The trauma risk adjustment model: a new model for evaluating trauma care

Area of Science:

• Trauma care research
• Medical outcome prediction
• Health services research

Background:

• The Trauma Injury Severity Score (TRISS) is a long-standing tool for retrospective trauma risk assessment.
• TRISS has significant limitations that may affect the accuracy of trauma care evaluations.

Purpose of the Study:

• To develop and validate the Trauma Risk Adjustment Model (TRAM) for mortality prediction.
• To compare TRAM's predictive validity and risk adjustment capabilities against TRISS.

Main Methods:

• The Quebec Trauma Registry and American National Trauma Data Bank were used for model derivation and validation.
• Logistic generalized additive models with cubic smoothing splines were employed to model hospital mortality.
• Model performance was assessed using discrimination and calibration metrics.

Main Results:

• TRAM demonstrated strong predictive validity in derivation (AUC 0.944) and validation (AUC 0.942).
• TRAM showed superior discrimination and calibration compared to TRISS in the validation phase.
• Implementing TRAM resulted in a significant change in hospital risk-adjusted mortality odds ratios.

Conclusions:

• The Trauma Risk Adjustment Model (TRAM) offers improved predictive accuracy over TRISS.
• Adoption of TRAM can enhance the validity of trauma care evaluations and outcome research.