Comparative validation of predictive score efficacy among trauma patients

Objective. This study aimed to evaluate the predictive efficacy of the BIG, TRISS, NISS, and APACHE II scoring systems in trauma patients treated at a tertiary care hospital in South India. Materials and methods. Six hundred ninety-nine trauma patients were included in this observational and comparative study conducted between July 2021 and January 2024. Data were collected and analyzed using IBM SPSS version 25.0. Descriptive statistics, area under the curve (AUC) comparisons, and independent t-tests were performed. Receiver operating characteristic (ROC) curves were generated, and mean ± standard deviation (SD) values for each score were compared between survivors and non-survivors to assess statistical significance. Results. The BIG score had an AUC of 0.869 (95% CI: 0.839–0.898; p = 0.0005). The NISS score showed strong predictive accuracy across all three timepoints: day 1 – 0.899, day 3 – 0.888, and day 5 – 0.882. The APACHE II score also performed well: day 1 – 0.880, day 3 – 0.872, and day 5 – 0.889. In contrast, TRISS demonstrated lower AUC values, ranging from 0.498 to 0.512. The BIG score mean ± SD was 22 ± 8 in survivors and 38 ± 12.5 in non-survivors (F = 54.39; p = 0.000). Both NISS and APACHE II scores showed statistically significant differences between groups at all timepoints. TRISS, despite being widely used, showed limited discriminative ability. Conclusions. NISS and APACHE II were the most effective scoring systems for predicting mortality in trauma patients. The BIG score also showed high predictive value. Although TRISS is commonly applied, this study found it less useful in this context.