Daily Yonder/Dartmouth Health Atlas This map shows the difference in per capita Medicare costs among 1,845 rural hospital service areas. Green areas have costs below the national average. Brown areas are above the national average. For a larger version of the map, click here.
In Whitefish, Montana, the average yearly cost of taking care of a Medicare patient over a three-year period ending in 2006 was $3,950.
Across the country in the Florida Panhandle town of Graceville, the cost of tending a Medicare patient during the same time was nearly $15,500.
People in Graceville are poorer than people in Whitefish, it’s true. But the difference in cost of caring for a Medicare patient in these two towns is astounding — more than four times more expensive in one rural Florida hospital than in one town in rural Montana.
The map above shows the wide range of costs in caring for Medicare patients among 2,990 rural and exurban hospital service areas. The map, the first of its kind, is based on a remarkable set of data collected by researchers at the Dartmouth Medical School. Doctors and economists there take a sample of Medicare costs from every hospital. They account for differences in race, sex and age from place to place, but not income. What they have discovered are large differences in medical costs from one part of America to another.
“This map illustrates that variation in health care spending is not just about big city versus rural areas,” Dartmouth College economist Jonathan Skinner told The Daily Yonder. Skinner said this “landmark map” showed that “even within rural there is a large variation in costs.”
Those costs are not explained by differences in population or in health outcomes, according Dartmouth researchers. “Regional differences in poverty and income explain almost none of the variation,” Skinner and a team of researchers wrote in a recent issue of the New England Journal of Medicine.
Places with sicker patients do spend more, Skinner writes, but that discrepancy accounts for only 18% of the total difference between the highest and lowest cost regions. Skinner calculates that 70% of the difference between high and low-cost areas is due to “discretionary decisions by physicians.” Doctors in some areas order more tests, hospitalize more patients and consult more specialists than do doctors in other places. (Remember that health outcomes are generally better in low cost areas, according to Dartmouth’s research.)
The Dartmouth data has become central to the national debate over health care costs and reform. A recent article in The New Yorker magazine compared a high cost health care region around McAllen, in the Rio Grande Valley of Texas, with a low cost region in and around the western slope city of Grand Junction, Colorado. The author, Atul Gawande, found that low cost areas had better health outcomes than high cost areas — and his article soon became must reading in the Obama White House, which is looking for ways to lower the overall cost of health care.
The New Yorker article and all other studies using the Dartmouth data refer to hospital regions. There are just over 300 hospital regions, all containing metropolitan areas. It was impossible to tell if the same wide variation existed among rural hospitals by examining only the large regions.
To determine costs in rural America, the Yonder used per capita Medicare costs from “hospital service areas” (HSAs). There are nearly 3,000 HSAs, and 1,843 of these contain a majority of people living in rural or exurban zip codes.
The map above shows that the same differences in cost described in Gawande’s New Yorker article from one city to another also exist within rural America.
The green areas on the map denote hospital service areas that spent below the nation of average of $8,176 per patient (averaged from 2004 to 2006) for Medicare patients. The brown areas spent above the national average. The white areas represent urban hospital service areas, and in a few cases, areas with no data (For a large portion of western Maine, for instance, there was no data).
Rural areas generally spend less than the national average on Medicare than do urban areas. Only 27% of rural hospital service areas had per capita Medicare costs above the national average from 2004-2006.
But the differences among rural areas are as great as the cost variations from one metro region to another. Moreover, the high cost rural service areas are generally in the same places where there were cities with high Medicare costs. Gawande wrote about the high charges for Medicare patients around McAllen, a metro area. Medicare costs in the rural regions of far South Texas are among the highest cost in the nation, also. Louisiana has the highest costs statewide; and six of the ten most costly rural hospital service areas are in Louisiana. Below is a list of the fifty most costly rural hospital service areas.
(To download an Excel chart showing per capita Medicare costs in rural and exurban areas, click here. To download a similar chart for urban hospital service areas, click here. To look at all the data and articles Dartmouth has produced, go here.)
One reason for the disparity in costs from one rural service area to another could be differences in health, Skinner told the Yonder. “It’s not surprising that people in worse health use more health care,” he said. Poverty doesn’t create higher costs, Skinner says. Poor health does.
Yet poor health alone doesn’t explain the differences in cost from one rural place to another. In Southeastern Kentucky, for example, there are a number of high cost hospitals, and, historically, the Kentucky mountains have had high rates of illness and disability. The hospital service area in Pineville, however, is low cost, even though it serves a poor population that is very similar to the high cost areas around it. Similarly, West Virginia has mostly low cost rural hospitals while rural hospitals in Mississippi are high cost. Yet both are generally poor states. “It’s not simply poverty that explains these differences,” Skinner said. “Poverty doesn’t have a lot of effect on spending; it’s the fact that people are sick.”
(The Dartmouth data has generated controversy since the New Yorker article appeared. Here is an article critical of the Dartmouth research. And here is a defense of the Dartmouth project.
In Whitefish, Montana, hospital administrator Jason Spring said he noticed that the Obama administration had promoted Grand Junction and the Mayo Clinic in Minnesota as examples of low-cost medicine, “and we were lower than all of them.” Below is a list of the fifty least costly rural hospital service areas. 
Daily Yonder/Dartmouth Health Atlas
Spring said that Whitefish had a “healthy population…a lot of people involved in fitness, skiing or hiking, a community really focused on health.” He also said his area of northwestern Montana had a “really strong primary care base here. (Primary care doctors) are managing those illnesses wisely.” Dartmouth research has found that high cost regions refer patients more frequently to specialists.
Spring also said the size of the community helped. Doctors in Whitefish talk to one another. “There’s strong communication between the specialties here because it’s a small enough community,” Spring said. Gawande found that doctors in low-cost Grand Junction and at Mayo were in constant contact about patients’ care. 
Boundary Community Hospital The North Valley Hospital is the second lowest cost rural hospital in the country, just above the hospital serving Bonners Ferry in the northern tip of Idaho. North Valley sits next to Glacier National Park. Pretty spectacular setting, and low cost care.
Spring’s North Valley Hospital is also owned by a local foundation, not by doctors. Skinner said he “wouldn’t be surprised if in some high cost areas you had high numbers of physician-owned hospitals.”
There are statistical reasons some hospital service areas could be showing extremely low or high costs. Veterans who qualify for Medicare but receive their treatment at VA facilities reduce the overall Medicare bill for a service area. If there are a large number of these older vets in an area, it could reduce the overall Medicare bill.
Since the Dartmouth data is based on a sample, it’s also possible that an extremely expensive patient could inflate a hospital service area’s average cost. To smooth out these fluctuations, the Yonder averaged three years of data.
What the map of rural hospitals shows ultimately, Skinner says, is that there are quite significant cost differences from one hospital to another that can’t be explained by poverty, age, sex or race. The nation is spending vastly more on health care in some areas than in others without any indication that the increased spending results in better health.
“Medicare spending in 2006 varied more than threefold across U.S. hospital referral regions,” Dartmouth researchers reported earlier this year in a publication subtitled More Isn’t Always Better. “Research has shown that some of the variation is due to differences in the prices paid for similar services, and some is due to differences in illness; but even after accounting for these factors, twofold differences remain. In other words, the differences in spending are almost entirely explained by differences in the volume of health care services received by similar patients.”
Those differences are as great or greater in rural America.