Abstract 20: Public Reporting of Mortality Rates for Cardiovascular Conditions Did Not Improve Patient Outcomes

Abstract

Background: Public reporting may improve provider performance, although evidence for its impact has been mixed. In the U.S., public reporting of hospital performance became widespread in late 2004. Led by the Centers for Medicare and Medicaid Services (CMS), this program initially reported process-of-care metrics only, but in 2008 expanded to include public reporting of 30-day mortality rates. We set out to determine whether public reporting of mortality rates led to improvements in mortality rates for cardiovascular conditions. Methods: We used national Medicare data from 2005-2011 to compare trends in risk-adjusted 30-day mortality rates for acute myocardial infarction (AMI) and congestive heart failure (CHF), two publicly reported conditions, for hospitals participating in the Hospital Compare program throughout the study period. We considered 2005 through 2007 as the process reporting period (pre-mortality-reporting period), and 2008 through 2011 as the mortality reporting period. We performed linear regression analyses with hospital random effects to determine if trends in mortality changed after the implementation of public reporting of mortality rates. For comparison, we also selected two common non-reported cardiovascular conditions, atrial fibrillation and stroke, and examined trends in mortality rates over the same time period. Results: There were 3,617 hospitals in our sample. Mortality rates for AMI were improving at a rate of -0.09% per quarter during process reporting, but the improvement slowed to a rate of -0.05% per quarter during outcomes reporting (+0.04% per quarter, p=0.016). For CHF, mortality was improving at -0.03% per quarter during process reporting, but worsened at +0.06% per quarter during mortality reporting (change of +0.09% per quarter, p<0.001). Interestingly, improvements in mortality rates also slowed for the non-reported cardiovascular conditions over the study period (Table). Conclusions: We found that mortality rates for publicly reported conditions were improving slightly during the period when only processes of care were being reported, but that these improvements slowed or reversed once public reporting of mortality rates began. These findings suggest that public reporting may be necessary, but is clearly not sufficient, to improve patient outcomes.