Predicting cardiac arrest after neonatal cardiac surgery

Abstract

Abstract Objective In-hospital cardiac arrest (IHCA) following cardiac surgery is a rare but consequential event with detrimental effects on patient outcomes, including morbidity, mortality, and long-term neurologic outcomes. Neonatal patients are the most vulnerable population. We aimed to create a model to identify neonates at the highest risk of suffering IHCA early in their postoperative course using readily available candidate physiologic and laboratory variables. Methods Single-center, retrospective cohort. Results Of 118 postoperative neonates, IHCA occurred within 48 h in 10% of the cohort (n = 12). Multiple strategies were employed in the development of a risk prediction model for IHCA. The best performing model contained vasoactive-inotropic score (VIS) at 2 h after admission, admission lactate level, and change in VIS from admission to 2 h post-admission. The model characteristics were training mode—area under the receiving operating curve (AUROC) 0.99 (95% CI 0.99–1.00), sensitivity 91.7%, specificity 98.1%; test model—AUROC 0.92 (95% CI 0.76–1.00), sensitivity 75.0%, specificity 97.2%. Conclusion We derived a risk prediction model for neonatal IHCA after congenital heart surgery that is simple and capable of predicting early IHCA within 2 h of postoperative admission to the cardiac intensive care unit. Pending external validation, our model may be used to identify neonates who may benefit from targeted interventions and prevent IHCA after cardiac surgery.