We are creating a brand new application in EHR domain. The client has requested the application to be FHIR compliant. While reading more we discovered FHIR is mainly used for interoperability between systems.
As the system is a prototype and not have much of the resource data (it may become a full fledge EHR in future). Do we really need to create each resource in our system. We prefer to keep it simple without adding overhead of the difficult structure? In future if any system/client requires FHIR data from our system, we can always build FHIR compliant API from the existing system data.
FHIR provides both a useful data model and API specification. I’d say, if the application is clinically relevant and you hope to support EHR integration in the future, starting with FHIR seems like a good investment. You don’t necessarily need to implement every resource or method, but consider applying as much of FHIR’s data model as is useful to you. If you need to adhere to a proprietary data model, I’d suggest mapping out how you want to translate your fields into FHIR in the future so that you can discover gaps between the two models – things like that are usually easier to fix upfront before the app is live.