FAQ: How We Ranked the Best Hospitals 2014-15

The facts and methodology behind the latest adult hospital rankings in 16 specialties.

Doctor with patient examining X-ray


By + More

The rankings are just a starting point, then.

That's right. Patients have to do their own research. We also appreciate that families have to consider the stress and expense of traveling to another city, as well as the willingness of an insurer to pay for care at a hospital outside its network.

How many hospitals were analyzed for the 2014-15 rankings? Were teaching hospitals the only ones considered?
For the 12 data-driven rankings, we started with virtually all nonfederal community hospitals in the U.S. of any size, a universe that comprised 4,743 hospitals. (We very much want to include military and VA hospitals but have been unsuccessful in persuading the federal government to release the necessary data.)

It is not true, as even "experts" in the professional literature often claim, that we only look at teaching hospitals. From the starting pool of 4,743, a hospital qualified for consideration in the data-driven specialties if it met any of four possible criteria: teaching hospital status, medical school affiliation, 200 or more beds, or 100 or more beds plus availability of four or more specific medical technologies we consider key to high-quality care, such as a PET/CT scanner and certain precise radiation therapies. This year 2,280 hospitals, or 48 percent of the initial number, met at least one of those criteria. 

In the four specialties in which ranking was determined only through the physician survey, any hospital with enough nominations over the last three surveys was ranked.

How many hospitals were ranked or regional high performers?
Across all 16 specialties, only 144 U.S. hospitals performed well enough to be nationally ranked in one or more specialties. Another 608 were regional high performers. Just 17 qualified for a spot on the Honor Roll by ranking at or near the top in six or more specialties (more on the Honor Roll below). 

What was the process following the initial screening?
Hospitals next had to show that they treated enough patients to be eligible to be ranked in specific specialties. The volume, which varied by specialty, was a certain number of Medicare inpatients discharged from 2010 to 2012 after certain specialty-related procedures and treatment for certain conditions, each at a specifically defined level of severity and complexity. The threshold for gastroenterology & GI surgery, for example, was 563 patients, 148 of them surgical. For pulmonology the total was 1,033 with no required surgical minimum.

A hospital that fell short could still make it through the gate if nominated by at least 1 percent of the physicians in a specialty who responded to the 2012, 2013 and 2014 reputational surveys.

A total of 1,907 hospitals were eligible in at least one specialty.

What determined whether a hospital was ranked?
We examined each hospital's performance in various data categories. Some statistics came from the federal Centers for Medicare & Medicaid Services' MedPAR data base. Other information came from the American Hospital Association and professional organizations. The connection with quality and safety is evident for some of the statistics used, like death rates. For other categories, such as the number of patients and the balance of nurses and patients, the link may be less obvious but is nevertheless supported by ample research. The physician survey also played a role, though it accounted for only about one-fourth of each hospital's score (more on that below).

In the four reputational specialties, most care is delivered on an outpatient basis, and so few patients die that mortality data, which carry heavy weight in the 12 other specialties, mean little. Hospitals therefore are ranked solely based on reputation.

Corrected on July 15, 2014: A previous version of this story incorrectly stated that 581 hospitals were regional high performers. The correct number is 608.

Corrected on July 15, 2014: A previous version of this story neglected to include Vermont as a state with high-performing hospitals that are unranked.

Corrected on July 15, 2014: A previous version of this story referred to 94 metro areas in which hospitals are recognized. The correct number is 92.

patient advice