How to process search with multiple chained/reverse chained parameters


I’m trying to understand this part of the spec Search - FHIR v4.0.1

From this:

Note that chained parameters are applied independently to the target resource. For example,

GET Patient?

may return Patients cared for by Joe from CA and Jane from MN: no one practitioner need satisfy both conditions. E.g. the chains are evaluated separately.

My interpretation is the search parameters end up in an OR condition: name = Joe OR state = MN

Now on the Reverse Chaining (Search - FHIR v4.0.1) when there are multiple _has parameters, my interpretation is that also is considered an OR, based on:

“Or” searches are allowed (e.g. GET [base]/Patient?_has:Observation:patient:code=123,456), and multiple _has parameters are allowed (e.g. GET [base]/Patient?_has:Observation:patient:code=123&_has:Observation:patient:code=456).

Note that each _has parameter is processed independently of other _has parameters.

Consider this search URL:


Should that be interpreted with an OR like: Condition.code=C43.0… OR Observation.code=2160-0… OR Observation.code=1743-4… ?

My interpretation from the specification is that the OR is the right way of processing these kinds of parameters, but a colleague did a test on the HAPI FHIR test server online and got that the logic was really an AND, because with the _summary=count he got less results when adding another condition to the search (first tested with one condition then added another one).

I can’t provide the example right now because the HAPI FHIR server is offline at the moment. I’ll try later.

What that will do is check that the Patient has a Condition with one of the specified conditions, an Observation with one of the first set of observation codes and an Observation with one of the second set of codes. In your case, that’s probably what you want.

The warning is to say that something like this:
is not going to do what you want - i.e. the search for an observation with a matching code and the search for an observation with a matching date will be independent. They are however an ‘and’, in that you’ll only get a Patient if you can find at least one Observation that meets the first criteria and at least one Observation that meets the second. There’s no expectation that the two Observations would have to be the same.

Thanks @lloyd that clarifies the spec. Thought, do you think the spec is clear enough about this in both chained and reverse chaining?

If you think further clarification would be helpful, feel free to suggest appropriate wording using the “propose a change” link on the page.

1 Like

Good suggestion, thanks.