Identifying potential predictive indicators of massive transfusion in pediatric trauma

Objective: Predictive scores were developed in adults to identify patients at risk for traumatic hemorrhage but have not been studied in children. Our objective was to identify clinical predictors of massive transfusion (≥40 ml/kg red blood cells or ≥80 ml/kg total blood product at 24 h) in injured children. Methods: We conducted a retrospective case–control study of children <18 years old who presented from 2005 to 2014 for trauma care. Cases were children who received massive transfusion. Two control groups were identified: (1) ‘random’ and (2) matched for age and injury severity score (matched). Variables included vital signs, injury mechanism, active bleeding, and laboratory values. Multivariate logistic regression models to predict massive transfusion were developed using only clinical findings (‘pre-arrival’ model) and then also including laboratory values (emergency department model). Results: Of 11,995 injured children, 44 received massive transfusion. We selected 132 random and 127 matched controls. Using random controls, the pre-arrival model included heart rate, Glasgow Coma Scale, temperature, and penetrating injury mechanism (AUC = 0.965). With matched controls, the pre-arrival model included Glasgow Coma Scale, penetrating injury, and active bleeding resulted (AUC = 0.812). The ED model using random controls included Glasgow Coma Scale, hemoglobin, and penetrating injury (AUC = 0.987). The emergency department model using matched controls included hemoglobin, temperature, prolonged partial thromboplastin time, and active bleeding on arrival (AUC = 0.734). Conclusions: We identified predictive models for children who receive massive transfusion that rely on only clinical findings pre-arrival to the emergency department and then incorporate laboratory tests for those in the emergency department.