Ian Cheong wrote:
> At 6:38 am +1100 22/12/06, Tim Churches wrote:
>> See
>>
>> http://www.nehta.gov.au/component/option,com_docman/task,cat_view/gid,122/Itemid,139/
>>
>>
>> Worth scrutinising and commenting upon, in writing, to NEHTA (but
>> perhaps comments can be CCed to this list to promote discussion firts?).
>> But don't only send comments to this list, send them to NEHTA as well as
>> formal responses, for goodness sake.
>>
>> and
>>
>> http://www.nehta.gov.au/component/option,com_docman/task,cat_view/gid,151/Itemid,139/
>>
>>
>> For path test and result codes, it seems like SNOMED CT is where it is
>> going to be at, and the Auspath codes (see ), which are based on LOINC
>> codes, are deprecated, which makes sense to me since LOINC, although
>> freely available, doesn't adequately cover all of health and health
>> care, and ultimately it doesn't make sense to have to build decision
>> support and other information systems which have to deal with one
>> terminology and corresponding set of codes for pathology, another for
>> procedures, another for more general diagnoses, yet another for
>> disabilities and so on.
>>
>> Usual grumbles about the use of Microsoft Excel files as the format used
>> to distribute the draft code sets, when three more clicks by the
>> relevant NEHTA functionary could have rendered the lists as CSV text
>> files, acceptable to everyone.
>>
>> Tim C
>>
> 
> The Auspath codes are here:
> http://www.austpath.uow.edu.au/index.cgi

Thanks, forgot to include that URL in my post.

> lack of revision could mean one of two things:
> 1. lack of use - which probably applies to request codes
> 2. good fitness for purpose - which probably applies to report codes
> sent routinely in HL7 messages daily.

Over the last 18 months we have been talking to a lot of labs with a NSW
presence, both public and private sector, regarding electronic
communicable disease reporting, and the take-up of and enthusiasm for
LOINC request and result codes seems very low. One large lab that we
have dealt with seems to use them, the rest don't.

We also found large gaps in AusPath result codes, and smaller gaps in
the request codes, with respect to microbiology (including NAT and
serology). Some of these gaps reflect gaps in LOINC, others reflected
gaps in AusPath which seems to be a slightly divergent subset of LOINC.
We did initially try to fill these gaps but no-one seemed to be
maintaining or monitoring submissions tot he AusPath web site, so we
gave up and decided to recommend SNOMED CT to labs for all coding
purposes with respect to comm disease reporting in NSW. The labs
greeting that suggestion warmly, although all indicated that it would
take time to change over, but we expected that.

> The NEHTA effort reads like 1 steps forward, 3 steps back.
> 
> Auspath request codes have:
> * preferred terms
> * synonyms

SNOMED CT has equivalents of these too, out of the box.

> * infinite extensibility based on usage - if you want a new code or
> synonym you can have it and someone will figure out a preferred term and
> linkage

NEHTA is supposed to be establishing a mechanism to allow local
extensions to SNOMED CT using a local (Oz) namespace assigned to it by
the College of American Pathologists (CAP), and shortly by SDO, the
international Belgium-based SNOMED CT governance body that will be
taking over from CAP. The idea is that new, local codes can be assigned
quickly and then, if indicated, they can be sent up for international
consideration by the SDO for inclusion in SNOMED CT itself in due course
as a single, world-wide code (which may mean translation or mapping from
the local code). I gather that NEHTA wants to operate this local SNOMED
CT code maintenance facility itself and has been recruiting (or trying
to recruit) suitably experienced people to do so. That's the only bit I
am dubious about - would have been better to outsource such operational
work to an existing centre of nosological excellence, like the National
Centre for Classification in Health at Uni of Sydney - they maintain
ICD-10-AM and several other widely used classifications, as well as
doing research into health classifications and being involved in
teaching medical information managers etc.

> Auspath report codes are based on a simplified set of LOINC codes,
> because proper LOINC codes go down to method of analysis, which makes
> them too unique for cross-comparison. LOINC codes are part of SNOMED
> last I looked. (Has it changed???)

This simplification was problematic for our requirements (notification
of communicable diseases to health depts).

> The general method for codifying reports is described in AS4700.2 and
> HB262 which is presently under revision I think. Not all reports are
> encoded as LOINC, as it depends on the nature of the report. Certainly
> individual biochemical analytes are LOINC coded. 

Ian will be pleased to know that our proposed NSW comm dis notification
HL7 message formats, with which the labs seem happy, are backwardly
compatible with AS4700.2 and HB262, but extend those to overcome the
ambiguities and close the interpretation gaps in those documents. We
have been planning to submit these extensions to Standards Australia
once our public health colleagues in other states and territories agree
to what we propose (they broadly do, I think). We have been keeping the
relevant NEHTA people in the loop the whole way along.

>From memory (it is over a year since we pored over AS4700.2 and HB262),
LOINC coding for request types is recommended but not mandated by the
standards, and results reporting is not specified at all. Or is it the
other way round? Anyway, for one anything can be used and for the other
LOINC is recommended but not mandated (and the standards are just
recommendations anyway, of course).

> Pap smears are SNOMED
> coded as one would expect. The general principle is the test/analyte
> name would have a LOINC code and the result value would be numeric or
> SNOMED coded. It should be a piece of cake to translate unique LOINC
> test codes to unique SNOMED Procedure codes, but the complexity of
> pathology reporting is not well described in the available NEHTA
> documents. It is well understood by the pathology informaticians who
> actively contribute to standards development in Australia and NZ.

Mapping test, result, specimen, specimen site and other relevant LOINC
and mostly lab-specific codes from the 5 or 6 biggest labs in NSW to
SNOMED CT, just for the sub-domain of communicable diseases
notification,  took an experience microbiologist and an assistant
several person-months (part of that was getting up to speed with SNOMED
CT). But only needs to be done once - maintenance of SNOMED CT subsets
thereafter should not be too taxing, we think.

> Perhaps someone can understand the rationale for flying a new flag when
> everyone is happy standing behind the old one.

Our impression was that almost everyone was indifferent to the old LOINC
flag, and that in general path labs saw the advantages of using a single
coding scheme (ie SNOMED CT) for requests, results, specimens, analytes,
test classes, anatomical locations, diagnoses, procedures, presenting
problems and so on. The future is one in which the clinical setting for
a path request is described in SNOMED CT codes in the electronic request
message to the path lab from the GP or elsewhere, and the lab then sends
back the results and documentation of the test and specimen as SNOMED CT
codes to the GP, also electronically. All these SNOMED CT codes can be
grist for electronic decision support and other value-adding tools at
the lab, and in the GP's information system, and these tools can have
access to a more wholistic picture of the patient encoded with a single
coding system, not just LOINC-coded path test results (and ICD-10-AM
coded diagnoses, MBS-coded procedures, ICPC-2 coded presenting problems
and so on).

None of which is to say that SNOMED CT is perfect - far from it - but it
is far more comprehensive than any other terminology and its
comprehensive nature confers very powerful "network effects" on it, I
think. But there is a lot of R&D work which needs to be done yet to make
SNOMED CT readily deployable in clinical information systems. In lab and
other "back-room" information systems, there is some slog work to be
done in deploying SNOMED CT, but not much R&D rocket science is needed.
Not so for clinical deployment, though. But Australia has the necessary
expertise and capacity to carry out that R&D, all that is needed are
some funding bodies with the vision to fund it. If it is left to "the
market", I fear the results will be disappointing and expensive, and
will need to be paid for in US$ and Euros.

Tim C

_______________________________________________
Gpcg_talk mailing list
[email protected]
http://ozdocit.org/cgi-bin/mailman/listinfo/gpcg_talk

Reply via email to