Government
What's in a number? Cost variance figure drives policy and courts controversy
■ As Congress begins work on national health system reform legislation, the Dartmouth Atlas Project's estimates of unnecessary health spending gain traction.
By Doug Trapp — Posted April 27, 2009
- WITH THIS STORY:
- » Exams, tests affect spending
- » Less-than-universal care
- » External links
- » Related content
Washington -- Listen long enough to the health system reform debate heating up on Capitol Hill and you're bound to hear a striking figure: Up to 30% of U.S. health spending might not improve -- and might even hurt -- health outcomes. Some experts say that costs the nation upwards of $700 billion a year.
The estimate has become a mantra for President Obama's budget director and is often cited by the congressional architects of the largest national reform effort in 15 years. The physician researchers behind the 30% figure -- members of the famed Dartmouth Atlas Project -- already have testified before Congress at least three times this year.
The researchers derived the figure from years of examining Medicare claims for regional variations in health care usage and spending. The highest-spending regions in the U.S. provide significantly more care to Medicare patients without a corresponding increase in quality, even after adjusting for patient characteristics and regional demographics, they reported in two Annals of Internal Medicine studies in 2003. More frequent use of office exams, tests, imaging and other diagnostic procedures account for a significant part of the differences in care, they concluded.
"This is a remarkable opportunity for us to improve efficiency of care and improve access to care," said Elliott Fisher, MD, MPH, director of population health and policy at the Dartmouth Institute for Health Policy and Clinical Practice and a professor of community and family medicine at Dartmouth Medical School in New Hampshire.
The researchers concluded that national health spending could be reduced by 30% if physicians in the highest-spending regions -- such as Los Angeles -- adopted the practice patterns of physicians in the lowest-spending regions -- such as Minneapolis. Lawmakers like the theory that ballooning health spending could be curbed without drastic cuts or other tough choices if some doctors changed how they deliver care.
But is the Dartmouth estimate on target? Other experts in the field generally praised the Dartmouth methodology and said it has improved the understanding of health care variations. But some said they aren't sure the 30% figure -- and the way to address it -- is so clear cut.
Marilyn Moon, PhD, vice president and director of the health program at the American Institutes for Research, said she gets nervous when people toss out numbers and say the nation can cover more people simply by reducing health system bloat. "Your bloat is somebody else's absolutely necessary care."
The origin of a mantra
The Dartmouth Atlas estimate of U.S. health system waste relies on two primary conclusions: Some Medicare patients receive more intensive care that doesn't improve their health outcomes; and physicians treat many patients the way they treat their Medicare patients.
John Wennberg, MD, MPH, and a colleague started the work that would lead to the Dartmouth Atlas in the early 1970s by examining variations in practice across small areas in Vermont. In 1992, Dr. Wennberg was a cofounder of the Dartmouth Atlas with Dr. Fisher and others. Dr. Wennberg led the team that divided the U.S. into 306 "hospital referral regions," or HRRs, where clusters of patients tend to seek care. The Dartmouth Atlas of Health Care demonstrated that after adjusting for age, race and sex, per capita Medicare spending varied widely among HRRs.
Because some remaining variations in spending are due to differences in health status and illness severity, Dr. Fisher then led a team that used patient surveys, chart reviews and U.S. Census and Medicare data to account for differences in the HRRs based on age, sex, race, income, education and health status of the region. The work also adjusted for the different prices Medicare pays around the country.
This work led to the 2003 studies -- the Dartmouth group's landmark work -- which concluded that higher-spending regions were not getting much for the money. Patients with one of three chronic conditions in the highest-spending regions received 60% more services than those in the lowest-spending ones, but they were no more likely to report better results on quality measures.
Physicians in higher-spending areas probably also provide more care to their privately insured patients, Dr. Fisher and his colleagues said. A 2008 Health Affairs Web article by the Dartmouth group compared care patterns in the last two years of life for chronically ill California patients on Medicare and similar, privately insured patients. The article concluded that similarly ill patients spent more time in hospitals in higher-spending regions than in lower-spending ones.
Some lawmakers and key health policy leaders have embraced the Dartmouth findings.
Peter Orszag, PhD, director of the White House Office of Management and Budget, began citing Dartmouth's work in late 2007 when he was still director of the Congressional Budget Office. Orszag has taken Dartmouth's 30% estimate one step further to conclude that up to $700 billion of the nation's $2.3 trillion in annual health spending does nothing to improve outcomes. Senate Finance Committee chair Max Baucus (D, Mont.), who is emerging as the congressional point man on health system reform, prominently cited Orszag's dollar figure in an 89-page reform proposal that Baucus released in November 2008.
Ted Epperly, MD, president of the American Academy of Family Physicians, said the Dartmouth work is valid. Primary care physicians must better coordinate health care in close consultation with specialists instead of simply handing off patients, he said. "Nobody's putting it all together. Nobody's connecting the dots."
New lines of research could help on this, said American Medical Association Board of Trustees Chair Joseph M. Heyman, MD. Variations in care can occur when the best diagnostic workup or treatment plan has yet to be established, he said. "Findings from comparative effectiveness research will help clarify these questions and narrow the range of variation."
Rep. Bill Cassidy, MD (R, La.), who worked in a public hospital in Baton Rouge for 20 years, wonders how much of the variation in care and spending documented by the Dartmouth Atlas Project can be explained by obesity rates, poverty statistics and other demographics. In any case, Dr. Cassidy said, the work is attracting more attention from lawmakers as they begin crafting national health reform legislation. "It's going to be very influential."
The skeptics
Even some researchers who said the Dartmouth methodology is generally sound still aren't sure that 30% is a good estimate of the unnecessary care in the U.S. health system.
Kenneth Shine, MD, executive vice chancellor for health affairs at the University of Texas, responded in print to the 2003 Annals studies. Dr. Shine said then that Dartmouth had significantly advanced the understanding of health care variations. He said he had seen dozens of examples of patients receiving costlier and more invasive care than necessary.
While he stands by that assessment, however, Dr. Shine is not ready to acknowledge that 30% of health care in the U.S. is ineffective. That theory needs to be studied thoroughly to see if less intensive practice patterns could maintain high quality in a high-spending region, he said.
Moon, with the American Institutes for Research, also credited the Dartmouth researchers with advancing the discussion on health care variances. But many possible flaws exist within their 30% estimate of health care waste that could mean the real figure is lower or even higher, she said. "That's a very, very bad number to put faith in. Health care is such a complicated area that it's hard to take any big aggregate number with great certainty."
Others are more skeptical. Tom Rosenthal, MD, the UCLA Medical Center's chief medical officer, said the Medicare claims used by Dartmouth researchers in the 2008 Dartmouth Atlas of Health Care do not capture some unique demographic and workload issues facing higher-spending hospitals. For instance, Dr. Rosenthal said UCLA -- one of the highest spending hospitals in the Dartmouth Atlas -- has far more patients waiting for heart transplants and requiring costlier care than does the Mayo Clinic in Rochester, Minn. -- one of the lowest-spending hospitals.
Dr. Rosenthal is finishing a study of patients who had congestive heart failure or chronic liver disease at the five University of California hospitals. He anticipates that patient risk factors will account for most of the variation in the hospitals' end-of-life care as indicated by some of the Dartmouth research. The UCLA group does not intend to disprove the Dartmouth findings that variations in practice patterns exist, but it is concerned that a policy approach relying solely on these data could harm patients and penalize those who care for the most complex cases.
One physician critic stands out in his challenge of the Dartmouth Atlas Project. Richard Cooper, MD, a professor of medicine at the University of Pennsylvania, questions some of the project's basic methodology.
The 2003 studies assigned hospital referral regions to one of five quintiles, based on the levels of Medicare expenditures per enrollee. Quality rankings did not vary much between them. But Dr. Cooper said this was because each quintile contained a heterogeneous mix of referral regions and that averaging obscured known quality differences in some areas.
Dr. Cooper said Medicare spending is a bad proxy for overall health spending, because while some of the high-spending regions were poor and spent much less on younger patients, others were affluent and spent more. Total spending is what affects community-wide outcomes, he said.
Dr. Cooper said people looking to save the system money should examine spending on poor people. "Look at their outcomes, look at their costs and look at the adverse effects on their lives. That's where we have to intervene."
Dr. Fisher defended the 2003 research, saying its innovative method of risk adjustment was reviewed by six peers before publication -- more than customary -- and has been found to be valid by many others in the years since, including the CBO. None of the criticisms undermine the 2003 Dartmouth work, he said, in part because those studies used patient chart information for some of the patients as well as claims data.
When translating such research conclusions into policy, some choices mix the mathematical with the philosophical. Take the question on whether a negative CT scan of an injured knee is wasteful spending, Dr. Rosenthal said. Lawmakers must ponder whether there is value in a patient feeling better after knowing the results. "Is that waste?"