Patients' Perceptions and Medicare Data

By now, most everyone has seen USA Today’s August 5, 2011 headline: “Deaths Plague Even Top Hospitals.” The point of the article, and accompanying articles, was the newspaper’s analysis of Medicare data on the relationship between patients’ perceptions of the quality of their care and objective measures. Although patients at 120 hospitals felt that these hospitals provided excellent care, the hospitals actually had higher than average death rates for heart attack, heart failure, or pneumonia.

The Medicare's analysis of data from 4,700 hospitals found that 323 (about 7%) had above-average death rates for heart attack, heart failure, or pneumonia. Two hospitals had high death rates for all three measures. Only 13 hospitals had lower than average death rates. The national death rate averages are 15.9% for heart attacks, 11.3% for heart failure, and 11.9% for pneumonia. On average, 67% of patients would recommend the hospital in which they received care, which means that about third of patients would not.

Of note, Medicare's database is limited to patients 65 and older and data was abstracted from records of patients hospitalized between July 2007 and June 2010. Therefore, the disconnect between patients’ perceptions and objective measures cannot be generalized to all patient populations. Further, patient surveys are notoriously inaccurate for many reasons. The return rate is low and it’s well documented that responders tend to be those who are very pleased and those who are very displeased with their care. And many times care is confused with customer service; it may be that patients are not getting the attention they feel they should, yet they attribute this to “poor care.” Also, the hospitals that higher than average death rates were generally small town community hospitals.

The USA Today headline will likely cause the curious to look up how their hospital compares to others (data can be found at http://www.hospitalcompare.hhs.gov/). The problem, though, will be what to make of the data.