PROMs / Score Panel Hierarchy Modeling

Hello everyone!

I am in the process of creating an archetype for the Seattle Angina Questionnaire-7 (SAQ-7) (the full questionnaire can be viewed here on page 644). You can see an overview of the panel hierarchy of the questionnaire on the LOINC website.

During the modeling process I was wondering about how to model the structure of the questionnaire including the panel hierarchy with superordinate and subordinate questions.

Option 1
Should I use a CLUSTER for question 1 to represent the superordinate question and slot the subordinate questions 1a.-c. as ordinals?

I doubt that this is the intended use of a CLUSTER but have seen this approach in several score/questionnaire archetypes of which none have been published yet. When querying for this Cluster with AQL you get the values of each ordinal as a response which is quite nice.

Option 2
Alternatively, the questionnaire could be modeled with ordinals only, but then the parent question might have to be repeated in the descriptions of question 1a-1c:
This seems to be more inline with the modeling guidelines.

But let’s say I want to render this questionnaire directly from an .opt-template using the description fields that include the full-text questions. Then the second approach would render the superordinate question 1 for each of the 3 first questions.

I hope I could get my question across in a understandable way :yum: and I am looking forward to hearing your input!

Best,
Jannis

1 Like

Hi Jannis!

I’d model this as a combination of your two options, by using a cluster to group questions 1 a, b and c, to avoid unnecessary repetition, but leave 2, 3, 4 and 5 on the root level. This is in line with how GOSE and YGTSS-R have been modelled.

4 Likes

Dear Silje

Thank you very much for your answer and the examples provided.
I am happy to see that clusters can be used in such cases!

Best,
Jannis

1 Like