Stream: IG creation
Topic: MR error for MeasureScore despite extension
Eric Haas (Jul 08 2020 at 01:32):
For DEQM we defined an DEQMAlternateScoreTypeExtension for MeasureReport.group.measureScore
For scoring other that Quantity so this example:
group:
- id: location-datetime
population:
- code:
coding:
- system: 'http://terminology.hl7.org/CodeSystem/measure-population'
code: measure-observation
display: Measure Observation
id: measure-observation
count: 1
measureScore:
extension:
- url: >-
http://hl7.org/fhir/us/davinci-deqm/StructureDefinition/extension-alternateScoreType
valueDateTime: '2019-10-07T14:29:28-07:00'
yields this error:
MeasureReport/date-of-last-power-outage: MeasureReport.group[0].measureScore (l61/c23) error A value is required when the Measure.scoring={0}
Which I believe needs to allow for the extension
Grahame Grieve (Jul 08 2020 at 01:33):
what kind of extensions would be allowed in place of the value? (@Bryn Rhodes )
Bryn Rhodes (Jul 08 2020 at 01:44):
Seems like it ought to align with possible value choices for Observation: http://hl7.org/fhir/observation-definitions.html#Observation.value_x_. J#22824 (to be applied) would allow dateTime, CodeableConcept, Period, Range, and Duration.
Bryn Rhodes (Jul 08 2020 at 01:44):
That extension is intended to provide support in R4 for what will be possible in R5 with that tracker applied.
Grahame Grieve (Jul 08 2020 at 01:45):
so any extension that has one of those value types? that doesn't sound right for me
Bryn Rhodes (Jul 08 2020 at 01:46):
How else could we limit it?
Grahame Grieve (Jul 08 2020 at 01:48):
I'm thinking from the pov of the validator. Eric is saying that this particular extension - which is specific to a particular IG - is an acceptable substitute for an actual value. I'm not in the habit of writing exceptions for particular IGs into the validator, so I'm wondering what extensions would be ok here..
Grahame Grieve (Jul 08 2020 at 01:49):
not about the acceptible value types for this particular extension
Bryn Rhodes (Jul 08 2020 at 01:51):
Are there other cases like this the validator already handles? Like for R5-uplift extensions?
Grahame Grieve (Jul 08 2020 at 01:54):
well, the validator knows about some R5 extensions. But there's nowhere where the validator decides that a requirement for a value is set aside by the presence of an extension
Bryn Rhodes (Jul 08 2020 at 01:57):
Isn't that what a data-absent-reason extension is doing?
Grahame Grieve (Jul 08 2020 at 02:02):
I don't see that extension here
Bryn Rhodes (Jul 08 2020 at 02:09):
Well I think data-absent-reason would be an acceptable extension, what if the MeasureScore is in fact null?
Bryn Rhodes (Jul 08 2020 at 02:12):
And is my understanding correct that an extension, in general, satisfies a cardinality requirement of 1..1 for an element? Or does the validator require an actual value child for 1..1?
Grahame Grieve (Jul 08 2020 at 02:36):
well, this particular error doesn't come from the direct cardinality requirements, but from the business rules you and I talked about.
Bryn Rhodes (Jul 08 2020 at 02:57):
Is there a way to characterize the types of extensions that would be allowed? I would think anything that still conceptually provided a value would be a reasonable extension.
Grahame Grieve (Jul 08 2020 at 02:58):
I don't think we have anyway to know that
Eric Haas (Jul 09 2020 at 22:51):
what about the DAR extension, I would think that would be kosher as well
Grahame Grieve (Jul 15 2020 at 02:41):
so we haven't resolved this.
Grahame Grieve (Jul 15 2020 at 03:06):
I've added some DEQM specific logic for this, but I don't think that this is at all appropriate. The extension should be moved to somewhere more general
Bryn Rhodes (Jul 16 2020 at 14:53):
I've added a tracker to define this extension in the base specification: https://jira.hl7.org/browse/FHIR-28077
Last updated: Apr 12 2022 at 19:14 UTC