UNDERSTANDING VARIATIONS IN SURGICAL MORTALITY IN MEDICARE PATIENTS
*Amir A Ghaferi, *Nicholas H. Osborne, *John D. Birkmeyer, *Justin B. Dimick
University of Michigan Health Systems, Ann Arbor, MI
OBJECTIVE: Wide variations in outcomes after major surgery are becoming increasingly apparent. We sought to determine whether the excess deaths at high mortality hospitals were due to higher complication rates or less success in managing complications once they occur.
METHODS: We studied all Medicare patients undergoing six major operations in 2005-2006: pancreatectomy, esophagectomy, abdominal aortic aneurysm repair, coronary artery bypass grafting (CABG), aortic valve replacement, and mitral valve replacement. We ranked hospitals according to risk adjusted operative mortality and divided them into quintiles. We then compared the incidence of complications and the mortality rates following these complications ("failure to rescue") across quintiles.
RESULTS: Hospitals in the highest mortality quintile had complication rates nearly 2-fold higher than the lowest quintile for pancreatectomy (24.4% vs. 44.0%) and esophagectomy (36.7% vs. 56.4%), but the other four operations showed no significant differences. However, there were large differences in the failure to rescue (FTR) rates between the highest and lowest mortality quintiles for all included operations. The largest difference was for pancreatectomy with a 10-fold higher rate of FTR between the highest and lowest mortality quintiles (3.9% vs. 48.1%). The smallest effect was for CABG with a 3-fold difference in FTR (4.1% vs. 12.9%).
CONCLUSIONS: Excess mortality at poorly performing hospitals appears to be attributable to large differences in failure to rescue, but only small differences in the incidence of major complications. Improvements in surgical quality will require strategies to improve the ability of low performing hospitals to manage major complications.