Friday, October 02, 2009

Study significant in what it doesn’t say

Congressman Alan Grayson put up a sign on the floor of congress listing the “republican plan” for health care as “die quickly.”  The “support” for this statement is a “study” he claims “proves” that thousands die because they don’t have health insurance.  The “study” is on his web sitediequck and he invited everyone to read it for themselves.

Well I did look at the study.  First of all, it says nothing about “republicans” but that’s not the point.  There’s some very significant fine print in that study!

The chart does indeed show an increased risk of death among those without health insurance as compared to those with private health insurance

Wait a second!  Isn’t the study missing something?  What about those who have government supplied health care?  They simply chose not to include any government health care plans in their study!  Why?

We can’t really know why for sure.  But there are some hints that suggest a reason:  If they had included government supplied health care, chances are it would show that having government health care would cause more deaths than either not having insurance at all or having private health care!  Let me explain why that’s likely.

Let’s simplify things with a little analogy.  Let’s say I want to study how far an athlete can throw various types of fruit.  Let’s say I want know which can you throw farther: an apple, an orange or a pineapple?  Simple enough, right?

So we do our experiment (study).  We carefully control for the strength of the thrower and randomize things so one person doesn’t always throw the apple first or the pineapple last to avoid skewing the results by having a tired throwing arm, and things like that.  No need to go into more details.  Let’s just say the results are in and we read the published, peer reviewed results:  The apple was thrown 1.4 times farther than the orange.  Great study, huh?

image But wait a minute, what happened to the pineapple?  In the fine print, it says we didn’t include the pineapple because the throwers couldn’t seem to throw the pineapple right.  Good excuse?

Why couldn’t they make a glove or something to throw the pineapple or a machine to throw all the fruits?  Or something?  Maybe they say they had to “adjust” for the different weights of the apple and the orange.  So if they could “compensate” for one of them, why couldn’t they adjust for confounding factors for the pineapple?  Something is wrong.

Also, from the comment in the notes “not able to throw the pineapple right;” doesn’t that suggest that the pineapple didn’t go as far as the others?  Isn’t that the implication?  Because if the pineapple flew farther than the others, the note would say something like “the pineapple seemed to be too aerodynamic and flew too easily” or something like that.

image Furthermore, let’s say there’s a political debate raging over the issue of if pineapples or apples fly farther.  Doesn’t that make it all the more suspicious when the pineapple is mysteriously excluded?

Back to the health study.  They write that the reason they excluded the “government health care plans” is that, well, those people must be really sick already.  Doesn’t that imply that the results were that people on government insurance did poorly, but they aren’t going to report that?

imageThey even mention they had to do all sorts of adjustments to “even out” various factors for the differences between the “insurance” group and the “no insurance” group.  Things like income, smoking rates, rates of obesity, different patterns of exercise among the groups, even assessments of their current health.  So, we are to believe that the study authors were able to compensate for all these factors, including current health, (the very reason cited for not including the government insurance in the results) but were somehow unable to do these same compensations to get an accurate view of government supplied health care?

I smell a rat!  What would congressman Grayson’s sign say if they included the government health care in the study?

graphmaybe

No comments:

Post a Comment