Tuesday, September 29, 2009

More on Senator Ensign's Comments

For preliminaries, read the post below. Essentially, Senator John Ensign (R-NV) argues that the United States health care system works better than those of any other comparable nations if we first take out car accidents and violent deaths.

There's some method to his madness, and it goes like this:

What is it that health care "produces"? If we can't give the output some measure, how do we know if the costs of a particular system are too high or not? Indeed, the whole discussion of high health care costs is always juxtaposed with the idea that the system isn't somehow producing what those high costs would make us expect.

Here's the deep-level problem: It's very, very difficult to measure the output of the health care system. Ideally we'd like to do that by measuring the health improvements or the prevention of worse health, for every single patient, and for both physical and mental health. Researchers do try to do this, but mostly they resort to something much cruder and ruder. Often this has to do with measures of mortality or morbidity. The reverse of these would of course be good things, and to the extent we can tie them to what the health care system does, we get a rough performance measure, right?

That's where Senator Ensign steps in. His point is that average life expectancy measures (which capture mortality in reverse) are very sensitive to anything that kills people at young ages. All such deaths cut off all those extra life years the average person would expect to have.

Because the United States has an excess death rate at young ages, when compared to other economically similar countries, the U.S. life expectancy figures look worse. That this excess death rate is due to factors (driving and violence) which are not something physicians and nurses can easily influence might mean that the system here isn't functioning too badly at all! Indeed, according to Senator Ensign, it functions just great (though uninsured Americans might disagree).

What he fails to state is that other mortality figures also show the United States in a bad light, and these are measures which cannot be explained away in the same way. The most important of these is the infant mortality rate (deaths of children up to the first year of life), and this rate has been shown to be fairly easily influenced by health care.

The whole topic of how to measure the output of health care is a large and cumbersome one, and soundbite-type discussions do it a disservice, sigh. At the same time, it's so very easy to draw a few pieces of data out of a political top hat and to come across as convincing. You do that by setting up straw-people left, right and center, and then you set them on fire.

The source of Ensign's ideas appears to be a study which controlled for traffic accidents and violent deaths but did not control for average incomes in a country. You get partial results with partial controls.