The Differential Effects of Intermediate Complications with Postoperative Mortality JAMIL BORGI, M.D.,* ILAN RUBINFELD, M.D., M.B.A.,* JENNIFER RITZ, R.N.,† JACK JORDAN,† VIC VELANOVICH, M.D.‡ From the *Department of Surgery and the †Office of Clinical Quality and Safety, Henry Ford Hospital, Detroit, Michigan; and the ‡University of South Florida, Tampa General Hospital, Tampa, Florida Most attempts at understanding perioperative mortality have been based on assessing individual patient risk factors, types of operations, and hospital characteristics. The hypothesis of this study is that there is a relationship between postoperative mortality and postoperative complications; therefore, understanding this relationship may provide a basis for prevention and rescue. Using the 2007 SemiAnnual National Surgical Quality Improvement Program Report, we obtained data for each reporting hospital’s rates of observed mortality, overall observed morbidity, observed cardiac, respiratory, renal complications, venothromboemoblic events (VTEs), surgical site in- fections (SSIs), and urinary tract infections (UTIs). Simple and multiple linear regression analyses were done comparing absolute rate of observed mortality with absolute rate of observed mor- bidity and each morbidity group. One hundred ninety-seven hospitals were included in the study. There were statistically significant associations between observed mortality rates and observed morbidity rates, cardiac complications, respiratory complications, and VTE rates. Renal compli- cations, SSIs, and UTIs showed no statistically significant association with observed morbidity. This study demonstrates that rates of observed morbidity, especially cardiac, respiratory, and VTE complications, are associated with observed mortality. These findings suggest that care providers should focus efforts at prevention and rescue of cardiac, respiratory, and VTE complications. A PPROXIMATELY 234 MILLION operations are done an- nually worldwide. 1 In developed countries, per- ioperative mortality occurs in 0.4 to 0.8 per cent of operations. 2, 3 Using data from the 2006 National In- patient Sample provided by the Agency for Healthcare Research, we estimated that of the nearly nine million inpatient operations done, over 137,000 deaths oc- curred. 4 Therefore, trying to understand the relationship between operations and perioperative mortality is of great importance. de Leval 5 suggests that each death be subjected to forensic analysis, implying that knowing the cause can lead to its future prevention. The National Surgical Quality Improvement Program (NSQIP) was established in 1994 after the success of the National Veterans’ Affairs (VA) Surgical Risk Study that began in 1991. First limited to VA hospitals, the NSQIP provided risk-adjusted outcome data in the VA system to compare and ultimately improve hospital performance. 6–10 In 2003, the 30-day post- operative mortality rate had declined approximately 34 per cent (from 3.16 to 2.08%) and the 30-day post- operative morbidity rate by 41 per cent (from 17.44 to 10.3%). 11, 12 The American College of Surgeons (ACS) launched its NSQIP private sector initiative in 1998. The ACS NSQIP published its first report in 2005 and since then has reported risk-adjusted data to its member hospitals on a semiannual basis. The NSQIP has grown to include over 200 hospitals with a large database of over 200,000 surgical cases used in its statistical analysis. The model has proven to be a valid method of assessing hospital morbidity and mortality. 11 The data provided include rates of 30-day postoperative mortality and specifics morbidities as well as risk-adjusted observed rates of these complications to the expected rates (O/E ratios). The hypothesis of this study is that postoperative mortality is usually preceded by postoperative compli- cations; therefore, understanding which complications are more likely to be associated with postoperative mortality may provide a basis for focused prevention and rescue. Methods This study was deemed exempt by the Henry Ford Health System Institutional Review Board because no Presented at the American College of Surgeons Clinical Con- gress, Chicago, Illinois, 2009. Address correspondence and reprint requests to Vic Velanovich, M.D., University of South Florida, One Tampa General Circle, Tampa General Hospital, F145, Tampa, FL 33601. E-mail: vvelanov@ health.usf.edu. 261