Statistics in baseball and medicine

I went to a baseball game with my father last week (go Sox!). Our pitcher wasn’t doing so great – his strike rate was hovering around 50%, and it was taking him almost twice as many pitches to get through an inning. I knew this because my dad kept pointing out statistics – live, updated – on the board. And that was a simple statistic, calculated as I watched. But they only got more complicated. I began to envision a gigantic database in my head. They reported the number of errors a player had made while in that position (so the database must have a field for position played). Then the number of times a pitcher pitched a strike when there was a full count (the database must have not just the number of balls and strikes, but the order). I was very impressed. But then I realized it couldn’t be that hard to design a good, if complex, database. I mean, they’re doing it for baseball.

But then I went back to work the next day. I’m a grad student doing health care quality research. We’re trying to answer a relatively simple question about medications and laboratory monitoring. Little has been reported on this information before because almost no one has even more basic data on prescribing rates and test ordering rates. Besides, prescribing databases are unreliable because patients may not fill the prescriptions, while claims data isn’t always reliable because some people pay out of pocket, for example. So we can’t answer simple questions like how many people got appropriate monitoring tests to better target interventions and improve care, much less the complex ones.

Imagine if we had the stats of baseball in health care. Imagine what we could do. Which is more important?