Oh man, where do I start?
I kind of swam in the Data Warehouse, and it was really important to me to get an ability to get as much of a ‘story of Joe’ as I could, because . . . well, that’s a LOT of data, and there’s a lot to weed out, each claim can be coded differently depending on what type of claim it was, and facilities use ICD-9/10 codes differently and the like, it’s a really big mess.
So, after learning what I could and stealing from some giants (like pulling the tables out of the CDPS grouper) and starting to do a general translation to simpler language I thought it would be a good idea to get in front of the ICD-9 to ICD-10 conversion. In case you haven’t guessed, ICD-10 is the newer one, and guess what country is slow to adopt new things? There were better de-identified data sources over in Europe and I figured ‘hey! I can use these to help trend things!’ I’d been particularly frustrated by how incomplete some of our data was, so I was hoping I could come up with a system where I could do good predictions.
Long story short, in a whole lot of realms it seemed like ‘Being in America’ kept bubbling up to the top when it came to most forms of poor outcome analysis. . .even generic decision trees in tools like Rapid-i found that one in no time flat, and I was frustrated because it made this massive repository of great info useless, you know? So I basically set up a little factor value that I’d run against combinations of conditions (+diabetes II, +CHF, -High Lipids and so on) so I at least wouldn’t have all my baselines be totally out of whack.
I ran into a similarly exasperating problem searching for root causes for high cost and problem pregnancies, where by far the strongest predictor of a high cost childbirth was that I HAD NO DATA ON THEM! I mean, we didn’t always get everything right away from private insurers, but pattern was strong going back, too . . . which kind of pointed to people needing prenatal care but not getting it because of our healthcare system, but obviously I wasn’t able to get any headway on that (Scott Walker got elected and hooo-boy did things get political!).
So, yeah. . . pretty questionably efficient system. I was less than pleased. I suspect mental health may also be a factor in the original variance, because I’d find some pretty strong correlations between physical outcomes and even some forms of depression (much less more consistently diagnosed conditions like our poor bipolar and schizophrenic folks had to deal with), but our history documentation and the difficulty in lower income folks getting to see a psychiatrist for anything other than meds do indicate a slightly different pattern.
I have no real idea what the roots are behind the drop I saw here in the US, probably a combination of it being a stressful place to live (with uncertainty about our future being treated as if it was a good thing) and a generally messed up and crappy system. Either way, outcomes are crap despite spending an insane amount of money.