摘要：Abbreviated Injury Scale (AIS)-based systems such as injury severity score (ISS), exponential injury severity score (EISS), trauma mortality prediction model (TMPM), and injury mortality prediction (IMP), classify anatomical injuries with limited accuracy. The widely accepted alternative, trauma and injury severity score (TRISS), improves the prediction rate by combining an anatomical index of ISS, physiological index (the Revised Trauma Score, RTS), and the age of patients. The study introduced the traumatic injury mortality prediction (TRIMP) with the inclusion of extra clinical information and aimed to compare the ability against the TRISS as predictors of survival. The hypothesis was that TRIMP would outperform TRISS in prediction power by incorporating clinically available data. This was a retrospective cohort study where a total of 1,198,885 injured patients hospitalized between 2012 and 2014 were subset from the National Trauma Data Bank (NTDB) in the United States. A TRIMP model was computed that uses AIS 2005 (AIS_05), physiological reserve and physiological response indicators. The results were analysed by examining the area under the receiver operating characteristic curve (AUC), the Hosmer–Lemeshow (HL) statistic, and the Akaike information criterion. TRIMP gave both significantly better discrimination (AUC
TRIMP, 0.964; 95% confidence interval (CI), 0.962 to 0.966 and AUC
TRISS, 0.923; 95% CI, 0.919 to 0.926) and calibration (HL
TRIMP, 14.0; 95% CI, 7.7 to 18.8 and HL
TRISS, 411; 95% CI, 332 to 492) than TRISS. Similar results were found in statistical comparisons among different body regions. TRIMP was superior to TRISS in terms of accurate of mortality prediction, TRIMP is a new and feasible scoring method in trauma research and should replace the TRISS.
