April 26, 2005

Pick a Metric, Any Metric

Steve Verdon has an interesting post on the problems with using life expectancy as a measure of the "goodness" of any given health care system. I agree, it's somewhat problematic. Lots of things affect life expectancy, and just because the numbers are low isn't proof that the health care system sucks. That reminds me of an interesting study once mentioned by Phillip Longman:
In a recent issue of Health Affairs, three researchers from the Robert Wood Johnson Foundation examined scores of studies dating back to the 1970s on what factors cause people to die prematurely. They reported that genetic predispositions account for 30 percent of premature deaths; social circumstances, 15 percent; environmental exposures, 5 percent; behavioral patterns, 40 percent; and shortfalls in medical care, 10 percent. As they note, these proportions are easily misinterpreted. Ultimately, nearly everyone's health is determined by a combination of factors. For example, while only about 2 percent of human diseases are caused by inherited genetic mutations alone, nearly everyone carries various genetic dispositions that, when combined with a hazardous environment or unhealthy lifestyle, can contribute to ill health. But this only underscores the relatively small role medicine plays in preventing premature death.
So yes, if we really wanted to boost life expectancy here in the U.S., we probably ought to focus more on those "social circumstances"—especially reducing poverty—along with getting people to exercise more and eat healthier stuff. (We might also want to try a bit of genetic engineering, though I won't wade into that right now.) Meanwhile, improving health care access for the poor and otherwise uninsured would certainly be a grand thing, and would probably save lives, but in the grand scheme of things probably wouldn't boost our life expectancy numbers up to the level of our OECD competitors.

So put a grain of salt on those life expectancy graphs. But do notice: It's still extremely hard to find statistics that cast the U.S. health care system in a favorable light. My favorite metric is the one used by the WHO, which tries to measure three goals: good health, responsiveness to health expectations, and fairness of financial contribution. For good health, the WHO looks at disability-adjusted life years (i.e. years without disability) rather than life expectancy. The surveys on responsiveness, meanwhile, try to control for different cultural interpretations of what constitutes "good health" and "quality care". And even here, the United States regularly does dismally among developed nations—32nd in good health, 15th in responsiveness, 54th in fairness, and 37th in overall performance. Now you could say, "Hey, that's not fair, the U.S. would be doing pretty well were it not for the fairness part," but that's an awfully odd thing to bracket off like that.

Again, you can poke holes in just about any statistic used. There are all sorts of contortions and mitigating factors and things not considered. Tech Central Station runs clever articles like this all the time ("Well perhaps this number can be explained by this, or perhaps this, or perhaps, or perhaps...") But at some point the pro-U.S. health care faction really ought to make at least some statistical case that our health care system, as a whole, does pretty well for itself. Because from what I can tell, in ranking after ranking and chart after chart, it falters.
-- Brad Plumer 1:38 PM || ||