Comparison of the predictive performance of the BIG, TRISS, and PS09 score in anadult trauma population derived from multiple international trauma registries

Abstract

Background: The BIG score (Admission base deficit (B), International normalized ratio (I), andGlasgow Coma Scale (G)) has been shown to predict mortality on admission inpediatric trauma patients. The objective of this study was to assess itsperformance in predicting mortality in an adult trauma population, and to compareit with the existing Trauma and Injury Severity Score (TRISS) and probability ofsurvival (PS09) score.Materials and methods: A retrospective analysis using data collected between 2005 and 2010 from seventrauma centers and registries in Europe and the United States of America wasperformed. We compared the BIG score with TRISS and PS09 scores in a population ofblunt and penetrating trauma patients. We then assessed the discrimination abilityof all scores via receiver operating characteristic (ROC) curves and compared theexpected mortality rate (precision) of all scores with the observed mortalityrate.Results: In total, 12,206 datasets were retrieved to validate the BIG score. The mean ISSwas 15 ± 11, and the mean 30-day mortality rate was 4.8%. With an AUROC of0.892 (95% confidence interval (CI): 0.879 to 0.906), the BIG score performed wellin an adult population. TRISS had an area under ROC (AUROC) of 0.922 (0.913 to0.932) and the PS09 score of 0.825 (0.915 to 0.934). On a penetrating-traumapopulation, the BIG score had an AUROC result of 0.920 (0.898 to 0.942) comparedwith the PS09 score (AUROC of 0.921; 0.902 to 0.939) and TRISS (0.929; 0.912 to0.947).Conclusions: The BIG score is a good predictor of mortality in the adult trauma population. Itperformed well compared with TRISS and the PS09 score, although it hassignificantly less discriminative ability. In a penetrating-trauma population, theBIG score performed better than in a population with blunt trauma. The BIG scorehas the advantage of being available shortly after admission and may be used topredict clinical prognosis or as a research tool to risk stratify trauma patientsinto clinical trials. © 2013 Brockamp et al.; licensee BioMed Central Ltd.