The Dartmouth Atlas shows wide differences in Medicare spending across health care regions that cannot be explained by differences in health care outcomes. It has been touted by the Obama administration as Exhibit A in the case for health reform. However, The New York Times the other day took the Dartmouth researchers to task in a front page article and columnist David Leonhardt piled on. I also weighed in at this blog and the Dartmouth folks defended themselves here.
I think the work of Elliott Fisher and Jonathan Skinner is very important. But behind the very public to and fro about their work is a very important policy issue that all sides allude to and none are confronting directly. To wit: Can we (meaning the government) use statistics on Medicare spending to force the health care system in general, and Medicare providers in particular, to deliver more efficient, higher-quality care?
The weight of the evidence I believe is fairly clear: No, we cannot.
Before getting to details it may be helpful to remind everyone that this health care discussion is very similar — almost word-for-word similar — to the public policy discussion over public education that has been underway for a quarter of a century. See my previous post here.
Education researchers have conducted Dartmouth Atlas-type investigations and documented wide disparities in outcomes among public school students that cannot be explained by student socioeconomic characteristics or by any inputs controlled by the governing authorities. Good education outcomes are not correlated with the amount of money we spend, pupil/teacher ratios, etc. If anything, the districts that spend the most money seem to have the worst outcomes!
Why is that? In general, a bureaucratic system is one in which normal market forces have been systematically suppressed. In such an environment, there tends to be a sea of (relative) mediocrity, sometimes punctuated by little islands of excellence. Further, the islands of excellence tend to be randomly distributed. They do not correlate with much of anything.
This picture not only describes most public school systems around the world, it also describes most health care systems. (In fact, I can’t think of a single exception.)
Now for 25 years, education policy experts have been asking this simple question: Is there a way for the funders of care (the buyer side of the market) to get low-performing schools to produce as well as the high-performing schools? In other words, can we use the power of the purse, the power of authority or any other governmental power to replicate excellence?
The answer appears to be: No. Every example we have of a good school is an example of excellence achieved because of the enthusiasm, energy, leadership, vision, etc. of people on the supply side of the market. We have no examples of good schools created by federal judges, school boards, city governments, state governments, etc. (unless you count magnet schools that siphon off all the best students).
The same thing is true in health. Recently, a group of researchers from Brookings and other think tanks identified some of the best-performing health regions in the country. How many of these were created by Blue Cross? Zip. How many were created by employers? Zero. How about Medicare? Nada. Further, the researchers could find no common parameters that could be easily duplicated. That is, some plans were capitated; others were fee-for-service. Some had electronic medical records; others did not.
Now back to Dartmouth. Based on Medicare data alone, Dartmouth researchers not only found wide discrepancies in spending, they concluded that up to 30% of spending could be saved if the rest of the country practiced medicine the way the lowest-spending areas practiced. Based on this observation, OMB Director Peter Orszag claimed that $700 billion could be saved if Medicare changed the way it purchases health care. (Notice the slip from a purely hypothetical statement that involves no practical implementation to a completely unsupported claim about what policy can actually achieve.)
Now aside from the obvious observation (at least it’s obvious to me) that we don’t know how to replicate low-cost, high-quality institutions, critics have made several points. The New York Times showed that low-cost hospitals can have poor quality as well as high quality. Some have argued that higher spending institutions often have better outcomes. Others have objected that focusing only on Medicare data ignores non-Medicare spending at the same institutions.
In their research on Medicare spending, Thomas Saving and Andrew Rettenmaier found that:
- Medicare and non-Medicare spending tend to be substitutes for each other. Where Medicare spending is high, private spending tends to be low and vice versa. This implies that much of what the Dartmouth Atlas calls excessive spending may just be cost-shifting to Medicare in areas where private payment is low.
- Among states, we could hypothetically reduce spending by 25% if all states practiced medicine as efficiently as the five lowest-cost states. But after adjusting for socioeconomic and other differences the savings drops to only 10%. When all health care spending is considered, the potential saving is only 5%.
And that 5% assumes you do everything right.
So how can we achieve efficiency in health care? Recognizing that all our examples of low-cost, high-quality care originate on the supply side of the market, it is to that side we must turn. Rather than have third-party payers create more rules, regulations and perverse incentives, we must liberate the providers from the rules, regulations and perverse incentives that already exist. Rather than impose such demand-side initiatives as managed care, pay-for-performance, value-based purchasing and the artificial creation of HMOs, PPOs and ACOs, etc., we must let the providers show the payers how to get the job done.
The providers have far more information than paying institutions will ever have. They have professional pride, training and experience. And (not to be overlooked) they actually know the patients they treat.
I have written about how to do this elsewhere.