Background: The equivalent Injury Severity Score (ISS) cutoffs for severe trauma vary between adult (ISS, >16) and pediatric (ISS, >25) trauma. We hypothesized that a novel injury severity prediction model incorporating age and mechanism of injury would outperform standard ISS cutoffs.
Methods: The 2010 to 2016 National Trauma Data Bank was queried for pediatric trauma patients. Cut point analysis was used to determine the optimal ISS for predicting mortality for age and mechanism of injury. Linear discriminant analysis was implemented to determine prediction accuracy, based on area under the curve (AUC), of ISS cutoff of 25 (ISS, 25), shock index pediatric adjusted (SIPA), an age-adjusted ISS/abbreviated Trauma Composite Score (aTCS), and our novel Trauma Composite Score (TCS) in blunt trauma. The TCS consisted of significant variables (Abbreviated Injury Scale, Glasgow Coma Scale, sex, and SIPA) selected a priori for each age.
Results: There were 109,459 blunt trauma and 9,292 penetrating trauma patients studied. There was a significant difference in ISS (blunt trauma, 9.3 ± 8.0 vs. penetrating trauma, 8.0 ± 8.6; p < 0.01) and mortality (blunt trauma, 0.7% vs. penetrating trauma, 2.7%; p < 0.01). Analysis of the entire cohort revealed an optimal ISS cut point of 25 (AUC, 0.95; sensitivity, 0.86; specificity, 0.95); however, the optimal ISS ranged from 18 to 25 when evaluated by age and mechanism. Linear discriminant analysis model AUCs varied significantly for each injury metric when assessed for blunt trauma and penetrating trauma (penetrating trauma-adjusted ISS, 0.94 ± 0.02 vs. ISS 25, 0.88 ± 0.02 vs. SIPA, 0.62 ± 0.03; p < 0.001; blunt trauma-adjusted ISS, 0.96 ± 0.01 vs. ISS 25, 0.89 ± 0.02 vs. SIPA, 0.70 ± 0.02; p < 0.001). When injury metrics were assessed across age groups in blunt trauma, TCS and aTCS performed the best.
Conclusion: Current use of ISS in pediatric trauma may not accurately reflect injury severity. The TCS and aTCS incorporate both age and mechanism and outperform standard metrics in mortality prediction in blunt trauma.
Level of evidence: Retrospective review, level IV.
Copyright © 2021 Wolters Kluwer Health, Inc. All rights reserved.