[marktangotango]

But access to the data that exists in those systems is a pretty big carrot for integration partners to help push adoption.

The scary reality is, they don't even have access to the data in those systems. I had a friend contracted at cerner working on a FHIR api end point. Something as simple as answering the question "what medications have been administered to a patient" was extremely difficult to pull out of their system. One would think there'd be a simple report for that information, but they didn't have it. The domain/system knowledge wasn't there either. That one simple api endpoint drug on for months as new data was "found" in the 5000+ table schema. Oh, and they don't use foriegn key constraints in production either.

[NoNotTheDuo]

I can vouch for some of that as well, although I disagree with some of it. The domain/system knowledge is there, but there's likely a relatively small group of people that know all of the ins and outs. And the group that knows about inpatient medications likely isn't the the same that knows about outpatient medications. And maybe the ordering of meds is different from the administration of meds.

I think the point about FK constraints is perhaps a little too generalized. In a 5000+ table schema (which is entirely accurate!), I'm sure there are some tables which don't have FK constraints enabled. But I know for a fact that there are FK constraints on some of the tables (because I created them!)