A Novel Trauma Composite Score is a More Reliable Predictor of Mortality than Injury Severity Score in Pediatric Trauma
Robert Keskey1, David Hampton1, Henry Biermann1, Justin Cirone2, Tanya Zakrison1, Jennifer Cone1, Ken Wilson1, *Mark Slidell1
1University of Chicago, Chicago, IL; 2Dartmouth-Hitchcock Department of Surgery, Lebanon, NH
Background: Injury severity score (ISS) predicted mortality varies between adult (ISS>16) and pediatric (ISS>25) trauma patients. We hypothesized that a novel ISS prediction mortality model, which accounted for individual ages and mechanism, would outperform standard ISS cutoffs.
Methods: The 2010-2016 NTDB was queried for pediatric trauma patients. Linear discriminant analysis determined the predictive capabilities of: ISS cutoff of 25 (ISS25), age-based ISS, shock index pediatric adjusted (SIPA), and our novel Pediatric Trauma Composite Score (pTCS). The pTCS consisted of significant variables (AIS, GCS, Gender, and SIPA) selected a priori for each age. Each model’s area under the curve (AUC) determined their prediction accuracy. Cut point analysis determined the optimal ISS for predicting mortality.
Results: There were 109,459 blunt (BT) and 9,292 penetrating traumas (PT) studied. There was a significant difference in median ISS (BT:9 [IQR:4,10] vs. PT:5[IQR:1,10], p<0.01) and mortality (BT:0.7% vs. PT:2.7%, p<0.01). ISS varied significantly across all ages (ISS:15-25) and the optimal ISS for BT was 22 (AUC:0.95) and 16 for PT (AUC:0.94). Model AUCs varied significantly (BT - SIPA:0.70 vs. ISS25:0.91 vs. pTCS:0.98, p<0.01; PT – SIPA:0.66 vs. ISS25:0.88 vs. pTCS:0.97, p<0.01). pTCS outperformed all models across all ages (Fig.1).
Conclusions: The pTCS which incorporates age, mechanism, and AIS, outperforms standard injury metrics in mortality prediction. This may improve injury benchmarking for quality improvement and trauma center validation.
Back to 2020 Abstracts