Abstract 13052: Stability of Hospital Rankings for Acute Myocardial Infarction

Abstract

Background: Although hospital mortality for acute myocardial infarction (AMI) varies widely, little is known regarding the trajectory of mortality over time at hospitals with high and low mortality at baseline. Methods: We studied 623,488 Medicare AMI patients at 1923 hospitals during 2007-2010. We used 30-day risk-standardized mortality rates (RSMR) to stratify hospitals as bottom 25% and top 25% in 2007 (baseline) and compared their performance on AMI mortality during next 3 years. We also ranked top 25% and bottom 25% hospitals into quartiles of RSMR, each year during 2008-2010 (0 = worst mortality quartile, to 3 = lowest mortality quartile), and summed quartile ranks for each year to yield a composite score (CS: range = 0 to 9). We examined which bottom 25% and top 25% hospitals achieved low ( < 1) and high CS (≤ 8). Results: Average RSMR in 2007 was 15.5% (range: 11.2%-21.6%). During 2008-2010, there was considerable overlap in AMI mortality at 481 hospitals that were top 25% (blue bars) and bottom 25% (red bars), respectively (Figure). Of bottom 25% hospitals, only 107 (22%) had persistently high mortality over next 3 years (CS < 1) where as 37 (8%) hospitals had marked improvement in AMI mortality (CS ≤ 8). Similarly, only 127 (26%) of top 25% hospitals had persistently low mortality over next 3 years (CS ≤ 8) and 37 (8%) had marked worsening in AMI mortality (CS < 1). Average RSMR at hospitals with persistently high mortality was 18.5% vs. 13.7% at hospitals with persistently low mortality, which led to an additional 834 Medicare deaths at high mortality hospitals. Patient and hospital characteristics were limited in differentiating hospitals with persistently high and persistently low mortality. Conclusions: Among hospitals with high and low mortality at baseline, performance on AMI mortality does not persist during subsequent years. Concentrating quality improvement efforts at a few hospitals with persistently high mortality could avoid a substantial number of AMI deaths.