• Heterogeneity

    This is a TIE-U post associated with Karoline Mortensen’s Introduction to Health Systems (UMD’s  HLSA 601, Fall 2012). For other posts in this series, see the course intro.

    Here’s a sure-fire way to make any debate more complex and, sometimes, win it. Just claim heterogeneity. “How does that work?” you ask. I’ll tell you.

    Suppose you say, “The problem with the US health system is we spend too much on stuff that doesn’t work. We should just do a lot more clinical studies, find out what doesn’t work, and stop doing it.”

    I actually agree with that, roughly. But I recognize that it is also too simplistic. (Hey, it was two sentences. By definition, it has to be simplistic. If anyone claims to have a solution to problems in health care that is shorter than a book, he’s lying or misguided. But I digress, (which you can tell because I’ve put this digression in parentheses.))

    One counter to what I just pretended you said is that not all treatments work equivalently on all people. A treatment that is, on average, not as good than another might be better than another for a specific type of person (e.g., with the right genetics or mix of conditions). People are not all the same. There is heterogeneity. If you buy that and you think it’s a large and important issue (sometimes it might be, but probably not always), that pretty much blows up the idea that we should always just test treatment X against Y in a randomized design, find out which treatment performs worse and throw it away (or not provide insurance coverage for it). Obviously, we have to do more careful, nuanced studies, at least in some cases.

    (I still think we should strive to know what works best on average and start there as first-line treatment unless we also have good, evidence-based reasons to deviate. In other words, heterogeneity is not a good argument for sticking our head in the sand. For more on this topic, see the recent Health Affairs post by Dana Goldman and colleagues, with which I don’t entirely agree but is a good overview of issues nonetheless.)

    Now let’s apply the same idea to something else I’ll pretend you say: “The problem with the US health system is that we use too much health care. We know a lot of it does no good. And it is costing us a fortune. We should just cut back.”

    I bet you’ve heard that before. Now apply heterogeneity, the fact that it isn’t true for all classes of people. The fact is that there is underuse as much as overuse. One of this week’s readings The Quality of Health Care Delivered to Adults in the United States, by Elizabeth McGlynn and colleagues (NEJM, 2003 and ungated) illuminates underuse. According to the study, about 55% of people receive recommended health care. That’s right, nearly half of Americans don’t get enough of the right kinds of health care.

    It turns out, I’ve covered that paper before, so let me save myself some effort and just refer you to that post, which wraps up:

    With this much underuse, a suspected significant amount of overuse, and also wrong use (which I’ll leave for another post) it’s unlikely any of us have received optimal health care. And yet, how many question our doctors’ suggestions or therapies? How many think they’re not getting good care? I’ll bet most people think their care is good and appropriate. All that bad stuff is happening to other people.

    I’ve never seen a poll on this, but I bet it is true that people tend to think their care is more appropriate than everyone else’s. So, sure, there’s heterogeneity and there’s also this (suspected) bias. It may be what leads to health reform ideas heavy on efficiency and light on equality: just cut back, for example, and particularly for those other people. Well, some of those other people aren’t getting enough. And you might be getting too much. Or vice versa.

    @afrakt

    Share
    Comments closed
     
    • Heterogeneity is a issue, for sure. Not only for effectiveness, but also very much for side effects.

      And it is not just an issue of genetic or developmental, i.e. persistent differences. As an example, it is common for individuals on NSAIDs for arthritis to find efficacy initially good, but fading after a year or two. A switch to a different drug in the same class often restores the efficacy, again for a year or two. It is truly helpful to have multiple choices in this class.

      I am all for comparative effectiveness (and comparative tolerability) clinical studies. But make no bones about it: they are difficult to design and execute, they are astonishingly expensive, they take a lot of time, and they will always leave some questions unanswered.