• Pre-test probability

    Several years ago a young man visited my clinic and requested a “complete set of blood tests”.  I asked what he meant by that.  He clarified that he wanted “all available blood tests done”.  I asked why he wanted such information and he confidently asserted that he just wanted to be sure that there was nothing wrong with him.  We spent some time discussing blood tests and I gently tried to explain that he very likely did not have enough money to run every conceivable test that could be purchased at that time.  When I tried to explain why that would not be a good idea anyway, the conversation became a little more difficult.  Ultimately, I endeavored to explain the statistical concept of pre-test probability. I find that this concept is poorly understood by many people.

    More here. It’s a good message, but at the time I read the post, I caught two errors in it, both serious. I noted them in the comments to the post. So, if after reading the post, you’re confused, read the comments. (Oddly, all my comments are entered twice even though I only did so once.)

    This, by the way, is why expansion of home testing is potentially problematic.


    Comments closed
    • This medical issue is related to a behavioral issue, namely that a relatively small part of the population consumes a large percentage of health care. Why is that? Sure, age is a factor, as is the distribution of disease. But so is the propensity to consume more of what has been tried. Or put another way, once someone becomes part of the health care treatment system, she is more willing (indeed, more demanding) to consume it. This phenomenon is difficult to measure because there can be many variables for consuming health care. It will be interesting to see how the millions who couldn’t afford health care respond when ACA’s universal coverage provisions become effective – how many heretofore non-consumers of health care will become obsessive consumers of health care after the initial introduction.

    • One word: Bayes.