Related Experiment Video
Updated: Jul 4, 2026
10:08
Pseudofracture: An Acute Peripheral Tissue Trauma Model
Published on: April 18, 2011
A trauma mortality prediction model based on the anatomic injury scale
Turner Osler1, Laurent Glance, Jeffery S Buzas
1Department of Surgery, University of Vermont, Burlington, VT, USA. tosler@silicondairy.net
Annals of Surgery
|June 4, 2008
Summary
A new trauma mortality prediction model (TMPM) offers superior accuracy in predicting patient survival compared to the standard Injury Severity Score (ISS). This empirically derived model provides better discrimination and calibration for trauma mortality.
Area of Science:
- Trauma care and injury research
- Biostatistics and epidemiological modeling
- Public health and patient outcomes
Background:
- Current trauma mortality prediction relies on the Injury Severity Score (ISS), a 30-year-old standard with known limitations.
- The ISS is nonmonotonic with mortality and performs worse than simpler models, hindering accurate trauma center comparisons.
- A reliable, empirically-based measure of overall injury severity is needed for accurate adjustment and comparison of trauma outcomes.
Purpose of the Study:
- To develop a statistically rigorous trauma mortality prediction model using empirical injury severity estimates.
- To compare the performance of the new model against the established Injury Severity Score (ISS).
- To introduce a more accurate tool for assessing trauma patient outcomes and guiding clinical decisions.
Main Methods:
- Utilized data from 702,229 patients in the National Trauma Data Bank (2001-2005).
- Derived empirical injury severities for Abbreviated Injury Scale (AIS) lexicon using regression models on 60% of data.
- Developed and validated three mortality prediction models, comparing them to ISS using discrimination and calibration metrics on the remaining 40% of data.
Main Results:
- Developed three novel trauma mortality prediction models based on empirical AIS injury severities.
- All new models demonstrated superior discrimination between survivors and non-survivors compared to ISS.
- The Trauma Mortality Prediction Model (TMPM) exhibited superior discrimination (ROC 0.901) and calibration (HL 58) over ISS (ROC 0.871, HL 296).
- Augmenting TMPM with age, gender, and mechanism of injury further improved performance (ROC 0.925, HL 18), outperforming ISS significantly.
Conclusions:
- Trauma mortality models utilizing empirical individual injury severity estimates outperform the standard ISS.
- The developed Trauma Mortality Prediction Model (TMPM) offers superior discrimination and calibration compared to the ISS.
- The TMPM is recommended to replace the ISS as the standard measure for overall injury severity in trauma research and practice.