Skip to content
 

Healthcare spending and life expectancy: a comparison of graphs

Yesterday I posted this graph, a parallel-coordinates plot showing health care spending and life expectancy in a sample of countries:

6a00e00982269188330120a76420ea970b-500wi.jpg

I remarked that a scatterplot should be better. Commenter Freddy posted a link to the data–you guys are the best blog commenters in the world!–so, just for laffs, I spent a few minutes making a scatterplot containing all the same information. Here it is. (Clicking on any of the graphs gives a larger version.)


healthscatter.png

(I was able to make the circles gray thanks to the commenters here.)

How do the two graphs compare? There are some ways in which the first graph is better, but I think these have to do with that graph being made by a professional graphic designer–at least, I assume he’s a professional; in any case, he’s better at this than I am! He also commented that he removed a few countries from the plot to make it less cluttered. Here’s what happens if I take them out too:

healthscatter2.png

(Unlike the National Geographic person, I kept in Turkey. It didn’t seem right to remove a point that was on the edge of the graph. I also kept in Norway, which was the highest-spending country on the graph, outside the U.S. And I took out Sweden and Finland–sorry, Jouni!–because they overlapped, too. Really, I prefer jittering rather than removing as a solution to overlap, but here I’ll go with what was already done in this example.)

What the scatterplot really made me realize was the arbitrariness of the scaling of the parallel coordinate plot. In particular, the posted graph gives a sense of convergence, that spending is all over the map but all countries have pretty much the same life expectancy–look at the way the lines converge to a narrow zone as you follow the lines from the left to the right of the plot.

Actually, though, once you remove the U.S., there’s a strong correlation between spending and life expectancy, and this is super-clear from the scatterplot.

The only other consideration is novelty. The scatterplot is great, but it looks like lots of other graphs we’ve all seen. This is a plus–familiar graphical forms are easier to read–but also a minus, in that it probably looks “boring” to many readers. The parallel-coordinate plot isn’t really the right choice for the goal of conveying information, but it’s new and exciting, and that’s maybe why one of the commenters at the National Geographic site hailed it as “a masterpiece of succinct communication.” Recall our occasional discussions here on winners of visualization contests. The goal is not just to display information, it’s also to grab the eye. Ultimately, I think the solution is to do both–in this case, to make a scatterplot in some pretty, eye-catching way.

P.S. I never know how much to trust these purchasing-power-adjusted numbers. Recall our discussion of Russia’s GDP.

P.P.S. And here’s the R code. Yes, I know it could be cleaner, but I just thought some of the debutants out there might find it helpful:

The first step was to download the data, open the Excel file and save the relevant data matrices as three .csv files. I don’t know how to extract information from .xls files in R, so I went into Excel and did that part manually. After that, I did the following:

library (arm)

read.page0) {
      years.keep

Sorry about all the blank lines; that’s just what “pre” does in html.

40 Comments

  1. http://models.street says:

    Andrew: regarding blank lines, it's actually not what <pre> does, there are many tags (newlines) put there by your blogging software. If you go in and edit the post by removing formatting from that section you might be able to recover the un-blank-lined version in a nice way.

    To harp on my favorite topic of nondimensionalization of models, I would like to see this data where the dollar figures are divided by per capita gdp for that country in dollars (or better yet, just use the two figures in their original currencies to avoid the additional confusion of purchasing power parity.

    Both sets of data are interesting, absolute quantity of resources used is one question, but another question is fraction of average income used. It isn't a big deal if you're spending say $5000 per year if the average person makes $50,000 as compared to a country where people spend $500/yr but the average income is only $2000 for example.

    An even better measure would be to divide healthcare dollars spent by some measure of minimum "cost of living" such as the annual budget for a healthy diet in that country. A 3D plot could plot the following data points:

    (HC$/Capita/yr) / (Food$/Capita/yr)
    (GDP/Capita/yr) / (Food$/Capita/yr)
    (Life Expectancy / Life Expectancy population weighted Avg)

    These are 3 nondimensional variables that all have something to say about spending and outcomes.

    Does anyone know where I can get GDP/Capita and/or Food$/Capita data? I would be happy to create such a graph if I could get the data reasonably easily.
    </pre>

  2. http://models.street says:

    argh, again the comment swallowed my <pre> tag and my tag

  3. Ben Hyde says:

    "a strong correlation between spending and life expectancy" … 200% more get maybe 7.5% more years, or $25K over a lifetime buys another year.

    Dropping six points confuses much of the effect.

  4. Shaun says:

    Some obvious comments:
    I've seen this data (or similar) presented numerous times over the past few months, and it is often accompanied by the explicit commentary that US citizens don't get bang (in terms of life expectancy) for their buck. I wonder has anyone constructed a cleaner dataset where only people who have spent their entire lifetime within the system are examined? This would exclude immigrants in the US, for example. Better yet, could we see life expectancy figures for only those who have had some form of health insurance throughout their life? That would be a fairer measure of what subscribers get in each country. Finally, the spiraling cost of US healthcare is a relatively recent phenomenon, and yet it is being compared with a metric that takes decades to change. How do we know that the present excessive cost of US healthcare will not translate into higher life expectancy in the future?

  5. Bob O'H says:

    There is the xlsReadWrite package which can be used to read Excel spreadsheets into R (in Windows). I don't think there's anything on CRAN that can do it, which is a shame.

    Also, could you use 3 letter codes for the countries, to reduce the clutter?

    Oh, and some of your medical sciblings might be interested in the graph too, so they can play "spot the outlier".

  6. Paul says:

    This graph seems to be preaching to the converted in that it considers life expectancy over the entire US population. However, some advocates of the current system (such as on Fox News) do not think that universal coverage should be a right. As mentioned by shaun, it would be interesting to see where the US plots after restricting in some way to covered individuals. It would also be interesting to see where those that are not covered plot — my understanding is that people who are not covered are often treated at emergency which, at least for the hospital, would be expensed at a high rate and then perhaps later written off.

  7. Thorfinn says:

    You could probably construct a similar graph for education. America spends far more on education than other countries, yet has worse outcomes. Could we cut the education budget by a half and do all right then?

    One comparison I'd like to would be to compare Swedish with Swedish-Americans, and so forth. I suspect common ancestry predicts life expectancy better than current country of residence.

  8. freddy says:

    glad you found the link helpful.

    I'd thought about making a parallel coordinates plot of lots of the (similar) variables contained in the dataset – because I was (slightly) suspicious that the original graph's variables were cherry-picked for maximal US-outlyingness.

  9. jtt says:

    There are several ways to read data in R directly from Excel. See the R Wiki at http://wiki.r-project.org/rwiki/doku.php?id=tips:… You can even copy the data to clipboard and read it in R from there. Personally, I like using the RODBC-package the most:
    <pre>
    library(RODBC)
    tot.exp=sqlFetch(odbcConnectExcel(
    "OECDHealthData_FrequentlyRequestedData.xls"),
    sqtable = "Total expenditure, % GDP", na.strings = "NA",
    as.is = T)
    odbcCloseAll()
    </pre>

  10. Simon Johnson says:

    Andrew,
    Do have any suggestions why Denmark has appears to have lower life expectancy than other European Union countries?

  11. Oskar Shapley says:

    I don't know about other people, but I'm seeing DEA in the graph. There's an interesting production technology frontier, where countries get the best results for money spent. I'm seeing who is inefficient and who is close to what is technologically possible.

    Obviously the USA is the worst efficiency performer, which actually can be calculated with this data and an inefficiency measure could be made public.

  12. Oskar Shapley says:

    I have too much time on my hands:

    http://i875.photobucket.com/albums/ab313/oskarsha

    #2006 data, except Turkey, where healthexpenditure is 2005

    hea = structure(list(healthexpenditure = c(3167L, 3608L, 3356L, 3696L, 1535L, 3357L, 2709L, 3423L, 3464L, 2547L, 1457L, 3207L, 3001L, 2673L, 2581L, 1491L, 4162L, 777L, 3611L, 2398L, 4507L, 920L, 2150L, 1322L, 2466L, 3124L, 4165L, 618L, 2885L, 6933L), lifeexpectancy= c(81.1, 79.9, 79.5, 80.7, 76.7, 78.4, 79.5, 80.7, 79.8, 79.6, 73.2, 81.2, 79.8, 81.4, 82.4, 79.1, 79.4, 74.8, 79.8, 80.1, 80.5, 75.3, 78.9, 74.3, 81.1, 80.8, 81.7, 73.2, 79.5, 78.1)), .Names = c("healthexpenditure", "lifeexpectancy"), class = "data.frame", row.names = c("Australia", "Austria", "Belgium", "Canada", "Czech Republic", "Denmark", "Finland", "France", "Germany", "Greece", "Hungary", "Iceland", "Ireland", "Italy", "Japan", "Korea", "Luxembourg", "Mexico", "Netherlands", "New Zealand", "Norway", "Poland", "Portugal", "Slovak Republic", "Spain", "Sweden", "Switzerland", "Turkey", "United Kingdom", "United States"))

    #uncomment this line to use log of expenditure, is better
    #hea[,1] = log(hea[,1])

    remove .999,])

  13. Peter Løhmann says:

    Short answer: Tobacco and alcohol.

    When you compare with Sweden, which in many ways resemble Denmark, tobacco and alcohol consumption explain 80 % of the difference in life expectancy among the two countries.

    Source:
    http://www.ugeskriftet.dk/portal/page/portal/LAEG
    (in Danish)

    English summary from said article:
    Summary
    Life expectancy and mortality in Denmark compared to Sweden. What is the effect of smoking and alcohol?

    Ugeskr Læger 2008;170(33):2423-2427

    Introduction: For many years life expectancy in Denmark has improved less than in other comparable western countries, e.g. Sweden. An unhealthy life style, in particular the consumption of alcohol and tobacco, has often been mentioned as a possible explanation.

    Materials and methods: Life expectancy and mortality in Denmark and Sweden has been compared by means of nationwide cause of death registries. Alcohol- and tobacco-related deaths are defined from death certificate diagnoses. The comparisons between the two countries are made by age standardised mortality rates and life expectancies for the period 1997-2001.

    Results: 50 years ago Denmark had one of the highest life expectancies in the world, but is now at the bottom of the list when compared to similar countries. Life expectancy in Sweden is now almost three years longer than in Denmark. Before the age of 75 there were a total of 3700 premature deaths among Danish men and 3400 among Danish women. Relative excess mortality was highest among Danish men aged 35-64 with a relative excess mortality at 40-50%. Among women excess mortality was 50-60% in the age group 35-74 years. Overall, alcohol and smoking account for almost the entire difference between Danish and Swedish men and for 75% of the difference between Danish and Swedish women.

    Conclusion: A very substantial part of the Danish excess mortality and low life expectancy compared to Sweden can be attributed to high mortality related to alcohol and tobacco consumption. A reduction of this difference in life expectancy does not seem realistic without a reduction in the consumption of tobacco and alcohol.

  14. Bill D says:

    Life expectancy is a terrible, terrible proxy for the efficiency of health-care delivery systems. It is completely muddied by lifestyle and other factors (e.g. obesity, teenage births, accidents, ect) that are unrelated to a country's health-care system. The fact that we insist on using this measure without ever acknowledging its severe shortcomings is very troubling.

  15. Lauri Garber says:

    Rather than bang for our buck, Can you factor out how much buck we are paying for our bang? How much does gun ownership contribute to health care costs and life expectancy in this country? What would the graph look like without it?

  16. Anonymous says:

    I don't see any muddling in the graph. Except for the US there seems to be a more or less linear correlation between life expectancy and health care expenses. Its only the US that is the outlier. While there may be other causes of this discrepancy rather than "efficiency" problems, its clear the US has serious issues that directly reduce its peoples life expectancy.

  17. Charles says:

    Certainly life expectancy isn't a perfect measure of a country's health care system, but saying that it shouldn't be used as a measure is like saying a sports team's win-loss record doesn't reflect the team's talent. At the end of the day all of the other statistics are meaningless if we come up short in the ultimate measure of success (in this case life expectancy).

    I agree with you that other factors are involved, but why wouldn't obesity and teenage pregnancy be considered aspects of a country's health care system. A health care system should promote a healthy lifestyle, not just save us from an unhealhty one.

  18. NA says:

    Why isn't there a circle around Norway?

  19. Kevin says:

    I agree that life expectancy is not an ideal proxy for the larger health care debate. Different cultures demand different things from their health care system. For example, the French are not too big on dental care, whereas Americans are not content living out their life expectancy with subpar dental care.

  20. Ian says:

    People DO correct for lifestyle differences when comparing health outcomes and the results turn out the same.

    Play around with these OECD data to start with.

    My wife conducted this type of analysis for a major insurance industry advocacy group. They used really expensive proprietary data sets. To the advocacy group's dismay, after correcting for all sorts of demographic and lifestyle differences, the US still does poorly in terms of outcomes.

  21. A Siegel says:

    Another element to throw into the mix: the US dollar is near historic low levels against many of the currencies here. Imagine that (unlikely) the US dollar were again in parity range with the Euro. The US costs per capita vs delivered results would look even worse. (Nat'l Geographic post states that it relies on 2007 numbers. Don't see that it is reliant on purchase parity as opposed to straight currency exchange, but might have missed note(s).)

  22. William Ockham says:

    Because there is no data in the file for doctor visits per capita (which is what the circle represents) for Norway.

  23. Tim says:

    After I showed a friend the original National Geographic graph, he commented that the healthcare spending figures include some bizarre items. He claimed that for this graph, buying a diet Pepsi counts as a "healthcare expenditure" as do buying designer track shoes and sunglasses. I've been looking around to find any sort of list of what's included in the figures — specifically to see whether diet soda or track shoes or sunglasses are included in the U.S. figures — and I can't find anything to confirm or refute. Anyone have pointers?

  24. Ellen says:

    Lifespan is still a good metric, even if it is not perfect. Lifestyle, diet, pollution, exercise, obesity, etc. have a huge impact on health care costs. None of these variables are isolated in real life. We could reduce our health costs AND increase our lifespans if our government stopped subsidizing corn syrup.

  25. Andrew Gelman says:

    Tim: Followed the link to the data in the above blog entry and you can see more on the sources. I didn't see anything there about diet soda, track shoes (designer or otherwise), or sunglasses. Maybe your friend was joking.

    Thorfinn: It would be fascinating to see a similar plot of educational outcomes vs. spending on education. If you send me the plot, I'd be glad to post it on the blog. Some decisions would need to be made in defining the data–for example, does spending on child-care/creche/maternelle/preschool/kindergarten count as "education"–but I'm sure people have done some of these international comparisons.

  26. Lee says:

    Have you seen Gapminder?
    A powerful way to view the same data over time

  27. Shoe says:

    Great comment about factoring in/out gun ownership and deaths.

  28. USA says:

    Good point. Compared to most of the countries on the list, the US has higher levels of alcohol and tobacco consumption. Also, our subsidy and price support structure favors HFCS over sugar. We also have higher rates of violent crime and auto accidents than most of these countries.

    We also produce the lion's share of the world's medical advancements. The other health care systems are only good at distribution, not innovation. It would be sad if we sold out the incredible pace of medical advancements out of greed and the false belief that someone else owes us a living.

    This sort of analysis purposely ignores other factors which contribute to longevity as well as factors within a health care system that contribute to better outcomes. What is the median level of health care in various countries? This would remove some of the distortions that occur from the less even distribution of wealth in the US.

    The US has much better outcomes for cancer than other health care systems. Surely, outcomes are a better metric for health care than longevity because they are less affected by lifestyle choices and more directly related to care received.

    There is still time to stop Congress from passing a bill that will add thousands of dollars to the cost of health insurance for our nation's families. Whatever the deficiencies of our current system, we can't allow a misguided attempt at "reform" to put quality health insurance out of the reach of hard-working families.

    We spend more on health care in America because we are a rich and free nation. We have more health care options. Restricting freedom is not the answer.

  29. RCW says:

    Actually, compared to most of the countries noted, the U.S. has lower rates of alcohol and tobacco consumption – significantly lower rates. Additionally, the overwhelming majority of medical advances from the U.S. are paid for by federal research funds, and that money is not included in the cost of American health care as it is normally and typically calculated. Many (most) of those 'advances' are from drug companies and most of them aren't actually advances in a strictly real-world sense (even when they are not merely redundant productions designed to generate knockoff revenue) and most of them are produced by drug companies which are foreign-owned.

    In short, innovation is something at which we once excelled in a relative sense, but now that the rest of the industrial world has caught up and surpassed us both in research spending and education, that mantra is a self delusional exercise.

    There is a higher incidence of cancer and marginally better outcomes for cancer in the U.S. system (the one high point, in fact), but most of what you stated is simply not factual.

  30. mht says:

    What is the life expectancy of US citizens WITH health insurance!! (as compared to the total population)

  31. William Cox says:

    Here's the same data graphed by Gapminder.

    http://bit.ly/7AWr09

    <a href="http://graphs.gapminder.org/world/#$majorMode=chart$is;shi=t;ly=2003;lb=f;il=t;fs=11;al=30;stl=t;st=t;nsl=t;se=t$wst;tts=C$ts;sp=6;ti=2006$zpv;v=0$inc_x;mmid=XCOORDS;iid=tR3MM-UTZ0B44BKxxWeAZaQ;by=ind$inc_y;mmid=YCOORDS;iid=phAwcNAVuyj2tPLxKvvnNPA;by=ind$inc_s;uniValue=8.21;iid=phAwcNAVuyj0XOoBL_n5tAQ;by=ind$inc_c;uniValue=255;gid=CATID0;by=grp$map_x;scale=log;dataMin=10;dataMax=7154$map_y;scale=lin;dataMin=23;dataMax=86$map_s;sma=49;smi=2.65$cd;bd=0$inds=i239_t001995,,,," rel="nofollow"&gt <a href="http://;http://graphs.gapminder.org/world/#$majorMode=chart$is;shi=t;ly=2003;lb=f;il=t;fs=11;al=30;stl=t;st=t;nsl=t;se=t$wst;tts=C$ts;sp=6;ti=2006$zpv;v=0$inc_x;mmid=XCOORDS;iid=tR3MM-UTZ0B44BKxxWeAZaQ;by=ind$inc_y;mmid=YCOORDS;iid=phAwcNAVuyj2tPLxKvvnNPA;by=ind$inc_s;uniValue=8.21;iid=phAwcNAVuyj0XOoBL_n5tAQ;by=ind$inc_c;uniValue=255;gid=CATID0;by=grp$map_x;scale=log;dataMin=10;dataMax=7154$map_y;scale=lin;dataMin=23;dataMax=86$map_s;sma=49;smi=2.65$cd;bd=0$inds=i239_t001995,,,,” target=”_blank”>;http://graphs.gapminder.org/world/#$majorMode=chart$is;shi=t;ly=2003;lb=f;il=t;fs=11;al=30;stl=t;st=t;nsl=t;se=t$wst;tts=C$ts;sp=6;ti=2006$zpv;v=0$inc_x;mmid=XCOORDS;iid=tR3MM-UTZ0B44BKxxWeAZaQ;by=ind$inc_y;mmid=YCOORDS;iid=phAwcNAVuyj2tPLxKvvnNPA;by=ind$inc_s;uniValue=8.21;iid=phAwcNAVuyj0XOoBL_n5tAQ;by=ind$inc_c;uniValue=255;gid=CATID0;by=grp$map_x;scale=log;dataMin=10;dataMax=7154$map_y;scale=lin;dataMin=23;dataMax=86$map_s;sma=49;smi=2.65$cd;bd=0$inds=i239_t001995,,,,

  32. Paul says:

    Some of the interesting information is embedded in the graph.

    Note that the US is a heterogeneous culture while Japan, Korea, Mexico etc are much more homogeneous. No African or south American countries are included so there is the potetial for a large bias in the results shown. Also, we do not have spending by age.

    The graph shows that japaese live significantly longer that Czechs although the spending is about the same . So we can cocludevthat this ethnic component is significant. That means that we need to read in ethnicity into the US results. With about 65 percent euro, 15 African, 5 east Asia, 5 Arab or Indian, etc it looks like spending doesn't much to extend life. This is in agreement with what we know about the percent spent at what age so the conclusion is now opposite of what we might think by a naive view of the selected data.

  33. Paul says:

    Well, it does turn out that ethnicity in the US is a large determinant of life expectency, eg African is about 6years less than euro. So that factored on would tend to make the life expectency of the US more like that of Slovakia.

    So them a better question would be how much longer a certain population might live with more health care spending. The Czech and Slovaks could be comated at the low end as well as the swedes and nowegians at the high end. In the middle we could compare the Brits and Irish.

    There is a clear conclusion. Eat fish.

  34. just passing through says:

    I’m lucky enough to have been knocking around inside the belly of this beast for a few decades. I’ve done pharmaceutical research in an academic medical school, analysis of data in a drug company, and studies of outcomes and effectiveness of all sorts of things for health insurers. My family and friends feature a bunch of health care professionals, including not just doctors, etc. but also “alternative” providers, such as naturopaths. I’ve had personal and family experience with the healthcare systems in both the US and Canada. So I’ve naturally got some opinions.

    Stratifying life expectancy between countries by age clarifies things powerfully. (See the medical literature; the famous JAMA paper in the mid 90s, for example.) The US does relatively worst with newborns, increasing monotonically in relative quality with age, until we do in fact become best in the world with the very elderly. This explains quite a bit; firstly, it tends to reinforce the use of life expectancy as a metric as it eliminates hypotheses regarding the effect of lifestyle, firearms, etc. (very few newborns smoke, drink, are morbidly obese, drive while drunk, or get shot). Secondly, it explains the persistently low showing of the US with respect to overall life expectancy, as the inevitable result of the huge leverage of untimely death in infancy; assuming a potential lifespan of 80 years, for example, death of a newborn represents a loss of 80 years, equal in effect to the death of 16 75 year olds representing a loss of 5 years each, and so on up the age scale; healthcare money and effort in the US are expended in inverse proportion to the return. Thirdly, it explains our outlier status regarding cost; the commonly effective interventions in infancy (and pregnancy) tend to be extremely cheap items like vitamin supplements (the news items this week regarding folic acid deficiencies for instance), while the interventions required for the aged tend to be hugely expensive. Culminating in the vast efforts and expense spent in high tech end of life interventions which end up delaying death by a month or so. This, of course, is not unrelated to the propensity of American doctors to go into lucrative and glamorous specialty medicine to the neglect of boring and relatively low paying primary care, raising costs at the same time it reduces average life expectancy. Although this is partially due to medical students’ interest in prestige and avoiding boring routine in favor of interesting problems to solve, the role of salary differential certainly can’t be disregarded as a motivator; of course, the three factors are highly interdependent.

    A similar principle applies to the pharmaceutical industry; Big Pharma doesn’t get big by peddling multivitamins to impoverished pregnant teens, not when there are the aforementioned cancer patients whose health plans can be persuaded to pay for the newest drugs which feature 5% greater efficacy and 200% higher price. Not 200% higher cost, mind you, the cost of any drug being overwhelmingly determined by the fixed costs of keeping the company in business as apportioned to each product by marketing, not according to some assignment of research expenses relevant to each drug, but rather according to what pricing will maximize gross income.

    Not surprisingly (although I’m often surprised how many Americans even in some segment of the health care biz seem to be surprised by the numbers), this correlates pretty well with where the health care money goes. A bit more than 25% to the relatively infrequent inpatient costs, such as services of tertiary specialists, high tech imaging, IV chemotherapy, etc. (with relatively little going to the actual hospital or other facility itself) and a similar amount to pharmaceuticals (other than inpatient drugs), with the remaining less than 50% covering all the vast bulk of routine and quasi-routine medicine; office visits, outpatient procedures, same-day surgery, vaccinations, even the often scapegoated pair of emergency room costs and insurance company overheads. (Again, this is not unrelated to the average profit margin of 25% for pharmaceutical companies, versus less than 5% for health insurers).

    Of course, this is getting dangerously close to the untouchable third rail of health care cost containment; limiting physician salaries, whether directly (highly unpalatable; it’s not as though they’re executive officers in AIG, after all) or indirectly (waiving student loan repayments for medical students who go into primary care after graduation, for instance, and thereby settle for a lower salary). While the low end of physician salaries relative to the average wage is a bit higher in the US than other Western countries, the upper end in the US has gone off the charts with respect to other countries. And yet, highly paid specialists in Canada or Britain seem to do pretty well financially, despite this financial shortfall; they’re apparently not driven to emigrate to our land of milk and honey to escape their impoverished situation, most doctors who do immigrate to the US being on the lower end of the pay scale where the difference in income represents a more significant difference in lifestyle.

    People seem more comfortable discussing government regulation of pharmaceutical prices, although obviously there is a split regarding pro or con. Those who fear stifling of innovation etc. need not be troubled; I have no doubt the pharmaceutical industry would be able to adapt to the hardship of having their prize cash cow reduced to the profitability of the rest of the countries in their herd, rather than just close up shop. As observed above, the US is not the source of pharmaceutical research for the world as some seem to believe (have they never noticed that these companies all seem to have Swiss or German names, to begin with?); and there is no particular reason why the companies are required to do their research here just because we're their most profitable market. It's not as if they don't sell the same products worldwide at various prices, no matter where they were developed. The actual clinical trials themselves are increasingly conducted in the third world where costs are minimal. In any event, there is a vast literature waiting for anyone with any curiosity regarding such things as the dwindling productivity of the drug industry’s research investments, the increasingly sparse research pipeline of promising drugs, the overwhelming trend to develop “me too” drugs offering little or no advantage over existing drugs (other than marketability), or the industry’s increasing reliance on new innovations in marketing and in patent law rather than in medical research.

    Although what if anything needs to be done is open to debate, the parameters of the situation are pretty clear, despite "the best healthcare system in the world" becoming a political shibboleth. This might be a good place to point out that those elderly patients who are so well served by our healthcare system, in that age group where Americans do have best in the world life expectancies, all enjoy universal coverage via Medicare (despite its imperfections); i.e., socialized medicine. Ironic isn't it; American socialized medicine handles the toughest cases on a best in the world level, while our private system can't manage a mediocre performance on routine and simple care.

  35. Paul says:

    Well, there is Medicaid also so the US does have a socialized system at two ends of the spectrum. That said, there is likely to be a problem with distribution that we don't see in the Northern European countries.

    Human systems are highly multi-dimensional and multi-relational so the lessons to be learned can be easily obscured by a premature modeling of the data. Measures of central dispersion such as variance are not going to be very good diagnostically for all but the simplest problems (i.e. not for data exploration on these systems).

    If someone has the complete data set, I'd be happy to analyze it with info-theoretic measures.

  36. BobS says:

    Dunno, but if you go to the Gapminder graph, mentioned above by Wm. Cox, and change the scale of the dollar axis to linear (using a little drop-down thingee in the lower right) you'll see a plot that's strikingly consistent with the idea that spending more than about $3K/year doesn't increase lifespan (and that the US has some kind of health problem that doesn't respond to money as well as other countries).

  37. Thinking Man says:

    For those that want to leave out the uncovered or immigrants, you're missing the point. The uncovered individuals lead to lower life expectancies and higher healthcare costs. If you have insurance you are more likely to go to the doctor timely and not end up in the emergency room for something that could have easily been treated early on.

    As for immigrants, are you really a Native American? Or are you of European descent like I suspect.

    The bottom line is what is the most cost-effective way to deliver healthcare? Imaging if you had to have emergency insurance so when the police or fire departments delivered services to you, they would have to bill your insurance company for reimbursement. Do you think this would be a cost effective way or would it inflate the cost of these services? Sure the people who used these services the most would pay more, but is that how we want it? "You live in the ghetto so you should pay more for police and fire."

    As for smoking, drinking and other life shortening behaviors: Studies have shown (in Denmark) that they reduce the lifetime healthcare costs for that individual since they won't need long term assisted living and nursing homes which are very expensive.

    As for guns, most gunshot wounds either kill you or you are healed. Explosions, car crashes, war, etc. lead to long term health needs due to diminished capabilities (quadriplegic, paraplegic, amputation, sever brain trauma, etc.).

  38. alpha754293 says:

    I think that the parallel-coordinate plot does not present information in as succinct of a manner as does the scatterplot.

    Having said that though, I think that rather than using the sizes of the circles to denote the annual doctors visits, you might have been able to use the line thickness AND the size of circle (since the size of circle can be deceiving when they're very similiar like that).

    I think that while the formatting is an aspect of the debate, the clear picture here is US fairs quite poorly in its health care system.

    In regards to purchasing-power-adjusted values, you can always process the data in a variety of ways. I've always taken that since the currency used is USD that the most basic conversion would be the exchange rate. However, if you were to say…compared the cost of various commodity and necessity items (such as bread, butter, etc.) and take into account those differences as well, and there you have it.

    You can always make the data/numbers to tell you what you want to hear.

    The trick is to convince people of your data analysis methodology.

  39. A slightly different approach to the same question, of demographics (ageing) and health spending, is to do a scatter diagram of two indicators, Total Health Spending/DGD versus Percentage of Population 65+years old.

    You can see the results for a cross-section of smaller European countries (based on WHO Health for All data) in the blog PPP Lusofonia:
    http://ppplusofonia.blogspot.com/2010/01/despesas

    The correlation is clear, the causality makes sense. If the total annual health spending on 80-year-old pensioners is 3-4 times the spending on 40-year-old active workers, the financial burden of health care will go up directly and progressively with the dependency ratio.

  40. Keith says:

    Sean, I think what you propose would actually make the US look like it gets much less value for the money than even these graphs. The US has a lot of immigrants from poorer countries like Mexico and China. Many of these immigrants are without medical insurance, but they do get care, just much less of it. The number of immigrants from the wealthier countries with national health care is significant and these tend to be legal immigrants who get good jobs here with health insurance. They get more and better care. I think the better way to look at this question is by stratifying between those in the US with health insurance coverage and those without regardless of how long they were in the system.

    First, when you look at average expenditure numbers for the US you have to remember that large portions of even the native born population are without health insurance coverage. Probably the vast majority of those without coverage have a much lower per person expenditure and those with coverage have a much higher per person expenditure.

    Second, this also works for life-expectancy. Those without insurance probably have a significantly lower average life-expectancy than those with health insurance.

    Third, even those with insurance, who have greater per person expenditures than even the national average number graphed above, probably don't have a life expectancy greater than that of Japan. And if you compared them to their socio-economic counterparts in Japan they may still have a significantly lower life expectancy.

    In sum, the insured population in the US probably has an average expenditure significantly greater than that graphed above with a life expectancy closer to the highest (Japan) but probably not quite as high. So the luckiest in the US (those with insurance) spend well over twice as much (maybe three times) as Japan and get the same or less benefit in terms of life expectancy.