Quality Assurance Committee Minutes for January 12-15, 1993 QA Meeting (Tuesday afternoon, January 12) Attending: Gary Dickinson, George Geelere, David Black, Wayne Tracy, Scott West, Stan Huff, John May, James FRy, Scott Joyce, Peter O'Brien, Dean Ellis, David Marotta, Cliff Bernstein, Mark Tucker The QA committee met in a joint session with the MEDIX Laboratory sub- committee which had previously scheduled to meet in Orlando. We discussed work on an object oriented framework, and how that related to MEDIX's role within the MSDS (Messaging Standards Development Sub- Committee, the group within the HISPP working on standards convergence) process. The following items came up during our discussion: o The standards committee need to agree on modeling conventions. We noted that MEDIX and some of the CEN work is using Coad/Yourdon. The group's consensus was that use of Coad/Yourdon Object Oriented techniques should be recommended to HL7. o Division of labor between multiple standards groups needs to be based on a shared high-level domain model. The division of subject areas between the standards groups (as shown by the minutes to the MSDS meeting) should be done in terms of this model. The model could be developed in a short session such as a weekend retreat bringing together domain and modeling experts from all the standards groups. The MSDS group should work to organize such a meeting. o The convergence process needs a coordinating/reconciling body which will point out discrepancies in the models developed by the individual groups. This body should draw its membership from all the cooperating groups. o The HL7 QA committee will plan to hold future meetings on the framework jointly with MEDIX. We will schedule a meeting for Monday and Tuesday of the next HL7 meeting (May 16 & 17 in Boston). o The issue was raised: "Who owns the model?". In response, it is owned by all the groups jointly, but it needs to be maintained in one place. MEDIX should take on the role of maintaining the repository for the model. o The need for a single data definition language was raised. Both ASN1 and the Object Management Group's CORBA are candidates for this role. o We need more joint development meetings to make the MSDS process work. These meetings should bring together people with domain and with modeling expertise. The QA committee should try to ensure that the right people are available to act as facilitators in building the model. The ADT meeting made it clear that we need to bubble modeling issues up to the QA committee so they can be resolved. The group voted to accept the following proposal for inclusion in the minutes: Members of the HL7 QA Committee met jointly with members of the MEDIX Laboratory modeling subcommittee to discuss how standards development might be best harmonized under the MSDS agreement drawn by HISPP. The group which included members from ASTM and a number of MEDIX Subcommittees in addition to HL-7 found itself in almost unanimous agreement on a basic strategy. This group offers its plan to the MSDS, and recommends its adoption by that group and the constituent standards bodies as the basis for advancing the HISPP objectives. REQUIREMENTS The group is agreed that the following is a minimum set of requirements: 1. There needs to be a common model "framework" document that lays out for standards developers how the model is to be developed; how it will be reviewed and coordinated; and how the content of the models is described in modeling tools and in a common syntax. The framework should be based upon object-oriented concepts, but will not require object technologies to implement the resultant standards. Both HL-7 and MEDIX have done substantial work to develop such a framework. These efforts are believed to be readily harmonized. The harmonization should include participation from the other cooperating standards group and from CEN-TC251 which has developed similar models in support of its standards efforts. The harmonized framework will be offered for ballot to the constituent standards bodies. 2. There needs to be a high-level object model of the whole standards domain to serve as the basis for affirming and clarifying the subject area assignments that were initially spelled out in the MSDS agreement. (The group believes this could be accomplished in a three-day weekend retreat.) 3. There needs to be a global data model maintained in a repository by a common secretariat to document both the evolution and final representation of the models as they are developed. 4. Each of the constituent standards bodies will develop the portions of the model under their purview, and will continually feed these developments to the secretariat for inclusion in the global model. A common committee will review these submissions to identify conflicts that may develop within the model, and to adjudicate these differences. CONCLUSION THEREFORE, the group recommends that a model coordinating committee be designated by MSDS to undertake the above tasks. A member of MEDIX, in its role to coordinate the Global Data Model, should chair the committee; and MEDIX should provide the secretariat services, including the global data model repository. The membership should be comprised of two to three members from each of the constituent standards committees. To be effective, these must be members whose primary objective is the development of a robust, global data model, and who bring a knowledge of data and object modeling to the table. Their first responsibility represent the objectives of the global data model to their constituent standards body, not vice-versa. QA Meeting (Thursday Morning, January 14) The committee met on Thursday morning (with a rather small turnout). The discussion covered Quality Assurance activities for Version 2.2, and the status of the HL7 Message Development Framework document. The group suggested the following areas of importance for QAing V2.2. 1. Reviewing the HL7 tables. This has been a source of concern and confusion in the past. We felt that the technical committees should indicate which tables were required for HL7 processing. These should have all their values defined within the standard. Other tables should either be valued locally, or should refer to a completely specified external standard, e.g., ICD9-CM. This would mean dropping the formulation in which some table values are indicated while allowing users to add other values locally. If the technical committee wanted to indicate possible values to nail down the data element definition, these should be clearly listed as examples. 2. Referential Consistency This involves checking references within the document to ensure consistency. 3. Consistent Format and Appearance This involves ensuring that the chapters have a consistent look and feel. Consistency can range from the level of explanation, and internal numbering formats, to consideration of whether all chapters take the same approach to messages and events. 4. Data Element Definitions We need to ensure that data element definitions are complete, clear, and communicate the meaning of the field to readers. Definitions also need to be consistent across chapters. In general, the group felt that QAing Version 2.2 is going to involve a lot of judgment, since there is no clearly defined framework that the contents of the release will be measured against. The QA process should not attempt to re-write the standard after the fact, but it should try to avoid the problems that arose from the lack of in-depth QA for V2.1. The meeting reviewed the Technical Committee balloting for our framework document, the HL7 Message Development Framework. The document has passed the technical committee balloting stage. There was one negative vote, which was withdrawn after discussion and changes to the draft document. At the same time, a number of committee members spoke against sending the document to Working Group balloting. The reason is that this document has been overtaken by events, and would be an entity- relationship based framework in an object-oriented world. Since this was a pivotal issue for the committee, and since there were many key players missing, we decided to defer a decision on this topic to a meeting on Thursday afternoon which could be better attended. QA Meeting (Thursday afternoon, January 14) The committee met on Thursday afternoon in a special session to discuss the HL7 Message Development Framework. We discussed what to do with the document in light of the fact that a number of the technical committees were working on modeling, but were using and Object Oriented as opposed to an Entity Relationship point of view. We discussed three options: 1. Put the document to a Working Group ballot, and start working on an Object Oriented framework for Version 3. This is the course that we had planned over the last 18 months since work started on the document. It would give the document the widest exposure, and highlight the useful material contained within it. Much of the discussion in the document, especially related to message development techniques and to naming standards, will remain relevant in an OO environment. On the other hand, people were concerned that it would be confusing to release an document of this kind at the same time as people were just starting to get excited about a different style of modeling. Why should people read an involved discussion of ER modeling when they had a different style on their minds. Therefore, release of an ER based model would be a flop, which would also hurt the later release of an OO model. 2. Shelve the document, and start work on our OO version. Since groups were already working on models, they were coming up with issues that needed to be dealt with right away. Therefore the emphasis should be put on solving people's OO related problems, and not on confusing them with yesterday's framework. 3. Don't put the document to a ballot, but do publish it as part of the Working Group minutes. This would not confuse people, but would make the document available as a reference. It would allow access to some of the useful material contained within the framework document. We voted for option 3. The document will be contained within the meeting minutes. Contact Mead Walker for a paper copy. Next Meeting The QA committee will be meeting on Monday and Tuesday of the week of the Working Group meeting to discuss OO modeling, and our updated framework document, The dates of this expanded meeting will be May 17 and May 18. There will also be a meeting within the usual Working Group meeting to discuss other issues.