FHIR Chat · Null Flavour vs Data Absent reason · fhir/infrastructure-wg

Stream: fhir/infrastructure-wg

Topic: Null Flavour vs Data Absent reason

Brian Postlethwaite (Sep 22 2020 at 22:25):

Looking for advise on data absent reason vs null flavour for a value omitted due to security masking...

 <subject>
    <extension url="http://hl7.org/fhir/StructureDefinition/iso21090-nullFlavor">
      <valueCode value="MSK"/>
    </extension>
  </subject>

or

 <subject>
    <extension url="http://hl7.org/fhir/StructureDefinition/data-absent-reason">
      <valueCode value="masked"/>
    </extension>
  </subject>

I was showing off some work to @Drew Torres and he pointed me at the second, which also makes sense...

Brian Postlethwaite (Sep 22 2020 at 22:32):

@John Moehrke do you have an opinion on which way this would go from the security group?

Grahame Grieve (Sep 22 2020 at 22:41):

we generally prefer the data absent extension over the older more generic nullFlavor extension that includes some really difficult concepts for FHIR

Lloyd McKenzie (Sep 22 2020 at 23:00):

The v3 one is primarily there for systems performing conversion from v3/CDA

Lloyd McKenzie (Sep 22 2020 at 23:00):

If we had been smart, we would have had the value set draw from the equivalent DAR codes where there were equivalents - or vice versa.

Lloyd McKenzie (Sep 22 2020 at 23:00):

But we weren't, so there are codes that mean the same thing :)

Lloyd McKenzie (Sep 22 2020 at 23:01):

That said, the extensions aren't normative yet, so theoretically we could adjust them - if someone thinks it's worth the pain?

Grahame Grieve (Sep 22 2020 at 23:02):

not that one. It would be a mapping.

Brian Postlethwaite (Sep 22 2020 at 23:32):

Thank you both for the speedy feedback, at least now I have an action to take...
It didn't get to prod yet - but only just.

John Moehrke (Sep 28 2020 at 15:19):

I don't have an opinion on the 'how'.. but from a privacy perspective it is often discouraged at including any indication that information was masked, as that indication itself indicates that something 'was there' and that what was there was 'sensitive'.. thus to indicate that something was masked is to violate privacy. So my preference would be to do neither of the alternatives and just not include the element at all. Recognizing that this sometimes can't be done due to cardinality rules, then I guess the alternatives would be needed and wouldn't expose the privacy problem I outline.

Last updated: Apr 12 2022 at 19:14 UTC