If you’ll forgive me for egregiously mixed metaphors, I want to draw attention to an implicit assumption among many health care reform advocates related to controlling healthcare spending: that if not for the politics involved, it would be fairly easy to rein in costs.
That’s because, the argument goes, there is easily identifiable inefficiency in the way we currently spend health care dollars. There are enormous regional disparities in, for instance, per capita Medicare spending. What is more, these differences are apparently unrelated to differences in the health of the underlying populations, and they don’t produce better outcomes. Rather, the differences reflect the ways that health care providers diagnose and treat patients in different parts of the country. So say the much-revered Dartmouth College health researchers, whose findings have been fairly uncritically embraced by many on the left.
Politics aside (the difficulty is that one person’s wasteful diagnostic test is another’s life-saving intervention), I always was suspicious of this argument. If there are excess profits to be made, then why is it that providers in only some parts of the country go after them or successfully extract them? Then a fascinating study came out that was mostly ignored but that should have raised questions about the Dartmouth research.
A potential problem with the Dartmouth research is that if there are unmeasured differences in health between patients who go to different providers, then the finding that greater spending is unrelated to outcomes could simply derive from people in worse health being very expensive to treat. The Dartmouth researchers use relatively crude measures to statistically control for these differences (because they are the only ones available).
MIT economist Joseph Doyle got around this problem by looking at patients who needed emergency care while they were visiting Florida. Because there is no reason to expect that unhealthy tourists are more likely to end up in higher-spending ERs, any differences in outcomes between those who went to high-spending hospitals and those who went to low-spending ones should reflect only the spending difference. Doyle found that higher spending did produce better outcomes.
Now MedPAC, the panel that monitors how Medicare reimburses providers and makes recommendations to Congress, has released a study that shows that disparities in Medicare spending are quite a bit smaller when other important factors — such as regional differences in wages and extra reimbursement related to medical education — are taken into account (hat tip to Mickey Kaus). If one looks only at per capita Medicare spending, high-spending areas of the country have costs that are 55 percent higher than low-spending areas of the country (I’m talking about the 90th and 10th percentiles, for those of you statistically inclined). After making MedPAC’s adjustments, however, that difference shrinks to 30 percent.
Thirty percent might still be considered a big number — in a perfect world adjusted spending shouldn’t differ at all — but other evidence in the MedPAC data gives reason to question the precision of any of these kinds of comparisons. I put the figures for all 404 geographic areas into a spreadsheet (which you can get from me if you’re interested — data wants to be free!) and looked at the top and bottom quarter of adjusted spending.
High-spending areas are dominated by the South, particularly the states stretching from Florida across to Texas and Oklahoma. They also include 15 of the 30 biggest metropolitan areas, including all of the biggest southern and midwestern metros, save Atlanta and Minneapolis, and none of the biggest northeastern or western metros, save Los Angeles, Las Vegas, Phoenix, Denver, and Pittsburgh.
On the other hand, low-spending areas are dominated by the West, particularly Alaska, Hawaii, Washington, Oregon, Idaho, and most of California (with the exception of Los Angeles and San Diego). Also overrepresented are small metropolitan areas in the upper Midwest and Dakotas, in New York, Maine, Virginia, and Georgia. None of the biggest ten metropolitan areas are represented in the bottom quarter, and only four of the biggest thirty are (San Francisco, Seattle, Portland, and Sacramento).
Compare these findings to those of the Dartmouth folks (Map 1). While many of the same conclusions show up in their map, there are some notable differences. Most importantly, California and the Boston-Washington corridor look like they spend a lot more in the Dartmouth map than they do in the MedPAC data (and the Mountain West states look like they spend a lot less).
If different sets of rankings differ as notably as these two do, then that says to me that there is a lot of noise in these rankings and that perfectly adjusted spending figures would potentially produce a distribution of areas that would look different from either set. In particular, I suspect that it would show that the vast majority of spending variation could be explained by factors that had nothing to do with inefficiencies.
The point is that even discounting the political difficulties of enacting policies that rely on comparative effectiveness research to weed out inefficiencies in healthcare spending, it’s not at all clear that regional variation in healthcare spending is proof that such inefficiencies exist. That’s not to say that there are no inefficiencies, but weeding them out won’t be as simple as making Florida providers act like Minnesota ones.
The views expressed in this piece do not necessarily reflect those of the Progressive Policy Institute.
