Risk factors and prediction model for inpatient surgical site infection after major abdominal surgery

Abstract

BackgroundSurgical site infections (SSIs) are a common source of postoperative morbidity and a marker of surgical quality. The ability to predict the incidence of SSIs is limited and most models have poor predictive value. We sought to identify risk factors associated with SSIs and develop a prediction model for SSIs after major abdominal surgery. MethodsA total of 1744 patients undergoing pancreatic, hepatobiliary, and colorectal resections between January 1, 2010 and August 31, 2013 at Johns Hopkins Hospital were identified. Risk factors for any inpatient SSI (superficial and deep) were evaluated using multivariable logistic regression. ResultsMedian patient age was 58 y (interquartile range 47, 68); surgical procedures included colorectal (59.0%), liver (26.2%), and pancreas (14.8%) resections. SSI occurred in 7.6% (n = 132) of patients. Factors associated with SSI included preoperative weight loss >4.5 kg (odds ratio [OR], 2.12; 95% confidence interval [CI], 1.06-4.25), emergency operations (OR, 2.05; 95% CI, 1.32-3.17), and colorectal resections (OR, 1.65; 95% CI, 1.13-2.43) (all P ≤ 0.003). Intraoperative and postoperative risk factors included estimated blood loss (EBL) >600 mL (OR, 2.23; 95% CI, 1.54-3.25), maximum respiratory rate (tachypnea) >20 breaths/min (OR, 1.74; 95% CI, 1.19-2.54), and perioperative transfusion (OR, 2.01; 95% CI, 1.33-3.04) (all P = 0.001). Intraoperative hypothermia, hyperthermia, bradycardia, tachycardia, hypotension, and hypertension were not associated with SSIs (all P > 0.05). After controlling competing risk factors, transfusion, EBL >600 mL, tachypnea, and colorectal resection were independently associated with SSIs (all P < 0.003). On the basis of the beta-coefficients in the multivariable model, an SSI scoring system was created by assigning 2 points for EBL >600 mL, 2 points for a colorectal resection, 3 points for tachypnea, and 3 points for a transfusion. The model showed good discriminatory ability to predict SSI (c-statistic = 0.7232; Akaike information criterion 875.37). ConclusionsA novel, simple 10-point SSI scoring system that incorporated perioperative risk factors such as blood transfusion, EBL, tachypnea, and the type of surgical procedure accurately stratifies patients according to SSI risk.