Our Methodology: Inside the 2012-13 Best Children's Hospitals Rankings

Here's how U.S. News selected the hospitals that do best at treating very sick children

By SHARE

The outcomes and care-related indicators were made up of scores of fine-grained measures, listed in the glossary. How we analyzed the data and put it together is fully described in the Best Children's Hospitals Methodology Report, a viewable and downloadable PDF file. Here are the basics of the three major components, each of which counted as one-third of a hospital's score:

Outcomes. Nothing matters more than outcomes—keeping kids alive, protecting them from infections and surgical complications, and improve the quality of life of those with chronic conditions. More research and better data collection and analysis have boosted the amount and reliability of outcome-related information that goes into the rankings. This year's additions, for example, include a measure of bloodstream infections caused by urinary catheters. Other kinds of data evaluate survival from various childhood cancers and from a complicated, multistage heart surgery, rates of bloodstream infection in intensive care units, and successful management of chronic conditions like asthma and cystic fibrosis.

Process of care. Previously, the process component meant one thing: a hospital's reputational score, based on a random survey of 150 board-certified pediatric physicians per specialty who were asked to name the five best hospitals in their area of expertise for children with serious or difficult medical problems, without considering location or expense. A hospital's score was the average of the most recent three years of survey responses.

To reputation we have now added several elements. One is a hospital's compliance with widely endorsed "best practices" such as conducting morbidity and mortality conferences to discuss unanticipated deaths or complications. A second is commitment to infection control through measures such as having a specialized "infection preventionist" on staff and tracking the correct use of antibiotics prior to surgery to prevent postop infections.

We have also changed how reputation is factored into a hospital's final score to reduce the tendency of high-reputation hospitals to outrank hospitals that, despite lower reputation scores, have records of significantly stronger performance on outcomes and other measures. The weight of reputation hasn't changed. It is still 25 percent of a hospital's overall 0-100 score. But how much of that 25 percent is credited to a hospital now depends entirely on the hospital's own reputational score. If it was nominated by 50 percent of the surveyed physicians, it receives 50 percent of 25 percent in the final score, or 12.5 "points."

This differs in two ways from past years. Before, the hospital with the highest reputational score in a specialty got the entire 25 percent (25 points in its final score) no matter how many or how few physicians nominated it, and the credit to other hospitals was scaled down accordingly. In this standard statistical practice, called normalization, hospitals got whatever part of 25 points corresponded to their reputational score times the top hospital's. If the top-scoring hospital had a reputational score of 75 percent and the next hospital got 45 percent, hospital No. 2 got 45 percent times 75 percent as its share of 25 percent, or 15 points in the final score.

In the revised methodology, however, a top-scoring hospital with a 75 percent reputational score receives 75 percent of the full 25 percent, or 18.75 points. A hospital with a 45 percent reputational score gets 45 percent times 25 percent, or 11.25 points in the final score. Last year's 10-point difference would narrow to 7.5 points.

More than 54 percent of the surveyed physicians submitted responses, a three-point increase from last year's already extremely high rate for such a survey.

Mix of care-related indicators. Surgical volume, nurse-patient ratio, and condition-specific clinics and programs are a few examples of an assortment of 38 measures, many with numerous submeasures, that were used across the 10 specialties.


Corrected on : *Editor's Note, 6/14/2012: Following publication, programming errors were discovered that produced misrankings of some hospitals in seven specialties. The rankings have been corrected and, for five specialties, extended beyond the top 50 to either 51 or 52. For more information including how the rankings changed, get details here.