Predicting cardiac surgical site infection: development and validation of the Barts Surgical Infection Risk tool

Objective: To develop and validate a new risk tool (Barts Surgical Infection Risk (B-SIR)) to predict surgical site infection (SSI) risk after all types of adult cardiac surgery, and compare its predictive ability against existing (but procedure specific) tools: Brompton-Harefield Infection Score (BHIS), Australian Clinical Risk Index (ACRI), National Nosocomial Infection Surveillance (NNIS). Design: Single-centre retrospective analysis of prospectively collected data. Patients and Setting: Data from 2,449 patients undergoing cardiac surgery between January 2016 and December 2017 from one European tertiary centre were included. Methods: Thirty-four variables associated with SSI risk after cardiac surgery, identified from the literature, were collated from three local databases. Independent predictors were identified using stepwise multivariate logistic regression. Bootstrap resampling was conducted to validate the model. Hosmer-Lemeshow goodness of fit test was performed to assess calibration of scores. A p-value of <0.05 was considered statistically significant for all analyses. Results: The B-SIR model was constructed from six independent predictors (female gender, body mass index (BMI) >35, diabetes, left ventricular ejection fraction (LVEF)