In health care, the treatment you receive often depends on where you live, not on the best course of action.
Older women in Buffalo, for instance, are twice as likely to get a mastectomy for breast cancer than patients in Burlington, Vt., but they are half as likely to undergo the surgery than women in Tupelo, Miss.
Instead of medical therapy, men over 65 with an enlarged prostate are twice as likely to get prostate surgery here than in Bend, Ore., but far less likely to get the operation than if they lived in Minot, N.D.
And elderly patients on average spend more than twice as many days in the hospital in Buffalo in the last six months of life than in Boise, Idaho, but almost half as many days as in Manhattan.
What does it all mean?
The huge differences in the cost, quality and quantity of health care across the United States cannot be completely explained by differences in the sickness of patients.
"We are a death-defying society. Patients expect to be cured. At the same time, doctors are not good at predicting the survival of patients, and there are not good incentives to do it either. It implies failure," said Dr. Jack Freer, an internist and ethicist at the University at Buffalo School of Medicine and Biomedical Sciences.
Doctors and hospitals often have good reasons for taking an aggressive or conservative approach.
But in many instances, more costly tests and procedures, as well as prolonged stays in hospitals and intensive-care units, don't lead to better results for patients. In addition, studies suggest that the more surgeons and hospital beds in a region, the more often patients are admitted and have operations.
This, in turn, creates significant variations across the country in the cost of treating comparably sick patients with the same illnesses.
"In health care, what's remarkable is that where you live is as important or more important than who you are," said Dr. David Goodman, co-author of the Dartmouth Atlas of Health Care.
Researchers at the Dartmouth Atlas have been at the forefront of studying these variations in medical practice using data about patients enrolled in Medicare, the federal health program for people 65 and older. The information for 2005, although not representative of all patients, remains among the best available to draw a picture of health care from community to community.
For Buffalo, two trends stand out:
Older patients here were hospitalized and underwent procedures like back surgery and knee replacements less often than patients elsewhere in the United States. That does not mean they didn't receive health care, only that their doctors used alternative treatments.
In the last two years of life, especially the final six months, older patients tended to spend more time in hospitals in Western New York and were more likely to die in the hospital instead of at home or in hospice or nursing facilities.
"Buffalo is notable for the high portion of patients who die in the hospital. This represents high use of hospital resources without necessarily better outcomes," Goodman said.
Some of those patients probably died in hospital beds that can "swing" from general inpatient use to a focus on the sort of palliative care that hospice provides. In addition, critics say the atlas might not account well enough for differences in patients.
Hospitals in the Buffalo area, nevertheless, ranked in the 85th percentile nationally for hospital stays during a terminal hospitalization, indicating a higher rate than 85 percent of the hospitals studied. Buffalo also was at the 82nd percentile for average number of days patients spent in the hospital in the last six months of life and at the 77th percentile for percent of deaths occurring in the hospital.
Drawing conclusions about the Buffalo area from the atlas is difficult, UB ethicist Freer said. The statistics carry the larger message that the country as a whole lacks an effective system to deal with aggressive but futile care at the end of life, he said.
Unneeded tests and procedures put patients at risk of infections and complications, and can bring misery to the final days of patients' lives. Yet deciding what is appropriate care is complex. Often, whether a therapy or procedure would be effective or futile remains unclear, and many patients improve with intensive treatment, Freer said.
So, who is getting the right amount of care and in which hospitals?
That is difficult to say. The Dartmouth Atlas, for now, does a better job of making information public to illustrate the larger disturbing trend of too many patients receiving excessive care, with its greater risks and costs, but without better results.
"End-of-life-care is part of the bigger problem of overuse, underuse and misuse of health care in general," said Bruce Boissonnault, president of the Niagara Health Quality Coalition, a leader in publishing data on hospital quality.
The gaps can't be easily dismissed by citing cultural differences between East Coast and West Coast, or North and South. Wide variations can be found within the state.
The rate of surgery, for instance, ranges from 84.1 cases per 1,000 Medicare enrollees in Buffalo to 101 per 1,000 in East Long Island.
Figures also show racial disparities. Consider leg amputations, an uncommon complication of peripheral vascular disease and diabetes.
In the Buffalo area, African-Americans are more likely to lose a leg than whites -- 1.91 amputations per 1,000 for black Medicare enrollees compared with 0.99 per 1,000 for whites. Yet, of the regions the atlas studied, Buffalo reflected the smallest difference between blacks and whites.
The disparities over amputations were significantly greater elsewhere, such as in Memphis, Tenn., where the rate for blacks was six times higher than for whites, suggesting blacks were not receiving adequate preventive care.
The atlas -- actually a collection of a handful of detailed studies -- raises questions that are not easily answered.
The geographic differences indicate that factors other than evidence-based decisions on what's best for patients contribute to whether patients get hospitalized, tested or undergo surgery.
Goodman recommends more integrated health services and reform of a medical reimbursement system that favors high-intensity care over basic primary care.
The atlas, although complicated to use, is available at www.dartmouthatlas.org.