Stream: questionnaire
Topic: Question about questions
Eric Haas (Apr 14 2020 at 17:01):
Jean Duteau (Apr 14 2020 at 17:03):
sorry all - didn't know about this stream
Jean Duteau (Apr 14 2020 at 17:08):
Questionnaire is definitional so QuestionDefinition == Questionnaire. Why is it too heavy? ifn you want to mix and match item for Qs you still need the source meta data for each question?
name/title/identifier/version/effectivePeriod - those are all things that are sort of irrelevant if I'm just defining one question that can be used in multiple different questionnaires.
And I'm fine with using the Questionnaire to represent a single question. I'm still interested to know the relationship of the new extension to the .item.definition element
Lloyd McKenzie (Apr 14 2020 at 17:36):
If you want to define questions independent of a Questionnaire, you can use StructureDefinition. You can either define a single StructureDefinition per data element or can point to the specific 'element' within the StructureDefinition that defines the question of interest. The StructureDefinitions could be FHIR resources or profiles, but could also be logical models.
Lloyd McKenzie (Apr 14 2020 at 17:37):
The item.definition points to the canonical of the StructureDefinition with a '#' followed by the Element id.
Jean Duteau (Apr 14 2020 at 17:40):
Lloyd McKenzie said:
If you want to define questions independent of a Questionnaire, you can use StructureDefinition. You can either define a single StructureDefinition per data element or can point to the specific 'element' within the StructureDefinition that defines the question of interest. The StructureDefinitions could be FHIR resources or profiles, but could also be logical models.
Right - that is sort of what the existing specification documents but I just can't wrap my head around how exactly to do that (thus the JIRA ticket I created requesting an example of that).
Lloyd McKenzie (Apr 14 2020 at 17:41):
We will have examples of this coming for SDC soon, but I don't think we've gotten to the population/extraction examples where this would be most relevant yet
Jean Duteau (Apr 14 2020 at 17:44):
I guess my biggest hurdle is when you say "you can use StructureDefinition". Do I just create a vanilla StructureDefinition instance? There is a little bit of discussion about how the ElementDefinition fields map to questions but it's not complete and I couldn't see how to do the simple - question.item.text, question.item.type, question.item.answerOptions.value[x].
Lloyd McKenzie (Apr 14 2020 at 18:33):
Look here: http://build.fhir.org/questionnaire.html#definition
Though it looks like we missed saying that ElementDefinition.code.text is the question text. @Grahame Grieve can you confirm?
Jean Duteau (Apr 14 2020 at 18:36):
Lloyd McKenzie said:
Look here: http://build.fhir.org/questionnaire.html#definition
Though it looks like we missed saying that ElementDefinition.code.text is the question text. Grahame Grieve can you confirm?
I did look there and it didn't help. :). type was there, but it seemed I had to define a valueSet to give my simple options?
Lloyd McKenzie (Apr 14 2020 at 18:38):
With ElementDefinition, yes you do. No way to define re-useable options without a value set.
Grahame Grieve (Apr 14 2020 at 21:13):
though you would typically make the value set a contained one
Paul Lynch (Apr 14 2020 at 21:15):
If the purpose of defining questions as ElementDefinitions is to re-use them in multiple Questionnaires, what would contain the ValueSet?
Grahame Grieve (Apr 14 2020 at 21:16):
the structure definition that contains the element definition
Eric Haas (Apr 14 2020 at 21:59):
how this approach lighter weight than my suggested approach of just creating a bunch of mini questionnaire that you reuse. Then you don't need excess baggage of a bunch of SD and valueset for each question.... what are the pros and cons?
Eric Haas (Apr 14 2020 at 22:00):
seems like an unnecessary layer of definitions
Eric Haas (Apr 14 2020 at 22:00):
to me
Grahame Grieve (Apr 14 2020 at 22:06):
it at least partially depends on whether you get reuse of the definitions elsewhere
Lloyd McKenzie (Apr 15 2020 at 02:42):
Mini-questionnaires impose requirements on link id uniqueness and also 'polute' the questionnaire space with a whole bunch of Questionnaire instances that aren't, really. There's no way to declare a Questionnaire as abstract/non-implementable. On the other hand, logical models are intrinsically distinct. As well, you can leverage existing profiles and resources for element/question definitions.
Eric Haas (Apr 15 2020 at 14:39):
As well, you can leverage existing profiles and resources for element/question definitions.
can do that with the Q you import into a root too, right.
Lloyd McKenzie (Apr 15 2020 at 15:07):
There aren't existing questionnaires for elements defined in the resources - so you'd have to define a separate instance. In general, the definition mechanism is preferred. Sub-questionnaires I would expect to be reserved for situations where you've actually got a collection of questions intended to be re-used as a group.
Eric Haas (Apr 16 2020 at 00:17):
@Paul Lynch does the NIH form builder use definitions to build Qs?
Paul Lynch (Apr 16 2020 at 00:20):
Not yet, but neither does it (yet) support sub-questionnaires.
Last updated: Apr 12 2022 at 19:14 UTC
