FHIR Chat · docs / Issue #46 Proper json for imaging-study with regar... · fhircast-github

Stream: fhircast-github

Topic: docs / Issue #46 Proper json for imaging-study with regar...

view this post on Zulip Github Notifications (FHIRcast) (Nov 20 2018 at 20:28):

gkustas opened Issue #46

Which is correct? The FHIR spec shows an actual element named "accession", whereas the FHIRCast example simply has a single identifier. I suspect the FHIRCast example is wrong, that is, unintentional.

Example in FHIRCast.org:

"key": "study",
"resource": {
"resourceType": "ImagingStudy",
"id": "8i7tbu6fby5ftfbku6fniuf",
"uid": "urn:oid:2.16.124.113543.6003.1154777499.30246.19789.3503430045",
"identifier": [
{
"system": "7678",
"value": "185444"
}
],

FHIR 3.0 specification:

"resourceType" : "ImagingStudy",
// from Resource: id, meta, implicitRules, and language
// from DomainResource: text, contained, extension, and modifierExtension
"uid" : "<oid>", // R! Formal DICOM identifier for the study
"accession" : { Identifier }, // Related workflow identifier ("Accession Number")
"identifier" : [{ Identifier }], // Other identifiers for the study

view this post on Zulip Github Notifications (FHIRcast) (Nov 20 2018 at 21:10):

isaacvetter commented on Issue #46

Hey @gkustas -

It's very possible that I messed up some of the FHIR examples in the FHIRcast spec, the FHIR spec is definitely authoritative for FHIR resources. Thank you for taking the time to write this up!

I'm not certain that the example is strictly wrong. The ImagingStudy resource has both optional accession and identifier elements.

Do we think we should update the example to include an accession number? Should that replace the identifier or supplement it? (If so, interested in submitting a PR?)

Isaac

view this post on Zulip Elliot Silver (Nov 20 2018 at 21:15):

What version of FHIR are you targeting? R4 has only identifier, of which an accession is a specific type of identifier.

view this post on Zulip Isaac Vetter (Nov 20 2018 at 21:15):

Ah, thanks, Elliot!

view this post on Zulip Elliot Silver (Nov 20 2018 at 21:16):

For that matter study uid is also an identifier.

view this post on Zulip Isaac Vetter (Nov 20 2018 at 21:18):

hmmm, i think targeting r4 for the initial FHIRcast specification ballot makes the most sense. I'll see if we can hijack this ticket to make that so.

view this post on Zulip Github Notifications (FHIRcast) (Nov 20 2018 at 21:20):

isaacvetter commented on Issue #46

@gkustas - @ElliiotSilver on zulip points out that in FHIR release 2, the ImagingStudy resource collapses both accession and uid into identifier.

We should probably update the FHIR examples to be r4-compatible, since we'll most likely be balloting FHIRcast when r4 is the the most recent version of FHIR.

That said, we can still look at adding an additional identifier that's specifically intended to be an accession number, to this example.

Isaac

view this post on Zulip Github Notifications (FHIRcast) (Nov 20 2018 at 21:28):

gkustas commented on Issue #46

Thanks @isaacvetter !

So then, how do we identify the accession number from any other identifier? Would this be the "system" element, and if so, where would we find the proper namespace for accession number. I assume it is something like this one (for ss#):

http://hl7.org/fhir/sid/us-ssn

I have no problem using it in this manner, but I'd like to start out right. Same thing for medical record number for patients. Namespace?

Maybe I should be posting that question on Zulip, huh?

Thanks again!

view this post on Zulip Elliot Silver (Nov 20 2018 at 21:30):

1. MRN for the patient isn't an identifier of the study.
2. Accession is identified using the identifier type "ACSN" and appropriate HL7 code system.

view this post on Zulip Elliot Silver (Nov 20 2018 at 21:33):

Encoding of accession, Study UID, and Series and Study UID are explicitly discussed in the R4 ImagingStudy: http://build.fhir.org/imagingstudy.html#notes

view this post on Zulip Github Notifications (FHIRcast) (Nov 20 2018 at 22:10):

isaacvetter commented on Issue #46

Hey @gkustas - well, either you should check out zulip or Elliot should respond on github. :smile:

He answered your question on this ticket's auto-generated zulip post.

I'm going to go ahead and close this ticket, but open a new one documenting that we should update all FHIR examples to be compatible with FHIR r4.

Isaac

view this post on Zulip Github Notifications (FHIRcast) (Nov 20 2018 at 22:10):

isaacvetter closed Issue #46

Which is correct? The FHIR spec shows an actual element named "accession", whereas the FHIRCast example simply has a single identifier. I suspect the FHIRCast example is wrong, that is, unintentional.

Example in FHIRCast.org:

"key": "study",
"resource": {
"resourceType": "ImagingStudy",
"id": "8i7tbu6fby5ftfbku6fniuf",
"uid": "urn:oid:2.16.124.113543.6003.1154777499.30246.19789.3503430045",
"identifier": [
{
"system": "7678",
"value": "185444"
}
],

FHIR 3.0 specification:

"resourceType" : "ImagingStudy",
// from Resource: id, meta, implicitRules, and language
// from DomainResource: text, contained, extension, and modifierExtension
"uid" : "<oid>", // R! Formal DICOM identifier for the study
"accession" : { Identifier }, // Related workflow identifier ("Accession Number")
"identifier" : [{ Identifier }], // Other identifiers for the study

view this post on Zulip George Kustas (Nov 20 2018 at 22:29):

Hi @Elliot Silver !

I hear you are the Epic guy... Thanks so much for responding. I had no idea the FHIRCast github was routed here.

Thanks for that link - it answers my question (and also increases the size of my payload considerably) :-) No problem. I'm thinking we can ignore everything but identifier.coding.type.code = "ACSN" and identifier.value = the actual accession number. @Isaac Vetter - perhaps the example in FHIRCast should use the same notation as the FHIR spec in this link?

I realize that MRN is an identifier for the patient resource which we also use in imaging study context sharing. There are no similar notes in the FHIR Patient Resource documentation for medical record number. Any idea what the type and namespace should be for that? Is it necessary (do you think) that we enforce that in FHIRCast?

view this post on Zulip Isaac Vetter (Nov 20 2018 at 22:32):

Hey @George Kustas !

view this post on Zulip Isaac Vetter (Nov 20 2018 at 22:33):

I had no idea the FHIRCast github was routed here.

And, in the future, an even better place to converse is in the regular #FHIRcast stream. This fhircast-github stream is really just for the most hardcore people who want to see every suggested change to the spec.

view this post on Zulip Isaac Vetter (Nov 20 2018 at 22:33):

perhaps the example in FHIRCast should use the same notation as the FHIR spec in this link?

Yes, very much so.

view this post on Zulip Elliot Silver (Nov 20 2018 at 22:40):

@George Kustas, not Epic -- I'm Change Healthcare, and HL7 Imaging co-chair.

For accession: Identifier.type.coding.code="ACSN"; Identifier.type.coding.system="http://terminology.hl7.org/CodeSystem/v2-0203".

For MRN: Identifier.type.coding.code="MR" or ; Identifier.type.coding.system="http://terminology.hl7.org/CodeSystem/v2-0203" however, this isn't as well defined (from what I've seen; there may be other guidance in various IGs, such as US core). A single patient may have many MRNs--several local MRNs, an CPI, an EMPI, etc. I think the patient MRN used would be an agreement between systems.

I would suggest that if a system sends you a particular MRN, you should be able to go back to that system to retrieve the patient with that MRN, so in some sense, it doesn't matter which MRN you use.

view this post on Zulip George Kustas (Nov 20 2018 at 22:51):

Thanks @Elliot Silver

I guess I confused you with someone else. I'm new to this group (obviously). Appreciate all the help I can get.

Last updated: Apr 12 2022 at 19:14 UTC


An HL7 Website -