Stream: patient administration WG
Topic: RelatedPerson relationship codes
Thomas Tveit Rosenlund (Jan 08 2021 at 14:32):
I have a hard time to understand the order of the relationships given in the RelatedPerson.relationship element, and if the HL7 FHIR itself states any specific order for the codes given in RelatedPerson.relationship element. The documentation of the RelatedPerson resource don't seem to specify any specific order of the relationship codes in the relationship element.
The definition states: "The nature of the relationship between a patient and the related person."
However, the codes in the preferred valueset seems to indicate a given order.
"child | The player of the role is a child of the scoping entity."
You can interpret this either way, but if you say that the "player of the role" is represented by the RelatedPerson instance and the "scoping entity" is represented by the patient reference this would be the "natural" order of the relationship.
Our problem in Norway is that the relationships are ordered the other way around when we read them from the source (Master Person Index of Norway). Would it be OK to state that these relationships is ordered the other way around. In other words that the code in the RelatedPerson.relation is actually the role played by the Patient instance?
We would not use the ValueSet preferred in the standard (https://www.hl7.org/fhir/valueset-relatedperson-relationshiptype.html) as the codes from the MPI is Norwegian specific.
René Spronk (Jan 08 2021 at 15:16):
IMHO it's from the viewpoint of the Related person, so 'child' means 'child of' (RelatedPerson is a child of the Patient). If the Folkeregister has it the other way around, you'll have to translate the child concept code over there into a FHIR parent concept code.
Richard Townley-O'Neill (Jan 10 2021 at 22:58):
In R4 RelatedPerson.relationship has the definition "The nature of the relationship between a patient and the related person." and the example Relatedperson-example-newborn-mom I read as being the mother of a new-born child.
So I think it is from the viewpoint of the patient.
Thomas Tveit Rosenlund (Jan 11 2021 at 06:45):
Thank you very much for the feedback @René Spronk @Richard Townley-O'Neill
Both have understood the relationship as beeing stated from the RelatedPerson to the Patient, or as Rene put's it "child of" "mother of" and so forth. The official example is also stated that way to, the RelatedPerson is the "natural mother" of the patient reference represented in the patient reference.
Is'nt the viewpoint then from the RelatedPerson to the patient @Richard Townley-O'Neill ?
I still have a feeling that the direction of the relationship is a property of the preferred ValueSet, and not really a property of of the RelatedPerson resource itself, or the relation element defintion. For that reason, should it not be possible to use other ValueSets that actually states the relationships in the opposite direction? Or have I missed something in the RelatedPerson definition?
Richard Townley-O'Neill (Jan 11 2021 at 07:11):
Thomas Tveit Rosenlund said:
Is'nt the viewpoint then from the RelatedPerson to the patient Richard Townley-O'Neill ?
My thinking is that the child could say "this related person is my mother", so I think of that statement as being from the point of view of the patient.
Richard Townley-O'Neill (Jan 11 2021 at 07:20):
But I agree that this is not clear. I used the example to decide which party was the first and which the second in the relationship.
Richard Townley-O'Neill (Jan 11 2021 at 07:23):
I still have a feeling that the direction of the relationship is a property of the preferred ValueSet, and not really a property of of the RelatedPerson resource itself, or the relation element defintion. For that reason, should it not be possible to use other ValueSets that actually states the relationships in the opposite direction? Or have I missed something in the RelatedPerson definition?
A useful value set needs to have both "mother of" and "child of". I do not see how you can tell which party comes before "mother of" and which after just from the values in the value set.
René Spronk (Jan 11 2021 at 08:49):
https://terminology.hl7.org/1.0.0/CodeSystem-v2-0131.html#contactrole2 "concepts which specify a relationship role that the next of kin/associated parties plays with regard to the patient" - at for that value set it's pretty non-ambiguous. The wording in https://terminology.hl7.org/1.0.0/CodeSystem-v3-RoleCode.html#v3-RoleCode-_PersonalRelationshipRoleType indicates the same thing, but one would have to understand HL7v3 jargon, talking about 'playing a role' and relationship with a 'scoper'.
Documentation on RelatedPerson.relationship doesn't specify the 'viewpoint' (this could be fixed by creating a Jira issue). This is a 'preferred' binding, so one could use ones own coding system / value set. Codes "father (related person is the father of the patient)" and "the patient is the son of the relatedperson" would both be valid as an expression of the relationship.
Thomas Tveit Rosenlund (Jan 11 2021 at 10:32):
Richard Townley-O'Neill said:
Thomas Tveit Rosenlund said:
Is'nt the viewpoint then from the RelatedPerson to the patient Richard Townley-O'Neill ?
My thinking is that the child could say "this related person is my mother", so I think of that statement as being from the point of view of the patient.
Thanks. Now I get it.
Thomas Tveit Rosenlund (Jan 11 2021 at 10:49):
René Spronk said:
https://terminology.hl7.org/1.0.0/CodeSystem-v2-0131.html#contactrole2 "concepts which specify a relationship role that the next of kin/associated parties plays with regard to the patient" - at for that value set it's pretty non-ambiguous. The wording in https://terminology.hl7.org/1.0.0/CodeSystem-v3-RoleCode.html#v3-RoleCode-_PersonalRelationshipRoleType indicates the same thing, but one would have to understand HL7v3 jargon, talking about 'playing a role' and relationship with a 'scoper'.
Documentation on RelatedPerson.relationship doesn't specify the 'viewpoint' (this could be fixed by creating a Jira issue). This is a 'preferred' binding, so one could use ones own coding system / value set. Codes "father (related person is the father of the patient)" and "the patient is the son of the relatedperson" would both be valid as an expression of the relationship.
Thanks again @René Spronk
That was my point, given the current definition of RelatedPerson, it is actually the ValueSet used for relationship codes that defines the order or direction of the given relationship code. It was actually a review of the preferred ValueSet (https://terminology.hl7.org/1.0.0/CodeSystem-v3-RoleCode.html#v3-RoleCode-_PersonalRelationshipRoleType) that made me reach out to the community in this matter.
In my case the current state of the definitions actually gives me just the flexibility I need to define another ValueSet containing the norwegian codes that express the relationship in the opposite direction of what is stated in the preferred ValueSet provided by HL7. The wording in the preferred ValueSet is quite clear to me but as you write you need to interpret "the subject" and "the scoper" correctly according to the FHIR model. I to believe some additional information concerning these matters would be welcome to most readers of the FHIR definition. I will create a Jira issue to propose some information is added to the documentation to better guide the reader in these matters.
Thomas Tveit Rosenlund (Jan 11 2021 at 10:57):
A quick search in Jira and I found that @Josh Mandel already asked for a clarification of the direction of the relationship.
FHIR-26994
Last updated: Apr 12 2022 at 19:14 UTC
