Any good hospital should be able to handle everyday procedures and conditions—repairing hip fractures, unclogging neck arteries, and removing gall bladders, to name three. Because routine operations are done so often, moreover, the numbers can easily be crunched to establish benchmarks and see how hospitals compare. That's what most public rankings and reports do. But measurements of garden-variety care have their limits. A medical center that does well with typical cases of peripheral vascular disease—circulatory blockages in the legs—may be the wrong place for elderly patients with recurring problems. They may be told the problem is untreatable and the next step is amputa-tion, when a top-notch surgeon might have a newer or better approach that could save the limb.
That patient, who truly needs outstanding care, is the target of the annual U.S. News "Best Hospitals" rankings, in which hospitals are judged not in routine procedures but in difficult cases across an entire specialty. This year, the 19th, hospitals are ranked in 16 specialties, from cancer and heart disease to respiratory disorders and urology. Out of the 5,453 hospitals put through a rigorous statistical mill, only 170 scored high enough to appear in any of the specialty rankings. (Data for military and veterans' hospitals are unavailable.) And only 19 of the 170 made the Honor Roll of medical centers ranked at or near the top in at least six specialties, as explained on Page 72. This year, for the first time, data and scores are available online for more than 1,500 hospitals that in the end fell short of being ranked.
Twelve of the 16 specialty rankings are driven largely by hard data; in four others, ranking is based on three years of nominations by specialists surveyed. To be considered at all for the 12 data-driven specialties, a hospital had to meet at least one of three requirements: membership in the Council of Teaching Hospitals, affiliation with a medical school, or availability of at least six of 13 key technologies such as robotic surgery. This year, nearly two thirds of all hospitals failed this first test.
If they passed, hospitals had to perform a certain number of specified procedures on Medicare inpatients in 2004, 2005, and 2006. The number varied by specialty—294 in orthopedics, for example. Or the hospital had to have been nominated by at least one physician in U.S. News surveys in 2006, 2007, and 2008.
That left 1,569 hospitals eligible for ranking in at least one data-driven specialty. Each facility received a U.S. News Score from 0 to 100 made up in equal parts of reputation, death rate, and care-related factors such as nursing and patient services. The 50 hospitals with the highest scores are ranked. Here is how the elements break down:
Reputation. For 2008, a random sample of 200 physicians for each of the 16 specialties was drawn from the American Board of Medical Specialties database. (For 2006 and 2007, the source was the American Medical Association Masterfile.) They were asked to list five hospitals they consider among the best in their specialty for difficult cases, without taking into account cost or location. The number for a hospital in the "reputation" column of the rankings is the combined percentage of responding physicians who listed the hospital in 2006, 2007, and 2008.
Mortality index. This ratio defines the ability to keep patients alive. It compares the number of Medicare inpatients with certain conditions who died within 30 days of admission in 2004, 2005, and 2006 with the number of deaths that would have been expected after adjusting for severity. An index number below 1.00 means the hospital did better than expected; a number above 1.00 means the hospital did worse than expected. Severity adjustments were made using 3M Health Information Systems APR-DRG software.
Other care-related factors. This information came from various sources, most prominently the American Hospital Association's 2006 survey of member and nonmember hospitals. It includes technology, volume, nurse staffing, and other patient-related information.