FHIR Chat · How to indicate extension on all Questionnaire.item

Stream: conformance

Topic: How to indicate extension on all Questionnaire.item

Sean McIlvenna (Apr 19 2018 at 16:40):

How do you indicate that an extension should be allowed on any Questionnaire.item, Questionnaire.item.item, Questionnaire.item.item.item, etc.?
Right now, I have a profile which has an extension defined on the top-level Questionnaire.item, but I don't think that inherently applies to all children, grandchildren, etc.

Sean McIlvenna (Apr 19 2018 at 16:41):

(this is in a StructureDefinition profile within the context of a Questionnaire)

Lloyd McKenzie (Apr 19 2018 at 17:34):

This ties into Monday's FHIR-I discussion. You'd use a contained StructureDefinition if you wanted to define behavior that applied recursively - or you'd have to declare the extension on Questionnaire.item and Questionnaire.item.item

Michel Rutten (Apr 23 2018 at 10:14):

Interesting question! Maybe we could introduce a custom syntax like contentReference? For example:
extension context = "#Questionnaire.item" => refers to the "inline" item component definition (on each level)
extension context = "Questionnaire[...].item" => refers to specific item node at a specific level

Michel Rutten (Apr 23 2018 at 10:15):

On the other hand, I'm reluctant in adding too many "hacks" in order to make contentReferences work...

Lloyd McKenzie (Apr 23 2018 at 14:32):

That wouldn't solve the issue. Questionnaire.item might have 3 invariants, one intended to apply to all items and the other two only intended for the root. Contained StructureDefinitions is the only clean way to differentiate.

Sean McIlvenna (Apr 23 2018 at 15:38):

Yet, a contained StructureDefinition seems like a LOT of work for a simple task

Ewout Kramer (May 07 2018 at 12:02):

Yes, so shortly we were thinking about having a new syntax for it. But that's also not trivial, since that's new methodology. I guess the last word hasn't been said about this, I expect we're going to review this issue next week in Köln...

Chris Grenz (Sep 04 2018 at 19:55):

@Ewout Kramer @Lloyd McKenzie @Michel Rutten Did this ever get resolved? If I remember correctly, on the FHIR-I call we were leaning towards getting rid of contentReference in favor of contained SDs...