Understanding Variations in Surgical Mortality: Differences in Complications or Failure to Rescue? Amir A. Ghaferi, MD AcademyHealth Annual Research Meeting June 28, 2009 Wide variations in surgical mortality • Surgical mortality is a public health issue – 50,000 operative deaths a year in US • Chance of death depends on where and by whom the surgery is performed – Surgical mortality rates vary widely across hospitals Strategies for reducing mortality • Efforts to reduce surgical complications – Pay for performance • Medicare’s Surgical Care Improvement Project (SCIP) – (Non)pay for (Non)performance • Not reimbursing for certain complications • Uncertain whether these efforts will reduce variations in mortality Explaining variations in surgical mortality rates Hypothesis #1 Operation Hypothesis #2 Postoperative complication High mortality hospitals have higher complication rates Death High mortality hospitals are not as effective in “rescuing” patients once they develop a complication Previous work assessing Failure to Rescue • In response to increasing use of mortality as a quality measure – Evaluated patients undergoing cholecystectomy and transurethral prostatectomy – Patient characteristics Complications – Hospital characteristics Failure to rescue Silber et al. Med Care 1992. Objective To determine the extent to which complication incidence and failure to rescue rates explain variations in mortality across hospitals Methods • Data source – 2005-07 American College of Surgeons National Surgical Quality Improvement Project (ACS-NSQIP) • Subjects – Patients undergoing inpatient general and vascular procedures with overall mortality >1% (n=84,730) • Outcome measures – Operative mortality – Postoperative complications – Failure to rescue (ie., case-fatality among patient with complications) Analysis • Ranked hospitals based on riskadjusted mortality rates • Divided hospitals into 5 equal groups (quintiles) • Compared incidence of major complications and “failure to rescue” rates across hospitals Variations in Mortality 6.9 7 5.8 6 4.6 5 Risk-adjusted mortality (%) 4 4.8 3.5 3 2 1 0 1 2 3 4 Hospital Quintiles 5 Patient severity across hospitals Median age Gender (%male) Non-white race Smoking status (within past year) ASA class ≥4 Diabetes Chronic Obstructive Pulmonary Disease History of Myocardial infarction Emergency operation Dialysis Albumin <3.5 Hypertension Expected mortality† Risk-adjusted mortality Very low Low Medium High Very high (n=17,379) (n=16,780) (n=17,923) (n=15,953) (n=16,695) 63.6 63.0 63.9 61.7 62.4 52.0% 52.0% 51.2% 54.4% 51.6% 18.9% 14.5% 14.6% 24.1% 26.4% 22.2% 23.2% 23.8% 26.6% 27.4% 15.7% 14.3% 14.3% 16.7% 15.9% 20.2% 19.4% 19.3% 21.3% 21.7% 8.4% 8.6% 7.9% 9.2% 9.1% 1.5% 1.8% 1.3% 1.8% 1.8% 18.3% 18.8% 19.3% 18.2% 19.5% 3.6% 4.1% 3.1% 4.6% 4.7% 26.9% 23.5% 28.1% 27.4% 27.0% 58.2% 56.9% 57.6% 58.6% 59.3% 5.2% 3.5% 5.4% 4.6% 4.8% 4.8% 5.1% 5.8% 4.8% 6.9% †Expected mortality derived from patient characteristics (age, sex, race, comorbidities) and hospital procedure mix. Complications and Failure to Rescue: All operations 35 Hospital Mortality (Quintiles) 30 27.6 26.9 26.9 Very low 24.6 25 Low 23.5 21.4 19.3 Patients (%) 20 18.4 18.0 18.2 17.5 16.2 14.9 14.7 15 12.5 10 5 0 All Complications Major Complications Failure to Rescue Medium High Very high Odds Ratios, best vs. worst hospitals Complication Incidence 1.0 1.0 Pneumonia Prolonged mechanical ventilation Unplanned intubation 1.0 1.7 1.2 1.7 1.9 1.3 Acute renal failure 1.7 1.0 Myocardial infarction 1.6 1.0 Pulmonary embolism Postoperative bleed Deep wound infection Failure to Rescue 1.0 1.7 1.2 1.7 2.3 0.9 Organ space infection 1.9 1.0 Septic Shock 2.1 1.5 Fascial dehiscence Stroke 1.0 0.7 1.0 3.0 Summary • High mortality hospitals have: – very similar complication rates – markedly higher failure to rescue rates • Findings consistent across individual operations and complications Implications • Policy efforts aimed at preventing complications may not reduce variation in hospital mortality • Reducing variation in mortality will require greater focus on the timely recognition and management of complications once they occur • Need to better understand processes of care and resources related to failure to rescue • Surviving Sepsis Campaign • Dedicated ICU intensivists • Higher nurse to patient ratios