FHIR Chat · Specimen.subject · Orders and Observation WG

Stream: Orders and Observation WG

Topic: Specimen.subject

Bob Milius (Jun 26 2018 at 18:21):

I have a question about use of Subject reference in Specimen.

Let's say our organization collects a buccal swab from a potential stem cell donor and we send it out to a contract lab for tissue typing. We have the potential donor name and demographic info in our system without a FHIR interface. We want to put the specimen info into a Specimen resource so that the reference can be used in a DiagnosticReport/Observation bundle that the lab sends us. The Specimen resource is de-identified and only has a specimen.identifier which we'll later cross-reference to identify the patient in our system. Can we do this? Subject is required for Specimen. What do I put in for Subject?

e.g.,

<Specimen>
    <identifier>
        <system value="http://myorg.com"/>
        <value value="123"/>
    </identifier>
    <collection>
        <collectedDateTime value="2016-11-10"/>
        <method>
            <coding>
                <system value="http://hl7.org/fhir/v2/0488"/>
                <code value="SWA"/>
                <display value="Swab"/>
            </coding>
        </method>
        <bodySite>
            <coding>
                <system value="http://snomed.info/sct"/>
                <code value="261063000"/>
                <display value="Buccal space"/>
            </coding>
        </bodySite>
    </collection>
    <subject>
        ???
    </subject>
</Specimen>

Eric Haas (Jun 26 2018 at 19:16):

We talked about this a dev-days. what about the specimen id - this is an opaque identifier still right?

...
    <subject>
        <identifier>[specimen id]</identifier>
    </subject>
...

Lloyd McKenzie (Jun 26 2018 at 19:50):

That's possible, but it would be better to make specimen.subject minOccurs 0. It's not needed for all uses of specimen, thus it shouldn't be required.

Eric Haas (Jun 26 2018 at 19:58):

I think a specimen needs a source. It is by its very nature a sampling of something so you need that something. This is de-id issue.

Eric Haas (Jun 26 2018 at 19:58):

what is wrong with an opaque ID

Eric Haas (Jun 26 2018 at 19:59):

especially for trace back to the actual patient.

Lloyd McKenzie (Jun 26 2018 at 19:59):

Every specimen has a source - just as every patient has a birth date. That doesn't mean that the source will be known or is relevant to all uses of the specimen.

Lloyd McKenzie (Jun 26 2018 at 19:59):

We don't force the inclusion of information that system may not have. Forcing someone to copy an element already elsewhere to satisfy a mandatory is poor design on our part.

Bob Milius (Jun 27 2018 at 14:36):

Wouldn't using specimen.id would be a chicken/egg thing? I don't know the id until I create the specimen resource. I could use the specimen.identifier for the specimen.subject.identifier. Or, I've thought about using 'de-identified' as the specimen.subject.display for all the specimens.

Eric Haas (Jun 27 2018 at 16:18):

" could use the specimen.identifier for the specimen.subject.identifier" that is what I meant.

Lloyd McKenzie (Jun 27 2018 at 18:20):

It's wrong to make specimen mandatory if there are valid use-cases that don't know/need the specimen. We only make things mandatory when we're confident that the resource is useless for all reasonable usecases without the element present. That's not the case here. We don't set minOccurs to establish good practice. Good practice is for profiles, not resource designs.

Lloyd McKenzie (Jun 27 2018 at 18:23):

GF#17424

Bob Milius (Jun 27 2018 at 19:18):

@Lloyd McKenzie I think you meant specimen.subject

Lloyd McKenzie (Jun 27 2018 at 19:40):

Yes

Lloyd McKenzie (Jun 27 2018 at 19:40):

(Thanks for correcting me :))

Andrea Pitkus, PhD, MLS(ASCP)CM, CSM (Jun 27 2018 at 23:14):

Bob, any laboratory (such as yours) sending a specimen to another laboratory (reference laboratory) is required to submit a lab order for testing with the patient information (name, identifiers, etc.) and all CLIA required data elements for the order. When the report comes back from the performing laboratory, there are additional CLIA requirements with the report information as indicated below including multiple identifiers as indicated below from CLIA law.

(c) The test report must indicate the following:
(1) For positive patient identification, either the patient's name and identification number, or a unique patient identifier and identification number.
(2) The name and address of the laboratory location where the test was performed.
(3) The test report date.
(4) The test performed.
(5) Specimen source, when appropriate.
(6) The test result and, if applicable, the units of measurement or interpretation, or both.
(7) Any information regarding the condition and disposition of specimens that do not meet the laboratory's criteria for acceptability.
(d) Pertinent “reference intervals” or “normal” values, as determined by the laboratory performing the tests

Lloyd McKenzie (Jun 28 2018 at 06:27):

The lab needs that on the order. It doesn't mean it needs it in the system that tracks specimen progression through the lab. And not all specimens will be patient-associated.

Bob Milius (Jun 28 2018 at 14:49):

@Andrea Pitkus, PhD, MLS(ASCP)CM, CSM I wouldn't call us a laboratory (we don't do any testing directly, but have contracts with external labs that do our testing). But I believe when we send specimens to contract labs we include the specimen identifier which we use internally as a 'donor id'. I believe this satisfies the second option in (1), i.e., "or a unique patient identifier and identification number"

Bob Milius (May 15 2019 at 14:52):

@Eric Haas Is is possible to search for subject.identifier using this? I can search for specimen.identifier, but wonder if I can do the same for subject using this construct?

Eric Haas (May 15 2019 at 15:42):

you can chain the search like this.

GET [base]/Specimen?subject:Patient.identifier=http://example.org/ids|1234 if the subject is a patient

or just plain old GET [base]/Specimen?subject:identifier=http://example.org/ids|1234 if want to look across all types.

Bob Milius (May 15 2019 at 16:22):

Thanks, Eric, but it's not working for me. Maybe an issue with HAPI? or STU3?
Here's my tiny example Specimen resource:

<Specimen>
    <identifier>
        <value value="001^123456789"/>
    </identifier>
    <subject>
        <identifier>
            <value value="001^123456789"/>
        </identifier>
    </subject>
</Specimen>

This works:
GET http://hapi.fhir.org/baseDstu3/Specimen?identifier=001^123456789

This doesn't:
GET http://hapi.fhir.org/baseDstu3/Specimen?subject:identifier=001^123456789

Eric Haas (May 15 2019 at 18:02):

The server may not support chaining for Specimen

Last updated: Apr 12 2022 at 19:14 UTC