Stream: implementers
Topic: No Observation.valueAttachment?
John Silva (Jul 25 2019 at 16:56):
We were using Observation.valueAttachment in STU3 and I just noticed that Attachment type was removed. What or how should we represent a document attachment (PDF via a URL and content type, etc.)
Lloyd McKenzie (Jul 25 2019 at 17:33):
Media
John Silva (Jul 25 2019 at 17:39):
Media over DocumentReference? I took a quick look at both; what goes into the decision making process to use one over the other? (and there's a 'big warning' about Media being 'trial use' --- do we get ourselves into trouble by using this immature resource?)
Eric Haas (Jul 25 2019 at 17:47):
For R5 we are getting rid of Media and merging it into DocRef and I am pushing for a valueAttachment extension for those cases where the value is actually an attachment and not a result derived from the attachment. so use either a DofRef or a valueAttachment extension.
John Silva (Jul 25 2019 at 18:15):
@Eric Haas -- I guess a valueAttachment extension would be the easiest for us since we already have mappers and code that expects it. I guess you also gave us a very good reason to stay away from the Media resource; we'd only be 'shooting ourselves in the foot' if we spent the effort to migrate to it only to find out a year (or so) later that it's been 'deprecated'. This is the kind of info that somehow needs to be communicated better; if I didn't ask a question here I would have went on my way choosing Media only to be frustrated that it was removed a year later. (not all implementers have the time to watch and read through these message threads to get this 'inside information')
Kathy Pickering (Sep 10 2019 at 19:37):
Is there a specific valueAttachment extension you recommend?
Eric Haas (Sep 10 2019 at 21:24):
None in currenct spec. @Andrew Torres do you have an extension SD you can share? I will be proposing we add valueAttachment back to Observation 1st - if unsuccessful then proposal to create one for R5 ( mea culpa for ever voting to remove it in the first place :-( )
Eric Haas (Sep 10 2019 at 21:28):
To Bolster my case at the HL7 Working Group Meeting where these discussion are made I'd like a straw poll:
Eric Haas (Sep 10 2019 at 21:28):
/poll Add valueAttachment back into Observation for R5?
Yes
No
No - create a standard valueAttachment extension for R5
...
Jenni Syed (Sep 10 2019 at 21:35):
My opinion has nothing to do with Media and Doc Ref relationships
Jenni Syed (Sep 10 2019 at 21:35):
But we also didn't want to see attachment go away
Eric Haas (Sep 10 2019 at 21:35):
that would be a yes
Jenni Syed (Sep 10 2019 at 21:36):
Well, I don't care about the "Now that the Media..." :)
Eric Haas (Sep 10 2019 at 21:36):
I'm trying to edit the question
Eric Haas (Sep 10 2019 at 21:37):
oddly won't go away
Michele Mottini (Sep 10 2019 at 21:37):
I'd go for the extension - so that we can use it also in R4
Eric Haas (Sep 10 2019 at 21:39):
or both ? what if we backport the R4 extension and the R5 element?
Eric Haas (Sep 10 2019 at 21:39):
I don't know how that would be committed to spec though
Eric Haas (Sep 10 2019 at 21:52):
pre WGM valueAttachement Poll: see above
Jenni Syed (Sep 10 2019 at 21:54):
There are a few observation extension (or generic extensions) in R4 that allow Attachments
Jenni Syed (Sep 10 2019 at 21:56):
EG: http://hl7.org/fhir/r4/extension-workflow-supportinginfo.html (ew, contained)
Jenni Syed (Sep 10 2019 at 21:57):
http://hl7.org/fhir/r4/extension-workflow-relatedartifact.html (don't think this quite fits either)
Jenni Syed (Sep 10 2019 at 21:57):
Just want to make sure we're officially voting against those :)
Eric Haas (Sep 10 2019 at 22:06):
this is about the attachment being the *actual result value *of the observation. not from which the value was derived ( in which case .derivedFrom would used and would reference DocReference. Those extensions you mention represent other information to help with the interpretation or to contextualize the observation.
an example would be a jpg of a gel for electrophoresis. I am not sure how useful that would clinically be but that is an example.
Eric Haas (Sep 10 2019 at 22:10):
another user brought up a use case where an irregular series of measurements was documented in a csv file and that is the result value.
a picture of my cats ugly absess is another.
Last updated: Apr 12 2022 at 19:14 UTC
