"We took what is simply a statistical method and put in quality data and what that method told us was which hospitals were of higher or lower quality," Lieberthal says.
"Then we validated that by looking at whether the quality scores in one year predicted the outcomes in the next year. We created a score for 2010—these are the higher and lower quality hospitals—and then we validated that that was correlated with mortality in 2011. We both applied the method and validated it by saying that it could predict future outcomes."
Lieberthal's study relied on Medicare data available on Hospital Compare and other data from the American Hospital Association.
"Our method was designed to take these different types of data, hospital characteristics, process measures of care, outcome measures, and satisfaction and then from those measures, to determine the score. We didn't decide ahead of time what was important. Our statistical method told us which were the important measures and which ones were correlated with quality," he says.
"Hospitals that scored the highest or 'always' on satisfaction measures such as patients received help as soon as they wanted it or nurses or doctors communicated well, that was correlated with lower quality as measured through our method."