Impact of Inflammation After Cardiac Surgery on 30-Day Mortality and Machine Learning Risk Prediction

OBJECTIVES: To investigate the impact of systemic inflammatory response syndrome (SIRS) on 30-day mortality following cardiac surgery and develop a machine learning model to predict SIRS. DESIGN: Retrospective cohort study. SETTING: Single tertiary care hospital. PARTICIPANTS: Patients who underwent elective or urgent cardiac surgery with cardiopulmonary bypass (CPB) from 2016 to 2020 (N = 1,908). INTERVENTIONS: Mixed cardiac surgery operations were performed on CPB. Data analysis was made of preoperative, intraoperative, and postoperative variables without direct interventions. MEASUREMENTS AND MAIN RESULTS: SIRS, defined using American College of Chest Physicians/Society of Critical Care Medicine parameters, was assessed on the first postoperative day. The primary outcome was 30-day mortality. SIRS incidence was 28.7%, with SIRS-positive patients showing higher 30-day mortality (12.2% v 1.5%, p < 0.001). A multivariate logistic model identified predictors of SIRS. Propensity score matching balanced 483 patient pairs. SIRS was associated with increased mortality (OR 2.77; 95% CI 1.40-5.47, p = 0.003). Machine learning models to predict SIRS were developed. The baseline risk model achieved an area under the curve of 0.77 ± 0.04 in cross-validation and 0.73 (95% CI 0.70-0.85) on the test set, while the procedure-adjusted risk model showed improved performance with an area under the curve of 0.81 ± 0.02 in cross-validation and 0.82 (95% CI 0.76-0.85) on the test set. CONCLUSIONS: SIRS is significantly associated with increased 30-day mortality following cardiac surgery. Machine learning models effectively predict SIRS, paving the way for future investigations on potential targeted interventions that may mitigate adverse outcomes.