Comparison of Trauma Mortality Prediction Models With Updated Survival Risk Ratios in Korea

Area of Science:

• Trauma Care Research
• Health Informatics
• Public Health

Background:

• Trauma injuries pose a significant disease burden in Korea.
• Nationwide trauma care status assessment has been insufficient.
• This study explores a diagnosis code-based method for injury severity measurement.

Purpose of the Study:

• To evaluate the feasibility of a diagnosis code-based injury severity measurement method.
• To assess the real-world usage and performance of the International Classification of Diseases-based Injury Severity Score (ICISS) model in Korea.
• To contribute to strengthening the Korean Trauma Care System.

Main Methods:

• Utilized datasets from the National Emergency Department Information System and Korean Trauma Data Bank.
• Developed six mortality prediction models, including Trauma and Injury Severity Score (TRISS) and extended ICISS variations.
• Employed stratified random sampling for training (80%) and validation (20%) datasets.
• Assessed model performance using area under the receiver operating characteristic curve (AUC) and Brier score.

Main Results:

• Extended ICISS models demonstrated feasibility, with performance slightly lower than the TRISS model (AUC 0.936-0.938 vs. 0.949).
• No statistically significant differences were found in Survival Risk Ratio (SRR) calculation methods.
• Replacing the Glasgow Coma Scale with the Alert, Voice, Pain, Unresponsive Scale in Revised Trauma Score (RTS) calculation did not degrade model performance.

Conclusions:

• A practical ICISS model is available and feasible for use in Korea.
• The developed ICISS model can aid in mortality prediction and severity classification.
• This model is expected to enhance the Korean Trauma Care System.