The acceleration may be due to the rapid progress that the 802.11g committee has made in working out technical issues to assure coexistence with older 802.11b technology and growing public demand for faster data rates than 802.11b can provide. As soon as this technology appears in laptops and tablets, the medical community will begin using it. The hospital will quickly become a bed for both 11.b and 11.g wireless technologies.

The concerns of using any wireless technology in medicine include: security of data, interoperability without interference and quality of service each of which is related to the other. There has been a lot of preliminary information and some vendor-generated misinformation about these issues, particularly in relation to the use of the new Wireless Medical Telemetry Spectrum (WMTS) band created by the Federal Communications Commission for medical telemetry use. As all of these technologies have evolved and matured, coexistence has been more or less worked out through cooperation among the various committees and ceases to be a practical issue at the application level in systems that are well-engineered and implemented.

Coexistence of 802.11b and 802.11 FH has proved to be possible if each type of AP is properly separated. Welch Allyn (Skaneateles Falls, New York) has been using 802.11 FH devices for patient-worn telemetry in hospitals that have 802.11b DS widely deployed for mobile devices and stationary workstations. Their quality of service performance is impressive, particularly compared to use of older, transmit-only uptuned devices operating in the WMTS band. One-way patient-worn telemetry systems uptuned from the VHF or UHF band to operate in the WMTS band can exhibit a lost communications rate of about 25 minutes per day, or 2%, according to some Welch Allyn studies. The actual dropout rate can be higher due to unlicensed interference from environmental noise, depending upon the site's RF characteristics and proximity to interference sources. Most currently available WMTS systems use fixed, narrow-band frequency communication technologies inherited from their uptuned predecessors. Although they utilize digital encoding, their one-way transmissions (from patient-worn transmitter to remote receiver) prevent them from recovering data lost or corrupted during transmission by simple techniques like request for retransmission.

Currently available uptuned VHF and UHF transmitters operating in the WMTS band do not efficiently use the very limited 6 MHz wide WMTS spectrum, allowing little more than about 120 devices (25 kHz radios with 25 kHz spacing) only 12% of the channels recommended by the American Hospital Association (Chicago, Illinois) to support hospital wireless medical systems in the future. While it may be possible to reuse a portion of these channels if departments covered by telemetry are sufficiently separated from one another, there is a management overhead cost to keep transmitters dedicated to separate areas along with the added service support inherent in this limitation. The larger the traditional antenna system, the greater the susceptibility to interference and the greater the service cost to support and troubleshoot it. This also is true for one-way distributed antenna systems; no matter what frequency band is used, including the WMTS band.

Each vendor operating in this band uses a different, proprietary transmission protocol that offers few tools for network security and diagnosis. Thus, quality of service and security to meet HIPAA requirements are serious concerns for systems, with the exception of GE Medical Systems (Waukesha, Wisconsin), that are operating uptuned transmitters in this spectrum. As a result, companies such as Welch Allyn have switched to using the ISM band, with 802.11 frequency hopping or 802.11b DS WiFi bidirectional radios for real-time vital signs transmissions. As they have done so, coexistence information has become more widely available and concerns originally voiced by vendors of uptuned VHF and UHF telemetry have become moot. Most patient-worn radios that use the ISM band have relatively low duty cycles and use a listen-before-transmit protocol. Proper separation of 802.11 access points allow multiple radios to co-exist in the same area. The limiting factor for network performance depends more on the throughput capacity of the AP and much less on the number of wireless devices in proximity to one another.

The Welch Allyn Micropaq and Wireless Propaq patient monitors have very small radio duty cycles, less than 3%, even when they are continuously connected to the wireless network. As a result, the wireless performance of these Welch Allyn radios is 30 to 100 times more reliable than traditional uptuned, one-way telemetry operating in the WMTS band. Welch Allyn has recently completed a reliability test of 11 hospital installations. Data shown in Table 4 were collected for seven days, representing more than 65,000 hours of patient monitoring. Dropped packet data includes time required for battery changes. Average dropped packets were 0.063%, or 30 times better than traditional one-way telemetry systems.

Five hospitals, one of which has a direct sequencing system, exhibited a dropout rate of 1.2 seconds packet every 10 hours. The worst-case hospital (dropped packet rate 0.15%) had the greatest number of Micropaqs (217) distributed over 120,000 square feet, which still exceeded a one-way traditional telemetry systems by 12-fold.

Such data as this, taken from real-world installations, pretty much dispel the concerns about viable coexistence of multiple 802.11 protocols in the same facilities when each is properly designed and aware of the others. This is the hospital's responsibility to assure, or whomever they hire to manage the design and installation of their wireless networks. Indeed, each presenter at TEPR who discussed the deployment of wireless systems indicated they used third parties to conduct or manage the engineering and installation of their wireless infrastructure, and then in some cases used the vendor of the wireless systems to come in after the system was installed and validate its operation. They all recommended that hospitals deploying new systems also adopt this methodology to minimize any problems.

Of the applications that use wireless technology, the most interesting is voice communications. Spectralink (Boulder, Colorado) is a well-known provider of in-facility wireless phones for voice communications, while Vocera (Cupertino, California) is a new supplier of voice-activated, provider-worn, pendant-type voice communications devices that are not unlike Star Trek communication badges. The Vocera devices operate in the ISM band using 802.11b protocol. The more conventional Spectralink wireless phone handsets can operate in the ISM bands' 900 MHz or 2.4 GHz spectrum or in another dedicated spectrum as the hospital's bandwidth availability may dictate. The contrast of the two approaches to voice communications was dramatic. Spectralink is migrating a conventional handset by upgrading its 24-character-only display to a pixel-mapped display and adding some paging features, but still seems content to have the nurse carry its phone, a separate Data Critical pager, yet another PDA or tablet PC to chart on.

In contrast, Vocera has essentially reinvented voice communications with a keyless pendant, using voice commands to dial by speaking a person's name, a person's role, either point-to-point or from one person to a group of other people. For example, the system has the notion of user-defined groups, such as a "code group." If a nurse using Vocera encounters a patient in cardiac arrest, she can command Vocera to contact the "code group," and the system will broadcast to all members of the group, which might include an anesthesiologist, an internist, a cardiologist literally the entire team with one call. The system is even smart enough to track down alternative members of the component groups, so that if the primary doctor on the team from the "cardiology" group is not available, Vocera can contact an alternate physician who is part of that same group all automatically all without the nurse having to use her hands to dial a lot of numbers. This is not intended to replace the code button that exists at every ICU bedside; however, if a telemetry patient arrests in a stairwell or somewhere else in the facility where no code button is available, this feature could be a lifesaver, allowing the nurse to continue to provide CPR to the patient while the code team is assembling and coming to her support.

Seeking optimal EMR solutions

The hospital and physicians' office market is realizing how critical and vital EMR/CPR systems will be to assure quality, expedite billing, maximize reimbursement for services and track outcomes to market themselves to hospital and third-party payer groups. Yet they are confused about which systems provide the best value and performance for their applications. There is a seemingly overwhelming number of such systems out there, and as with voice communications, some of the more entrenched EMR suppliers are not necessarily offering the best available solutions. One key to an optimal solution depends on communicating to physicians and other users the proper balance between expedient front-end data entry and flexibility and power in extracting consolidated and population outcome data from the back end of these same systems. This remains a major vendor challenge and barrier to the more widespread adoption of EMR/CPR systems. The key to this is the back-end vocabularies, nomenclatures and ontologies that define the schema for how history of present illness (HPI), lab results, symptoms and other data is collected on the front end.

Back-end nomenclatures are key to building systems that actually do something that is clinically useful and sustainable in the clinical charting space, whether in hospitals or alternate-site settings. Structured vocabularies are simply too vast for any one CPR vendor to maintain their own terminologies directly, yet several of the CPR/EMR companies showing their products do just that. While such comprehensive vocabularies and pick-lists are critical across the broad spectrum of alternate-site medicine, in any one specific practice only a small subset of the entire conceptual content of the vocabulary or nomenclature may actually be used.

Using a vendor-generated limited vocabulary and ignoring the well-structured and more elaborate vocabularies available from several sources can be penny-wise but pound-foolish. Using a structured language infrastructure is the key to getting useful data out of the information captured by all of these systems. Without it, the potential to accurately resolve charting expressions and variations and map synonyms to a single element that can be reported are insurmountable. Snomed CT alone supports 344,000 clinical concepts organized into hierarchies, plus 913,000 descriptors for these concepts. It maps more than 1.385 million relationships between concepts and provides a useful crosswalk to other elements. These products link the diverse findings from the HPI with procedures, measurements, labs, specimens, body parts and other data in an integrated knowledgebase structure. Small vendors have no chance of developing such an infrastructure or maintaining it without a product like Snomed CT, Medcin or Apelon, and those that try are doomed to waste a lot of time and resources in doing so, or to end up with a system that cannot easily or adequately report the data it collects.

The complaint and barrier to wider clinical adoption of these systems has been that they "slow doctors down" and make it hard to complete as many patient exams with the electronic charting as they could with paper charting and dictation. Non-customized, rigid workflow systems do in fact slow physicians down. Other systems, with user-customizable workflow and charting nomenclature customized to the actual type of practice, are at least a break-even arrangement on workflow after a couple of weeks of going up the learning curve in using them. One such system we saw was JMJ Technologies' (Marietta, Georgia) Encounter-Pro EMR. This versatile and growing company offers systems tailored to cardiology, dermatology, family medicine, internal medicine, ob/gyn, ophthalmology and pediatric practices and is a frequent award winner in these types of events. The company now has in excess of 3,000 users in more than half of all U.S. states and is growing rapidly. One system can't fit all clinical practice needs, and successful systems like JMJ's can be customized to meet the needs of the specific type of practice they are being sold to.

Systems designed to make getting data easy and fast tend to be unstructured or not precisely structured, and as a result are difficult to extract meaningful, accurate, population-wide data out of. On the other hand, highly structured nomenclatures that make getting data in precisely and resolve it to non-ambiguous elements that can be searched, accumulated and output, tend to slow down physician input and therefore patient encounter, often crippling office efficiency in terms of patient encounters per physician per day to suffer a bit. But they do provide well-organized records and outstanding statistical outputs and reports a higher-quality EMR patient record than less-structured systems.

There are systems that operate on the fringe of each extreme. The Advanced Imaging Concepts (AIC; Louisville, Kentucky) system that images the patient records and keeps the images and depends on a limited subset of data captured in the transcription process to populate the EMR record is a breeze at expediting input, but very limited in the consolidation of practice-wide, cross-provider or patient population statistical outcome data. It therefore has severely limited utility in evaluating outcomes and best medical practices. They are easy to get physicians to use as they make minimal intrusion into the paper-based practice modality that doctors already are accustomed to doing. Patient charts, however, are legible, retrievable and useful in several parts of the physician office at the same time due to their electronic nature. AIC therefore is growing as an EMR provider.

Over time, however, the sweet spot in the trade-off between usability and data usefulness is progressing toward better usefulness, particularly when a practice selects systems that utilize standard nomenclatures like Snomed as its hidden backend knowledgebase structures. The key is to limit the massive clinical and conceptual space of these powerful knowledge bases to the practical clinical space actually encountered in a specific practice.