Health Affairs: Does publicly reporting hospital data miss the mark?

The goal of websites that publicly report hospital data is to improve the quality of care; however, a study published March 5 in Health Affairs showed that these sites, such as Hospital Compare, may not meet overarching goals. In fact, researchers found that public reporting for at least three conditions—heart failure, pneumonia and heart attack—had little or no impact on patient mortality, and didn't make patients flock to higher performing hospitals.

“Public reporting of quality measures has been widely implemented in healthcare to increase transparency and accountability, spur quality improvement, and steer patients toward high-quality providers,” Andrew M. Ryan, PhD, of the Weill Cornell Medical College in New York City, and colleagues wrote. “Despite implementation in nearly all U.S. hospitals, it is unknown whether the Hospital Compare program has improved patient outcomes for these three conditions [heart attack, pneumonia and heart failure].”

And while it has been said that hospital performance may be impacted by these public reports, it remains unknown whether they can reduce risk-adjusted mortality. Additionally, the authors said it remains unclear whether Hospital Compare websites lead patients to choose hospitals with lower mortality rates.

To better understand the impact these quality care websites have on clinical practice, Ryan et al used 2000-2008 Medicare claims data to estimate the effect Hospital Compare had on 30-day mortality rates for myocardial infarction (MI), heart failure (HF) and pneumonia. All patients were admitted between Jan. 1, 2000, and Nov. 30, 2008.

The Centers for Medicare & Medicaid Services (CMS) began its public reporting initiative in 2005. And because of rather low rates of reporting, in 2004 Medicare payments to hospitals were made to be conditional on reporting for 10 of the 17 indicators, which helped increase reporting nearly 98 percent in acute care hospitals. The Hospital Compare website was launched in April 2005 and displays process-of-care measures such as whether patients presenting at hospitals with MI received aspirin or whether pneumonia patients received influenza vaccines. The website began reporting hospital mortality rates in June 2007.

During the current study, the authors used linear regression models to estimate the incremental changes in mortality associated with the start of the Hospital Compare website. Additionally, the researchers tested whether or not quality improvements occurred after conditions were publicly reported to the Hospital Compare website.

After Hospital Compare, risk-adjusted mortality was found to improve during the study period for all three conditions studied—MI, pneumonia and heart failure. However, the authors said that this could be due to the “continuation of pre-existing trends.” The researchers noted that only for HF was there a “substantial deviation” from the trends prior to the launch of Hospital Compare.

Relative risk ratios for risk-adjusted mortality after Hospital Compare were 0.83 for MI, 0.87 for HF and 0.78 for pneumonia. But, when the authors adjusted for the pre-existing trends in mortality for the three publicly reported diagnoses, these risk ratios decreased and were 0.97 for MI, 0.92 for HF and 1.01 for pneumonia.

“In the analysis that further adjusted for trends in risk-adjusted mortality in non-reported conditions, Hospital Compare was no longer associated with reduced mortality for heart attack and was associated with increased mortality for pneumonia: Relative risk ratios were 1.01 for heart attack, 0.97 for heart failure, and 1.07 for pneumonia,” the authors wrote.

“Prior studies provide mixed support for public quality-reporting programs as a means of improving patient outcomes,” the authors noted. “Survey evidence has shown that quality report cards are underused by patients and even explicitly ignored by referring physicians.” However, in the current study the authors found that public reporting did not result in patients choosing higher-quality hospitals.

The authors did question why there was a mortality reduction for HF but not MI or pneumonia. “This finding also raises the larger issue that hospital improvement on the process measures does not appear to have reduced mortality for the publicly reported diagnoses,” the authors said.

During the study the authors found a higher risk-adjusted mortality rate for pneumonia, which they attributed to the fact that it may have been more difficult for hospitals to further reduce pneumonia from the steep decline in pneumonia seen between 2000 and 2005.

“These findings do not imply that public reporting in general, or Hospital Compare in particular, cannot improve patient outcomes. It only signals that the specific design of Hospital Compare during the period we evaluated did not substantially reduce mortality,” the authors summed.