Polytrauma scoring revisited: prognostic validity and usability in daily clinical practice

Purpose Scores are widely used for the assessment of injury severity and therapy guidance in severely injured patients. They differ vastly regarding complexity, applicability, and prognostic accuracy. The objective of this study was to compare well-established with more recently developed trauma scores as well as intensive care unit (ICU) scores. Methods Retrospective analysis of severely injured patients treated at a level I trauma centre from 2010 to 2015. Inclusion criteria: Age >= 18 years, Injury Severity Score >= 16 and ICU treatment. Primary endpoint was in-hospital mortality. Several scores (ISS, APACHE II, RTS, Marshall Score, SOFA, NISS, RISC II, EAC and PTGS) were assessed to determine their predictive quality for mortality. Statistical analysis included correlation analysis and receiver operating characteristic (ROC). Results 444 patients were included. 71.8% were males, mean age was 51 +/- 20.26 years. 97.4% sustained a blunt trauma. The area under the ROC curve (AUROC) revealed RISC II (0.92) as strongest predictor regarding mortality, followed by APACHE II (0.81), Marshall score (0.69), SOFA (0.70), RTS (0.66), NISS (0.62), PTGS (0.61) and EAC (0.60). ISS did not reach statistical significance. Conclusions RISC II provided the strongest predictive capability for mortality. In comparison, more simple scores focusing on injury pattern (ISS, NISS), physiological abnormalities (RTS, EAC), or a combination of both (PTGS) only provided inferior mortality prediction. Established ICU scores like APACHE II, SOFA and Marshall score were proven to be helpful tools in severely injured trauma patients.