Hi JV, Yes I am a doctor (GP) but no longer doing clinical work. I was not saying that a relational DB would not work, just that without O-R mapping tools or Row modelling/shredding techniques, it is very difficult to support the extensibility/adaptability required by clinical systems, to cope with both inter-specialty differences and a rapidly changing domain. Archetypes provide an excellent way to model and work with these requirements in code.
As a design framework OpenEHR seems to me to be on exactly the right lines. What remains unclear is the most efficient way of expressing OpenEHR concepts at persistence level. In a sense this has nothing to do with OpenEHR but it is obviously one of the trickier parts of any implementation. The Object-Relational Impedance issue is by no means unique to healthcare, and one can many attempts across the industry to allow persistence layers to easily mirror the huge advantages brought by OOP modelling and programming. My impression is that there is as yet no clear winning formula which completely reconciles the efficiency (speed and space) of traditional RDBMS design with the extensibility of object orientated methods. I am a big fan of OpenEHR but in answering Bert's original query, issues of the best way to manage persistence have yet to be resolved. In that sense OpenEHR is not as complete as he may have understood it. Ian