Post details: Cuba, US, Health

Cuba, US, Health

posted by anwar on 05.04.13 at 07:07, Economics, null, 6 comments Permalink

Collin's post about Cuba got me thinking about health care...

..if you don't like these measures for health care, which ones do you propose we should use for comparisons?

From the CIA World Factbook:

Infant Mortality Rate (deaths/1000 live births)
-World 50.31
-USA 6.63
-Cuba 6.45
-EU 5.3
-Canada 4.82
-Japan 3.28

Avg. Life Expectancy at Birth
-World 64.05
-Cuba 77.04
-USA 77.43
-EU 78.1
-Canada 79.96
-Japan 81.04

Comments:

Comment from: anwar [Member] Email · http://nonplatonic.com/anwar.php
Some ideas of better metrics (the idea is that these need to be relatively easy to measure):

%Chance of living to age 18

[this weeds out long term effects (that may not necessarily be dependant on the health policy) like radiation poisoning, health problems from smoking, etc]

# of able-bodied years (time in workforce?)

This might have some problems, as you will get skewed results in countries where there are incentives to work less/retire early.
Permalink 04/13/05 @ 09:30
Comment from: ben [Member] Email · http://ben.nonplatonic.com
Comparing the numbers without error bars is pretty meaningless... though the samples are probably the entire population of a country, so the errors are very small.

I've seen the median used as a single indicator because it is less susceptible to bias from any outliers.

Then there's the question of what sort of quality we are considering. I would guess that in countries with socialized healthcare, the care an individual receives varies less than in a country without socialized healthcare. My guess would be that the US has higher quality healthcare, but it is not available to everyone. I wonder what happens if you consider the statistics for the US population with health insurance, or with some minimal quality of health insurance.

Then there are dietary and exercise considerations...
Permalink 04/13/05 @ 13:37
Comment from: marco [Member] Email · http://www.cs.berkeley.edu/~barreno
Ben wrote:
Comparing the numbers without error bars is pretty meaningless... though the samples are probably the entire population of a country, so the errors are very small.

It depends on what you're counting. If the numbers truly come from the entire population (counting all deaths/births or whatever with reasonable accuracy) and if the statistic is something like "Infant mortality rate for 2004," then the statistics are population statistics so error bars would be meaningless. However, if you're trying to say something about the sustained rate over a period of years, then you could treat each year's rate as a data point, or you could use sampling to estimate the birth and death rates over a period of 20 years.

I do wonder how realistic it is to assume that the statistics are compiled by counting all births and deaths (or whatever you're counting). It's true that, at least in this country and probably most developed countries, birth records and death records are routinely kept. But there are plenty of people outside of the system, especially the very poor and illegal immigrants, who wouldn't necessarily be recorded. I wonder how they come up with numbers for those people.
Permalink 04/13/05 @ 14:07
Comment from: devin [Member] Email · http://www.hwaethwugu.com/blog
Ben wrote:
Comparing the numbers without error bars is pretty meaningless... though the samples are probably the entire population of a country, so the errors are very small.
Marco wrote:
I do wonder how realistic it is to assume that the statistics are compiled by counting all births and deaths (or whatever you're counting). It's true that, at least in this country and probably most developed countries, birth records and death records are routinely kept

As Marco points out, sample error is not the issue; systematic error is, especially in developing countries. In fact, Cuba's reported mortality rate rose during the early 60s. This is because afte the revolution, the Ministry of Public Health improved data gathering. (Source: Waitzkin, Howard. _At the Front Lines of Medicine_, a book I'm trying to slog through right now). In the case of Cuba, the Former Soviet Union, et. al. there is also the issue of whether you trust the government to accurately report their mortality statistics.

Permalink 04/13/05 @ 14:18
Comment from: collin [Member] Email · http://nonplatonic.com/collin.php
So Marco, you contend that even in the US the government is aware of every birth? This is the problem with statistics (as opposed to probability theory), in the real world you cannot measure an entire population. I'm not saying that there are huge error bars on something like the infant mortality rate in a country like the US, but they're still there. Basically my point is that there will allways be sample error.

And as for something like "Life Expectancy at Birth," Jesus H. Christ that's complicated. Seriously, how is this computed? If it's just "for what age x are half the people of that were born in year now-x still alive" then "life expectancy at birth" is a horrible misnomer. I can't think of a sussinct explination of the extrapolations needed to make that phrase meaningful, but hopefully you can see my point. Ergo, I vote that error bars are needed for these measures to be meaningful. Do any of you know how you construct error bars for a sample when you don't know the size of the population? Things like this have never been explained to me with the rigor I want in order to believe them.
Permalink 04/13/05 @ 21:02
Comment from: ben [Member] Email · http://ben.nonplatonic.com
My original point with the error bars was more that you need to know something about the relative sizes of the populations.

Also, I think the best way to model it is to treat the population as drawn uniformly from an infinite distribution since drawing a full population from a finite distribution would conglomerate additional stochastic processes with example selection. I don't know if that made any sense... I'm not sure how to explain what I'm thinking.
Permalink 04/14/05 @ 01:32

Leave a comment:

Your email address will not be displayed on this site.
Your URL will be displayed.
Allowed XHTML tags: <p, ul, ol, li, dl, dt, dd, address, blockquote, ins, del, span, bdo, br, em, strong, dfn, code, samp, kdb, var, cite, abbr, acronym, q, sub, sup, tt, i, b, big, small>
Options:
 
(Line breaks become <br />)
(Set cookies for name, email & url)
Comment from: anwar [Member] · http://nonplatonic.com/anwar.php
Some ideas of better metrics (the idea is that these need to be relatively easy to measure):

%Chance of living to age 18

[this weeds out long term effects (that may not necessarily be dependant on the health policy) like radiation poisoning, health problems from smoking, etc]

# of able-bodied years (time in workforce?)

This might have some problems, as you will get skewed results in countries where there are incentives to work less/retire early.
Permalink 04/13/05 @ 09:30
Comment from: ben [Member] · http://ben.nonplatonic.com
Comparing the numbers without error bars is pretty meaningless... though the samples are probably the entire population of a country, so the errors are very small.

I've seen the median used as a single indicator because it is less susceptible to bias from any outliers.

Then there's the question of what sort of quality we are considering. I would guess that in countries with socialized healthcare, the care an individual receives varies less than in a country without socialized healthcare. My guess would be that the US has higher quality healthcare, but it is not available to everyone. I wonder what happens if you consider the statistics for the US population with health insurance, or with some minimal quality of health insurance.

Then there are dietary and exercise considerations...
Permalink 04/13/05 @ 13:37
Comment from: marco [Member] · http://www.cs.berkeley.edu/~barreno
Ben wrote:
Comparing the numbers without error bars is pretty meaningless... though the samples are probably the entire population of a country, so the errors are very small.

It depends on what you're counting. If the numbers truly come from the entire population (counting all deaths/births or whatever with reasonable accuracy) and if the statistic is something like "Infant mortality rate for 2004," then the statistics are population statistics so error bars would be meaningless. However, if you're trying to say something about the sustained rate over a period of years, then you could treat each year's rate as a data point, or you could use sampling to estimate the birth and death rates over a period of 20 years.

I do wonder how realistic it is to assume that the statistics are compiled by counting all births and deaths (or whatever you're counting). It's true that, at least in this country and probably most developed countries, birth records and death records are routinely kept. But there are plenty of people outside of the system, especially the very poor and illegal immigrants, who wouldn't necessarily be recorded. I wonder how they come up with numbers for those people.
Permalink 04/13/05 @ 14:07
Comment from: devin [Member] · http://www.hwaethwugu.com/blog
Ben wrote:
Comparing the numbers without error bars is pretty meaningless... though the samples are probably the entire population of a country, so the errors are very small.
Marco wrote:
I do wonder how realistic it is to assume that the statistics are compiled by counting all births and deaths (or whatever you're counting). It's true that, at least in this country and probably most developed countries, birth records and death records are routinely kept

As Marco points out, sample error is not the issue; systematic error is, especially in developing countries. In fact, Cuba's reported mortality rate rose during the early 60s. This is because afte the revolution, the Ministry of Public Health improved data gathering. (Source: Waitzkin, Howard. _At the Front Lines of Medicine_, a book I'm trying to slog through right now). In the case of Cuba, the Former Soviet Union, et. al. there is also the issue of whether you trust the government to accurately report their mortality statistics.

Permalink 04/13/05 @ 14:18
Comment from: collin [Member] · http://nonplatonic.com/collin.php
So Marco, you contend that even in the US the government is aware of every birth? This is the problem with statistics (as opposed to probability theory), in the real world you cannot measure an entire population. I'm not saying that there are huge error bars on something like the infant mortality rate in a country like the US, but they're still there. Basically my point is that there will allways be sample error.

And as for something like "Life Expectancy at Birth," Jesus H. Christ that's complicated. Seriously, how is this computed? If it's just "for what age x are half the people of that were born in year now-x still alive" then "life expectancy at birth" is a horrible misnomer. I can't think of a sussinct explination of the extrapolations needed to make that phrase meaningful, but hopefully you can see my point. Ergo, I vote that error bars are needed for these measures to be meaningful. Do any of you know how you construct error bars for a sample when you don't know the size of the population? Things like this have never been explained to me with the rigor I want in order to believe them.
Permalink 04/13/05 @ 21:02
Comment from: ben [Member] · http://ben.nonplatonic.com
My original point with the error bars was more that you need to know something about the relative sizes of the populations.

Also, I think the best way to model it is to treat the population as drawn uniformly from an infinite distribution since drawing a full population from a finite distribution would conglomerate additional stochastic processes with example selection. I don't know if that made any sense... I'm not sure how to explain what I'm thinking.
Permalink 04/14/05 @ 01:32