Predictive analytics characterize cardiovascular function and outcomes following cardiac surgery
摘要
Using supervised machine learning, predictive models of cardiovascular and respiratory instability were previously created and display risk scores (CoMET score), characterizing the risk of clinical deterioration in the ensuing 12 h. We assessed the relationship of CoMET scores with vasoactive-inotropic scores (VIS) and short-term clinical outcomes in pediatric patients following cardiac surgery.
MethodsRetrospective cohort study of all patients admitted to the pediatric cardiac intensive care unit following cardiac surgery at a single center over an 18-month period. We performed comparative analyses to assess the relationships between CoMET scores and both VIS and postoperative clinical outcomes.
ResultsThere were 296 postoperative courses in 262 patients during the period of study. Both cardiovascular and respiratory CoMET scores during the first 48 postoperative hours were correlated with VIS (p < 0.001). Increased cardiorespiratory CoMET scores during the first 48 postoperative hours were associated with longer postoperative cardiac intensive care unit length of stay (p = 0.004), decreased ventilator-free days (p < 0.001), and increased incidence of necrotizing enterocolitis (p = 0.039), cardiac arrest (p = 0.19), and in-hospital mortality (p = 0.003).
ConclusionsDisplay of risk scores of cardiovascular and respiratory instabilities following cardiac surgery may help identify patients at risk for worse clinical outcomes, allowing for earlier intervention, potentially improving outcomes.
ImpactUsing previously established predictive models of risk of cardiovascular and respiratory instability, risk scores were correlated with vasoactive-inotropic scores in pediatric patients immediately following cardiac surgery. Increased risk scores following cardiac surgery were associated with increased intensive care unit length of stay, decreased ventilator-free days, and increased incidence of necrotizing enterocolitis, cardiac arrest, and in-hospital mortality. Display of risk scores of cardiovascular and respiratory instability following cardiac surgery may help identify patients at risk for worse clinical outcomes, allowing for earlier intervention, potentially improving outcomes.