Stream: implementers
Topic: Combining chaining & reverse chaining
Lloyd McKenzie (Apr 23 2018 at 18:29):
Is Observation?patient._has:ResearchSubject:identifier=123 legal? What I want to find is all observations made for a particular study subject. Studysubject points to Patient and so does Observation. But I haven't seen an example that confirms this is legal...
Alexander Zautke (Apr 23 2018 at 19:18):
I guess this is the same problem as in https://chat.fhir.org/#narrow/stream/4-implementers/topic/search.3A.20chaining.20_has.20and.20normal.20chained.20parameters
The summary of that discussion was that there is currently no way to combine chaining & reverse chaining if the chaining is at the beginning of the query.
Or at least this is how it is currently being implemented.
Lloyd McKenzie (Apr 24 2018 at 02:31):
If this isn't possible, then we'd have to get implementations to put a "researchSubject" extension on all of the Observations, Procedures and other records that are related to a given ResearchSubject. And if they don't, we wouldn't find them on queries. That's not super-desirable. Alternatively, we'd have to update the Patient to have an extension to point to all associated research subjects, creating a bi-directional association. That's not so great either.
Lloyd McKenzie (Apr 24 2018 at 02:33):
@Grahame Grieve do you think combining chaining and reverse chaining should be possible? Is this something where GraphQL is necessary?
Christiaan Knaap (Apr 24 2018 at 07:27):
The real problem here is the format of the search. It lacks the link from ResearchSubject to Patient. This is the correct format:
<base>/Observation?patient._has:ResearchSubject:patient:identifier=123
Alexander Zautke (Apr 24 2018 at 09:19):
Sorry for the confusion. I just thought, because no one objected the last time, that it would not be possible.
Maybe such an example should be added to the reverse chaining section?
This would also help to align behavior across implementations, for example, Vonk currently returns a result in such a case, AEGIS WildFHIR returns no results and HAPI gives an error stating that the parameter chain is invalid.
Matthew Koch (May 24 2018 at 18:28):
Is this thread relevant? https://chat.fhir.org/#narrow/stream/4-implementers/subject/Being.20RESTFul.20in.20clinical.20research
Last updated: Apr 12 2022 at 19:14 UTC
