A View From the Consensus Panel

Highlights of Moving Toward International Standards in Primary Care Informatics: Clinical Vocabulary

by Stephen Kay, Ph.D.


This paper reports on a method for facilitating progress and achieving consensus in the contentious area of clinical terminology. The process was facilitated by management consultants, but the model of consensus that took form during the meeting emerged from the participation of the individuals present. The outcome could not have been predicted at the outset, but the process succeeded in achieving an unlikely result: an agreed, shared vision on terminology for primary care.

A successful meeting, "Moving Toward International Standards in Primary Care Informatics: Clinical Vocabulary," was held on November 1 and 2, 1995, in New Orleans, after the American Medical Informatics Association (AMIA) Fall Symposium. The meeting focused on the value of clinical vocabularies, specifically from the viewpoint of primary care, as practiced throughout the world. Issues of clinical terminology and four specific concept representation systems were considered over a 2-day period.

The group met under the auspices of the AMIA Family Practice/Primary Care Working Group and Working Group V of the International Medical Informatics Association (IMIA). It comprised 79 participants from 9 nations around the world. This paper gives an early impression of what happened during that meeting. It is an informal commentary meant to complement the more formal proceedings. 

In particular, this paper introduces what the author has called "The Mississippi Model of Consensus."

Agenda, Participants, Structure, and Process

The meeting took place from 3:30 p.m. until late on the first day and from 7:30 a.m. to midday on the second day. The first day was focused on dissemination and education, and the second day, on consolidation and agreement. Both days were integral to the process of achieving consensus.

The early part of the agenda on the first day gave an opportunity for a number of presentations. The first was a general introduction, which prepared the way for 12 15-minute intensive presentations on specific concept representation systems. The 12 presentations were arranged in sets comprising 3 parts: an overview, critique, and application. The systems reviewed were the International Classification of Primary Care (ICPC), Read Codes, Systematized Nomenclature of Human and Veterinary Medicine (SNOMED), and Unified Medical Language System (UMLS). The choice of systems was not exhaustive. Other approaches, most notably GALEN (Generalised Architecture for Languages, Encyclopaedias and Nomenclatures in Medicine), were omitted because of meeting time constraints.

The attendees were expected to participate actively throughout the two days. It is useful to characterize these participants as four distinct parties: the presenters, the audience, the consensus panel, and the coordinators. The presenters, who assumed the role of "system champions," naturally formed the first party. The second party comprised the main audience. On the first day, their role was essentially a passive one of listening and learning more about each system. At the end of each presentation set, members of the audience were encouraged to question the presenters, express their opinions, and clarify what they had heard. In this respect, their role was identical to that of the third party, the consensus panel, except that the panel's primary task was to summarize both the output of the presentations and the resulting discussion of each system. The coordinators, who made up the fourth party, were charged with making the process of the meeting productive. The fourth party comprised the chairs of the meeting; those presenting an international survey of clinical vocabulary usage in primary care; and the two facilitators.

The facilitators retained their role throughout the proceedings. All the other participants of the meeting forsook their first role to be organized into seven groups on the second day. Small groups considered the summary paper from the consensus panel produced at the close of the first day. 

One useful innovation of the meeting was the employment throughout the proceedings of a stenographer, who recorded verbatim the spoken content of the meeting. The text was made available immediately as a modifiable ASCII file. Not only did this permit a faithful record of the event for the production of the proceedings, but it also permitted the consensus panel to accomplish its business before the start of the second day.

Consensus Panel

The five members of the consensus panel were preselected. The principal selection criterion was a reputation for impartiality rather than domain expertise; they, too, were expected to learn from the proceedings. The main meeting finished at 9:30 p.m. on the first day, and the panel continued its work until after midnight.

The facilitators, after initial discussions with the panel, decided on a trial format for producing the report. This was adapted again once the presentations were completed and became the final template for the panel's report. The purpose of the report was to gather and structure the material from the sessions on the first day. Its completion signified the finish of the business of the first day.

During that first day, each panel member was given a specific task: four had the responsibility of focusing on a particular system, and the fifth member was primarily concerned with the wider context. After the other participants retired, the panel met to produce its initial summary report, which was to be an interim document to be used as input for the second day. 

The report had as its central components a "framing issue," in which the raison d'etre for the meeting was expressed; a "vision," which looked at the long-term goal of the meeting; a list identifying the "critical elements" of a controlled vocabulary, as it emerged during the first day of the meeting; and a resume of the strengths and weaknesses of each of the four systems presented. Deliberately, no value judgments were made as to the best system, but UMLS was singled out as being distinct from the other three.

Consensus Development

The formal proceedings of the second day began at 7:30 a.m. After a brief introduction of the interim document, seven discussion groups of approximately equal numbers were formed to discuss and modify it. Each group had a chairperson and a rapporteur. At midmorning, the entire meeting reconvened to discuss each group's changes to the summary document. The facilitators asked each group in turn for its response and modified the document online to reflect the input. The facilitators then asked the audience to consider the amended text and affirm whether the consensus of the meeting regarding the modified component was high, medium, or low. 

The purpose of this paper is not to consider the entire proceedings but rather to focus on the statement of vision. Development of the vision statement illustrates the process in action and the character of the meeting. In describing this process, the author considers a model of consensus that seemed to emerge from the meeting.

Mississippi Model of Consensus

During the meeting, a participant expressed his frustration at being given a "hose pipe" when all he actually required for his clinical practice was a "water fountain." The author subsequently developed this water metaphor in the meeting. It is expanded here to incorporate both the micro and macro processes of accepting a common clinical controlled vocabulary. Hence, the Mississippi Model is one that maps the natural development of a river by maturity stages to the development of consensus during the meeting and beyond.

The meeting started with distinct and separate streams. Remote from each other, these rushing streams at least seemed to have a common direction—down. At this early stage, however, there was no evidence of convergence or consensus. Rather, there was considerable distance between the streams. A barrier was created by the champions, who were perceived as being competitive and out to convince the audience of the virtues of their preferred system. There was a common belief among the audience that the sole object of the exercise was to declare and recommend an outright winner. The futility of this course became clear, and sadly, there seemed little hope of any profitable outcome from the meeting.

As the meeting continued, however, the discussion of the distinct streams came together by effectively focusing beyond them, on the clinical requirements of the audience. To allude to "water fountains" and "hoses" was at least to acknowledge the valuable existence of the streams, that drought conditions had passed, and that mechanisms for controlling and directing the flow were now appropriate so as to prevent drowning and/or wastage. This "mature stage" of development came about only because the audience had, in general, been given sufficient insight from the presentations to be able to debate the subject of controlled vocabularies in a reasonable manner. Admittedly, the course of the discussion meandered, and the streams were still identifiable as separate entities. However, at least one could now imagine them as being tributaries of the same broad river, rising from different sources certainly, but eventually contributing to a main stream flowing in the same general direction.

The next stage of development involved reaching consensus on the framing issue and vision. The framing issue was broadened to recognize that primary care is not an isolated island, independent of what else might happen to the patient. Not surprisingly, given the subject of the meeting, there was a high level of consensus concerning the framing statement. The progression of the following vision statement is more interesting, as it made the participants question the assumptions that had been implicit in the framing issue.

The initial vision statement read:

"Our vision is for primary health care providers to easily and accurately record all patient information in such a manner that the information collected will be comparable to information collected elsewhere in the world."

The clinicians present at the meeting were focused on what could be achieved in their daily practice and for the patient before them. The framing issue had two parts that were not obviously correlated: how was the development of international consensus likely to benefit the local situation? The initial vision statement centered on the effective recording, understanding, and sharing of information, and ultimately, the benefit to be gained from comparing clinical data across "fading boundaries (1). "The advantages to be gained were seen as involving long-term research and epidemiology, the benefit of which would only eventually cascade back to practice. However, primary health care providers are pragmatic, preferring immediate gains to future promises. This is not to say they are not academic or interested in research; they are. They felt the vision as expressed was necessary but not sufficient for their specific purposes. Their fading boundaries are different. At a routine level, such boundaries typically exist between different types of clinicians involved in primary health care, and they require a direct focus on the subject of care at a point in time. The clinicians also believed that their current practice should be used to shape the form and content of the vocabularies; they did not want to passively accept what they are given.

The revised vision statement read:

"Our vision is for primary health care providers to record data about patients easily and accurately at the point of care in such a manner that clinically relevant information is available for the primary purpose of supporting the care of individual patients. Additionally, this information should be comparable with information collected elsewhere in the world."

The changes to the vision statement also represent the tension between the service-oriented and research-oriented aspects of primary care. The revised statement reflects the need for both. 

As the collaboration process continues, the revised statement may undergo further changes before being published. It may, for instance, be changed to reflect some of the following common beliefs that were voiced: that the vocabulary for primary care is the core from which others then specialize; that patient focus is essential; and that health and morbidity are preferable to mortality as starting points for describing our common condition.


The "old age" stage of consensus development was not reached in the meeting, as lack of time precluded the process for completing the full program. Perhaps this is symbolic. It would be wrong to believe or assume that we had finally "arrived." It would also be wrong to claim too much from a relatively small meeting of collected enthusiasts. 

The sea awaits the outcome. In this analogy, the sea is the whole terminological domain and debate, which is, of course, much deeper, wider, and bigger than the Mississippi. Without doubt, navigation will remain an issue for many years. There is still a considerable distance to travel to the sea, and the precise route has yet to be determined. It is clear from the meeting, however, that one of the world's biggest rivers is ultimately heading for the sea, and that the sea would be well advised to make itself ready to accommodate the volume and disturbance inevitable from such a significant and broad outlet.


  1. Hammond WE, Bakker AR, Ball MJ, editors. Information systems with fading boundaries. Proceedings of the International Medical Informatics Association Conference, Working Group 10 Conference; 1994 Aug 27-30; Durham, NC. International Journal of Biomedical Computing, Elsevier Science; 1994.
Acknowledgments: To all at the meeting, particularly Moon Mullins and Bob Price for initiating events; Nancy Lorenzi and Bob Riley for facilitating the process so effectively; and colleagues on the consensus panel—Bob Elson, Philip Hagen, Mike Hagen, and Michael Kidd.
Page last reviewed November 1995
Internet Citation: A View From the Consensus Panel: Highlights of Moving Toward International Standards in Primary Care Informatics: Clinical Vocabulary. November 1995. Agency for Healthcare Research and Quality, Rockville, MD. http://archive.ahrq.gov/research/findings/final-reports/pcinform/dept6.html