Stream: implementers
Topic: Patient.link types "refer/seealso"
Simone Heckmann (Sep 21 2016 at 20:40):
In PA we're durrently discussing the Patient.link.type ValueSet:
http://hl7-fhir.github.io/valueset-link-type.html
We're wondering what the usecases for "seealso" and "refer" are and what the disambiguation is.
Also: how do these link types relate to the concept of "Person" as the linking element for multiple active Patient records?
If anyone uses "seealso"/"refer" link types and has feelings against them being removed from the valueSet, please speak up!
Igor Sirkovich (Sep 22 2016 at 02:07):
Hi Simone, we are using "refer" in Ontario: our MedicatioDispense resource has a refercne to a "point-in-time" Patient resource with information that was captured by a pharmacy at the time of dispense. This Patient resource in turn has a refernce to our "master" Patient resource.
Simone Heckmann (Sep 22 2016 at 13:38):
@Igor Sirkovich: So are you linking two Patient resources across different systems, stating that they are both the same person?
Or is this more like having a duplicate Resource in one system with one being "correct/active" and the other being an inactive, possibly incorrect/outdated relic?
Simone Heckmann (Sep 22 2016 at 13:47):
You are using "refer" type link bidirectionally. Interestingly the "refer" doesn't seem to satify your usecase very well, since it states
The patient resource containing this link is in use and valid but *not* considered the main source of information about a patient.
However they can't be both "*not* considered the main source". In our discussion, we did recognize the requirement for bidirectional references, so we propose to rename "replace" to "replaced-by" and introduce an additional type "replacing" for the backreference from the active to the inactive Patient.
Simone Heckmann (Sep 22 2016 at 13:50):
We are wondering, if there is a use case for having linked Patient resources within the same system and have them both "valid and active".
Simone Heckmann (Sep 22 2016 at 13:52):
It sounds to me like your point-in-time copy of the patient should be inactive to prevent this Resource from accidentially being used instead of the master record, no?
Igor Sirkovich (Sep 22 2016 at 14:04):
@Simone Heckmann, our use is not bi-directional: it goes in one direction from the "point-in-time" inactive Patient resource to the current/active resource. We have clear explanation in our specification that the former (inactive) resource should only be used for investigation (call a pharmacy) if MedicationDispense is "suspicious", i.e. this might be an error/wrong identity on a dispense or a fraud (someone pretended to be a different person to get a drug). It's not about police investigation here, but rather about a clinical need to remove the wrong dispense event from the patient's list of medications.
Last updated: Apr 12 2022 at 19:14 UTC
