Re: [Gnumed-devel] Test results imports, and encounters

[Jim Busser]

On 24-Jun-09, at 2:44 PM, Karsten Hilbert wrote:

On Wed, Jun 24, 2009 at 02:21:11PM -0700, Jim Busser wrote:

So maybe we are now only talking whether to key results

creation -- when originated by an importer -- to a single

encounter per import session per patient (instead of

re-using a generic encounter per patient across all

imports).

Oh, OK. I would vote for one encounter per importer run. And

again, the importer could implement its own logic to decide

whether to re-use a pre-existing encounter of "its" type

within a certain time frame in order to aggregate successive

re-runs due to, say, intermittent problems.

A good example of this might be when the elements of a lab order (several tests) were resulted, and therefore delivered and imported, asynchronously. The importer might be coded in a way that allowed it to determine whether any results that were being added were part of the same order (lab order id) for which some results had already been received. Re-use of the earlier encounter raises no problem of ambiguous arrival of results, since they will each have their own timestamps in test_result to reflect having been created at different times.