IN A NATION AS LARGE AND DIVERSE AS THE UNITED STATES, it’s not surprising that health care costs vary by region. What is surprising is that higher spending doesn’t buy better care. That’s the finding of Elliott Fisher and his colleagues. Fisher, a professor at Dartmouth Medical School in Hanover, N.H., and co-director of the VA Outcomes Group at the VA Medical Center in White River Junction, Vt., was recently named director of the Dartmouth Atlas Project, which analyzes regional variations in practice and spending for the Medicare population. In July, Fisher’s team announced a partnership with the Brookings Institution, the Washington, D.C., think tank, to foster policies that improve health care while controlling escalating costs.

Q: It’s not hard to guess which Atlas Project finding surprised you most.
A: Right, that despite a twofold or threefold cost difference, people in high-cost areas such as New York City and Los Angeles are less likely than those in Rochester, N.Y., or Portland, Ore., to receive important treatments such as angioplasty, mammography and cholesterol-lowering drugs.

Q: What accounts for the higher costs?
A: The difference in spending is almost entirely due to greater use of discretionary services. Patients in high-cost areas see their physicians more frequently, are more likely to see specialists, spend 70% more time in the hospital and receive more minor diagnostic tests, imaging and other procedures.

Q: Could the concentration of elite medical schools and teaching hospitals in high-cost cities account for some of the difference?
A: Patients at teaching hospitals with the most expensive practices—such as UCLA and New York University Medical Center—have no better outcomes than those at low-cost centers. At UCLA it costs $72,793 to take care of seriously ill Medicare patients in their last two years of life. At the Cleveland Clinic, the average cost is $35,455, with no discernible difference in results. We’re looking into why that is.

Q: Don’t more doctor visits and tests mean better care?
A: On average, there’s no evidence that the extra visits and tests in high-spending regions are providing any benefit. In fact, low-spending areas actually do a better job of making sure that patients receive necessary services, such as treatment for heart attacks and screening for breast cancer. And the five-year mortality rate for patients with heart attacks, hip fractures or colon cancer is slightly lower in the low-cost regions.

Q: How do you explain that?
A: I think there are two likely reasons: poor communication and the harm that comes from unnecessary care. In high-cost regions, patients are three times as likely to have 10 or more physicians involved in their care. Physicians in those regions are more likely to say that lack of communication impedes high-quality care. Too many cooks spoil the soup. And receiving unnecessary treatment can be risky: Hospitals are dangerous places if you don’t really need to be there.

Q: Should we return to the days when one physician handled all or most of a patient’s problems?
A: It’s not that easy. Technologically, we’ve advanced way beyond the capacity of one individual to know everything about a patient or provide care. What we need is better coordination to ensure continuity of information and decision-making.

Q: Health care is a hot presidential campaign topic right now. What should the next president do?
A: Most of the candidates are focusing on expanding insurance coverage. This is important, but they should also think about how to improve the delivery system.

Q: Where to we start?
A: We need to move away from fee-for-service payment toward a system that rewards better—but not necessarily more—care. We should eliminate incentives for superfluous procedures and services while creating incentives for physicians to work together to effectively treat an entire population.

Q: Managed care is the answer?
A: When it was tried in the early 1990s, managed care made a lot of people nervous, and for good reason. It was adopted in the absence of comprehensive performance measures, and the incentives were for hospitals and physicians to provide less care. Any such system needs to be counterbalanced by vigorous efforts to track patient care in multiple dimensions: Was it safe, timely, effective, efficient and equitable? One of the reasons for our partnership with Brookings is to develop and implement performance measures and payment reforms that move the United States toward a high-quality—and affordable—health care system.