FHIR resources in specific scenarios
December 19, 2013 Leave a comment
Taking the example of Grahames ‘Q&A’ format, the following question was asked on the FHIR List forum:
Based on what I have seen, FHIR defines a set of resources and all possible relationships between resources, but doesn’t define which/how resources should be used in a particular scenario – the latter needs to be defined as a Profile. And FHIR per se doesn’t have Profile recommendations for any scenario. Implementers need to define their own Profiles and figure out how systems interoperate on a semantics level.
and answered by Lloyd:
Sort of. First, FHIR doesn’t actually define all possible relationships between resources. It merely defines the “common” ones – those deemed to be within the 80%. Many other types of relationships are possible through the use of extensions. (E.g. encounter is associated with a single patient in the core resource, but with extensions could support multi-patient “group” encounters, for example)
You can approach interoperability with FHIR in a couple of ways. In one approach, you just expose everything you do as a FHIR resource and anyone who wants to consume your content can simply pick and choose from what you expose (whether as a dump in a bundle or by hitting your FHIR REStful service). In the second approach, you expose a particular subset of content (or perhaps even enhance the capability of your system to support additional data elements) as defined by a particular profile on FHIR. These profiles could be created by a variety of groups. Some (e.g. a FHIR equivalent of CCDA) will likely be created by HL7 international. Some may be produced by IHE or other international SDOs. Some will be created by HL7 affiliates and national programs. Some will be created by large healthcare organizations.
The creation of profiles will take time and will be driven by specific use-cases. They don’t yet exist. So for early adopters, you sort of have to either take the first approach or go through the process of creating your own profile, ideally with a group of like-minded early adopters. If you’re taking the second approach, you need to identify the area of healthcare that you’re targeting, look at existing capabilities and business needs. Inpatient care and Outpatient care look quite different. (So does human vs. veterinary or individual vs. public health.) That’s why your initial question of identifying the “minimum” is so challenging to answer. The minimum depends very much on what area of healthcare you’re targeting, what problem you’re trying to solve and even what country/region you’ll be working in.
I thought it was worth sticking it up here so I can find it later! Profiles are going to be important…
About David Hay
Recent Comments