Prediction of bleeding in paediatric cardiac surgery using clinical characteristics and prospective coagulation test results

Bleeding caused by coagulopathy is common in children undergoing cardiac surgery and causes adverse outcomes. Coagulation testing assists selection of treatments to stop bleeding but has an uncertain role for predicting bleeding. We aimed to evaluate how well prospective coagulation testing predicted excessive bleeding during and after cardiac surgery compared to prediction using clinical characteristics alone. A single centre, prospective cohort study in children having a range of cardiac surgery procedures with coagulation testing at anaesthetic induction and immediately after cardio-pulmonary bypass. The primary outcome was clinical concern about bleeding (CCB), a composite of either administration of pro-haemostatic treatments in response to bleeding or a high chest drain volume after surgery. In 225 children, CCB occurred in 26 (12%) during surgery and in 68 (30%) after surgery. Multivariable fractional polynomial models using the clinical characteristics of the children alone predicted CCB during surgery (c-statistic 0.64; 95% confidence interval 0.53, 0.76) and after surgery (0.74; 0.67, 0.82). Incorporating coagulation test results into these models improved prediction (c-statistics 0.79; 0.70, 0.87 and 0.80; 0.74, 0.87 respectively). However, this increased the overall proportion of children classified correctly as CCB or not CCB during surgery by only 0.9% and after surgery by only 0.4%. Incorporating coagulation test results into predictive models had no effect on prediction of blood transfusion or post-operative complications. Prospective coagulation testing marginally improves prediction of CCB during and after cardiac surgery but the clinical impact of this is small when compared to prediction using clinical characteristics.