Hello Andrew,

I think the discussion is hampered by the lack of reference to the XDS
documents. I have looked at them in reasonable depth over that last
year or so, and do not feel they are appropriate for the Australian
setup. The US has lots of large grouped "Practices" or HMO's and they
have what XDS calls "Clinical Affinity Domains" where the patients
records are common to many users, and this is why there is no huge
security component to the model. The idea is that each department
stores its own documents and there is a registry of documents you can
query to find where documents exist for a particular patient, and then
go out and get documents of interest from a department repository. The
documents are often not standards based documents

This is not the Australian setup, In Australia we transfer documents
between lots of small "Enterprises" and there is a need for a strong
security model. You don't go to a registry and see what is available,
and get it, you either are sent it or not, or you find out from the
patient that is exists and request it from the source. The XDS model
could be used within a large hospital to locate documents that exist
in hospital departments and the security model there is part of the
organisation's infrastructure.

The XDS document specifically states "The placing and tracking of
Orders (eg Drug prescriptions, radiology orders, etc) is not supported
by XDS."

There is no structure for discovery of addressing or what type of
information is to be shared, or security. In fact what contents of
shared documents are basically blobs, but the framework is orientated
to HL7 CDA, which is not in use in Australia.

it also states that "The management of dynamic information such as
allergy lists, medication lists, problem lists etc is not addressed by
XDS".

So what XDS is is a system to index unstructured documents with a
Clinical Affinity domain with an established trust model. In that
domain it functions, but in the Australian Specialist - GP - Hospital
domain I cannot see it being a good model.

I would like to see Point to point encrypted, authenticated
transmission of structured data with rich discovery and trust
infrastructure and standards based messaging.

In the Web services hype there is a tendency to try and reinvent the
wheel. We have a messaging specification that supports Orders and
results and medications with atomic data. It is responsible for the
majority of Data transmission that occurs today - and it occurs
millions of times a day now and seems to work. That is of course HL7
V2, and we already have international and local standards that are
published and actually available for free. HL7 is not a document
structure it is a full blown messaging specification that has survived
20 years... and will continue to survive because it works. If you want
XML documents, no problem, HL7 V2 has well defined XML encoding specs.

All we lack is the transport layer to move this around and the
terminology and standards on how to use the terminology to achieve
more semantic interoperability.

wrt your radiology example:

HL7 V2 defines an OBX type called a reference pointer that allows (and
Medical-Objects has it working now and demonstrated it at HIC 2005)
radiology companies to send a link to the DICOM image that can be
accessed directly from the radiologist, if you choose to click the
link. The IHE document repository actually adds extra layers to this
model, for no benefit. For dialup users a store and forward mechanism
exists now, there is no need for XDS, for broadband users there is no
need for this, why not push the data directly to them? You get assured
delivery (or documented non delivery) with no middle man.

I agree that not all GP surgeries have the capability of running a
webserver, but we have no GPs doing this, we can proxy their SOAP/HTTP
interface and they just connect to the proxy and await incoming
connections. The client they run is lightweight and they do not have
to open firewall ports. My notebook runs three of these clients as
part of our beta testing. The proxy function is a service that doesn't
involve storing their data and is distributed and replaceable. If they
want to open a port on their ADSL model an direct it to our realtime
client and avoid the need for a proxy then they can do that as well
(It is a SOAP enabled webserver as well as a client of the proxy)

With the XDS model you will end up having 100's of repositories that
you will have to monitor. The notification in the examples I have seen
is often email, so you poll your email server, to be notified of a new
document, which means you check the registry and then connect to the
repository to collect the data. Its hardly a good point to point
model. The next stage, that I would love to see, is live realtime
messaging between providers for eg appointment availability etc and
XDS doesn't support this at all.

XDS adds a ebXML layer to unstructured data in a non realtime mode and
underlying it is no security or service discovery infrastructure.

Medical-Objects has studied the specification with the initial idea of
implementing it, but after reading it closely could not see the value
in it, and there are standards based solutions for many of the things
it offers large organisations (eg Distributed queries rather than
document registries)

Debate is needed, but information about what we are debating is the
first step. I would encourage people involved in the debate to read at
least the first 20 pages of the XDS standard so that we know what it
really is.

go to this address to find the documentation

http://www.ihe.net/

Andrew McIntyre

Medical-Objects.

Wednesday, May 10, 2006, 4:01:50 PM, you wrote:

ANS> Gavan,

ANS> Your question regarding whether it is appropriate to use a US-centric 
ANS> radiology model is very pertinent. The IHE model is, IMO, excellent for 
ANS> handling the kind of huge blocks of data inherent in the retrieval of 
ANS> images. Having only a 'manifest' stored centrally makes the process 
ANS> extremely efficient. That said, the IHE model does offer tremendous 
ANS> flexibility because it allows the storage at source, or in a centralised 
ANS> repository. This makes it feasible for pathololgy to *choose* to keep 
ANS> results on their server, for prescriptions to be centralised, while at 
ANS> the same time facilitating comms between small providers (such as 
ANS> specialists and GPs) when neither has the ability to host a repository 
ANS> in their practice.

ANS> I urge all the decision makers in this debate to consider diligently the 
ANS> implications on small providers and consumers of a SOA/WebServices/IHE 
ANS> (take your pick) model. Yes, pathology and radiology dominate in terms 
ANS> of volumes, and no model can succeed IMO without their seal of approval, 
ANS> but the Holy Grail of interconnecting small players continues to be 
ANS> overshadowed by grand designs.



ANS> Gavan Lim-Joon wrote:

>>eClinic has been running a PKI secured HTTPS 'web service' nationally for
>>more than five years now for pathology delivery.  This particular service
>>doesn't need to be online continually, and there are means to interoperate
>>amongst providers without the needs for a registry.
>>
>>I think that what is perhaps missed is the fact that the IHE standards are
>>all driven by big US RADIOLOGY vendors who have an interest in setting a new
>>industry up (ref http://hl7.org.au/2004-IHE.htm Charles Parisot, Co-chair,
>>IHE IT Infrastructure Technical Committee; GE Healthcare).  So the issues I
>>put are (A) is it applicable to the Australian environment and (B) is
>>Radiology the model to use for various health services?  That's the debate
>>I'm waiting to see happen...
>>  
>>
>>---------
>>Gavan Lim-Joon
>>Chief Technology Officer, eClinic Pty Ltd
>>[EMAIL PROTECTED] / www.eclinic.com.au
>>(03) 9381 4567 x106 / 040 234 8186   
>>  
>>



-- 
Best regards,
 Andrew                            mailto:[EMAIL PROTECTED]

Andrew McIntyre
Buderim Gastroenterology Centre
www.buderimgastro.com.au
PH: 07 54455055 FAX: 54455047

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

Reply via email to