Background: The surgical treatment of acute colonic diverticulitis is associated with significant morbidity and mortality. However, patient and operative characteristics associated with mortality in this patient population are unclear. We hypothesize that demographic and perioperative variables can be used to predict postoperative mortality.The purpose of this study was to identify perioperative variables predictive of postoperative mortality after emergent surgery for acute diverticulitis.
Methods: Patients with diverticulitis undergoing colostomy and/or partial colectomy with or without primary anastomosis were retrieved from the American College of Surgeons National Surgical Quality Improvement Program database for years 2005 to 2008 inclusive. Only patients undergoing emergent surgery for acute diverticulitis were included. Univariate analyses were performed to compare demographic characteristics, preoperative laboratory values, comorbidities, and intraoperative variables. Variables with a significant (p < 0.10) difference between survivors and nonsurvivors were included in a stepwise logistic regression model to determine predictors of 30-day mortality. Concordance indices (c indices) for postoperative mortality were calculated using 2005 to 2008 data to determine predictive accuracy and validated on 2009 data.
Results: A total of 2,214 patients met inclusion criteria. Mean age was 61 years, and 50% of patients were male. Thirty-day mortality was 5.1%. Nine preoperative variables were significantly associated with postoperative mortality on multivariable analysis. The c index of this nine-variable model was 0.901. Renal dysfunction, hypoalbuminemia, American Society of Anesthesiologists class, and age were chosen to create a simpler model, with a c index of 0.886 for 2005 to 2008 data and 0.893 for 2009 data.
Conclusion: Four readily available perioperative variables can be used to predict 30-day mortality after emergent surgery for acute diverticulitis.
Level of evidence: Prognostic study, level II.