It's definitely worth quantifying the cost of providing medical care, but doing it in a more fine grained way than at the department level.
I think you could have been more explicit about (what you now say was) your core point.
Yes it does vary dramatically, but there's no way there's 3 orders of magnitude of variation and definitely not 2 orders. There might be a single order from top to bottom, a 2 hour surgery versus a 20 hour surgery or a 15 minute visit with a doctor versus a couple of hours with a doctor.
Lawyers have the same problem and while they do bill by the hour, they can generally give you a pretty good idea of what the more basic things might cost to handle. I'm not suggesting that they could give you a +-3% estimate for your SCO vs IBM lawsuit, but if it's a DUI or a will or a trust or whatever, once they've done it a few times they know how the game is played.
I totally get that people aren't courts and doctors aren't lawyers, but I think it's a decent counter-example to the notion of "it's hard so why bother trying?"
But what if it's not so serious, and not an emergency? People have rotator cuff issues that they take years to get fixed regularly. Or a hip replacement? That's rarely an emergency but rather a solution to long term pain that's not painkillers.
I think the other thing a lot of people would like to see is some kind of doctor and hospital report card so that you could compare on price AND quality. Which would be great as it'd force the terrible surgeons who regularly do such a bad job (accidentally or on purpose) that their patients have complications, die, etc out of that line of work.
Obviously there would have to be some kind of risk classification system so that doctors and hospitals that take on high risk patients wouldn't be punished for doing so, but I think that some kind of "hot or not" style risk assessment system could be developed whereby doctors spend 15 minutes a day reviewing risks anonymously for other doctors and each case gets rated 5 times and that determines the risk score to hopefully remove all bias.