Stratification of Bleeding Risk Using Thromboelastography in Children on Extracorporeal Membrane Oxygenation Support*

Objectives: Patients undergoing extracorporeal membrane oxygenation are at high risk for bleeding and thrombotic complications. Current laboratory methods for assessing the coagulation system may be imprecise and complicate clinical decision-making. We hypothesize that thromboelastography may be more strongly associated with bleeding events than traditional methods and can aid extracorporeal membrane oxygenation coagulation management. Design: In a retrospective study, 40 patients with congenital heart disease requiring extracorporeal membrane oxygenation support yielded a total of 159 patient days of data for thromboelastography analysis. Setting: Pediatric cardiac ICU at a single institution. Subjects: Pediatric patients (<= 18 yr) with congenital heart disease requiring extracorporeal membrane oxygenation support. Interventions: None. Methods: Thromboelastography was performed on whole blood samples collected 6-12 hours following extracorporeal membrane oxygenation initiation and daily for the duration of extracorporeal membrane oxygenation. Bleeding during each 24-hour period was defined as need for re-exploration or need for blood transfusion. Associations between thromboelastography variables and bleeding over each 24-hour period (bleeding vs nonbleeding days) were assessed using mixed effects logistic regression and classification and regression tree analysis. Measurements and main results: Bleeding occurred in 25 patients (63%), contributing 87 bleeding days (55% extracorporeal membrane oxygenation days) for analysis. The probability of bleeding within the 24-hour period was not associated with activated partial thromboplastin time (p = 0.6) or anti-Xa levels (p = 0.3) on that day. The strongest correlate of bleeding was a maximum amplitude less than 55.4 mm on thromboelastography (odds ratio, 3.28; 95% CI, 1.63-6.60; p < 0.001). Bleeding occurred on 73% versus 35% of extracorporeal membrane oxygenation days for maximum amplitude less than 55.4 mm versus greater than or equal to 55.4 mm, respectively. Bleeding occurred on all days when a combination of maximum amplitude less than 55.4 mm and a reaction time greater than 12.9 minutes was present. The lowest risk of bleeding (28% of patient days) was associated with maximum amplitude greater than or equal to 55.4 mm and plasma fibrinogen greater than 345 mg/dL. Conclusions: Thromboelastography-derived variables maximum amplitude and reaction time, along with plasma fibrinogen levels, can help predict bleeding events in children on extracorporeal membrane oxygenation support. Research based on larger patient samples is needed to confirm the specific thresholds identified for bleeding risk stratification for extracorporeal membrane oxygenation anticoagulation management.