Horst, You do write some great stuff!I, along with other colleagues, spend quite some time each year on standards; the processes are far from what one might expect, and certainly far below my expectations as a disciplined thinker; generally none of the bodies do any implementation work to test their specifications. The IETF is a well-known exception; for me, Corbamed was probably the best quality forum on the one meeting I went to, mainly due to the engineering culture of the OMG (but with quite a few clinicians present at meetings, and excellent technical representation such as Dave Forslund). In short: I applaud your scepticism.
I am seriously worried about these standards, their complexity, the continuous chages, their failure to succeed in getting to be used. I have studied them the best I can, and I feel they are hopeless in the way they are implemented.
However, we have made some progress in our efforts in the standards arena: CEN TC 251 is becoming interested in implementation experience, and not just on 4 year time scales. I'm not sure about HL7. But I do know one thing - the presence of openEHR as a means of testing the specifications in real projects, and offering the results back to the standards orgs has made a few people nervous. Some even fear that it might be used instead....personally, I make no such grandiose claims, and simply aim to help it keep going and turn more specs into working software. This community will be (or can be) an important factor in that endeavour.
Just imagine some country starting a EHR and e-health project. To succeed the end usersd must be happy to use them, and must benefit from them immediately. If things are complicated due to the fact that there is conflict in what standards to implement and when, and this affects the speed of implementation, the whole thing may fail.a) can standards be 'insinuated' into existing systems? This is the messaging mantra, and assumes that every system's data is transformable to such messages.
I wonder if standards cannot be implemented to allow communication between systems that already exist, and can be implemented and insinuated into prevailing systems? Will some simple guidelines to the end user software - eg: data should be exportabl;e as XML etc. may not suffice?
b) can simple XML do it? Probably not, if you want to deal with the complex structures that naturally occur in many types of clinical data, and also make it interoperable vertically, i.e. between applications such as decision support and the back-end.
Both these have been tried many times, and only bring basic advantages; and if they are not complete specifications, companies come in and implement their own flavours, creating whole new worlds of incompatibility...
Excellent end user software in EHR, not having any standards are being used and liked by many clinicians around the world. I feel that end users will go their own way in the prevailing atmosphere relating to "standards".You are right, it does, and I can't always say I blame them. One approach to interoperability is to simply make sure every site runs exactly the same software; Vista is halfway there (but I am told localisations can make for different data schemas)...
Will it really matter to the patient? to the clinician? If so to what extent?I suggest that where it will matter is in shared care (can all the carers of a diabetic patient see the additions to the EHR of the others?) and decision support (can various DS modules interrogate the EHR in a reliable way?).
- thomas beale
