BBI Contributing Editor

SEATTLE, Washington — Toward an Electronic Patient Record (TEPR) is a conference — and an idea — in transition. It has become a vendor-driven forum rather than a true scientific conference, and this inbreeding masks the ability to see paradigm shifts in the market. The notion of electronic patient records as an end in themselves seems flawed, yet hotly pursued at TEPR. The educational sessions were another matter. The apparent strategy to implement the EPR one piece at a time has resulted in two very different approaches. One is the highly structured, electronic capture, storage and processing approach. The other is the loosely structured, free text or voice recognition, electronic transcription and optical archiving approach. Vendors that reflect both of these solution poles were active at TEPR.

One of the most interesting of the free-form, capture, transcribe, scan and store approaches was exhibited by Advanced Imaging Concepts (AIC; Louisville, Kentucky) in its Impact.MD product. Its approach to automating patient care is to transparently automate the back office records storage of the physician's practice, whether the doctors change their front-end practice method or not. AIC and the companies that embed its products in theirs have done this quite well. The approach is simple: Give doctors what they want. As one doctor/user put it during TEPR, "We don't have to file papers; we don't have to spend money on space to store paper charts anymore. We just scan it into the system and it's there. It's at our fingertips when we want it."

The scanning approach to medical records storage and management, when well done, overcomes the fears of many doctors about EMRs. It is cost-effective. It is controllable. It doesn't require a lot of training and can be done with existing office staff. It does provide rapid access to patient records, and it can allow physicians to continue to work in the manner they are accustomed to (with paper) as they gradually adopt a direct, electronic methodology to collecting patient information. The AIC approach accommodates early adopters of electronic records, as well as the late and reluctant adopters, who hold onto the paper record until they die, retire or feel uncomfortable being among the last adopters of a new paradigm. As such, AIC is a nearly perfect solution to half of the medical records problem in physician offices — it fixes the back-office records storage and retrieval problems. But it ignores the other important part — mining the rich data content of patient medical records and using it to modify the paradigm of care delivered at the point of care. Yet it is these front office point-of-care encounters where medication errors are caught, where charting to support billing is needed, and generally where changes are needed to raise the bar, so that treating sickness can be transformed into "health" care. To make this transformation the data resident in the patient's chart, no matter how it is stored, managed and retrieved, is required, and that is the next challenge for systems like AIC's and others that embrace the scanning approach.

What remains for AIC and others is how to mine data contained in its optical images. This will involve at least two steps: First, converting these images into a character-based, codifiable format, and second, indexing and cataloging such free-form data into medical concepts and frameworks that are unambiguously searchable. Neither of these tasks will be easy to solve. AIC seems poised to bite off the optical character recognition step next. This step will add a step into the medical records back-office process, however, requiring more time that will in turn reduce the cost-effectiveness of the solution somewhat. Even when this has been successfully accomplished, the matter of resolving essentially free-form information into viable medical concepts will remain.

Insurance and formulary eligibility validation

Scanning also has a place as an adjunct to the optimization of the physician practice front office. Card Scanning Solutions (Las Vegas, Nevada) makes a contribution with its MedicScan products. This is a scanner and companion software that allows the practice to scan a patient's insurance card and optimizes the process of getting it into the chart and making it accessible. Once attached to the USB port of any Windows-compatible PC, the scanner senses the insertion of an insurance card, capturing the front and back sides of the card in a few seconds and converting it into a predefined, compressed image that is automatically routed to the windows desktop or to a patient's record (optional software) for inclusion in the chart. Additional data can be annotated to this record to facilitate retrieval. This optimizes the initial capture of insurance information and facilitates expedited validation of that information on each subsequent patient visit.

Finally, to automate the completion of the patient form for each encounter, there are a variety of mechanisms, ranging from patient-carried, healthcare payer-issued ID cards to Internet-based patient medical demographics files that can be downloaded and merged onto the complaints visit form, eliminating the need for the patient to continually fill in things like their name, address, birthdate, insurance carrier, policy numbers, telephone numbers, etc. At most the patient can edit this information retrieved from their ID cards or Internet demographic files, and simply check off the symptoms, presenting complaints or services scheduled, etc. This expedites the time patients spend filling out forms and enhances their view of the efficiency of the physician practice. The conventional wisdom at TEPR was that cards were dead (in the U.S.) and being displaced by Internet-based patient demographic and summary medical information files maintained by third parties. However, neither of these mechanisms has yet been widely used, as they represent automation usually associated with the practice management systems, and many of these systems don't accommodate either of these mechanisms, as they have not been widely used by payers or patients. This is an area of patient education and automation that would be an excellent topic for the patient and physician to discuss upon the first visit and adoption into the practice, or upon the conversion to automation upon a yearly visit thereafter.

Storing medical concepts

To effectively retrieve medical information, one has to effectively store it in the first place. While that sounds simple enough, it is not — particularly in a multi-physician practice setting. It is even more complicated if the practice is multi-specialty. Human beings are by nature non-precise in their verbal (and written) expressions, particularly when they are in a hurry (as during a busy time in the office, seeing patients). As a result, variations in physician terminology creep into the medical records. Add in an office nurse, used to using nursing terminology, and there can be variations in free-form charting. For example, one doctor notes the patient has an elevated temperature, another that the patient has a fever and a third that the patient is febrile. The nurse may chart a complaint of "temp," and the patient write his systems as "I feel hot." All of these descriptions ultimately must be reduced to one code and stored in the clinical knowledgebase documenting this patient encounter.

Only when these diverse descriptions are so resolved does it become easy to extract information from the clinical knowledgebase about all patients with the code for fever. If instead of being stored under one code, such observations are stored under several descriptors, then the caregiver must know and concatenate all synonyms for fever when querying the database longitudinally (across multiple patients) to correlate fever with some other medical condition of interest. The same is true for immunizations, diagnosis, drugs administered and other data that a practice may wish to correlate. If the electronic system is rules-based to help the practice recognize and correct oversights such as immunizations, then the rules need to know which codes to look at and correlate to trigger the physician alert that some action or intervention is required for this patient, now — while he or she is in the office and available.

To assist the vendors making EMRs in this task, products such as SnoMed CT or Medcin from Medicomp (Chantilly, Virginia) exist. These are EMR back-end systems, and they were at TEPR this year as in the past. SnoMed is the product of the American Association of Pathologists (Northfield, Illinois). TEPR was the venue for the introduction of the new CT version of SnoMed. The CT version is the first to integrate the previous RT (reference technology) model of SnoMed with the UK Reed codes, a project that SnoMed has been working on for over two years. The inclusion of the clinical codes developed by Dr. Reed and widely used in the UK and across Europe, expands the utility of SnoMed in its clinical space for charting. SnoMed is an outstanding candidate for a back-end clinical repository of drugs administered, diseases, complaints and treatments prescribed, but it is not the only contender. Medicomp offers Medcin, a complementary front-end product. When used in combination with "pick" or "drop-down" lists, these products allow for synonymous terminologies to be neatly resolved to a single storage element in a clinical record, and thereby unambiguously queried, retrieved and consolidated with other data. This is the key advantage over the formatted, structured way of charting vs. the systems that accept essentially free-form descriptions of the clinical encounter — descriptions that can vary widely among caregivers in the same practice.

HIPAA compliance

The bottom-line advice to all hospitals from the Health Insurance Portability and Accountability Act (HIPAA) presenters at TEPR was to file for an addendum because the vendors are not ready to offer a solution without informational gaps. John Quinn of CapGemni Ernst & Young (New York) told the audience that the two pervasive myths still abounding among hospitals is that "the vendors will fix my HIPAA compliance problems" and "sending data through a clearinghouse will preclude remediation." Quinn said that there is a shift occurring among providers from a "do it for me" attitude to a "do it with me" attitude toward consultants. He said that most providers have read the HIPAA regulations now and are no longer intimidated by them. Providers must realize that HIPAA is simply the beginning of security needs and that it is a long-term process rather than a quick fix. For their part, vendors must begin to deliver comprehensive HIPAA solutions that include all elements — encryption, provider validation, access tracking, logs and other elements. Providers currently have to acquire partial solutions from many vendors and integrate them in-house, a process that often is not successful. A web site for electronically filing a HIPAA addendum is available at www.hhs.gov/hipaa/hipaa2/ascaform.asp. That site will accommodate filings for multiple providers as long as the basic information required is the same.

Growth in handheld medical platforms

If there were questions about the potential of handheld devices for medical applications, they were put to rest at TEPR. Handheld devices will become an integral part of the medical computing infrastructure of the future. This was made clear by a presentation of Mark Bard, a principal at Manhattan Research (New York), in his "Taking the Pulse" presentation. Bard's research, conducted with Deloitte Consulting, showed that 95% of physicians have access to the Internet during the last year as part of their practice, up from 62% only a year ago. Of these, 21% felt the Internet had become essential to their practices. An additional 45% were poised to adopt, while 34% were reluctant technology laggards. Of all physicians, 85% already had a computer in their offices. Of these, 88% were daily users of this resource, and 91% of those respondents used it for professional purposes. Of the professional users, 91% had office Internet access as compared to only 74% of all doctors. Doctors access the Internet from a variety of locations. Of the professional users, accessing the Internet from home accounts for 57% of their use, while accessing it from their offices accounts for an additional 40% of their use, compared to other physicians who access the Internet from home 67% of the time, and less from their offices (31% of the time).

Use of PDAs (portable data assistants), such as the Palm Pilot or Compaq Ipaq, already had been adopted by 30% of physicians. Another 23% were interested in adopting it in the near future, while 16% were not sure, and 31% had no plans to use PDAs. This probably splits along age-of-physician lines, as younger doctors entering practice already have been using PDAs in med schools and consider them a familiar tool to enter practice with. Younger physicians may also be more inclined to adopt this technology than their older colleagues.

When you look at what physicians use PDAs for, 84% use them to manage their personal schedules, while 68% use them to manage their professional schedule. 59% use PDAs for accessing drug information. As physicians bring the PDA into their practices, 19% use it for online access, 17% for writing or entering clinical notes on patients. 8% are using it for mobile email access while 6% use it to transmit drug orders to pharmacies, and 2% use it to retrieve lab results. These lower values of PDA use are more a function of missing office infrastructure and applications than the PDAs themselves. For example, regarding PDA use for electronic prescribing, 6% of physicians surveyed currently do so, but another 28% are interested in doing it in the future. Several factors are driving this interest. First, 65% of physicians feel that electronic prescribing will help reduce the incidence of medical errors, while 58% feel it will help improve accuracy and legibility of orders. And 55% associate this with fewer callbacks from pharmacists. 53% feel this will reduce time and overhead, particularly as 50% associate a PDA as a tool to select common drugs prescribed from "pick" lists. Almost half (47%) say they want to be able to more readily verify the patient's approved formulary coverage. All of these factors account for the interest in using PDAs for prescription ordering, refills and management.

On the technology side, PDAs are evolving as they are combined with cell phones and IEEE 802.11 LAN cards. Products like PDA-Secure will keep information on PDAs secure, and Thin client-server applications (that don't store any patient data on PDAs in the first place) minimize the issues with PDAs accessing patient data. Now that cellular encryption is being improved, wireless links are becoming more prevalent in PDA. The new Compaq (Houston, Texas) Ipaq 3870 with built-in Bluetooth LAN (IEEE 802.15) is appearing. The next two generations of PDAs will move even further ahead, as will the availability of new, low-power radio and DSP chips from vendors such as Intel, Transmeta and others.

Overall, this year's TEPR gathering was a surprisingly good showcase for new vendor products, even if the educational sessions were a bit basic, misinformed and too vendor-biased.