FHIR does not have the hubris to try and standardize all of healthcare in any given use-case. The domain is just too complex and all of its predecessors have failed in this ambition.
I remember in the earlier days of our software company we used PowerDesigner for physical data models. PowerDesigner saved the model in files with the .PDM file extension. Our designers tried to be as generic as possible and we jokingly talked about a model that could cover all: WORLD.PDM.
We never managed to capture the world in a single model. Before we went bankrupt or nuts we found out there are less elavated but more achievable goals in life.
FHIR negates the feasability of a model that captures the entire healthcare information domain. In stead, it introduces very generic building blocks (Resources) and a mechanism to mold these building blocks into something useful in a given context (plus something to extend these building blocks, but that is outside the scope of this blog post). This mechanism is called profiling. Proponents say: brilliant. Adversaries say: goodby to interoperability.
In part III of my interview with David McCallie (VP medical informatics at Cerner) he expresses concerns (not saying that he is an adversary though) about the complexity of profiles and the way they are te be managed. Listen to David McCallie on profiling: