Gaining Access to Data

By Robert B. Bruegel, PhD, and David J. Rothwell, MD

A functional controlled medical terminology (CMT) that standardizes the terms used in medical language continues to be a specific major concern for computer-based patient record (CPR) developers, and for the health care IT community in general. Development of a CMT has been described as one of the remaining “grand challenges” for all parties involved. Yet despite major efforts, the goals outlined in CMT development projects and programs remain unmet.

The focus of these efforts has been the development of a formal representation of the concepts required for the practice of medicine. According to this view, the terms and concepts used for patient care must be transformed into a deeper formal representation, one that is free from the ambiguity found in natural language expressions. At the same time, the representation must be consistent with re-use across multiple applications and be “transformable” between multiple languages.

This formal representation of medical concepts is intended to respond to the needs of the artificial intelligence community and for decision support applications in medicine. In short, the goal is to develop a terminology that will satisfy the needs of people and, at the same time, become consistent with and transformable to a terminology for machines (i.e., a “people” terminology and a “computer” terminology). Until we reach such a broad goal, however, practical and workable interim methods for collecting, displaying and accessing patient care information must still be developed and implemented.

Background

In two articles published in ADVANCE for Health Information Executives (February 1998; January 1999), we presented an overview of some of the major issues that any interim clinical terminology effort must address in order to be considered a serious contender for adoption. These issues were:

1. the breadth and depth of coverage of clinical terms and concepts;

2. the flexibility and adaptability of the terminology;

3. the ability to capture both ongoing information relevant to the process of care as well as “snapshot” information at the end stage of care;

4. the ability to capture and describe a broad range of patient perceptions, issues and concerns relevant to utilization, satisfaction and disease management;

5. the ability to support both structured data entry and textual data, whether dictated or made available through voice recognition;

6. the ability to explicitly address the issue of patient and provider confidentiality;

7. the ability to relate to and support the rapid adoption of the Internet and Internet-based standards;

8. the ease (or difficulty) of integration into clinical information systems;

9. the ongoing cost for development, distribution and maintenance, since these costs ultimately will be reflected in the cost of the organization’s clinical information systems; and

10. the flexibility and adaptability to support rather than restrict the growth and development of medical practice and medical knowledge.

This article examines the latest developments in the evolving field of medical terminology.

Supporting multiple views

Given the complex, interdisciplinary nature of health care, IT experts must develop techniques that represent the knowledge and practice of each of the health disciplines. The ability for any medical terminology to enable and support multiple views of patient information is thus a key additional element. Such information must be available for immediate use for clinical care of individual patients and for other related purposes (including data aggregation for epidemiologic studies, resource management and reimbursement).

Importance of XML

In our previous article, we described the increasing potential of XML, the Extensible Markup Language. An XML document can best be thought of as a self-describing data structure. XML documents include Document Type Definitions (DTDs), which define the valid content and structure of an XML document. XML applications also use DTDs to process the data elements included in the document.

An XML DTD may allow validation of a document’s structure — that is, assurance that the input structure conforms to the “rules” for that document type as defined in the DTD. XML seems well on its way to being adopted as a standard for e-commerce and other major Internet applications, with virtually all major vendors, including Microsoft, now supporting XML. This has significant importance for health care IT, since it means that the amount of development effort directed toward XML by this larger community lessens the need for efforts within the community.

As a result of these evolving trends, the interest in XML within the health care IT community has grown rapidly. Of special note is HL7′s version 3.0 adoption of XML as a messaging format. Based on the continuing overall growth in the acceptance and power of XML, we expect that other such “XML-ization” efforts will continue to be announced throughout the industry.

Medical Language Processing

The not-for-profit Health Language Center (HLC) has concentrated its efforts not simply on utilizing XML, but on utilizing the power of the convergence between the development of a health-related XML and the maturation and practical availability of the Medical Language Processor (MLP). Emerging from this work is a set of natural language parsing technologies for health care based on the pioneering work of Naomi Sager, PhD, and her colleagues. The results of these efforts are being commercialized by Aurum Language Systems, Inc.

Medical language parsing is especially significant because it has the potential to free health care providers from the severe restrictions of current approaches to data entry. Up to the present, the continued evolution of computerized records has been limited by the gap between the need for structured data representation in the CPR and the significant difficulties in obtaining such structured data — particularly methods that provide selection from lists of data elements offered in numerous pull-down screens (i.e., structured data entry [SDE]). CPR systems have thus been caught between the ease of use and flexibility of free text, with it’s lack of underlying data structure, and the major time/use limitations imposed by virtually all approaches to structured data entry. A functional MLP has the potential to break this bottleneck.

Development of SHML

Based on the belief in the power and potential of the combination of XML and the MLP, the HLC over the past three years has designed and developed the Structured Health Markup Language (SHML), which would utilize the potential of XML and its related standards. Creation of a “markup language” categorizing the content of a CPR with corresponding DTDs for widespread use is a major goal of the HLC.

SHML is not a traditional medical terminology; rather, it is a catalog of all of the terms and phrases that are used to document medical care.

SHML catalogs (“marks up”) all of the terms used to describe a medical encounter including not only traditional medical terminology, but also those terms expressing uncertainty, negation, exact and referential time, severity, quantity and other qualifiers. Currently, more than 40 distinct SHML categories have been created, each a descriptor of medical content in the CPR (and each with multiple subcategories).

As noted earlier, gaining rapid access to patient care data for different purposes and “views” has become increasingly difficult. This is partially due to the increased size of the CPR, but also and more importantly, because the “data of interest” frequently is present only in text form. MLP, in concert with the SHML, has the potential to resolve this problem — enabling the freedom, flexibility and provider acceptance of free text — while providing discrete data principally available from structured data entry.

With a fully functioning MLP, providers can dictate patient care records using their own language — without being restricted to pull-down menus. The resulting text — whether dictated and transcribed or converted directly by voice recognition from speech to text — is then “tagged” to a set of SHML categories that are internally consistent and highly structured. As the MLP identifies each meaningful phrase, medical concept and it’s modifiers in the text, SHML captures the medical sense(s) of each phrase and places the sense(s) into an XML structured database — while retaining the context of the original text.

Early tests of the SHML and Aurum MLP indicate that it is possible to parse dictated text taken from progress notes and discharge summaries and to tag the text found in the documents using the SHML. These SHML-tagged documents are then available for viewing through a special XML-based browser developed by the HLC. SHML tags — linked with “view” templates built into the browser — allow users of the browser to sort the tagged data elements into “presentation formats” (i.e., templates) to suit the needs of any group accessing information in a CPR. The HLC browser is designed to enable each group to create its own views of the structured data.

Work on this effort has been extremely promising, and has progressed to a point where the HLC plans to demonstrate working prototypes of the combined approach at the upcoming TEPR 2000 meeting (to be held in San Francisco, May 6-12). Based on continuing review and experience with these prototypes, we will present a detailed description of the working system, particularly the operation of the XML browser program. This work will also be described in a subsequent article. *

Dr. Bruegel, co-founder of Clinical Reference Systems, is president of the Health Language Center (HLC). He can be reached at (303) 499-1685.

Dr. Rothwell, a former co-editor of SNOMED, is chairman of the HLC. He is the chief developer of the SHML.

IOM Study Points to Critical Need for CPR

As noted in the accompanying article, a controlled clinical/medical vocabulary/terminology is critical to the successful enterprise-wide deployment of computer-based patient record (CPR) systems.

Recent “findings” by the Institute of Medicine (IOM) regarding critical events/errors further point to the need for a CPR — something that captures all pertinent patient data/information and that cannot be readily changed. What is most interesting — perhaps significant — is that the recent findings make no mention of the landmark CPR study conducted by the IOM with the report published in 1991 (updated in 1997), “The Computer-based Patient Record: An Essential Technology for Health Care.”

The reason for this “oversight” could be the lack of a coordinated, comprehensive, funded policy coordinated by both the federal government and the vendor community. The original IOM report called for federal government support, public/private R&D support, and public/private collaboration on reimbursement mechanisms related to the CPR. While the European Union has such a policy, the United States does not . . . hence, perhaps, the oversight.

The 1991/1997 IOM report presented seven recommendations for a national CPR policy building on private-sector efforts with the clout of the federal government to enforce such a policy. This is not to say that exclusive control of development is vested in the federal government; quite to the contrary, the private sector’s collaboration is essential to the sound, practical approach to CPR development.

The concepts and design criteria espoused in the accompanying article by Drs. Bruegel and Rothwell will be essential — nay, critical — to the successful promulgation of such a national CPR policy. The issue of controlled clinical/medical vocabulary/terminology has presented a significant barrier to the CPR developer community in the absence of a national CPR policy. Therefore, we see a tendency to downplay, work around or ignore the subject in many current CPR/CPR-type systems.

For example, we have found that most vendors consider ICD-9-CM (84 percent) and CPT-4 (73 percent) as “clinical coding systems,” whereas, in reality, they are “billing coding systems” (August 1999 issue of ADVANCE). A clinical coding system is required to support the controlled clinical/medical vocabulary/terminology referenced above. Most vendors do not support clinical coding systems as defined. This short-sighted approach will be difficult and expensive (maybe impossible) to correct.

Careful consideration and analysis are critical to the future deployment of full-blown, robust CPR systems. Developers will do well to pay attention to the R&D described in the accompanying article and investigating the potential for incorporation into their respective CPR product offerings.

– William F. Andrew, PE

Editor’s Note: Mr. Andrew, a widely recognized CPR system researcher and author, is president of William F. Andrew & Associates, Inc. He can be reached at (863) 299-4767. The results of a comprehensive survey of the CPR/CPR-type system vendor market, conducted by Mr. Andrew, will appear in the April 2000 issue of ADVANCE.

Similar Posts:

• Coders Prepare for Today, Tomorrow
• Medical Records: 5 Reasons Your EMR Transition Will Pay Off
• Diagnosis: inaccuracy
• Emerging Career Opportunities for the Future
• ABC Coding Solutions Announced Addition of Intergrative Medcial Coding System