Update: a friend has pointed out an error in my math, read the comments for better math. The spreadsheet has been updated. Also, the US is still behind.
Recently, I've heard some very poor arguments about why birth and death rate statistics make the whole "life expectancy" number not comparable from country to country. This is, obviously, an attempt to claim that the current US health care system is at least as good, if not better than, those countries that do better in life expectancy for less money. Before I get into proving them wrong with numbers, I'll point out that the people making these arguments are likely in one of two camps. Either they are "true believers", that is, a group that believes counter-arguments to something they dislike, regardless of how little proof there is, because "the enemy of my enemy is my friend", and quite possibly because their world-view is tied up into their beliefs (as opposed to reality). Or they are "mouthpieces", that is, a group of people who know the arguments to be false, yet still use them as an attempt to derail policy or conversation about policy through misinformation.
To get deeper into what birth/death rate statistics I'm talking about, the claim is that because of the higher US murder rate, and because we consider "live" much smaller premature births in the US, that our life expectancy numbers are unfairly biased downwards, so comparisons against health care are not comparable by using life expectancy. I intend to show that murder rates and infant mortality rates make little difference in the numbers, so the claim of not being comparable due to those numbers is bullshit.
First, we will start with a "nominal age", that is, an age at which we assume everyone dies if they aren't murdered or die in childhood. It turns out that, thanks to statistics, this works the same as if there was a more complicated distribution that just happened to average out to 80 years old. I use 80 because life expectancy for industrialized nations is about that, but we can use any age; the only difference is that higher ages scale the difference between the nominal age and expectancy up.
We will then have a "homicide rate", that is, the number of murders per 100,000 people. When we perform the calculation, we will assume that all people are murdered when they are born. This biases the calculated pseudo "life expectancy" number down, but because we'll do this for all countries we calculate, this is fair. We get these numbers from:
http://en.wikipedia.org/wiki/List_of_countries_by_intentional_homicide_rateWe have an "infant mortality rate", that is, the number of infant deaths per 100,000 people. We obviously assume these children die when they are born. We get these numbers from:
http://en.wikipedia.org/wiki/List_of_countries_by_infant_mortality_rateWe will also, for the sake of argument, replace "infant mortality rate" with "under-five mortality rate", also getting these numbers from:
http://en.wikipedia.org/wiki/List_of_countries_by_infant_mortality_rateWe also need "birth rate", so as to determine how much "infant mortality rate" and "under-five mortality rate" affect our statistics. We get these numbers from:
http://en.wikipedia.org/wiki/List_of_countries_by_birth_rateThis stuff should be used as reference, please see the comments for better math, which still shows the US behind.
First, we are going to find out the proportion of the population that is murdered. That is simple; we take our "homicide rate" number, and divide it by 100,000. In the case of the US, that is 6.8 / 100000 = .000068 . We'll call this h_prop.
We then find out the portion of the population that is born in any given year. For the US, that is 14.0 per 1000 population, which we can scale up to 1400 per 100,000 population (to be consistent), which we then scale back down to .014 . But, we need to know how many of these children die before their time. In the US, the infant mortality rate is 6.3 per 1000 live births, or .0063, which we multiply by .014 in order to get the actual proportion of our population that dies as an infant; .014 * .0063 = 0.0000882 , we'll call this im_prop. Doing the same number for under-five mortality gets us 0.0001092 , we'll call this ufm_prop.
We now have enough numbers to do our calculations.
age * (1.0 - h_prop - im_prop) -> ifm_exp -> life expectancy assuming only infant mortality and murder.
age * (1.0 - h_prop - ufm_prop) -> ufm_exp -> life expectancy assuming only under-five mortality and murder.
80 * (1.0 - .000068 - .0000882) = 79.987
80 * (1.0 - .000068 - .0001092) = 79.985
Looking at the worst-case under-5 mortality rate, (80-79.985)*365 = 5.17 days. Yep, the homicide rate in the US combined with under-5 mortality rate, assuming that murders happen when a person is born, adds up to a change in life expectancy of 5.17 days.
I created a google spreadsheet, which you can view and make a copy of if you have a google account, which will perform these calculations for you:
https://spreadsheets.google.com/ccc?key=0AiQlFAGAyO9SdEdaUm9zVnNVY1J0Q1JMMGJLT2xZcFE&hl=enAs I just showed you, using simple math, the argument that birth/death rate statistics make the numbers not able to be compared from country to country is just plain bullshit. Making claims of them being not comparable and not running the numbers is at best lazy, but really, it shows how little some people care about actually making this an argument about apples vs. apples. This is apples vs. death squads, apples vs. murders, and any other nonsensical comparison that can be done to derail the real discussion; health care in America could be better, and it is shameful that people are making arguments based in obvious lies, especially when the lies hurt people who aren't getting health care (you know, the 30 million people without health insurance). Shameful.
Update: a friend has pointed out an error in my math, read the comments for better math. The spreadsheet has been updated. Also, the US is still behind.