« Book Forum reminder -- The Logic of Life | Main | Markets in everything, food fight edition »
The health care graph
Here is Ezra Klein, here is Paul Krugman on the same. If we put the partisanship aside, and view this as raw statistics, what lessons can be drawn? The biggest surprise is Japan -- a country whose health care institutions are not generally popular -- at number two. Spain and Italy and #4 and #5 are less extreme examples of the same point. Do the Germans and Danes really kill so many extra people through their health care systems? Would you really rather get sick in Greece?
Nothing in this post is intended as apology for the United States health care system, but if we are going to look at the numbers let's consider all of them. If there is any lesson about the French -- who are a clear first -- it is that they do something right for health care apart from having so much government involvement. What might that be? What do we learn about what makes for a good health care system? Is there a correlation between health care performance and policy? I don't see it, maybe there is one, but I'm wondering if people are willing to draw lessons from this diagram consistently or not.
I might add I find it easy to believe that American health care institutions make a disproportionate share of stupid errors, or are responsible for lots of patient mistreatments, so I am not trying to undo our presence on the right hand side of this graph. I do, however, walk away suspicious of the concept of "amenable" mortality.
Addendum: It's much worse than I thought, read this, which includes a free link to the supposedly gated study.
Second addendum: Out there on the mea culpa watch, or not, here is DSquared.
Posted by Tyler Cowen on January 11, 2008 at 05:17 PM in Medicine | Permalink
Comments
If you look at the details of the study, you'll find they don't actually study health care at all, just disease incidence. So this just says that US has some less healthy life-styles than some other countries, which we already knew. A more detailed analysis of the paper is here:
http://www.coyoteblog.com/coyote_blog/2008/01/uncovering-some.html
Posted by: Paul at Jan 11, 2008 6:04:42 PM
The underlying study is behind a for-pay wall so can't assess it. The
obvious issue with any by-country mortality comparison is the issue of ethnic
similarity - which differs vastly between the much more heterogenous
USA and all of the relatively homogenous comps. Do a blend of Asian,
African, Latin American and European mortalities and compare THAT
to the USA. In this chart, the minimal difference between US
and UK/Ireland is also worth reflecting on.
If all you have is a hammer, every problem looks like a nail. If you
are a wonk, every problem looks like it's caused by bad policy or
at least the absence of a good policy. When it may be
policy neutral, or caused by non-health-policy, like gun usage, or it
may be due to local differences that get buried in the national lumping(e.g., US cities may have worse mortality than large European cities, or the Southeast US may have worse mortality
than Europe but the north central US may be as good).
it is one of the big problems in the health policy debate that there has never as far as I know been a solid multivariate regression analysis done for mortality differences; the USA is just to big and heterogeneous for that, I guess.
Posted by: mt57 at Jan 11, 2008 6:06:56 PM
As for medical mistakes on the job, I'd bet on a conditional Intrade contract that is due to the very long work hours many residents, nurses and doctors put it. I'm sure the additional time and money required to become a licensed health care practitioner in the USA has a lot to do with that, but there are probably good reasons those markets aren't clearing.
All-in-all I'd also bet much of Europe's centrally-planned health care system is better than America's centrally-placed health care system. That really doesn't surprise me, given the size difference of the countries involved.
For the most part, however, I'd bet the anti-market bias in health care is the result of political entrepreneurship on the part of much of the health care industry. In America, there is still room for more financial involvement in medical care, so political "entrepreneurs" can take advantage of that. Elsewhere, this "market" has already been exploited.
Posted by: G at Jan 11, 2008 6:12:54 PM
Re: Denmark. It is generally accepted that cancer treatment has been a weak spot in the Danish system. Too many small hospitals? Lack of national integration (Please note that the Danish and Swedish systems are run by regional councils, not a state agency)? Bad procedures at hospitals?
Smoking and drinking also take their toll compared with Norway and Sweden. On the other hand, Danes are less suicidal than Swedes.
Reference: Nordic Statistical Yearbook 2007 (pdf - big file). Check pages 83-93.
Posted by: Jacob Christensen at Jan 11, 2008 6:50:09 PM
As a health care professional I'm sensitive to these numbers and have studied them before. The conclusions I draw are this. America persues more lost causes than many other countries. A telling statistic is the share of health care dollar spent in the last 30 days of life. This number is dramatically higher in the US than any other country. In essence, in the US we have difficulty getting away from losers. Be it the 85 year old with renal failure or the fifty year old alcoholic with liver failure our goal is to treat (and cure) all comers. Not so in many other countries. (Liver transplants for alcoholics are highest in the US). If you exclude the sickest patients most likely to have complications the US moves to the left on that graph (likely past France). So our problem is not the care but patient selection.
Posted by: roland at Jan 11, 2008 6:58:20 PM
In the UK they have an institution (NICE) that screens treatments for cost effectiveness. I doubt if someone who dies in the UK because they are denied a treatment that is not cost-effective is classified as a death due to "health care"---but it probably should be.
Posted by: Chuck at Jan 11, 2008 8:29:11 PM
I posted this concern over on Thoma's blog and simply got 1 sneer from the religious left. This data is without a doubt incomplete. This means nothing, without a tally of the number of "amenable" deaths prevented in these countries.
Case in point, since the PhD leaves me to speculate, lets say the U.S. treated 300 patients / 100,000 population successfully. And France treated 100 / 100,000 population successfully. Would you rather have an "amenable" disease in the U.S (300/410 chance of surviving) or France (100/165).
Posted by: Jay at Jan 11, 2008 10:27:08 PM
"it is that they do something right for health care apart from having so much government involvement."
why do you automatically exclude the government involvement?
"I might add I find it easy to believe that American health care institutions make a disproportionate share of stupid errors, or are responsible for lots of patient mistreatments,"
I personally find this very hard to believe. Is a doctor really that much better in france than in the US?
Posted by: Jon at Jan 11, 2008 11:23:33 PM
Just two observations.
1) Several years ago France was rated as having the world's best health care system
in the world by the WHO. Do not know what their criteria were, but these numbers fit
in. When we were there and needed health care, it was excellent. Note: doctors
actually visit patients in their homes, even those who are foreign visitors. I am
old enough to remember when my family's doctor visited the house when were sick when
I was a kid. Can someone explain just exactly why it is that they do not do so anymore?
Is this perhaps one of those examples of backward-bending supply curves for the overpaid?
2) I picked up the latest UN Human Development Report while I was at the AEA meetings in
New Orleans. Denmark is tied with the US at 29th in life expectancy in the world at 77.9
as of the 2005 numbers in there. Frankly, I do not know why Denmark is down in the garbage
dump with the US on this, given its mostly good performance on most social indicators.
Posted by: Barkley Rosser at Jan 12, 2008 12:13:57 AM
Here is Tyler's post from 9/24/06
"New Jersey fact of the day
Life expectancy for an Asian female living in Bergen County, New Jersey: 91 years.
Here is the source, via Jason Kottke. Yes, lifestyle and attitude matter. This is one indication that the American health care system isn't as bad as it is sometimes made out to be."
What is the life expectancy of an Asian female living in a comparable environment in, say, Japan? Maybe mid to high-80s, but probably not 91. In any case, THESE are the kind of like-to-like comparisons that should be made. The same study that Tyler got his Asian female example from showed other ethnic/geographic cells in the US with life expectancies in the low 60s.
Given this diversity in the US, it's a bit ridiculous for a study to just lump it all together and compare it with societies that are much more homogeneous. Plus Coyote blog effectively exposes the holes in the Health Affairs paper. But this doesn't keep Krugman (careful distinguished scientist that he is) et al. from dancing around the maypole with it.
Posted by: Bill Kruse at Jan 12, 2008 12:50:48 AM
Tyler,
My comments are here (go to the Update):
http://www.medhumanities.org/2008/01/on-preventable.html
Posted by: Daniel Goldberg at Jan 12, 2008 1:36:17 AM
mt57 and Bill Kruse:
I don't see how you want to use the heterogeneity of the US population as an excuse to justify the US statistic! How does being heterogeneous excuse you of a bad average?
A bad average is bad nevertheless. The US demographic is indeed a sum of its parts and thats what you have to live with!
A multivariate study highlighting regional differences might be nice to have but the finer nature of the distribution does not change the magnitude of the average.
Posted by: Raul at Jan 12, 2008 5:40:47 AM
Barkley:
I come from a country where house visits by a family doc. are still the norm. You wonder why you lost that in the US?
It's an artificial doctor shortage propagated by the vested interests of current doctors! You let practicing docs have too much say in the certification process. Tell me, what incentive does a practicing doc. have to allow more to graduate and ruin his margins?
Reform those licensing boards etc. to make doctor "production" more market oriented. If needed "import"!
Posted by: anon at Jan 12, 2008 5:53:32 AM
I would not hesitate for one second before choosing the american system. The tiny increased risk of dying has to be balanced against living in a free country and not a socialist state such as France. I am also not worried at all about my choice being better for health too in the long run.
The problem is that so many people are dead set on forcing me to give up my chance to make a choice.
Posted by: Erik at Jan 12, 2008 5:58:13 AM
Are these statistics correlated with obesity rates? Looks like they may be.
Posted by: Charlie at Jan 12, 2008 9:01:07 AM
I don't understand why people seem to think France has a highly socialist healthcare system. It doesn't. French healthcare is far more market-oriented than Canadian or British.
France does not have a single-payer healthcare system.
If you need medical attention in France you basically get it from a private doctor, clinic or hospital. They compete for business, so they have a strong incentive to provide good care.
You then pay them, but are partly reimbursed by the government, by about 75%. The rest of the money either comes out of your pocket, or your own private medical insurance.
Over 80% of French people have private health insurance to cover this top-up.
This means both you and your private insurer have an incentive to reduce costs. Not a strong one, and correspondingly French healthcare is relatively expensive.
More information here and here.
Posted by: TheophileEscargot at Jan 12, 2008 10:42:28 AM
Are these stats correlated to account for access to health care? Potentially, someone with an incurable condition (a very serious heart failure, for instance) might be making it to the hospital in time to die in the U.S. and not in Greece. This would make the death "amenable to health care" in the U.S. but not in Greece even though it's actually a credit to the U.S. system that the patient made it to hospital at all.
In other words, I'm guessing that the number of deaths "amenable to health care" in Sudan is close to zero because health care is, for the most part, so inaccessible that only patients with stable conditions are able to make to hospital.
Posted by: Geoff Hamilton at Jan 12, 2008 12:08:35 PM
Raul,
The heterogeneity of the US population does matter, and the simple averages are misleading, IF the US or any other country has a disproportionately high (or low!) number of people more susceptible (because of lifestyle or things we're born with) to the kinds of diseases amenable to treatment covered in the Health Affairs paper. Maybe the US has a disproportionately high number of Asian females and its real score should be way worse! Certainly Japan has a high percentage of Asian females, and I think this probably goes a long way toward explaining its low average score in the paper.
It's amazing that the data presented in the Health Affairs paper are being leaped on by Kevin Drum and his ilk as another damning indictment of US healthcare (with the implication, of course, that we gotta have single payer, single payer) when, as pointed out by Coyote Blog, the paper doesn't really study health care at all.
Posted by: Bill Kruse at Jan 12, 2008 12:50:08 PM
You couldn't manage a second post for the dsquared bit? It's really not clear to me why the two studies are necessarily so contradictory. The 150,000 (violent) deaths number is fairly misleading, given the large increase recorded in nonviolent deaths. Since both studies used the same methodology, I'm not sure why you think an apology is needed, considering all dsquared did was argue against nonsensical statistical arguments.
Posted by: graeme at Jan 12, 2008 1:20:09 PM
Why did you stop at the US in the chart?
Why didn't you break the figures out by race for each country?
Posted by: cynic at Jan 12, 2008 2:19:21 PM
Correlated with obesity -- yes, very probably, but not caused by it. As someone noted in the discussion, the absolute death rate from a certain condition is not enough to gauge the performance of the health system per se. The incidence rates of the condition are also important. As a contrived example, consider two 3rd world countries, one has malaria mosquitoes and the other doesn't. Deaths from malaria are certainly amenable to health care, but the first country will look much worse than the second even if their health care systems are the same (or absent). In fact, the first country might even have a better health care system and still look worse on the absolute death rate graph.
Caveat: this is *not* a defense of the U.S. health care system, of which I (luckily?) have no experience anyway.
Posted by: A Tykhyy at Jan 12, 2008 3:12:55 PM
It's funny people mention ethnic diversity.
France doesn't strike as less diverse than any one they seem to outperform except the US and may be UK and New Zealand.
If it was that relevant, they will be last or second to last among european countries.
"You then pay them, but are partly reimbursed by the government, by about 75%."
Actually reimbursement:
- has different rates, from 100% for lifelong treatment like diabetes to 25% for some less important procedures (dental is one).
- is caped. that's why additional private insurance plans often promise coverage BEYOND 100% (of the government recomended price). But then again, more often that not, paying beyond the government-recomended price is really paying for extra-medical features.
Posted by: nu at Jan 12, 2008 3:27:12 PM
Why give so much credibility to a study with a weird definition not really subject to accurate measurement?
Just because Krugman gives it instant credibility does not mean the paper deserves credibility.
Posted by: save_the_rustbelt at Jan 12, 2008 9:28:18 PM
As an indicator of the quality or performance of national health care systems, the chart is meaningless, because it doesn't take into account differences in the prevalence of "causes of mortality amenable to health care" between different countries.
For example, gunshot wounds are presumably considered a cause of mortality amenable to health care (at least in some cases). I suspect the mortality rate from gunshot wounds in the U.S. is much higher than in most European nations, because we have so many guns and so much gun violence.
Does that mean the U.S. health care system is worse at treating gunshot wounds than those other countries? Of course not. If we have twice the rate of gunshot wounds as country X we're probably going to have a higher mortality rate from gunshot wounds as country X even if we are a lot better at treating them.
Posted by: Jason at Jan 12, 2008 9:37:33 PM
Several years ago France was rated as having the world's best health care systemin the world by the WHO. Do not know what their criteria were, but these numbers fit
in.
If you're referring to the WHO World Health Report, I believe the sole health indicator used in the rankings was disability adjusted life expectancy. Which is meaningless as a measure of the performance of a health care system, because there are so many other things that influence life expectancy more than the quality of health care services.
The French health care system is going bankrupt. The public insurance fund has been in deficit every year since 1986. See this recent article from the Canadian Medical Association Journal on what it calls the "hard choices" facing France over the future of its health care system.
Posted by: Jason at Jan 12, 2008 9:53:20 PM
