Following the discussion of a few days ago, may I propose an other perpective about GEHR and HL7 ? In my view we need to make a more clear distinction between the content of the information and the way we can include it in messages. In the past most standardization efforts of HL7 and the CEN, did focus primarily on message exchange problerms, but I see now 2 kinds of issues : ( 1 ) PERSISTENT MEDICAL INFORMATION ARCHITECTURE : - We live in a world where large networks begin to be implemented, up to now at least inside large hospitals, but later may be also at regional or even national levels. We need to be able to think in an environment where everything becomes on-line, soon or later. When I need a view, or in general any set of selected information from the global medical record, the system will get it immediately from one or several distributed storages sytems; (this of course on condition that rigorous access rules will be enforced, but this is not the point in the current discussion). In this network context, the most critical issue is the architecture of medical objects. We need a clear understanding of the essential properties of generic medical objects (so called 'medical items'), including many optional attributes, including specialized class derivations, including links between the objects in order to be able to navigate in the medical record (relation to the original contact, to the medical problem,, to the responsible department, etc....). Inside such a network everything is normally assumed to be linked by means of pointers. - I see GEHR as one of the best approach to achieve this kind of architecture. It is dealing with the content's architecture, but not at all with any transportation issues. ( 2 ) TRANSPORTATION ISSUUES : - Just a simple comparison with the use of internet : when you look at many internet pages, sometimes you may want to save a copy of a document in your local system. You may keep it in its original format, i.e. in this example in HTML format, or you may want to make a conversion to any proprietary format in function of a given legacy system. This is your problem. You may receive Emails, i.e. a notification that some event did occur somewhere else, and of which you were not otherwise aware of. In these emails you may receive an advice to take a copy from a given URL. You are free to accept this advice or not. You may also want to execute automatically all this kind of advices. - In fact, it is true that many systems are not connected and need message exchange. It is of course very necessary to be able to send a set of objects to an other environment. This require solutions to additional issues, in order to package your message in an appropriate container. The main problem is then to send coherent sets of objects, as defined above. For example the receiver need to understand a minimum about the related patient's identification, about the context of data acquisition, etc... The receiver could do nothing with pointers refereing to things, not visible in his new isolated environment. If the context and auxiliary dictionaries are not available, the receiver will probably want to receive a plain text version. CONCLUSION : I believe we should clearly identify content and transportation issues separately. Using this new philosophy, we will become able to cope with both situations, as well the network approach, as well the traditional interchange approach. Etienne Saliez MD clos des Pommiers 4 La Hulpe B-1310 Belgium [EMAIL PROTECTED]
