FHIR Chat · missing required field

Can I confirm my recollection that the extension "data absent reason" - http://hl7.org/fhir/extension-data-absent-reason.html - can be used in thse situations where there is a required field that cannot be populated. For example, the patient tells you that they did react to a medication, but the can't recall what it was...

Grahame Grieve (May 23 2016 at 12:50):

yes you can provide that instead of a value

Chris Grenz (May 23 2016 at 14:34):

Do the committees fully understand the implications of this? It's a very nuanced use of "required"...most will assume that a required element will always have an @value (or relevant children).

Eric Haas (May 23 2016 at 15:20):

DAR is used within the element so it is a relevent child.

Chris Grenz (May 23 2016 at 15:42):

I understand that it's both legal and relevant. My question is: is its use obvious/expected or surprising? Personally, I was surprised the first time I saw DAR or NullFlavor as a satisfaction of "required". The documentation for elementdefinition.min is silent about this.

Eric Haas (May 23 2016 at 15:48):

I call this a "naked" extension. And it took me by surprise too.

Chris Grenz (May 23 2016 at 15:56):

I think that most committees are taking "required" very seriously and won't set something to required unless the resource has no meaning without it. So adding the "NI" null flavor really defeats the purpose. As an implementer, it's completely unacceptable for a truly required field to have a NULL value, even if I know what flavor of NULL it is! If NULL is acceptable, then it's not required!

Chris Grenz (May 23 2016 at 15:59):

There's also some security implications here - none that can't be accounted for, but if you're depending on the validator to ensure that a field is present.... Again, legal by the letter of the law, but surprising.