[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Gnumed-devel] HL7 OBX result messages that involve multiple lines e
Re: [Gnumed-devel] HL7 OBX result messages that involve multiple lines esp Microbiology
Mon, 11 Feb 2008 13:12:15 +0100
> OBX contains a field
> 004 Observation Sub ID
> Which is used to enumerate multiple OBX records when they
> 1) pertain to the same LOINC code, for example when there are
> multiple pathologists' comments:
IMO those should be rolled into the relevant parent OBX note_test_org.
This is where the importer will actually have to have some intelligence
as to the *content* of the data.
Welcome to the idiosyncrasies of HL7 :-(
> 2) describe Microbiology results that describe sites, potentially
> multiple organisms, and their antibiotic susceptibilities (see also
> HL7 V2.3 Section 7.4.5):
> Here is one format and example in which reported first are all the
> Culture results; the Observation Sub-ID contains the id of the
> culture (i.e. 1, 2, 3 etc); after this, the Organism is reported,
> followed by all the antibiotics, in the order of the culture result:
> OBR|1||02-123456-URC-0^bcb|URC^URINE CULTURE^L||200011081429|
> OBX|2|FT|6463-4^Culture^L|1|Light growth of: Klebsiella sp.|||N|||F|||
> OBX|8|FT|6463-4^Culture^L|2|Light growth of: Staphylococcus aureus|||
> The above would carry little meaning as individual items in a GNUmed
> test_result table...
I agree. However, if I were to write an importer I'd attempt to roll
the above into some tabular format to be put into val_alpha (note that
there can be multiline results in val_alpha).
Note: site == "Urin" ;-))
> They seem to need to be linked to the test panel
> from which they originated (in this case, Urine culture) so maybe we
> need a foreign key to the test_profile as a nullable column in
I'd rather handle this particular case as a test_result of test_type ==
"Urine Culture" with a multiline val_alpha.
There's one argument against doing so in a simplistic way: computability.
The way it is structured in the original HL7 it, well sort of, lends
itself to clinical reuse -- GNUmed might check AB sensitivity against
currently prescribed antibiotics. One way to preserve that information
would be to use a well-defined structure for the multiline text, say,
XML or StructuredText (as in a wiki).
> Do we see some value or need (or none) to the OBX sequence counter
> (above, OBX|1 through OBX|18) and similarly any need to store the
> 'Observation SubID" which in this case associates the individual
> species (Klebsiella sp versus Staphylococcus aureus) with their
> specific antibiotic profile?
I absolutely see value in preserving the information it carries.
We'd need to analyze more instances of this to develop a suitable
relational structure if needed.
Psssst! Schon vom neuen GMX MultiMessenger gehört?
Der kann`s mit allen: http://www.gmx.net/de/go/multimessenger