FHIR Chat · data-absent-reason extension vs "nullFlavor"s in value set · IG creation

Stream: IG creation

Topic: data-absent-reason extension vs "nullFlavor"s in value set


view this post on Zulip Sarah Gaunt (Nov 12 2021 at 00:23):

We (@David deRoode) are wondering what current best practice is around using the data-absent-reason extension vs specifying "nullFlavor" type values in a value set? Are there cases where one would be better than another or is one way always preferred?

view this post on Zulip Eric Haas (Nov 12 2021 at 00:39):

Sarah Gaunt said:

We (David deRoode) are wondering what current best practice is around using the data-absent-reason extension vs specifying "nullFlavor" type values in a value set? Are there cases where one would be better than another or is one way always preferred?

Bottom line is to use codes instead of DAR for coded elements:

see http://hl7.org/fhir/us/core/general-guidance.html#missing-data

We got our hands slapped for originally trying to DAR coded elements, because you can't use an extension for extensible and required bindings for min=1 coded elements - the bindings prohibit a DAR work-around. Besides it makes more sense to express the DAR concept as a code in element itself. You are screwed if you have a min=1
and required binding and no code available. ( Had trackers to make this guidance more explicit and obvious, but I don't think it has been applied yet - need to follow up on that)

view this post on Zulip Sarah Gaunt (Nov 12 2021 at 02:41):

Eric Haas said:

We got our hands slapped for originally trying to DAR coded elements, because you can't use an extension for extensible and required bindings for min=1 coded elements - the bindings prohibit a DAR work-around. Besides it makes more sense to express the DAR concept as a code in element itself. You are screwed if you have a min=1
and required binding and no code available. ( Had trackers to make this guidance more explicit and obvious, but I don't think it has been applied yet - need to follow up on that)

Yikes - thinking I might have done that in a few places... Better go and check.

view this post on Zulip Frank Oemig (Nov 13 2021 at 09:02):

What about creating value sets that include the null-flavour codesystem?

view this post on Zulip Lloyd McKenzie (Nov 14 2021 at 20:06):

@Frank Oemig That's what's recommended. Though we generally encourage people to draw on the DAR code system rather than the null flavor code system.


Last updated: Apr 12 2022 at 19:14 UTC