On 2010-08-17, at 2:12 PM, Jim Busser wrote:
> this second scenario especially seems to make the case for logical, computational, and clinical legitimacy to keep OBR and OBX meaningfully associated?
I came across a different use case of
one OBR
multiple identical OBX
where a "single test (protocol)" involves the patient bringing into the lab 3 chemically-treated, self-sealing "cards" each of which contains a smudge of the person's feces (yes, that's "bowel movement") applied using a wooden applicator stick... yuk)
Anyway this is referred to as a test for "occult blood" and may be returned as
OBR-Occult Blood
OBX1-Occult Blood1
OBX2-Occult Blood2
OBX3-Occult Blood3
which would seem to make the case that each OBX be imported as its own row (could the subid be stored in GNUmed as a subfield within our test_reult fk?)