HL7 delivers healthcare interoperability standards

Implementers converge on one way to do CDA

March 18 2007

The CDA profiles for for structures defined by IHE and ASTM are converging. The IHE PCC profiles, which specify how clinical documents should be structured to be IHE conformant, are being adapted to ensure that they are consistent with CCD, the HL7 CDA representation of the ASTM CCR (Continuity of Care Record).

For full background the email is below:

Email from: Boone, Keith W (GE Healthcare)
Sent: 18 March 2007 16:09
To: lists.hl7.org
Subject: PPC Technical Framework Release 2.0

The IHE PCC Technical Framework will be updated this year to contain entries that conform to CCD, and to support migration of XPHR to CCD. In preparation for that activity, I have merged the profiles from this year into the technical framework.

There are a few issues that come up with one of the profiles (PPHP), having reviewed what some of the changes entail.

The "Impression" section is supposed to contain a list of not just problems, but also risks that need to be managed for the patient. This is fine in that I can see these as being models as concerns, with the subject of concern being the procedure and the risks. Issues occur with various modeling choices to identify the concomittant risks (e.g., post-operative bleeding), due to constraints on HL7 CDA, Care Structures, and Clinical Statements.

One possibility is to model the Risk as an observation, in event mood that the "patient is at risk of", which requires a vocabulary to establish that the observation is a "risk of...". The key issue here is that controlled vocabularies do not have sufficient precoordinated terms that identify risks to address the needs of the profile, and separating this out into code = "risk of", and value="condition" runs into the danger of applications incorrectly recording risks as actual patient problems -- I think I can live with that, but I'd like comments.

We need to be able to support coding in a few vocabularies, some of which do not support post-coordination, and most of the applications would not support it either, so that option is just not going to work.

Another option would have been to use an observation in RSK mood, which would be perfect, except for that doesn't exist in CDA, Care Structures or Clinical Statements space. If HL7 constrains the mood represented, I believe that it must include the contrary mood (GOL and RSK are contrary moods), unless there is a specific reason to prohibit it.

With what is left, the only options left are to record risks as non-goals. That's so ugly as to not even be worth consideration.

If any of you have good suggestions about how to resolve this problem, please let me know.

Those of you in HL7 Patient Care or Clinical Statement land might want to take a look at GOL/RSK and GOAL/RISK usage in your models. For example, using condition tracking structure today, I cannot say that the subject of concern is the risk of X without resorting to a vocabulary based solution, because you've eliminated risk from your model in both Observations (moodCode='RSK'), and act Relationships (typeCode = 'RISK'). I've already added this to the CDA R3 open issues list.


Last modified 13/09/07