FHIR Chat · subsumption in CQL · cql

Stream: cql

Topic: subsumption in CQL

Erich Schulz (Jun 28 2016 at 02:18):

@Bryn Rhodes just moving our inititial thread...

Erich Schulz (Jun 28 2016 at 02:19):

i found this example for a Concept:

Erich Schulz (Jun 28 2016 at 02:19):

define Concept1: Concept
{
   Code '66071002' from "SNOMED-CT:2014",
         Code 'B18.1' from "ICD-9-CM:2014"
} display 'Type B viral hepatitis'

Erich Schulz (Jun 28 2016 at 02:19):

could this be what I'm looking for?

Erich Schulz (Jun 28 2016 at 02:20):

before i post a comment on the spec I'd be keen to discuss a bit first

Erich Schulz (Jun 28 2016 at 02:20):

(I'm not sure I've yet found the right forum for these conversations)

Bryn Rhodes (Jun 28 2016 at 15:46):

Agreed, subsumption is an important use case, but the design is based on the premise that it's not a _critical_ one, because you can always create a valueset that contains the subsumed codes (which in fact is what measure definitions often do, precisely to avoid the potential ambiguity introduced by using subsumption).

Bryn Rhodes (Jun 28 2016 at 15:47):

So with CQL we've focused first on making sure we support just the basic operations correctly, expansion and membership, which we define strictly as equality-based membership.

Bryn Rhodes (Jun 28 2016 at 15:50):

And no, the Concept type in CQL is not for representing subsumption, it's conceptually equivalent to the CodeableConcept in FHIR, so by definition, all the Codes in any given Concept should carry equivalent meaning.

Erich Schulz (Jun 28 2016 at 22:38):

ok thanks

Erich Schulz (Jun 28 2016 at 22:39):

I think I need help from the likes of @Michael Lawley @Donna Truran and @Grahame Grieve

Erich Schulz (Jun 28 2016 at 22:40):

the question being "how to represent ideas like asthma and diabetes etc etc in a succinct, standard manner within the text of CQL logic?"

Erich Schulz (Jun 28 2016 at 22:40):

(with goal of allowing expression of shareable knowledge artifacts)

Bryn Rhodes (Jun 28 2016 at 23:14):

I can tell you that the way we're doing that for quality measures is to define a value set that specifies the codes of interest, and use that value set within a retrieve:

Bryn Rhodes (Jun 28 2016 at 23:15):

valueset "Diabetes": 'blah.blah.blah'
define "Diabetes Diagnoses": ["Diagnosis": "Diabetes"]

Bryn Rhodes (Jun 28 2016 at 23:17):

The value set is an identifier that may itself be a definition that uses subsumption to define the concept, so that functionality is potentially still available, even if all you have available is expansion and membership.

Erich Schulz (Jun 29 2016 at 00:43):

i guess blah.blah.blah is what needs specifying

Erich Schulz (Jun 29 2016 at 00:43):

otherwise we're curly braces all over again

Last updated: Apr 12 2022 at 19:14 UTC