Chart: Life Expectancy at birth and health care spending per capita

Chart: Life Expectancy at birth and health care spending per capita
  11/21/2013
  Aaron Carroll
Life Expectancy at birth and health care spending per capita, 2011 (or nearest year):
One of these things is not like the others.
@aaronecarroll

Share this...
Twitter
Facebook
email

  Explaining Research
  life expectancy, spending
item.php
- Comments (21)
- by Scott Collyer on November 21st, 2013 at 11:12
  This must be the result of the “best health care delivery system in the world” that the Anti-Obamacare crowd keeps telling me about.
  [top]
  - by Steve on November 21st, 2013 at 17:53
    I think they say it’s the best health care in the world, not the best “delivered” or the one that causes people to be healthiest. Doctors and hospitals provide health care, they can’t make people live much longer.
    [top]
  - by Ray Calloway on December 4th, 2013 at 09:05
    The point this graph makes to me is that we (USA folk) go to see our doctors and hospitals 2-4 times as much as we should. We want to rely on our health system for every little cut, scratch and manerism anamoly instead of just dealing with it the old fashion way. What does that get us?…higher costs naturally and a lower life expectancy compared to countries that don’t, or can’t, use their health systems as much as we do. It’s time for us to think about this chart and what it really means. It does not mean we have a higher cost health system, we simply use it more often than needed and the results are what they are…And I agree with the other comments, the graph should be pointing downward on the right half as it approaches the USA plotted point. I don’t like it… but that’s what it should do.
    [top]
    
    by Cardinal Fang on December 4th, 2013 at 11:01
    We don’t go to the doctor twice as much. We just pay twice as much when we go. Check the archives.
    [top]
- by Cardinal Fang on November 21st, 2013 at 11:32
  Ignoring the, ahem, outlier. it looks like if a country gets to around the $2200 level of spending, more spending isn’t correlated with better life expectancy.
  It would be interesting to see this chart with more recent data. That outlier on the left (USA! USA!) will still be an outlier, but I wonder what’s going on with all the other countries.
  [top]
  - by Cardinal Fang on November 21st, 2013 at 12:02
    *The outlier on the right, I mean.
    [top]
  - by Bill Gardner on November 21st, 2013 at 13:52
    Exactly. This is a famous graph, but the fitted line has always bothered me. There’s a fairly sharp inflection at the point CF identifies.
    And, of course, the implication that spending above $2200 / capita gets you nothing is unwarranted. You don’t get additional life expectancy. But you may get a hip replacement instead of a cane or walker.
    [top]
    
    by Cardinal Fang on November 21st, 2013 at 15:08
    Whether the extra spending gets you anything is an empirical matter. Do Norwegians get more hip replacements than Swedes and Belgians? I don’t know the answer. It would be interesting to see similar graphs with other measures: maternal mortality as a function of spending, hip replacements (age adjusted) as a function of spending, wait times for urgent care as a function of spending.
    But my main point was what Bill identified. I don’t like the fit of that curve.
    [top]
    
    by Tom Cloyd on December 3rd, 2013 at 10:38
    The fit reflects the considerable influence of the outlier. Outliers always distort regression curves, because we’re dealing with covariance and variance is about deviation form the mean, SQUARED.
    Normally, to deal with this, one removes outliers, but in this case that simply isn’t possible.
    The fit is correct, to my eyes.
    [top]
    
    by MikeK on November 25th, 2013 at 14:40
    It’s a simple log fit line; literally 3 seconds in Excel. Doesn’t change the fact that the US is well into the realm of diminishing if not negative returns on incremental expense.
    [top]
    
    by Cardinal Fang on November 25th, 2013 at 15:03
    The issue is not whether that’s the right log fit, but whether doing a log fit on that data makes any sense.
    [top]
    
    by Kevin on November 26th, 2013 at 16:17
    A log fit is the simplest possible mathematical fit that succeeds in capturing the diminishing returns that is clearly visible in the data. There are many other functional forms that could have been used, but they all would have involved more complex assumptions; e.g. what is the maximum possible average lifespan that would presumably be attainable as spending approaches infinity. Also, the log fit has no “turning point”: its slope bends smoothly, ever increasing at an ever decreasing rate.
    [top]
- by joe on November 21st, 2013 at 13:18
  Aaron don’t be like those who throw up cancer survival rates and say ‘AH HA! See America is the best!’ Without mention of lead-time bias, or the many criticisms of using 5-year rates to gauge quality.
  Same goes with life expectancy and infant mortality. They’re are criticisms, correct? How about at least a mention of adjusting for violence? Maybe, I’m wrong, but Tyler Cowen (or someone says) if you adjust for violence, then we’re #1. I’m not saying he’s right. But if you just throw one of these plots up, you contribute to the superficial analysis that we need FAR less off.
  [top]
  - by bob on November 21st, 2013 at 16:09
    Actually, he already covered those exact points about metrics in the article series from 2010 about the rating the US system.
    [top]
    
    by joe on November 21st, 2013 at 16:15
    Then the criticism is one of omission. Many people are looking at these plots without any context. I mean, a link, a sentence, anything. This post is misleading many people, and you can avoid this with a single sentence.
    [top]
- by Clint on November 21st, 2013 at 16:59
  So, what do we want Americans to spend their money on if not health. Isn’t it a natural extension of having a very high GDP?
  What I get form this graph is that Americans spend more money on health care than others and have one of the highest life expectancy rates. You can’t say more based on the data given. You can’t conclude the health care system is not working from this graph.
  A Canadian surgeon/doctor said the US provides 70% of the World’s medial innovation. I don’t know what the real number is, but I like medical innovation. I like non-life expectancy extending innovations.
  [top]
  - by joe on November 22nd, 2013 at 00:19
    Well under the assumption that the graph has no serious flaws then the point would be that the U.S spends so much more without much benefit. You ask rightly, what should the U.S. be spending this money on if not healthcare? The answer? Anything else. Well, anything else if one of two things are true:
    1) This is pure waste. If we can spend half as much like other countries and get the same results then we SHOULD NOT be spending this on health care.
    2) The marginal gains do not justify the cost. If it cost an additional trillion dollars to keep a person alive for another year, CLEARLY the answer to ‘What else should we spend it on?’ is obvious.
    But you do bring up the interesting point of innovation. America is paying a lot to innovate. So the question is if those innovation dollars are really worth it when comparing it to the opportunity cost. If all our money is going to the newest patented asthma drug that is just as good as the one 10 years old, then we are clearly overspending.
    I too like life-expectancy extending innovation, but it’s not clear that it’s our health care system that’s delivering that. We have a fantastic university system, the richest nation in the world, and the third most populated, you can bet a whole lot of that 70% innovation would still exist even under a single payer system, and to the extend that we lost some innovation and made other countries pick up the tab, Maybe that’s worth it.
    [top]
- by Floccina on November 21st, 2013 at 18:07
  We should emulate Italy?
  
  Per Capita Medical Spending and Life Expectancy by US State
  Med. Spending Life expectancy
  Utah 5,031.00 80.20
  Arizona 5,434.00 79.60
  Georgia 5,467.00 77.20
  Idaho 5,658.00 79.50
  Nevada 5,735.00 78.10
  Texas 5,924.00 78.50
  Colorado 5,994.00 80.00
  Arkansas 6,167.00 76.00
  California 6,238.00 80.80
  Alabama 6,272.00 75.40
  Virginia 6,286.00 79.00
  …
  Alaska 9,128.00 78.30
  Massachusetts 9,278.00 80.50
  District of Columbia 10,349.00 76.50
  
  Nah that is too hard surely the rest of the USA is more like Utah and Colorado than like Italy so lefts just emulate Utah and Colorado.
  Yeh Aaron you and I have solved the problems with healthcare in the USA. Cool.
  [top]
- by Floccina on November 21st, 2013 at 18:31
  It looks like Russia is also an outlier:
  The new mixed economy Russia has switched to a mixed model of health care with private financing and provision running alongside state financing and provision. The OECD reported that unfortunately, none of this has worked out as planned and the reforms have in many respects made the system worse.[76] The population’s health has deteriorated on virtually every measure. The resulting system is overly complex and very inefficient. It has little in common with the model envisaged by the reformers. Although there are more than 300 private insurers and numerous public ones in the market, real competition for patients is rare leaving most patients with little or no effective choice of insurer, and in many places, no choice of health care provider either. The insurance companies have failed to develop as active, informed purchasers of health care services. Most are passive intermediaries, making money by simply channelling funds from regional OMS funds to healthcare providers.
  Main source: OECD: Health care reforms in Russia
  [top]
  - by progrolib on November 22nd, 2013 at 11:49
    It sounds like the Russians have been trying to replicate the old US health care system!
    [top]
- by Blissex on November 27th, 2013 at 16:44
  This graphs and several others are explained in detail in a book by Prof. Angus Deaton, “”The Great Escape” (from early death and from poverty).
  He points out that good health depends mostly on investment in a clean, healthy environment, rather than in medical care, and that many of the most effective bits of medical care that works are cheap.
  But still healthy water, unpolluted air, good sewers, cleaning, unspoilt and varied food are by far the largest causes of improvements in health throughout recent history.
  In it is a very good book as far as it goes, and it goes pretty far, explaining in an easy way simple principles that most experts know. It has a couple of enormous flaws, the biggest of which is the omission of the role of cheap, cheap oil in making median incomes grow.
  [top]