Arkady
President
As most people know, US life expectancy is horrible, despite the US spending more on healthcare than any other nation, by a wide margin. However, one of the conservatives here pointed at that there's an argument that US life expectancy is actually just driven down by other factors that aren't really health related (commonly-cited factors being traffic accidents, suicide, and murder). One study attempted to adjust for these factors, and suggested that once you do so, you find the US is actually first in life expectancy, not 19th as the standard data would show.
It's an interesting argument, and I always appreciate it when a conservative approaches a question with numbers and scientific studies, rather than just anecdotes and rhetoric. However, there are at least two layers of problems with that approach. First and most obviously, simply throwing out traffic accidents, suicide, and murder, would be expected to disproportionately help the US. All three of those things hit the young at a higher rate than the old. And, in the US, the young are more likely to be uninsured than the old (both because of Medicare eligibility over age 65, and because older people are more likely to have the kinds of steady jobs that have health benefits). If the way the US healthcare system leaves many people uninsured results in worse care, we'd expect it to be most prominent with the young, who are more likely to be uninsured. And in the case of accidents, suicide, and murder, poor healthcare can, in fact, have an impact on the death rates. The difference between a shooting and a fatal shooting (or a car accident and a fatal car accident; or a suicide attempt and an actual suicide) is sometimes going to be the quality of the care the person receives after the fact. And with suicide you have the added issue that a medically neglected mental problem can mean more suicide attempts.
So, simply disregarding suicides, murders, and fatal traffic accidents could be expected to disproportionately extract from the life expectancy calculation things that should, in fact, be held against the American healthcare system. If we're suffering more traffic deaths, suicides, and murders in part because of failings in our medical system, we shouldn't be allowed to simply define those things out of existence for purposes of assessing our medical system.
Aside from that, though, it appears that the calculation the conservative poster was talking about was flawed to begin with. Rather than throwing out those problematic deaths and recalculating life expectancy (which would only have pulled the US up slightly, from 19th to 17th), they instead did a regression analysis involving GDP-per-capita, to try to figure out a guess for what life expectancy would have been if deaths from those causes had been equal. As you can imagine, trying to back into the answer by finding a correlation with GDP-per-capita doesn't tell you much of anything other than what countries have high GDP-per-capita. They took a lot of criticism for that and, to their credit, they don't seem to be denying the flaws... they've just fallen back to the weaker argument that whatever the flaws of their argument, it's at least suggestive that official life expectancy stats might be skewed by other factors):
http://blogs.wsj.com/numbersguy/does-the-us-lead-in-life-expectancy-223/tab/article/
http://theincidentaleconomist.com/wordpress/how-flawed-is-life-expectancy/
From "M. Simon" on the Amazon.com book review, here's a discussion of the actual formula used in the regression analysis:
Be aware that one of the most surprising conclusions in the book - that the U.S. healthcare system, when you take away injuries and accidents, actually has the best life expectancy in the world - is based on a very clever twist of the underlying data.
Instead of looking at real-world life expectancy and then taking away deaths due to injury and accident (a method that puts the U.S. at 17th) Mr. Ohsfeldt uses an equation that begins with an ESTIMATE of life expectancy BASED ON GDP PER CAPITA (a measure of how rich the country is), not on actual real-world data. Then he adjusts for injuries and accidents. Of course the U.S. remains at the top of the list, as we are one of the richest countries in the world.
Here's his equation, from the book:
LifeExpit = 50.78 + 3.020 * log(GDPPCit) - 0.077 * [mean(Trans)]
- 0.137 * [mean(Falls)] - 0.133 * [mean(Homicide)]
- 0.0326 * [mean(Suicide)] + year-effectsit
The equation includes "log(GDPPCit)", which is GDP per capita in country i, year t, and it includes factors for transportation accidents, falls, homicide, and suicide. Mr. Ohsfeldt suggests in his writing that the equation and graph are based on real world life expectancy data, which figure into the equation nowhere at all.
It's an interesting argument, and I always appreciate it when a conservative approaches a question with numbers and scientific studies, rather than just anecdotes and rhetoric. However, there are at least two layers of problems with that approach. First and most obviously, simply throwing out traffic accidents, suicide, and murder, would be expected to disproportionately help the US. All three of those things hit the young at a higher rate than the old. And, in the US, the young are more likely to be uninsured than the old (both because of Medicare eligibility over age 65, and because older people are more likely to have the kinds of steady jobs that have health benefits). If the way the US healthcare system leaves many people uninsured results in worse care, we'd expect it to be most prominent with the young, who are more likely to be uninsured. And in the case of accidents, suicide, and murder, poor healthcare can, in fact, have an impact on the death rates. The difference between a shooting and a fatal shooting (or a car accident and a fatal car accident; or a suicide attempt and an actual suicide) is sometimes going to be the quality of the care the person receives after the fact. And with suicide you have the added issue that a medically neglected mental problem can mean more suicide attempts.
So, simply disregarding suicides, murders, and fatal traffic accidents could be expected to disproportionately extract from the life expectancy calculation things that should, in fact, be held against the American healthcare system. If we're suffering more traffic deaths, suicides, and murders in part because of failings in our medical system, we shouldn't be allowed to simply define those things out of existence for purposes of assessing our medical system.
Aside from that, though, it appears that the calculation the conservative poster was talking about was flawed to begin with. Rather than throwing out those problematic deaths and recalculating life expectancy (which would only have pulled the US up slightly, from 19th to 17th), they instead did a regression analysis involving GDP-per-capita, to try to figure out a guess for what life expectancy would have been if deaths from those causes had been equal. As you can imagine, trying to back into the answer by finding a correlation with GDP-per-capita doesn't tell you much of anything other than what countries have high GDP-per-capita. They took a lot of criticism for that and, to their credit, they don't seem to be denying the flaws... they've just fallen back to the weaker argument that whatever the flaws of their argument, it's at least suggestive that official life expectancy stats might be skewed by other factors):
http://blogs.wsj.com/numbersguy/does-the-us-lead-in-life-expectancy-223/tab/article/
http://theincidentaleconomist.com/wordpress/how-flawed-is-life-expectancy/
From "M. Simon" on the Amazon.com book review, here's a discussion of the actual formula used in the regression analysis:
Be aware that one of the most surprising conclusions in the book - that the U.S. healthcare system, when you take away injuries and accidents, actually has the best life expectancy in the world - is based on a very clever twist of the underlying data.
Instead of looking at real-world life expectancy and then taking away deaths due to injury and accident (a method that puts the U.S. at 17th) Mr. Ohsfeldt uses an equation that begins with an ESTIMATE of life expectancy BASED ON GDP PER CAPITA (a measure of how rich the country is), not on actual real-world data. Then he adjusts for injuries and accidents. Of course the U.S. remains at the top of the list, as we are one of the richest countries in the world.
Here's his equation, from the book:
LifeExpit = 50.78 + 3.020 * log(GDPPCit) - 0.077 * [mean(Trans)]
- 0.137 * [mean(Falls)] - 0.133 * [mean(Homicide)]
- 0.0326 * [mean(Suicide)] + year-effectsit
The equation includes "log(GDPPCit)", which is GDP per capita in country i, year t, and it includes factors for transportation accidents, falls, homicide, and suicide. Mr. Ohsfeldt suggests in his writing that the equation and graph are based on real world life expectancy data, which figure into the equation nowhere at all.