Development and Validation of a Korean Trauma and Injury Severity Score (K-TRISS) Model for Predicting Trauma...

Area of Science:

• Trauma research
• Injury severity scoring systems
• Public health

Background:

• The Trauma and Injury Severity Score (TRISS) has been a standard for predicting trauma outcomes since 1987.
• Existing TRISS models may require adjustments for local populations and conditions.
• This study addresses the need for a localized prediction model for trauma patients in Korea.

Purpose of the Study:

• To develop a Korean-specific TRISS (K-TRISS) model for predicting trauma outcomes in the Korean population.
• To compare the predictive performance of the new K-TRISS model against established international models.
• To enhance the K-TRISS model by optimizing the representation of the age variable.

Main Methods:

• Analysis of adult blunt trauma patient data from the Korea Trauma Data Bank (KTDB) (2017-2021).
• Development of K-TRISS-1 coefficients using logistic regression.
• Comparative analysis of predictive abilities using Area Under the Receiver Operating Characteristic (AUROC) curves, C-statistic, and bootstrap verification.
• Enhancement to K-TRISS-2 by treating age as a continuous variable.

Main Results:

• The K-TRISS-1 model showed comparable predictive ability to existing MTOS (1987) and NTDB (2009) models.
• The enhanced K-TRISS-2 model, utilizing continuous age, demonstrated a significantly higher AUROC value (0.9336) compared to K-TRISS-1 (0.9232), MTOS (0.9210), and NTDB (0.9190).
• K-TRISS-2 exhibited excellent predictive power confirmed by C-statistic and bootstrap validation.

Conclusions:

• The developed K-TRISS model, based on KTDB data, offers a localized approach to trauma outcome prediction.
• Converting age to a continuous variable significantly improved the predictive accuracy of the K-TRISS model.
• The K-TRISS model, particularly K-TRISS-2, provides a valuable tool for assessing trauma severity and outcomes in the Korean context.