More care not necessarily better care, study says

Study focuses on care of chronically ill

Budget-conscious quality managers might want to take a good, hard look at the findings in the latest report from the Dartmouth Atlas Project, in Hanover, NH. It indicates that providing chronically ill Medicare beneficiaries more care at a higher cost does not translate into higher quality care. To the contrary, the study shows that beneficiaries in high-utilization areas end up with lower quality of care.

The project used a new, free database built with Medicare data that helps provide benchmarks for highly efficient care. (The database, funded by the Robert Wood Johnson Foundation, the long-time principal underwriter of the Dartmouth Atlas Project, is available to the public at

Using this database, the researchers calculated significant savings to the Medicare program if all U.S. hospitals provided care at the levels of highly performing health systems.

"The database has 100% of Medicare hospital claims," notes Megan McAndrew, MBA, MS, editor of the Dartmouth Atlas Project. "We also have samples of part B claims, census data, AMA data, and so on. The reason why we study Medicare is it's the only uniform national claims database."

The study is based on records of more than 4.7 million beneficiaries who died from 2000 to 2003 and had at least one of 12 chronic illnesses, including: solid tumor cancers, lymphomas and leukemia, chronic pulmonary disease, coronary artery disease, congestive heart failure, peripheral vascular disease, severe chronic liver disease, diabetes with end organ damage, chronic renal failure, nutritional deficiencies, dementia, and functional impairment.

Using three health care regions identified as "highly efficient" as benchmarks (Salt Lake City, UT, served primarily by Intermountain Healthcare; Rochester, MN, served largely by the Mayo Clinic; and Portland, OR, the largest and most metropolitan region in a state that has made improvement in end-of-life care a public policy goal), researchers used the database to compare care provided in all other regions of the country. Highly efficient health care was defined as high-quality/low-cost.

An ongoing effort

Researchers determined that Medicare could save an estimated $40 billion — or nearly one-third of what is already spent on chronically ill Medicare beneficiaries — if all U.S. hospitals practiced at the high-quality/low-cost standard set by the Salt Lake City region. By the Mayo Clinic benchmark, savings would have been $19 billion; by the Portland benchmark, $38 billion.

Founded in 1993, the Dartmouth Atlas Project "is an ongoing project, started as a way to determine where people go for health care," explains McAndrew. "The original concept was to be the managed competition model of health care purchasing; if you put people into markets and cooperatively buy insurance, then you'd have to follow patterns of wherever people seek care."

The project pointed out that the model did not work, but in the process, found itself headed in a new direction. "Once we figured out by matching zip codes of residents in hospitals we had defined all these markets, we saw we could compare them," McAndrew relates. "So, we started to compare what happens with different populations — classic epidemiology."

While epidemiologists usually study the incidence of disease, however, the project studied surgery and hospitalizations. "What you find is there are these really remarkable variations in what happens to people, and it depends on where they live, rather than on their medical needs or desires for care," McAndrew shares. "For most conditions, medicine is practiced very idiosyncratically in different parts of the country — and even in contiguous markets. In some areas of Florida, for example, heart surgery practices vary radically within a few miles."

This particular study was driven by the fact that "over time, in a broad sense, people couldn't see any benefit for individuals who lived in regions where they got a lot more services, and where Medicare spent a lot more money per capita — for example, twice as much in Miami as in Minneapolis," says McAndrew.

It also was argued that other studies hadn't significantly accounted for differences in population health status. "This report looked at people with at least one of 12 medical conditions who had died. We worked backwards and examined what had happened to them in the last two years of their lives — the differences in the health care resources they consumed," says McAndrew. "We found enormous differences in the likelihood of how many days in their last years of life they would spend hospitalized, in the ICU, and in how many specialists they saw."

Besides the disconnect between dollars spent and health care quality, the report also noted that individuals in high-spending areas had lower patient satisfaction levels.

"This data, from the state of California, came from the California Health Care Foundation [CHCF]," says McAndrew. She asserts that California is a valid model for the rest of the country. "You can find markets at both ends of the spectrum; for example, L.A. is like Miami, and San Francisco is like Minneapolis," she explains.

When the CHCF did its survey, she continues, "they found an inverse correlation between how happy patients were with care in high-intensity places. In additional studies, researchers here asked providers in high-intensity regions what their level of satisfaction was, and those in the highest-intensity areas said they had the toughest times; they used many more referrals, saw more hospital days [per patient], and so on."

Seek the best model

It's very difficult, McAndrew concedes, for someone right in the middle of things to recognize system flaws unless they compare their system with others. "Otherwise, you do not have a frame of reference for what other people are doing and what works," she says. "But you need to care about what the best model is for the best management of people with severe chronic illness."

Quality improvement professionals, she notes, "say every system is designed to get the results it gets; but the way Medicare is designed is in reverse. It encourages you to do more stuff; they won't pay for diabetic counseling, but they pay really well for amputations."

One step toward a better model, she notes, is for government to think hard about how it incentivizes different types of care. "On a moral and medical basis, academic medical centers should also take some responsibility for trying to figure out what the optimal model of care is," she adds.

Meanwhile, she says, hospital quality managers should study the models used in places like Oregon, "which has very consciously tried to manage end-of-life care."

But it doesn't have to be a statewide effort, she continues. Quality managers might consider models such as Salt Lake City's Intermountain Health Care, the Mayo Clinic in Rochester, MN, or the Danville, PA-based Geisinger Health System. "They've really thought about where to put their resources and what the best model is," McAndrew observes. "They've thought about how to do it better — which is also less expensive and closer to what people say they want."

The three aforementioned organizations are group practice models, McAndrew notes. "They are pretty well run by doctors," she says. "Historically, they were started by people who went out and trained with the Mayo brothers and then started rural multispecialty practices. What they have are salaried physicians, so they are not in head-to-head competition with other tertiary care hospitals."

If you are not locked in "mortal combat" with other facilities, she adds, "It becomes easier to share patient records."

Tools for benchmarking

The Dartmouth Atlas Project web site has a number of tools available on it for quality managers interested in benchmarking. "These tools enable you to 'slice and dice' the data in all different ways," says McAndrew. "Just yesterday someone called from New Jersey and wanted to know how many fewer hospital days they would have in their state if they looked more like Intermountain Health."

Using the tools, for example, you can compare things like Medicare spending and excess hospital days between your area and others, says McAndrew. "What you can also find, importantly, is population-based rates. What we're talking about is trying to use the methodology to define which patients go to which hospital. We can track patients who are loyal to your hospital — where they might get 90%-95% of their care."

Such data, she continues, can give you insight into your hospital's management style and quality factors associated with the physician staff. "It gives you a very legitimate look at how patients who are loyal to your hospital are being managed," McAndrew asserts.

Finally, she says, the report shows that some of the well-publicized "crises" in health care don't have to be as serious as they seem to be. "Everyone says Medicare is going broke, and we have a shortage of doctors," she notes. "We say, if everybody looked like Intermountain Health, they wouldn't need more doctors. And if Medicare were saving 30% of the money we estimate it has wasted, it would be a lot longer before it was financially in trouble."

For more information, contact:

Megan McAndrew, Dartmouth Atlas Project. Phone: (603) 650-1971. E-mail: