United States Health Care Spending

The vast discrepancy between what we spend on health care and what every other prosperous (or not-so-prosperous) country spends on health care–and the little good it does us–is so well-known that it’s not going to change any minds when it comes to health care reform. Opponents of reform have come up with their rationalizations (more spending on technology, someone has to subsidize cheap drugs for the rest of the world, etc.), some of which contain grains of truth. But even if people aren’t listening any more, that doesn’t make it any less true.

Ezra Klein brings us the latest reminders. Here’s the most amazing graph from National Geographic:

That’s a clever trick, putting the outlier above the title of the chart. I’ll have to try it sometime.

By James Kwak

54 thoughts on “United States Health Care Spending

  1. Obvious question: is that at PPP or market-exchange rates?

    Otherwise very interesting (says someone from the most average country on the chart – the UK).

    I’m also curious as to why they haven’t stretched the right hand scale as far as the left one (even excluding the USA as an outlier). This has the net effect of making Switzerland (2nd in both rankings) look bad, having a fairly steeply downward sloping line.

    Also, what is Hungary doing wrong – does anyone have any insight (perhaps from anecdotal experience?)

  2. Wow, talk about using graphics to lead one to a very misleading impression. I just lost a lot of respect for NG.

  3. Because life expectancy is affected by a host of other variables (especially lifestyle choices) that have nothing to do with the quality of healthcare being delivered. If my memory is correct, when you look at actual health-care outcomes (e.g., cancer survival rates), the U.S. has better results than most other countries, including those with universal health coverage.

  4. Because the scaling is ill-adjusted. Almost all countries are “pointing” at the middle. If (ignoring the USA) you set the best and worst outcomes and inputs level, you’d have a more informative graph.

  5. Here’s some interesting additional data from the NG staff, which was excluded for clarity and redundant/missing data purposes:

    HEALTH CARE SPENDING (per person in U.S. dollars)
    Norway: $4,763
    Netherlands: 3,837
    Belgium: 3,595
    Germany: 3,588
    Ireland: 3,424
    Iceland: 3,319
    ————-(OECD average: $2,986)
    Greece: 2,727
    Italy: 2,686
    Turkey: 618

    Italy: 81.2
    Iceland: 81.2
    Norway: 80.6
    Netherlands: 80.2
    Germany: 79.8
    Ireland: 79.7
    Belgium: 79.5
    Greece: 79.5
    ————-(OECD average: 79.2)
    Turkey: 72.1

    Belgium: 7.6
    Germany: 7.5
    Iceland: 6.5
    Netherlands: 5.7
    Turkey: 5.6
    Italy: no data
    Norway: no data
    Ireland: no data
    Greece: no data

  6. On consideration, they may have put the 1st and 3rd quartiles alongside (but they haven’t marked it to show that’s what they’ve done).

  7. Cancer survival rates are higher in the US, because we love throwing as much money at these problems as possible. But there’s a reason why cancer, a relatively expensive and often financially bankrupting disease, is often cited by “best health care system in the world” proponents, because it’s one of the highest cost, most studied diseases in the US.

    America still has one of the highest bankruptcy rates due to health costs, one of the highest preventable death rates, etc.

    As for the scale of the graph, It looks good to me. Putting the U.S. way outside as an outlier may be provocative and visually making a statement, but I would say it’s a good way to deal with an outlier pretty far out. The scales are consistant, which is a favorite tweak of misleading graphs, i.e., the distance between mexico and the swiss is the same as the U.S. and swiss, as it should be.

  8. Fair enough, subject to two caveats:

    (1) our willingness to spend money on cancer research and treatment is not necessarily a bad thing if it produces results; and

    (2) that still doesn’t answer the criticism that life-expectancy is very blunt measure of the effectiveness of a country’s health-care system.

    I’m not trying to defend the current system because I believe there are significant reforms that need to be made; but to the extent we are criticizing the chart, I think the use of life expectancy is misleading.

    System-wide, we spend money on seemingly superfluous treatments; but, as much as anything, that is a result of distortions in our system for financing health care. For example, we call our current health-care insurance system “insurance,” but in reality it operates more like a pre-paid medical plan in which we pay amazingly high premiums to avoid bearing the costs of even run-of-the-mill procedures that ordinarily would not be covered by an “insurance” scheme.

    Again, please don’t think I’m defending the current system; but arguments for reform should not rely on misleading, though very creative, graphics.

  9. Interesting that conservatives are willing to have us pay high prices for pharmaceuticals to “subsidize cheap drugs for the rest of the world,” but unwilling to allow us to subsidize health care for those who cannot afford it.

    Is it possible that what’s doable depends on whose ox gets gored?

  10. This should be a scatterplot. This type of infographic should always be a scatterplot. It is just far more informative.

  11. Fully understood. Given how far the U.S. cost is outside the norm, almost any indicator you put on the right side will be dwarfed, though.

    What other metric would you use, other than Life Expectancy?

  12. Sorry I do not understand how Mr Kwak could say the information is well-known (pls see dailyhowler.com).

    This information is unheard of for most people and rarely published.

  13. Nah, I don’t have to propose any affirmative solutions of my own; I’ll just take pot-shots at other people’s work. ;-)

    It’s a good question, and I don’t necessarily have a good answer. Wouldn’t it be possible to take various performance metrics (not only outcomes like disease survival and successful organ transplants, but also customer-service metrics like waiting time to see specialists) and create an index to track actual medical outcomes and service? I realize that such an index might not present a complete picture of our health-care system, but it would help narrow the analysis to measuring what we get in exchange for what we pay.

    I suppose that my concern about using life expectancy is that politicians and pundits alike are using as a proxy to justify the current reform efforts; and not only does it seem like a poor proxy, but the current reform proposals (e.g., the so-called “Cadillac tax”) don’t really seem particularly well crafted at remedying the underlying problem.

    I try to take a fairly humble, agnostic approach to all this, because I’m not a health-care professional, and my economics credentials are limited to a B.A. (on the basis of which I spout a lot of B.S.). While the point about life expectancy is well taken, it’s worth challenging ourselves to ask whether there isn’t another, potentially better, metric. (I cynically feel as though politicians and reform-advocates have latched onto blunt measures like infant mortality and life expectancy because they make for a good “scare points,” not because they are the best measures of the quality of medical care in the U.S.)

  14. Keep in mind that the US is subsidizing the rest of the world’s pharma spending. Countries with single payer systems are able to negotiate lower costs. In effect the US is paying for most of the pharma R&D.

    Another issues is Diabetes. 1/4 of all medical spend in the US is on diabetes related illnesses. I don’t know what the numbers are globally but I would bet that rationing really kicks in with elderly diabetes patients.

    As for life expectancy,,, Its a really bad metric largely due to the nature of our immigrant population. Many immigrants are coming from countries with much lower life expectancies which then actually improve they enter our health care system. A more interesting metric would be life expectancy delta with native born Americans weighted at 0.

  15. Canada, France and Australia also have large immigrant populations. Shouldn’t their stats be affected ?

  16. Would someone post specific concrete examples of ways to reduce health care spending.

    Saying we spend too much and should spend less is not a solution.

    I submit taxpayer paid tuition for medical school students(This would increase supply of doctors?) and the elimination of punitive damages.

  17. The most comprehensive data-set about health-care outcomes I’ve seen is this OECD publication:


    It includes data about “avoidable complications” of diabetes, specifically leg amputations. The US has by far the worse rates of any OECD country. According to the WHO, quoted in the accompanying commentary, 80% of those could have been prevented, given good long-term care. In other words, the fact that the US spends a lot on diabetes seems to be negatively correlated with the quality of American diabetes care.

    As for cancer treatment, according to the OECD survey the US is ahead in one type of cancer, but not in general. And the differences among the top performers seem very small in each case — nowhere near directly proportional to associated spending.

    The US is indeed first in the five-year survival rate for breast cancer (90.5% for the US, to Iceland’s 88.3% and Canada’s 87.1%). But it is not ahead in survival of either colorectal cancer (Japan), or cervical cancer (Finland). Again, however, the associated differences among the top performers are very small.

    Another striking statistic, concerning life-expectancy. Life expectancy at birth in the US is well-below average, among all OECD countries. Life-expectancy at age 65 in the US, though, is dramatically improved. Now, I’m no statistician, but it’s hard not to notice that what happens to Americans at age 65 is Medicare.

  18. PhillyRunner,

    Actual results are quite mixed, even when it comes to cancer care. You can find very comprehensive statistics here, from the OECD:


    The US tops the list for breast-cancer five-year survival (90.5%, to Iceland’s 87.3 and Canada’s 85.6%). But it does not do so for either colorectal cancer (Japan) or cervical cancer (Finland).

    As for other ilnesses, especially those requiring long-term care, the US outcomes are often rather worse than those of other wealthy OECD countries.

    But, do have a look at the OECD tables. There is a lot of extremely interesting information.

  19. FIQ,

    I saw your other post, and I appreciate you providing that information. I’m going to look forward to reviewing it in detail.

    Thanks for giving me some knowledge.


  20. What no one has mentioned here, and what is truly disturbing. Even a cursory analysis of the pending reform legislation will actually exascerbate the average cost, rather than reducing it, although at worst it should be flat-lining (that is, for the many to be included who are young and healthy, those would be offset by those who have uninsurable health at present). What’s truly tragic, is that it does almost nothing to control premiums which insurers charge, and we will be gamed to death.

    And, insofar as pharma manufacturers go, most other countries not only allow the import of drugs from the US, but encourage it (to control their manufacturers), but here, we don’t allow import of drugs from overseas (because of our pharma’s powerful lobbies and campaign contributions, Congress consistantly has voted against this with all kinds of contrived arguments). So, we pay as much as 100 times as much as other countries for the drugs sold here.

  21. Not sure Hungary is doing anything wrong, but the former eastern block countries’ ranking on healthcare spending is consistent with their GDP per capita.

  22. The other thing that this graph does not demonstrate is the “consumer component”. What variance between countries is due to ethnic, genetic, and lifestyle choices versus quality of medical care? There are some statistics regarding obesity but it is still unclear the health impacts of bad habits, especially in the US. This graph is not an apples to apples comparison because diet, exercise, and other lifestyle choices are not constant in, say, the US versus Sweden. I would argue that a major reason for the inefficiency of the American healthcare system is the lack of accountability. If I smoke, drive a motorcyle without a helmet, eat McDonald’s every meal, and practice unprotected sex (all a disturbingly large percentage of Americans) how does that skew the above graph? Rhetorical question, obviously. To me, as a healthcare provider, this graph means nothing other than we spend a lot of money on healthcare. Actually, it means we spend a lot of money on healthcare related industries (ie insurance, drugs, etc). I’m biased because I know that providers get 20% of the till.

  23. Edward Tufte (author of the classic “the visual display of quantitative information”) would be screaming at this useless plot. The lines connecting the left axis and right axis serve absolutely no purpose other than to display the frequency of visits per year as the thickness of the line. This should have been a scatterplot with variable plotted point diameters. You’d still get the desired effect – US spending is off the chart – without the horrible visual clutter.

  24. Franken amendment puts a floor of 85% (75% on individual plans) of premiums that must be spent on health care. This doesn’t control costs, but it does cap profit + overhead.

    Does it make the final cut?

  25. how about following the NHS model of the independent board that sets standards for treatments? From what I understand they mandate the types of treatments funded by the NHS where the one being proposed in the I believe this article has been linked here before, but the Atul Gawande article a few issues back in the New Yorker does a realy good job walking through some of the cost saving pilot programs in the Senate bill. The basic gist of the article is “there’s no clear cost saving silver bullet currently so pilot as much as you can and go from there”. He draws parallels with how this model was the same government spurned innovative engine that helped with US agricultural issues at the turn on the 20th century. It’s a good read if you have the time and speaks to this more clearly and rationally than most things I’ve seen on the topic:

  26. oops… I changed my post up above and forgot to delete the first couple of lines. I rarely post here and when I finally do I completely screw it up :)

    Oh well.

    Here’s what I meant to post:

    I believe this article has been linked here before, but the Atul Gawande article a few issues back in the New Yorker does a realy good job walking through some of the cost saving pilot programs in the Senate bill. The basic gist of the article is “there’s no clear cost saving silver bullet currently so pilot as much as you can and go from there”. He draws parallels with how this model was the same government spurned innovative engine that helped with US agricultural issues at the turn on the 20th century. It’s a good read if you have the time and speaks to this more clearly and rationally than most things I’ve seen on the topic:

  27. I understand how this graph is skewed from a pure ‘apples-to-apples’ comparison when looking at life expectancies of other countries {we smoke, they don’t / we are fat, they are not / etc, etc} but I think viewing this that narrowly misses something that this graph says {to me at least}.

    There is something deeply flawed with how far out of whack our society is versus lots of other societies out there. If we have rough habits, then why don’t other countries have the same? Do we have too much ‘rugged individualism’ to the detriment of our society? Why are we SO fat? At a base level, what the hell is wrong with America’s overall health I guess?

    People flipped out when it was floated to put a 3 cent tax on sodas… say what? What’s so objectionable about that? Put a 50 cents tax on it, maybe it’d reduce diabetes and save us some money not counting what we’d collect to put towards things that could help the country. Obviously I have a liberal point of view here but some of the inherent issues related to the general ‘health of America’ don’t seem like they are ready to be solved by the market anytime soon.

    Pass this f-ing bill already and get this thing going. According to the CBO I won’t be affected by it anyways {I have private insurance so my premiums will stay the same} unless me or any family members become catastrophically ill – in that case a private insurer can’t rescind my policy and there won’t be a cap on what they pay out towards my medical treatment so I won’t go bankrupt.

  28. Health Care Spending
    I agree this chart opens our eyes even more on this
    For your info, although the spending in Switzerland is
    lower, everybody complains about the high cost of health care here. But then everybody seems to believe that, having paid so much for the health insurance policy, he is entitled to the very best treatment available.
    This created a situation in which all the main players, i.e. pharmaceutical companies, hospital, doctors and patients are accusing each other of cheating. This discussion goes on and on with no real result except higher health costs every year.
    My question to you is the following:
    Since we are living in a FIAT money world with an eagerness to increase the GDP figure every year, does it really matter whether we spend this paper money on health matters or any other goods?
    Thanks for your thoughts!

  29. I would argue that a major reason for the inefficiency of the American healthcare system is the lack of accountability. If I smoke, drive a motorcyle without a helmet, eat McDonald’s every meal, and practice unprotected sex (all a disturbingly large percentage of Americans) how does that skew the above graph? Rhetorical question, obviously.

    Right, its a good thing that people don’t smoke in othe countries or drive unsafely (obviously foreigners don’t have motorcyles so there’s no helmets to worry about). If McDonalds ever gets around to start opening franchises overseas or worse if Amerca ever loses its monopoly on unprotected sex, the whole planet is doomed.

    Oh and you have accountability ass-backwards. Here in the US, citizens are responsible for securing their own healthcare. In every other industrialized country, elected officials are held accountable if quality medical care is not provided to its citizens. By your logic, Britain with its “free at the point of service” NHS would have the highest costs and the worse outcomes, when in fact Dominic J and his fellow Britons are hugging the median for both costs and outcomes.

    Finally, life expectancy is the best measure for healthcare performance for exactly the same reason that homicide rates are the best measure of police department performance– its easy to fudge the numbers for minor things (e.g. “customer service metrics” or “property crimes”), not so easy to fudge a death certificate.

  30. Interesting chart, regardless of the flaws. I live in the UK, but lived in the US most of my life.

    My personal experience is that the UK system is vastly superior to the US one, especially since we were members of an HMO for years – I spent hours waiting to see a doctor for 2-3 minutes at our HMO. Despite the many problems of the NHS, the overall experience is generally much more humane than the American system. It’s nice to see doctors who actually look you in the eye, ask questions, and seem genuinely interested in your answer.

  31. Thanks for the great link.
    My own take on the chart is the steepness of the lines between spending and life expectency. Just eyeballing it, it appears South Korea knows how to extract value with regard to health care.

    The truth of the matter is, lowering the amount we spend by 80% perconet would have almost no measureable effect on outcomes. As extending a child’s life has a disproportionate effect on the average, we could raise life expectance easily while cutting the cost. But 2 year old’s don’t vote.

  32. Let’s put it this way—it’s well known to people who don’t watch reality TV and people who don’t think a statue of Ronald McDonald is a landmark worthy of taking a photo together with.

  33. The only thing worse than the graph is thinking that it’s because America is stupid, lazy, and careless. If it’s true, health care would be a problem far behind the collapse of our society as we know it.

  34. Wow. Good job of missing the entire point. In America, every person wants every last possible expensive medical treatment for their stage 4 lung cancer regardless of expense. In other countries with more cost efficient systems, the public has a greater grasp and acceptance that unlimited health costs for a limited benefit is not sustainable for society. Perhaps you’ll recall the outcries about “rationing of care” and “death panels”. My point is that Americans make bad choices. Yes, people in other countries make the same bad choices, Mr. Obvious. However, in America, we still expect every last heroic, expensive measure to treat our stage 4 lung cancer following 30 years of smoking. Who cares how much it costs society? That, sir, was my point regarding accountability.

    BTW, life expectancy is not the best metric for healthcare performance and homicide rates are not the best metric for police departments. Do you think the police department in Red Oak, Iowa is better than the police department in New Orleans just because the homicide rates are lower? As with most systems, accounting for the independent variables and then following change is the best metric for quality. Change in life expectancy over time (and change in homicide rate over time) is a much better way of looking at the impact of healthcare or police work on a population. James, can we see that graph over the past, say, 20 years? I think you’ll find some of the same information (ie American healthcare is stagnant) but at least it means something.

  35. The central problem with what has passed for the “debate” about healthcare in the U.S. – and in many of the interesting comments posted here, is what separates the U.S. from most of the rest of the OECD countries (all of which perform better on standard health measures than the U.S.). All but the U.S. are social democracies characterized by parliamentary government (i.e. corporatist welfare states rather than as is U.S. a liberal welfare state). We are trapped in an ideology of atomistic individualism (or possessive to use McPhearson’s term), a kind of millenialism with the Framers as bringers of the golden tablets). Jeffersonian anti-statism is there from the origins and, for the last thirty years, anti-statist rhetoric has ruled both parties and dominated public discourse (such as it is). It is no wonder that the U.S. dis-invests in its citizens while trumpeting the values of ‘unfettered free markets’. We have dis-invested in infrastructure, in human capital (where else in the OECD do students have to take on the debt load to get a post-secondary education?)-the only major investments are in weapons (we outspend the rest of the OECD by more than 10 to 1)and risky financial ventures.
    The “healthcare problem” is not to be resolved by recourse to the usual blaming of individuals (e.g. a tax on the obese, or shooting diabetics) but by examining how the U.S. is failing as a modern state and society.

  36. Strikes me as odd that supporters of healthcare reform do not mention reducing the large amount the US spends of medical research compared to other countries. The budget for the NIH alone is $30 B; the nonprofit and private sector spending on research is even more, probably about an additional $60B. I remember the UK gov’t set a goal a few years ago to spend £1Billion on medical R&D… but I don’t think they’ve reached that level yet.
    If healthcare r&d were reduced, there would be a smaller pipeline of expensive technolgies in future years. I’m not sure I support this idea, but it seems to me a logical discussion point for the reform minded.

  37. Small quibble, edw: British GDP is about 1/7th of US GDP, so obviously you can’t expect anything close to parity. And, presumably, the UK Government number you quote should be compared to the NIH number, not the total R&D number (public and private) that you estimate.

    If you make all these adjustments, UK government spending on R&D comes out slightly less than half of comparable spending in the US — still a significant difference, but not nearly as spectacular as the raw numbers you quote make it out to be.

  38. Sorry, I should have said “slightly less than a quarter” — not “slightly less than half”.

  39. The discrepancy between US and Mexico is incredible. Is it possible that the US spends more on health care for Mexicans than Mexico does?

  40. The St. Petersburg Times PolitiFact.com reminds us that under the proposed healthcare reform bill, healthcare costs will continue to rise.

    “So we’ll be spending more over the next 10 years, even more than if we didn’t pass the bill, according to the CMS report. But let’s talk a minute about the “rate of growth” of health care spending. That’s what Obama is talking about when he says he wants to “bend the cost curve.” According to the projections in the CMS report, the rate of growth in national health care expenditures will spike a bit in the middle years of the health reform plan. That’s when the provisions that result in 33 million more people getting health insurance will kick in. But in the later years of the 10-year projections, the rate of growth rises slower than under current law (albeit only slightly: in 2019, for example, it’s the difference between rising at a clip of 6.9 percent under the proposed plan as opposed to 7.2 percent under current law).”


    In the next 10 years we’ll have to get US Average Life Expectancy at Birth well above 100 years in order to get a return on spending equivalent to the rest of the world.

  41. US life expectancy would be >80 if not for our homicide and vehicle death rates. Life expectancy at birth is an insufficient proxy for quality of health care system while ignoring causes of death.

    No information added by psuedo X-axis. Should be a scatter plot.

Comments are closed.