Performance of the Paediatric Trauma Score on survival prediction of injured children at a major trauma centre: A retrospective Colombian cohort, 2011–2019

BACKGROUND: Despite improvements in children's health due to a reduction in infections, trauma continues to cause many deaths among adolescents. Strategies to mitigate morbidity and mortality from trauma include severity scores to classify and refer patients to the appropriate hospitals to provide better management; however, these strategies have not been assessed in Colombian children. This study aimed to describe the characteristics and outcomes of injured children and evaluate the performance of the Pediatric Trauma Score (PTS) in predicting survival at a major trauma centre in a Colombian city. METHODS: This was a retrospective cohort study of children aged <18 years who were treated for injuries at a hospital in Colombia. The primary outcome was 30-day mortality. A simple logistic regression model was used with PTS as the predictor variable and vital status at discharge as the outcome variable. PTS performance was assessed by discrimination using the area under the receiver-operating characteristic (AUROC) curve and by calibration using the Hosmer-Lemeshow (HL) goodness-of-fit test. FINDINGS: A total of 1047 children were admitted. The median age was 12 years (interquartile range [IQR]=5–15); 73·7% were male, and 66·1% had blunt trauma. The most frequent cause of injury was traffic accident (31·5%) followed by assaults (29%). Mortality was 5·9%; 61·3% of these deaths occurred in adolescents between 15 and 17 years of age and 71% of deaths in this age group were due to injuries from a firearm. The PTS had a median of 7 (IQR=5–9), an AUROC of 0·93, and good calibration (HL=7·97, p = 0·33). INTERPRETATION: The highest proportion of trauma and death occurred among adolescents. Interpersonal violence was the most frequent cause of death in this age group. The PTS showed good predictive power for survival, with excellent discrimination and good calibration. FUNDING: None.