Stream: implementers
Topic: Retrieving Subset of Single Resource
Sean Moore (May 16 2016 at 19:59):
Is there a way within FHIR to ask a server to return only a subset of individual data elements for a given resource? For example, could a client compose a query to only retrieve the Gender of a Patient?
Sean Moore (May 16 2016 at 20:00):
(I'm asking theoretically. I realize this wouldn't be good practice in many, many cases).
Grahame Grieve (May 16 2016 at 20:00):
see _elements on the search page. This can also work for a read operation
Chris Grenz (May 16 2016 at 20:00):
Or, for a more formal system, support GF#10000
Sean Moore (May 16 2016 at 20:02):
@Chris Grenz I take it you and your team are using this for real?
Chris Grenz (May 16 2016 at 20:03):
Yep - in UAT now
Grahame Grieve (May 16 2016 at 20:03):
I'm doubtful this will make it into the spec per se though
Chris Grenz (May 16 2016 at 20:05):
There's some significant use cases for it...will have to do something.
Chris Grenz (May 16 2016 at 20:06):
We're billing FHIR as not just an interop standard but as something that can be directly leveraged by apps at web scale. This is going to require some capability to tune the API in a predictable way....
Sean Moore (May 16 2016 at 20:07):
Can you talk a little bit more about how you're using this currently? Is doing something such as getting only the patient's gender and no other demographics your use case?
Grahame Grieve (May 16 2016 at 20:07):
Chris, got to write up the white paper ;-)
Chris Grenz (May 16 2016 at 20:09):
We're doing custom clinical registries - basically allowing practices to put a mask/perspective over the standard FHIR model using standard FHIR profiles. So, a practice may trim each resource to a particular set of elements and/or define filters and slices specific to their needs.
Chris Grenz (May 16 2016 at 20:09):
Biggest use case is including/excluding extensions. However, for IRB governed research, excluding "core" elements is also pretty common.
Chris Grenz (May 16 2016 at 20:11):
Also, we've got an app that allows users to make targeted edits to the record. This app reads/writes using limited profiles (we only allow a user to write using a profile to which they're authorized).
Chris Grenz (May 16 2016 at 20:11):
@Grahame Grieve - working on it...
Sean Moore (May 16 2016 at 20:12):
That's interesting. The use I've heard about from others is to limit the repeated metadata that comes back on a query, say, for observation - since you know (or at least, you think) the LOINC code and other data is the same, you can just retrieve the metadata
Chris Grenz (May 16 2016 at 20:12):
That'd work just fine too
Chris Grenz (May 16 2016 at 20:13):
(assuming you've dealt with the security implications elsewhere)
Sean Moore (May 16 2016 at 20:14):
Right.
Sean Moore (May 16 2016 at 20:15):
The feedback that prompted my question was in regards to exposing the common clinical data set through FHIR - we want to expose all the demographic related information via the PAtient resource, but we've gotten comments that we should implement each element - Name, gender, etc. - as it's *own* API.
Sean Moore (May 16 2016 at 20:15):
That feels ridiculous in my mind.
Grahame Grieve (May 16 2016 at 20:16):
I agree. why did people ask for that?
Sean Moore (May 16 2016 at 20:16):
The snarky answer is that they don't understand APIs.
Sean Moore (May 16 2016 at 20:16):
I'm not sure I have a non-snarky answer.
Grahame Grieve (May 16 2016 at 20:16):
lol
Chris Grenz (May 16 2016 at 20:16):
;)
Josh Mandel (May 16 2016 at 23:31):
I think the non snarky answer is to point out that ONC has specifically clarified this point, and for EHR Certification even they don't suggest that you need to demonstrate separate APIs for name, birthdate, etc. And to point out that Argonaut is defining APIs that align with MU requirements as well as FHIR resourceful boundaries in a pretty natural way @Sean Moore
Ewout Kramer (May 19 2016 at 09:26):
I wish we would have such clear regulations in my home country...
Grahame Grieve (May 19 2016 at 21:15):
right. then you could talk about people are all different because violate regulations, instead of just how they're all different
Last updated: Apr 12 2022 at 19:14 UTC
