> The culture seems to be that individual departments get to decide their own software, based only on their needs.

I think that's a good thing by itself, various parts of a hospital are incredibly specialized and a one size fits all solution will never work. The alternative is some much higher level of management deciding what everyone buys.

HL7 needs to be scrapped though, it does nothing to achieve inter-operability, it's just there to make hospital managers think that it exists and keep smaller players out of the market.

I probably should have expanded on that a bit. In many cases the individual departments don't even communicate purchase intent to IT.

So things like "optional" integration modules go unpurchased. Contract terms don't address basic integration needs, etc.

Or, a specialized department has a software bake off, finds 3 solutions of equal value to them, and picks one at random. Not knowing it's the one least likely to fit in the larger picture.

Or, crossover software that serves more than one department is purchased by only one, with no discussion.

Basically, for whatever reason, healthcare is just more territorial and segregated than most other companies I've worked in.

You 100% hit the nail on the head. It's really, really amazing to see how absolutely disconnected the buyers and the users are.

I was doing training at one hospital where, when they admitted a patient to the floor from the ED, had to print out the entire EMR, because the EMR in the ED and the floor was not compatible. Insanity.

That kind of thing's not so rare though. Hospitals have lots of systems and paper works quite well.

The solution, obviously, is for everyone to be a stakeholder and everyone to get a formal veto on the matter.

By all stakeholders I mean:

  * The department needing the software.
  * The department integrating the software (usually IT).
  * The department(s) consuming the data.
  * Any other departments that interact with the above dataflow (IE does it contradict things they are/plan to do?).
  * Whoever signs off on the check.

The formal veto would include written problems with the proposal "this is not a good idea as proposed because Z" and/or "for this to work we need X as a requirement, but it isn't in the proposed spec".

I don't know that there's any particular formula that works. Other environments I've been around are also territorial, just less so.