Development of a model for predicting major infection following pediatric heart surgery

Abstract

Objective: The aim of this study was to develop a risk prediction model for major postoperative infection (MPI) after pediatric heart surgery and to validate the model of the Society of Thoracic Surgeons (STS). Materials and methods: We analyzed a retrospective cohort of 1,025 children who underwent heart surgery with cardiopulmonary bypass (CPB) from 2000 to 2010. We used a logistic regression model, which was validated. Results: Of the 1,025 patients, 59 (5.8%) had at least one episode of MPI (4.8% had sepsis, 1% had mediastinitis, 0% had endocarditis). Hospital mortality (63% vs. 13%; p < 0.001), as well as duration of postoperative ventilation (301.6 vs. 34.3 hours; p < 0.001) and intensive care unit stay (20.9 vs. 5.1 days; p < 0.001) were higher in patients with MPI. The predictive factors found were age, sex, weight, cyanotic heart disease, RACHS-1 3-4, Ross-modified functional class IV, previous hospital stay, and previous history of mechanical ventilation. The proposed model had a c-statistic of 0.80 (95% CI: 0.74-0.86) and was considered as clinically useful. The STS model showed a c-statistic of 0.78 (95% CI: 0.71-0.84) and a Hosmer-Lemeshow of 18.2 (P = 0.020). A comparison between the two models was made using an accurate Fisher test. Conclusion: A model with good performance and calibration was developed to preoperatively identify children at high risk for severe infection after cardiac surgery with CPB. The STS model was also validated and was found to have a moderate discrimination performance.