There is no 'need' to use CDA if it is not a requisite or need from the project.
For example in Swedish National they have started with 13606 patient
summary extract, now they are working on using openEHR archetypes
(this is the last thing I know, maybe someone else from this list can
give you a more updated information) and in the future they are
supposed to use full openEHR systems.
that is true. In a pure technical sense, it is hard to find a reason for
using something like CDA because it makes life difficult if you want to
have structured data (as opposed to just narrative blocks of data under
headings), and if you do do structured data, you are stuck with
RIM-based classes, which are not systematically designed (there is a
real mixture: SubstanceAdministration sits alongside Observation and
Act). Obtaining anything that looks like a normal object (UML) model for
CDA is hard - the ones I have seen are all based on specific XSDs, which
already compromises the ability to write good, generic OO software. In
addition, it provides no semantics for distributed versioning, for an
EHR, for an Extract (in which multiple documents could be put). Further,
integrating with systems that use CDAs more or less means having to
build a transformer for each one - not much different from the HL7v2
situation.
CDA people today I think would say that if you really want to use it
seriously for structured data, either wait for release 3, or else use
'green CDA'. My impression is that both of these are evolving things
that are trying to get to where openEHR already is. I think that a solid
future for health computing will be based on openEHR & 13606, improved,
aligned and augmented with service definitions, and with a rich open
library of archetypes and templates.
My original reply was a bit rushed... if you have to use CDAs that are
already exist and are defined, then you have a couple of strategies. It
is almost a guarantee that such CDAs will be level 2, not level 3, i.e.
contain header + headings + narrative content (cf level 3 = structured
content). These are not complicated things to process, so I am not
specifically advising against that, if that is what is needed.
Many people for some years now have proposed that in situations like
this, the best way to add controlled structured content based on shared
models like archetypes is to put openEHR Entries (i.e.Observations,
Evaluations etc) inside the CDA as opaque content (from memory, XML
lumps inside EDs or so). This 'openEHR-inside-CDA' concept I think may
have been trialled somewhere. It allows you to use single-source content
modelling (i.e. archetypes + templates) within a CDA container. I know
technically it seems a strange idea, since you could just use the full
openEHR COMPOSITION, or even better, EHR_EXTRACT (which gives you the
equivalent of multiple CDA documents in one transmission) - and then the
total structure could be templated from top to bottom... but these are
the things that appear to be necessary to keep everyone happy.
As for 13606 v openEHR I would say 'watch this space'. There are some
simplifications that are likely to appear in the openEHR reference model
in version 2.0; I would think that there are improvements available in
13606. There is now a lot of interest in producing (finally) a single
combined 13606/openEHR for the ISO 13606 revision, starting May 2012. It
won't be fast though (it's ISO after all). Right now, I would go where
there are more tools & technology that correspond to your particular needs.
Please note: I'm a newbie to HL7 and TC215, please forgive me in advance for any question doesn't make sense.
I was at the recent TC215 meeting in Chicago and there was some mentioning about revising ISO 13606 due to five years review process of ISO. Are there experts in this reflector working or have interest to revise the 13606? I am very interested in this space and would like tag on to those are already working on this.
Thanks!
--Wo Chang
Digital Data Preservation, Project Lead
Digital Media Group, Manager
Information Access Division (IAD)
Information Technology Laboratory (ITL)
National Institute of Standards and Technology (NIST)
Email: wchang@nist.gov
Phone: (301) 975-3439
Indeed it does make sense. There is a lot of interest to ensure the
industry lessons from openEHR and 13606 implementers can be added into
this revision of 13606.