Risk stratification models for congenital heart surgery in children: Comparative single‐center study

Objective Three scores have been proposed to stratify the risk of mortality for each cardiac surgical procedure: The RACHS-1, the Aristotle Basic Complexity (ABC), and the STS-EACTS complexity scoring model. The aim was to compare the ability to predict mortality and morbidity of the three scores applied to a specific population. Design Retrospective, descriptive study. Setting Pediatric and neonatal intensive care units in a referral hospital. Patients Children under 18 years admitted to the intensive care unit after surgery. Interventions None. Outcome measures Demographic, clinical, and surgical data were assessed. Morbidity was considered as prolonged length of stay (LOS > 75 percentile), high respiratory (>72 hours of mechanical ventilation), and high hemodynamic support (inotropic support >20). Results One thousand and thirty-seven patients were included, in which 205 were newborns (18%). The category 2 was the most frequent in the three scores: In RACHS-1, ABC, 44.9%, and STS-EACTS, 40.8%. Newborns presented significant higher categories. Children required cardiopulmonary bypass in more occasions (P < .001) but the times of bypass and aortic cross-clamp were significantly higher in newborns (P P = .016). Thirty-two patients died (2.8%). A quarter of patients had a prolonged LOS, 17%, a high respiratory support, and 7.1%, a high hemodynamic support. RACHS-1 (AUC 0.760) and STS-EACTS (AUC 0.763) were more powerful for predicting mortality and STS-EACTS for predicting prolonged LOS (AUC 0.733) and the need for high respiratory support (AUC 0.742). Conclusions STS-EACTS seems to stratify better risk of mortality, prolonged LOS, and need for respiratory support after surgery.