correlation does not equal causation.
What you will no doubt see in the media is this:
[MSNBC]: New data show that insuring people will save thousands of lives!
[Fox News]: Ridiculous! Silly liberals. The uninsured tend to be people who pick their noses, and nose-picking is what causes death--not the lack of insurance!
(Nose-picking is just an example here--fill in whatever third factor you like.)
The authors do appear to understand the distinction between correlation and causation, and use careful wording.* From the abstract:
"Uninsurance is associated with mortality."
"...is associated with" is much different than "causes," but I fear that "causes" is what the journalists will say.
This isn't to say the paper is without merit. The methodology is interesting, and so is the finding. They looked at people who didn't have insurance at a single point in the past, then looked at their mortality rate in subsequent years. What makes it somewhat interesting is that:
"After additional adjustment for race/ethnicity, income, education, self- and physician-rated health status, body mass index, leisure exercise, smoking, and regular alcohol use, the uninsured were more likely to die (hazard ratio=1.40; 95% CI=1.06, 1.84) than those with insurance."
This makes it more interesting since some obvious factors other than insurance status at the time are eliminated as explanations for the difference in death rate. Of course, there are probably lots of other factors that weren't controlled, but it is an interesting finding.
* On a personal note, I don't quite trust the authors--they've done a lot of cherry-picking in their other work. I haven't analyzed this paper, but the timing of their publication of this correlation finding is not a coincidence. They know quite well that most people will view it as a causation paper.
Posts
No comments:
Post a Comment