I have a question about the general rules for the semantics of repeated elements in the FHIR model. Does FHIR have an opinion on whether it is “good practice” or not to put a meaning into the order of the values in a repeated element. Specifically, lets say we have an element
element1 0..* string
and two resources each with the values [“value1”, “value2”] and [“value2”, “value1”] in that element. Can these two resources be considered equal semantically?
For context, I’m investigating how to map FHIR resources into a relational database model in the best way. I’ve tried to find any description on the issue of values in repeated elements but have not found any documentation on it.
I guess this is also related to whether duplicated values are allowed in a repeated element. The natural way to support duplicates in a relational model would be to impose an ordering of values.
ElementDefinition has orderMeaning. If that element is present, then the order has semantic meaning. If not, then the order has no meaning - and no meaning can be inferred by downstream profiles or implementation guides.
There is no prohibition against duplicates, though systems might impose business rules that do make prohibitions (and determine what constitutes a ‘duplicate’). For example, a patient might have two name repetitions that are identical other than the HumanName.period. (And the receiving system might not support ‘period’ and throw it away.) That’s not necessarily non-conformant.