Resolution: Not Persuasive
FHIR Core (FHIR)
Clinical Decision Support
DeviceRequest (was DeviceUseRequest)
NutritionOrder (was NutritionRequest)
RequestOrchestration (was RequestGroup)
All of the 'instantiates' elements should go away as we're now using extensions (four different ones) to meet the use-case previously met by these elements.
Bryn believes there are other places where this same pattern exists. He will submit a new tracker item for consideration in R6 that enumerates the places where a choice of repeating canonical|uri either exists or should exist where a new data type would be useful.All of the 'instantiates' elements should go away as we're now using extensions (four different ones) to meet the use-case previously met by these elements. Bryn believes there are other places where this same pattern exists. He will submit a new tracker item for consideration in R6 that enumerates the places where a choice of repeating canonical|uri either exists or should exist where a new data type would be useful.
Bryn Rhodes/Rick Geimer: 16-0-3
There are a number of resources which contain the following two elements:
This occurs 16 times (across 14 resources). Since this seems to be a well-established pattern, and since these elements are similar in purpose, it might be beneficial to formalize this construct.
Unfortunately, instantiates[x] cannot be used because the cardinality is 0..* on many of these (and perhaps some instances should use both kinds).
This is why we should consider creating a new datatype to represent this pairing, similar to CodeableReference.
One question that arises is if it ever make sense to have both a canonical and uri representation paired (indicating different representations of the same thing)?
- If yes (pairing a canonical and uri could make sense), then this is much like CodeableReference - so there is precedent to introduce a type like this.
- If no (it should always be a canonical or a uri), then the datatype would either have two elements w/ an invariant OR would have a single target[x] element (canonical | uri). Introducing such a thing likely requires more thought.
If this has been discussed in the past, I apologize for bringing it up again! But if not, it is probably worthy of some discussion.
Here is a listing of where we find this pattern in the FHIR R5 Ballot spec:
- is voted on by
BALLOT-41988 Affirmative - Chris Moesel : 2022-Sep-FHIR R5 STU
- mentioned in