Monday, May 04, 2009

Healthcare Metric--A Preliminary Proposal

Debating a recent post, commenter "Thai" has suggested that the United States is both relatively inefficient and relatively ineffective in providing healthcare. He challenged me to find cross-country comparative statistics. Here's a first try; reader assistance and analysis encouraged.

Introduction: I've previously addressed oft-cited trans-national healthcare metrics. And I've been largely dismissive--debunking misleading rankings based on infant mortality and life expectancy. On the other hand, I've highlighted America's superior ranking in life expectancy after accounting for various cultural differences and anecdotal evidence suggesting that America has the highest cancer survivor rates.

This time, I created and calculated my own cross-country metric, trying to capture three concepts: availability, cost-effectiveness and technology. Ideally, the figures would reflect the price at which sophisticated and modern treatments broadly are dispersed to the public. Call it the efficiency of providing high-tech healthcare per person.

I relied on numbers from the OECD, specifically its "Health Data 2008 - Frequently Requested Data," released last December. The sample set thus is limited to relatively rich nations.

I'm not positive my approach is valid or descriptive. But I decided to post the results, along with my reasoning, hoping readers would analyze and refine the methodology.

Outcome: On Saturday, I posted (in summary form) a comparison of country healthcare expenditures, per capita, measured at purchasing power parity. The previous day, I posted a ranking of countries based on the dispersal of magnetic resonance imaging (MRI) units per million population. For my metric, I (essentially) divided Saturday's chart by Friday's chart, set the United States at "100" and calculated the ratio of each nation's figure to that number. The final result is here:

source: NOfP chart from OECD Health Data 2008

Note that all data is from 2006 except Sweden, Switzerland and Ireland (2005) and Finland and Germany (2004). I omitted countries where either expenditure or MRI unit data predated 2004.

As can be seen, the U.S. is ranked 9th out of 28, i.e., relatively efficient. Japan and Korea in the top two spots could reflect those nations' presumed comparative advantage in technology production. I can't explain the extremely good rankings for Italy, Greece and Turkey, countries rarely associated with anything efficient. Yet, significantly, the relative positions in this chart vary significantly from those solely based on expenditure per capita, partially set forth Saturday, a result I expected and view as logical. Similarly unsurprising to me is the poor rankings of Germany, the U.K., Canada and France.

Syllogism: My underlying theory was to use the dispersal of MRI units to the populace as a proxy for the spread, and thus theoretically, the availability, of high-tech healthcare. I ignored other available data such as the number of hospital beds per capita, because that reflects less sophisticated healthcare delivery.

Dividing MRI units per million into expenditures per capita produces (sort of) the money (at PPP) required for each country to provide each's particular penetration of MRI machines. But not really, because it assumes that all healthcare expenditures are used for MRI machines--obviously not the case. That means that the dollar results themselves aren't meaningful. But, the ratios of costs between countries might be a proxy for the efficiency of making high-tech healthcare services available to the equivalent number of people. A proxy of a proxy, if you will.

Yes, technology isn't just a diagnostic machine or even MRI units in particular. Indeed, I'm not claiming MRIs are more important than basic healthcare services such as primary care or the number of nurses per patient. Instead, I sought to use the data available in a way that accounted for technology, availability and cost-effectiveness. Such a methodology may be a bug--but don't assume it's not also a feature.

More fundamentally, I'm not convinced that dividing per capita by per million--a unit "per population2"?--is sound. Or of using costs converted to PPP then benchmarked twice--both per population and to the United States. Analysis/suggestions welcome.

Thai and others may object to tying dispersal of a particular technology to availability of a service to the public without accounting for either equity or waiting time. But that's what I did--trying to control for cultural differences (such as crime and diversity) discussed in part here and here. In that light, I note that the study he cited oversampled from minority racial/ethnic groups.

Further, notwithstanding Thai's "hint," waiting times for specialists and treatments are shorter in the United States compared with at least some single-payer or socialized systems. Except, perhaps, in sectors and states where healthcare already has been burdened with extra social mandates.

Spreadsheets available upon request.

Conclusion: I've attempted to assess transnational efficiencies in the widespread availability of high-technology aspects of healthcare. By my measure, the United States performs better than most rich European countries.

I admit I am only an amateur in both healthcare and statistics. So help me learn more.


OBloodyHell said...

> Thai and others may object to tying dispersal of a particular technology to availability of a service to the public without accounting for either equity or waiting time.

There's not an absolute connection between cost/availablity of a service and the number of machines available to provide that service, but I suspect there will be a strong correllation, even more so in this case because the machines in question are not cheap.

Unless the nation in question is deliberately subsidizing the building of MRIs to build up a domestic industry in them (and I'm unaware of any nation specifically doing this -- overbuilding its supply of MRIs), why would a hospital purchase one if not to increase availability of the service for the medical professionals which work there?

And I do believe that any decently sized hospital has sufficient use for one to desire one, even if they don't have the money to buy one, or a clientele who can pay for them, either way. So a nation almost can't have "too many" MRIs.

Thai said...

@OBH Re: "So a nation almost can't have "too many" MRIs."

The thing I continue to fail to see in your comments is where you get the money to pay for statements like this...

Carl, my faith in you has not been in vain.


You have to admit that one of the most amazing things about the internet is the ability to have such wonderful conversations with otherwise complete strangers. And like me or not, I have made you think of this issue differently (personally I think in a good way but who knows).

Of course I now see where you were going and this is not amateur at all. Indeed, it is a lot more clever than many attempts I have seen to get at stuff like this.

AND (without in the least meaning to sound critical) I also agree with your concerns re: "...I'm not convinced that dividing per capita by per million... is sound."


I do think it worth doing the same kind of thing for a variety of technologies/procedures/medications/vaccinations, etc... in medicine to get a broader picture than just MRI's.

But I do like the approach, I really do and am going to think about it a bit more.

FWIW, you honestly should send this to the CRS. The primary data is out there and someone just needs to play with it a little more.

Re: your comments ascribing to me beliefs of inefficient or ineffective relative care, I think this might be a bit strong but I do think America's superior outcomes are often more the result of $ than simply skill and free markets (though we do have considerable skill) and I certainly see how my belief can be interpreted this way... Kind of a Clintonian "it kind of depends what the meaning of the word relative means". ;-)

And I don't object at all that you did not account for other factors as I would be very surprised you, or anyone else, has such data.

re: "oversampled from minority racial/ethnic groups".

This is a little unfair to say the least.

#1- minorities are Americans (which I know you know are)
#2- I was responding to your single example of when a rationing rule effected poor Sarah's teeth with data suggesting many pregnant American women also have this problem here in the US, nothing more.


IF you are going to say the relatively better position of how America compares is clear in this graph, so too must you also recognize the same of Iceland, Finland, Italy and Sweden, etc... (which you do).

... Personally I don't think Americans will ever embrace the Japanese and Korea care delivery models but I could be wrong.


You have made your point re: my original point:

"Yet, significantly, the relative positions in this chart vary significantly from those solely based on expenditure per capita".


And while I might add a nuanced "all medicine is local", clearly I need to think this issue through some more.


Even you must see SOME common ground SOMEWHERE in this with Liberals like me.

Carl, again, nice job and regards :-)

Carl said...


Thank you.

BTW, I think you've misunderstood my point about oversampling minorities. I was suggesting--perhaps too subtle--that because the United States is a far less homogeneous nation than most other OECD countries, oversampling minorities oversamples the cultural factors (e.g., immigration) for which I'm trying to control.

Thai said...

Carl, I thought you might enjoy reading another post about the endless are you a lumper or are you a splitter stalemate.

For I like NICE (e.g. I split), whereas you lump and don't like top down approaches.

I am not saying we need NICE AND the NHS (or CER vs. cer) but we do need agreement around this stalemate.

Conservative have an opportunity to propose their own alternative, so do you have one?


Carl said...

I previously laid out my proposal--competition at the consumer level among private insurers.

Carl said...

Two additional links: anecdotal and proposed policy.