Re: [Gnumed-devel] HL7 OBX result messages that involve multiple lines esp Microbiology

[James Busser]

On 13-Feb-08, at 6:04 PM, Karsten Hilbert wrote:

> > Seeing below, it appears that LOINC 19803-6 is used above as "Site"
> > and they have re-interpreted 6463-4 to imply "Culture". The problem
> > is that there is one OBX per antibiotic making none of their LOINCs
> > unique to the test_type, also "Urine culture" has no LOINC so we are
> > back to the problem of the test_code or test_panel
> It is probably best to really have test_panel.
>
> > and it feels very
> > arbitrary how to reliably get the multiple OBX into one multipline
> > val_alpha unless the observation subid (OBX|1, OBX|2 etc) can
> > reliably be the basis to trigger this.
> Sure, why not. OTOH HL7 likely isn't a fully relational structure and
> much depends on ordering of lines. IOW, all OBX followed by something
> other than OBX belong to their preceding OBR.

Not sure of the above since there can be multiple distinct OBX within  
a single OBR. Also I was clumsy above since multiple distinct OBX are  
communicated
OBX|1
OBX|2
where the 1,2... are simply numeric counters, the subid is actually a  
few | fields to the right in OBX 004

Maybe more correctly (?)

	all OBX having _code_ that immediately follow an OBX that had an  
*identical* _code_ *and* also have an OBX subid >1 go into note_test_org

This will help to differentiate the situation of true multiple  
distinct OBX having same code e.g. when a patient submits multiple  
(e.g. stool) specimens where the same test is performed on all specimens