Survival bias in pediatric hemorrhagic shock: Are we misrepresenting the data?

Abstract

Background Studies of hemorrhage following pediatric injury often use the occurrence of transfusion as a surrogate definition for the clinical need for a transfusion. Using this approach, patients who are bleeding but die before receiving a transfusion are misclassified as not needing a transfusion. In this study, we aimed to evaluate the potential for this survival bias and to estimate its presence among a retrospective observational cohort of children and adolescents who died from injury. Methods We obtained patient, injury, and resuscitation characteristics from the 2017 to 2020 Trauma Quality Improvement Program database of children and adolescents (age < 18 years) who arrived with or without signs of life and died. We performed univariate analysis and a multivariable logistic regression to analyze the association between the time to death and the occurrence of transfusion within four hours after hospital arrival controlling for initial vital signs, injury type, body regions injured, and scene versus transfer status. Results We included 6,063 children who died from either a blunt or penetrating injury. We observed that children who died within 15 minutes had lower odds of receiving a transfusion (odds ratio [OR] = 0.1, 95% CI = 0.1, 0.2) compared to those who survived longer. We estimated that survival bias that occurs when using transfusion administration alone to define hemorrhagic shock may occur in up to 11% of all children who died following a blunt or penetrating injury but less than 1% of all children managed as trauma activations. Conclusion Using the occurrence of transfusion alone may underestimate the number of children who die from uncontrolled hemorrhage early after injury. Additional variables than just transfusion administration are needed to more accurately identify the presence of hemorrhagic shock among injured children and adolescents. Level of Evidence Prognostic and Epidemiological, Level III