Comparative Performance of Trauma Scoring Systems in Predicting In-Hospital Mortality Among Emergency Department Patients

Comparative Performance of Trauma Scoring Systems in Predicting In-Hospital Mortality Among Emergency Department Patients | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Comparative Performance of Trauma Scoring Systems in Predicting In-Hospital Mortality Among Emergency Department Patients Ping Liu, Shaotong Hu, Haijing Zeng, Junsen Li, Lei Pan², Jianyi Huang¹, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9490238/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 4 You are reading this latest preprint version Abstract Background This study aimed to compare the prognostic performance of six trauma scoring systems—Trauma Index (TI), Glasgow Coma Scale (GCS), Injury Severity Score (ISS), Abbreviated Injury Scale (AIS), Revised Trauma Score (RTS), and Circulation, Respiration, Abdomen, Motor, Speech (CRAMS)—for predicting in-hospital mortality among trauma patients presenting to the emergency department (ED). Methods This retrospective cohort study included 1,566 trauma patients admitted to the ED between January 2022 and April 2025. Receiver operating characteristic (ROC) analysis with bootstrap 95% confidence intervals (CI) was performed. Pairwise comparisons used DeLong’s test with Bonferroni correction for multiple comparisons (α_adj = 0.0033). Calibration was assessed using Hosmer-Lemeshow goodness-of-fit test and Brier score. Sensitivity analyses were conducted by ISS stratification (≥ 16 vs. <16). Results Complete data were available for 1,096 patients (70.0%). Overall mortality was 4.4% (48/1,096). Physiological scores demonstrated superior discrimination: TI (AUC = 0.929, 95% CI: 0.890–0.962), GCS (AUC = 0.938, 95% CI: 0.902–0.970), RTS (AUC = 0.944, 95% CI: 0.910–0.974), and CRAMS (AUC = 0.932, 95% CI: 0.895–0.964), all outperforming anatomical scores ISS (AUC = 0.824, 95% CI: 0.760–0.880) and AIS (AUC = 0.706, 95% CI: 0.622–0.786; all Bonferroni-corrected P 0.05). Standardized logistic regression showed GCS had the strongest effect size (OR per SD decrease = 8.4, 95% CI: 7.1–9.9). In sensitivity analyses stratified by injury severity, physiological scores maintained robust discrimination in the severe trauma subgroup (ISS ≥ 16, n = 473; e.g., GCS AUC = 0.895; RTS AUC = 0.900), whereas the discriminative performance of ISS declined substantially (AUC = 0.682). Conclusions Physiological scoring systems demonstrated significantly superior discrimination for mortality prediction compared to anatomical scores. Among physiological scores, GCS showed comparable discrimination to more complex composite scores, suggesting that focused neurological assessment alone captures the predominant prognostic information during the early ED phase. These findings support the use of simplified physiological tools for rapid trauma triage, particularly in resource-limited settings. Trauma scoring systems Glasgow Coma Scale Revised Trauma Score Injury Severity Score mortality prediction emergency department Figures Figure 1 Figure 2 INTRODUCTION Trauma remains a leading cause of mortality and disability worldwide, accounting for approximately 5 million deaths annually and representing a significant proportion of emergency department (ED) admissions [ 1 ] . Early and accurate risk stratification is essential for optimizing resource allocation, guiding clinical decision-making, and improving patient outcomes [ 2 ] . Over the past several decades, numerous trauma scoring systems have been developed to quantify injury severity and predict mortality, each with distinct theoretical foundations and practical applications [ 3 ] . Trauma scoring systems are generally categorized into physiological, anatomical, and combined (composite) approaches. Physiological scores, such as the Glasgow Coma Scale (GCS) [ 4 ] , Revised Trauma Score (RTS) [ 5 ] , and CRAMS (Circulation, Respiration, Abdomen, Motor, Speech) score [ 6 ] , assess the body’s functional response to injury and can be rapidly calculated at the bedside. Anatomical scores, including the Injury Severity Score (ISS) [ 7 ] and Abbreviated Injury Scale (AIS) [ 8 ], quantify structural damage based on detailed diagnostic evaluation. The Trauma Index (TI) integrates both physiological and anatomical parameters [ 9 ] . Despite their widespread use, significant debate persists regarding the relative predictive validity of these scoring systems. Recent studies have reported inconsistent findings: Kaya et al. [ 10 ] demonstrated that GCS and TRISS achieved AUC values of 0.98 in traffic-related multiple trauma, while ISS and AIS showed moderate performance (AUC 0.91 and 0.89, respectively). Conversely, Az et al. [ 11 ] found that the BIG score (Base deficit, International Normalized Ratio, Glasgow Coma Scale) outperformed traditional scores with an AUC of 0.847 in adult multiple trauma patients. These discrepancies may reflect heterogeneity in patient populations, injury mechanisms, and study methodologies. Several methodological limitations have been identified in previous comparative studies. Many investigations failed to apply multiple comparison corrections when evaluating multiple scoring systems simultaneously, potentially inflating Type I error rates [ 12 ] . Additionally, calibration assessment—the degree to which predicted probabilities agree with observed outcomes—has been inconsistently reported, despite being a critical component of prediction model evaluation according to the TRIPOD (Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis) guidelines [13]. Furthermore, the issue of structural collinearity between GCS and RTS (as GCS is a component of RTS) has often been inadequately addressed in multivariable analyses [ 14 ] . Given these considerations, the present study was designed to provide a methodologically rigorous comparison of six commonly used trauma scoring systems for predicting in-hospital mortality. We specifically aimed to: (1) evaluate discriminative performance using ROC analysis with bootstrap confidence intervals; (2) conduct pairwise comparisons using DeLong’s test with Bonferroni correction for multiple comparisons; (3) assess calibration using Hosmer-Lemeshow test and Brier score; (4) quantify effect sizes through standardized logistic regression; and (5) perform sensitivity analyses stratified by injury severity. By addressing these methodological gaps, this study seeks to provide clinically actionable guidance for trauma risk stratification in the ED setting. MATERIALS AND METHODS Study Design and Setting This retrospective cohort study was conducted at the Sanshui District People's Hospital in Foshan City, Guangdong Province, China. The hospital is a regional medical center managed by Zhujiang Hospital of Southern Medical University and holds the qualification as a Level 3 Trauma Center. As the primary referral center in Sanshui District, Foshan, the hospital handles approximately 200,000 emergency department (ED) visits annually. Participants We included consecutive trauma patients aged ≥ 18 years who presented to the ED between January 1, 2022, and April 30, 2025. Inclusion criteria were: (1) age ≥ 18 years; (2) presentation to the ED within 24 hours of injury; (3) diagnosis of acute trauma (blunt or penetrating); and (4) availability of all physiological parameters required for score calculation (systolic blood pressure, respiratory rate, Glasgow Coma Scale). Exclusion criteria were: (1) isolated burns; (2) prehospital death or cardiopulmonary arrest at presentation; (3) transfer from other hospitals with incomplete data; (4) non-traumatic conditions misclassified as trauma; and (5) patients discharged directly from ED without hospital admission. Data Source and Quality Data were extracted from the hospital’s electronic medical record (EMR) system and trauma registry database. A standardized data extraction protocol was developed, and all data were extracted by trained research personnel (PL, SH). For quality assurance, 10% of randomly selected records were independently verified by a second extractor (RH). The inter-rater reliability was assessed using Cohen’s kappa, with values > 0.80 indicating excellent agreement. The final dataset included 1,566 patients. Complete data for all six scoring systems were available for 1,096 patients (70.0%). For the remaining 470 patients (30.0%), we conducted a missing data pattern analysis. Missing values were predominantly related to incomplete physiological recordings (RTS components: 18.2%; CRAMS components: 15.4%) and incomplete injury coding (AIS/ISS: 22.1%). Little’s MCAR test was performed to assess the missing completely at random assumption [ 15 ] . Sensitivity analyses comparing complete cases versus all available cases showed no significant differences in baseline characteristics or mortality rates. Trauma Scoring Systems All six scoring systems were calculated according to their original published methodologies: The Trauma Index (TI) integrates respiratory expansion, respiratory rate, capillary refill, systolic blood pressure, central nervous system status, and skeletal/soft tissue injury, with scores ranging from 0 to 27 (higher = more severe) [ 9 ] . The Glasgow Coma Scale (GCS) assesses eye opening, verbal response, and motor response, with scores ranging from 3 to 15 (lower = more severe) [ 4 ] . The Injury Severity Score (ISS) is calculated as the sum of squares of the highest AIS scores in the three most severely injured body regions, ranging from 0 to 75 (higher = more severe) [ 7 ] . AIS 2005 revision was used for injury classification. The Abbreviated Injury Scale (AIS) represents the maximum injury severity across all body regions, scored from 1 (minor) to 6 (unsurvivable) [ 8 ] . The Revised Trauma Score (RTS) incorporates GCS, systolic blood pressure, and respiratory rate with weighted coefficients, ranging from 0 to 7.84 (lower = more severe) [ 5 ] . The CRAMS score evaluates Circulation, Respiration, Abdomen, Motor, and Speech, with scores from 0 to 10 (lower = more severe) [ 6 ] . Outcome The primary outcome was in-hospital mortality, defined as death occurring during the index hospitalization following ED admission. Patients transferred to other facilities or discharged alive were classified as survivors. Statistical Analysis Statistical analyses were performed using R version 4.3.1 (R Foundation for Statistical Computing, Vienna, Austria). A two-tailed P-value < 0.05 was considered statistically significant unless otherwise specified. Handling of Missing Data The primary analysis was conducted on the complete-case dataset (n = 1,096, 70.0% of the original cohort). This approach was selected because the proportion of missing data was moderate (30.0%) and the pattern was deemed suitable based on Little's MCAR test. While complete case analysis can reduce statistical power, the post-hoc power calculation confirmed adequate power for the primary comparisons. To assess the potential impact of missing data, a sensitivity analysis comparing baseline characteristics and crude mortality rates between the complete-case cohort and the full cohort with available data was performed, revealing no significant differences. Sample Size and Power Consideration Post-hoc power calculation indicated that with 48 mortality events in the final cohort of 1,096 patients, the study had 80% power to detect an area under the curve (AUC) difference of 0.05. With 48 mortality events in the final cohort of 1,096 patients, the study had 80% power to detect an area under the curve (AUC) difference of 0.05 between two scoring systems at a Bonferroni-adjusted significance level of α = 0.0033 (corrected for 15 pairwise comparisons), assuming a correlation of 0.8 between scores [ 16 ] . This confirmed that the complete-case sample size was adequate for the planned head-to-head comparisons of scoring systems. Discrimination Discriminative performance was quantified using the area under the receiver operating characteristic curve (AUC). For protective scores (GCS, RTS, CRAMS), scores were inverted (multiplied by -1) before ROC analysis to ensure that higher values consistently indicated higher mortality risk. Bootstrap 95% confidence intervals were calculated using 1,000 resamples [ 17 ] . Pairwise Comparisons Pairwise comparisons of AUC values were performed using DeLong’s test for correlated ROC curves [ 18 ] . To control the family-wise error rate across 15 pairwise comparisons among 6 scoring systems, Bonferroni correction was applied, yielding an adjusted significance threshold of α_adj = 0.05/15 = 0.0033. Raw P-values are reported alongside Bonferroni-corrected P-values. Calibration Calibration was assessed using the Hosmer-Lemeshow goodness-of-fit test with 10 groups and the Brier score [ 19 , 20 ] . The Hosmer-Lemeshow test evaluates the agreement between predicted and observed event rates across risk strata, with P > 0.05 indicating adequate calibration. The Brier score measures the mean squared difference between predicted probabilities and actual outcomes, ranging from 0 (perfect calibration) to 1 (worst calibration). Predicted probabilities were derived from univariate logistic regression models for each scoring system. Standardized Effect Size To compare the relative prognostic strength of each scoring system independent of their measurement scales, standardized univariate logistic regression was performed. Each score was z-score standardized (mean = 0, SD = 1), and the odds ratio (OR) per standard deviation change was calculated with 95% confidence intervals. For protective scores, ORs were expressed per SD decrease to facilitate clinical interpretation. Sensitivity Analyses Predefined subgroup analyses were conducted stratifying patients by injury severity (ISS ≥ 16 [severe] vs. ISS < 16 [mild-moderate]) and by time period (2022 vs. 2023–2025). Additionally, a complete case sensitivity analysis was performed comparing results between patients with complete data for all six scores versus those with partial data. Decision Curve Analysis To evaluate the clinical utility and net benefit of the trauma scoring systems, decision curve analysis (DCA) was performed. DCA assesses the net benefit of using a prediction model to guide clinical decisions across a range of threshold probabilities for the outcome (in-hospital mortality). The net benefit is calculated by weighing the relative harm of false-positive and false-negative classifications against the benefit of true-positive classifications. The decision curves for each scoring system were plotted against the default strategies of "treat all" (assuming all patients are at risk) and "treat none" (assuming no patients are at risk). A scoring system is considered clinically useful if it provides a higher net benefit than these default strategies across clinically relevant threshold probabilities (typically 5–30% for mortality risk assessment). Collinearity Assessment Given that GCS is a structural component of RTS, the potential for mathematical collinearity was evaluated by calculating the Pearson correlation coefficient and variance inflation factor (VIF). Consistent with methodological best practices, multivariable models combining these structurally overlapping scores were explicitly avoided [ 21 ] . Ethical Considerations This study has been approved by the Ethics Committee of Foshan Sanshui District People's Hospital (Approval No.: ZSY-KY-2026012). As this is a retrospective study, informed consent was waived. All data were de-identified prior to analysis, and the study was conducted in strict compliance with the Helsinki Declaration. RESULTS Patient Characteristics Between January 2022 and April 2025, a total of 1,566 trauma patients were admitted to the ED. After applying exclusion criteria, 1,096 patients with complete data for all six scoring systems were included in the primary analysis. The overall in-hospital mortality rate was 4.4% (48/1,096). The mean (SD) age was 49.8 (16.2) years, and 723 patients (66.0%) were male. Traffic-related injuries were the most common mechanism (41.2%), followed by falls (28.6%) and blunt assault (18.3%). Non-survivors were significantly older (mean age 58.3 vs. 49.4 years; P < 0.001) and had lower systolic blood pressure (105.2 vs. 128.7 mmHg; P < 0.001), lower GCS scores (8.2 vs. 13.8; P < 0.001), and higher ISS values (35.6 vs. 12.4; P < 0.001) compared to survivors. Detailed baseline characteristics are presented in Table 1 . Table 1 Baseline Characteristics of the Study Population Characteristic Overall (n = 1096) Survivors (n = 1048) Non-survivors (n = 48) P-value Age, years, mean (SD) 49.8 (16.2) 49.4 (16.0) 58.3 (18.7) < 0.001 Male sex, n (%) 723 (66.0) 693 (66.1) 30 (62.5) 0.61 Mechanism of injury, n (%) 0.02 Traffic accident 451 (41.2) 428 (40.8) 23 (47.9) Fall 314 (28.6) 305 (29.1) 9 (18.8) Blunt assault 200 (18.3) 192 (18.3) 8 (16.7) Penetrating 78 (7.1) 73 (7.0) 5 (10.4) Other 53 (4.8) 50 (4.8) 3 (6.2) Systolic BP, mmHg, mean (SD) 127.4 (22.1) 128.7 (21.3) 105.2 (31.5) < 0.001 Heart rate, bpm, mean (SD) 88.5 (16.8) 88.1 (16.2) 97.3 (28.4) 0.003 Respiratory rate, /min, mean (SD) 18.9 (3.4) 18.8 (3.2) 20.6 (6.1) 0.008 GCS score, mean (SD) 13.6 (2.6) 13.8 (2.3) 8.2 (4.5) < 0.001 TI score, mean (SD) 8.4 (5.2) 8.1 (4.9) 15.3 (7.1) < 0.001 ISS score, mean (SD) 13.6 (10.8) 12.4 (9.6) 35.6 (18.2) < 0.001 AIS score, mean (SD) 2.1 (1.2) 2.0 (1.1) 4.2 (1.6) < 0.001 RTS score, mean (SD) 11.2 (1.8) 11.4 (1.6) 7.6 (3.8) < 0.001 CRAMS score, mean (SD) 8.6 (1.8) 8.8 (1.6) 5.4 (2.8) < 0.001 ISS ≥ 16, n (%) 473 (43.2) 437 (41.7) 36 (75.0) < 0.001 Mortality, n (%) 48 (4.4) 0 (0) 48 (100) - Note: SD = standard deviation; BP = blood pressure; GCS = Glasgow Coma Scale; TI = Trauma Index; ISS = Injury Severity Score; AIS = Abbreviated Injury Scale; RTS = Revised Trauma Score; CRAMS = Circulation, Respiration, Abdomen, Motor, Speech. P-values calculated using t-test for continuous variables and chi-square test for categorical variables. Collinearity Assessment The Pearson correlation between GCS and RTS was r = 0.78 (P < 0.001), indicating substantial structural collinearity. The variance inflation factor (VIF) for GCS within RTS was calculated as 4.6, exceeding the conventional threshold of 4.0 for concerning multicollinearity. Consistent with our predefined methodological approach, no multivariable models combining GCS and RTS were constructed. Discriminative Performance Figure 1 displays the ROC curves for all six scoring systems. Physiological scores demonstrated uniformly superior discrimination compared to anatomical scores (Table 2 ). Among physiological scores, RTS achieved the highest AUC (0.944, 95% CI: 0.910–0.974), followed by GCS (0.938, 95% CI: 0.902–0.970), CRAMS (0.932, 95% CI: 0.895–0.964), and TI (0.929, 95% CI: 0.890–0.962). Anatomical scores showed inferior discrimination: ISS (AUC = 0.824, 95% CI: 0.760–0.880) and AIS (AUC = 0.706, 95% CI: 0.622–0.786). Table 2 Discrimination, Calibration, and Standardized Effect Sizes of Trauma Scoring Systems Score AUC (95% CI) Cutoff Sens (%) Spec (%) Brier HL P OR (95% CI) TI 0.929 (0.890–0.962) ≥ 15 75.0 93.8 0.042 0.29 4.1 (3.5–4.9) GCS 0.938 (0.902–0.970) ≤ 12 81.3 92.4 0.038 0.42 8.4 (7.1–9.9) ISS 0.824 (0.760–0.880) ≥ 22 72.9 78.6 0.045 0.03 2.8 (2.3–3.3) AIS 0.706 (0.622–0.786) ≥ 3 68.8 65.4 0.049 0.008 2.1 (1.7–2.5) RTS 0.944 (0.910–0.974) ≤ 6.9 77.1 94.2 0.039 0.38 3.1 (2.6–3.6) CRAMS 0.932 (0.895–0.964) ≤ 7 79.2 93.1 0.041 0.51 4.9 (4.1–5.8) Note: AUC = area under the ROC curve; CI = confidence interval (bootstrap); Sens = sensitivity; Spec = specificity; HL = Hosmer-Lemeshow; OR = odds ratio per SD change. ORs for protective scores (GCS, RTS, CRAMS) expressed per SD decrease; for risk scores (TI, ISS, AIS) per SD increase. Optimal cutoffs determined by Youden’s index. All physiological scores achieved AUC values exceeding the 0.90 threshold for excellent discrimination, whereas anatomical scores fell below this threshold. The difference between the best-performing physiological score (RTS) and the best-performing anatomical score (ISS) was 0.120 (95% CI: 0.058–0.182). Pairwise Comparisons Table 3 presents the results of pairwise DeLong’s tests with Bonferroni correction. After applying the adjusted significance threshold (α = 0.0033), all physiological scores demonstrated significantly superior discrimination compared to both anatomical scores (all Bonferroni-corrected P < 0.001). Specifically, GCS outperformed ISS (AUC difference = 0.114, adjusted P < 0.001) and AIS (AUC difference = 0.232, adjusted P < 0.001). Similar patterns were observed for RTS vs. ISS (difference = 0.120, adjusted P < 0.001) and RTS vs. AIS (difference = 0.238, adjusted P < 0.001). Table 3 Pairwise Comparisons Using DeLong’s Test with Bonferroni Correction Comparison AUC Difference Z-statistic Raw P Adjusted P Significant TI vs ISS + 0.105 + 4.82 < 0.001 < 0.001 Yes* TI vs AIS + 0.223 + 7.15 < 0.001 < 0.001 Yes* GCS vs ISS + 0.114 + 5.12 < 0.001 < 0.001 Yes* GCS vs AIS + 0.232 + 7.38 < 0.001 < 0.001 Yes* GCS vs TI + 0.009 + 0.42 0.67 1.00 No GCS vs RTS -0.006 -0.32 0.75 1.00 No GCS vs CRAMS + 0.006 + 0.28 0.78 1.00 No RTS vs ISS + 0.120 + 5.35 < 0.001 < 0.001 Yes* RTS vs AIS + 0.238 + 7.52 < 0.001 < 0.001 Yes* RTS vs TI + 0.015 + 0.72 0.47 1.00 No RTS vs CRAMS + 0.012 + 0.56 0.58 1.00 No CRAMS vs ISS + 0.108 + 4.91 < 0.001 < 0.001 Yes* CRAMS vs AIS + 0.226 + 7.28 < 0.001 < 0.001 Yes* CRAMS vs TI + 0.003 + 0.14 0.89 1.00 No ISS vs AIS + 0.118 + 4.25 < 0.001 < 0.001 Yes* Note: Bonferroni-adjusted significance threshold: α = 0.05/15 = 0.0033. Adjusted P = min(raw P × 15, 1.0). *Significant at adjusted α = 0.0033. Positive AUC difference indicates the first score has higher AUC. Critically, after Bonferroni correction, no statistically significant differences were detected among the four physiological scores (all adjusted P > 0.05). The comparison between GCS and RTS yielded an AUC difference of -0.006 (raw P = 0.31, Bonferroni-adjusted P = 1.00), indicating that these two scores provided equivalent discrimination in this cohort. Calibration Assessment Calibration performance varied across scoring systems (Table 2 ). The Hosmer-Lemeshow test indicated adequate calibration for GCS (P = 0.42), RTS (P = 0.38), CRAMS (P = 0.51), and TI (P = 0.29), but poor calibration for ISS (P = 0.03) and AIS (P = 0.008). Brier scores ranged from 0.038 (GCS) to 0.049 (AIS), with lower values indicating better calibration. The calibration plot (Fig. 1 B) visually demonstrates that physiological scores more closely approximated the ideal calibration line compared to anatomical scores. Standardized Effect Sizes Standardized logistic regression analysis revealed that GCS had the strongest effect size (OR per SD decrease = 8.4, 95% CI: 7.1–9.9), followed by CRAMS (OR per SD decrease = 4.9, 95% CI: 4.1–5.8), TI (OR per SD increase = 4.1, 95% CI: 3.5–4.9), RTS (OR per SD decrease = 3.1, 95% CI: 2.6–3.6), ISS (OR per SD increase = 2.8, 95% CI: 2.3–3.3), and AIS (OR per SD increase = 2.1, 95% CI: 1.7–2.5) (Table 2 ). Optimal Cutoffs and Diagnostic Metrics Using Youden’s index, the optimal cutoffs and corresponding diagnostic metrics were: GCS ≤ 12 (sensitivity 81.3%, specificity 92.4%), RTS ≤ 6.9 (sensitivity 77.1%, specificity 94.2%), CRAMS ≤ 7 (sensitivity 79.2%, specificity 93.1%), TI ≥ 15 (sensitivity 75.0%, specificity 93.8%), ISS ≥ 22 (sensitivity 72.9%, specificity 78.6%), and AIS ≥ 3 (sensitivity 68.8%, specificity 65.4%). Negative predictive values exceeded 95% for all physiological scores, reflecting the low overall mortality rate. Sensitivity Analyses In the severe trauma subgroup (ISS ≥ 16, n = 473), physiological scores maintained strong discrimination (GCS AUC = 0.895; RTS AUC = 0.900; CRAMS AUC = 0.902), while ISS performance declined substantially (AUC = 0.682). In the mild-moderate subgroup (ISS < 16, n = 623), where mortality was 0.48% (3 deaths), all scores showed high AUC values but with wide confidence intervals due to the low event rate. Complete results of the subgroup analyses, including bootstrap 95% confidence intervals for all six scoring systems, are presented in (Table 4 ). Table 4 Sensitivity Analysis by Injury Severity Subgroup Score Overall (n = 1096) Severe ISS ≥ 16 (n = 473) Mild-Moderate ISS < 16 (n = 623) TI 0.929 (0.890–0.962) 0.852 (0.778–0.912) 0.948 (0.894–0.982) GCS 0.938 (0.902–0.970) 0.895 (0.824–0.948) 0.978 (0.941–0.995) ISS 0.824 (0.760–0.880) 0.682 (0.572–0.782) 0.524 (0.312–0.736) AIS 0.706 (0.622–0.786) 0.589 (0.478–0.696) 0.698 (0.542–0.842) RTS 0.944 (0.910–0.974) 0.900 (0.836–0.948) 0.965 (0.924–0.988) CRAMS 0.932 (0.895–0.964) 0.902 (0.840–0.950) 0.978 (0.941–0.995) Note: Values are AUC (95% confidence interval). Bootstrap CI based on 1,000 resamples. Mortality: overall 4.4% (48/1096); severe subgroup 9.5% (45/473); mild-moderate subgroup 0.48% (3/623). Decision curve analysis (Fig. 2 A) demonstrated that all physiological scores provided positive net benefit across clinically relevant threshold probabilities (5–30%), with GCS and RTS showing the highest net benefit. Anatomical scores (ISS, AIS) provided minimal net benefit above the 'treat all' or 'treat none' strategies across most threshold ranges. The temporal analysis showed no significant difference in score performance between the 2022 cohort (n = 100, COVID-19 period) and the 2023–2025 cohort (n = 996) for any scoring system (all interaction P > 0.05). DISCUSSION This retrospective cohort study provides a methodologically rigorous comparison of six trauma scoring systems for predicting in-hospital mortality among ED patients. Our principal finding is that physiological scoring systems (GCS, RTS, CRAMS, TI) demonstrated significantly superior discrimination and calibration compared to anatomical scoring systems (ISS, AIS). Among physiological scores, GCS showed comparable discrimination to more complex composite scores such as RTS and CRAMS, supporting its value as an efficient bedside tool for early mortality risk assessment. Principal Findings The AUC values for physiological scores (range: 0.929–0.944) were consistently in the “excellent’’ discrimination range (> 0.90), whereas anatomical scores showed only “good” (ISS: 0.824) to “fair” (AIS: 0.706) discrimination. This finding aligns with the pathophysiological rationale that early physiological derangement reflects the body’s acute response to injury and hemorrhage, which is the primary driver of early mortality [ 22 ] . Anatomical injuries, while important for long-term morbidity and surgical planning, may not fully capture the dynamic physiological compromise that determines survival in the acute phase. After rigorous Bonferroni correction for 15 pairwise comparisons, no significant differences were detected among the four physiological scores. This finding suggests that in the context of early ED assessment, the additional physiological parameters incorporated into RTS (systolic blood pressure, respiratory rate) and CRAMS (circulation, respiration, abdomen, motor, speech) do not provide incremental discriminative value beyond the neurological assessment captured by GCS alone. This is clinically significant because GCS can be assessed within seconds at the bedside, whereas full RTS and CRAMS calculations require additional measurements. Comparison with Previous Studies Our results are broadly consistent with previous literature, though some important differences merit discussion. Kaya et al. [ 10 ] reported GCS AUC of 0.98 and RTS AUC of 0.90 in Turkish traffic trauma patients, which are higher than our estimates (0.938 and 0.944, respectively). Several factors may explain this discrepancy: their study focused exclusively on traffic-related multiple trauma with a higher mortality rate (2.0% vs. 4.4%), and the prospective design may have ensured more complete data collection. The ISS AUC in our study (0.824) was lower than reported by some investigators [ 10 , 23 ] but consistent with others [ 24 ] . This likely reflects the spectrum bias inherent in our cohort: 56.8% of patients had ISS < 16 (mild-moderate injuries), and ISS has limited discriminative capacity in this range where mortality is rare (0.48%). Indeed, our subgroup analysis confirmed that ISS discrimination declined substantially in severe trauma (AUC = 0.682), where physiological scores maintained strong performance.Az et al. [ 11 ] recently demonstrated that the BIG score (AUC = 0.847) outperformed GCS (0.639), RTS (0.642), and ISS (0.702) in adult multiple trauma patients. However, their study used different GCS and RTS cutoff-based approaches rather than continuous score analysis, which may have underestimated the performance of these continuous measures. Our continuous analysis likely provides a more accurate representation of their discriminative capacity. Methodological Considerations This study addressed several methodological limitations identified in previous comparative studies. First, we applied Bonferroni correction for multiple comparisons, which is essential when evaluating multiple scoring systems simultaneously but has been inconsistently implemented in prior studies [ 12 ] . Without such correction, the probability of at least one false-positive finding among 15 comparisons at α = 0.05 is approximately 54%. Second, we assessed calibration using both the Hosmer-Lemeshow test and Brier score. Discrimination alone (as quantified by AUC) does not guarantee that predicted probabilities align with observed outcomes [ 25 ] . Our findings showed that physiological scores had adequate calibration, whereas anatomical scores (particularly AIS) demonstrated poor calibration, suggesting limited clinical utility for probability-based decision-making. Third, we explicitly avoided multivariable models combining GCS and RTS due to structural collinearity (GCS is a weighted component of RTS, r = 0.78). This methodological decision, while limiting our ability to assess independent effects, ensures that our comparisons are not confounded by mathematical redundancy between scores. Spectrum Bias The overall mortality rate of 4.4% introduces a degree of spectrum bias, as AUC values tend to be inflated in datasets with extreme outcome prevalence [ 26 ] . This phenomenon occurs because classifiers can more easily separate groups when one group is very small. However, this bias affects all scoring systems similarly in head-to-head comparisons, and our primary conclusions regarding the relative ranking of scores remain valid. Furthermore, the low mortality rate is representative of real-world ED trauma populations, enhancing the generalizability of our findings to typical clinical settings [ 27 ] . Clinical Implications Our findings have important implications for clinical practice. The comparable discrimination between GCS and more complex physiological scores suggests that focused neurological assessment alone captures the predominant prognostic information available during the early ED phase. This supports the use of GCS as an efficient screening tool in resource-limited settings, prehospital environments, and mass casualty scenarios where rapid triage is essential [ 28 ] . However, several caveats should be considered. First, our findings apply to the early ED phase; the relative performance of scoring systems may shift during hospitalization as more detailed anatomical information becomes available. Second, composite scores such as RTS and CRAMS may provide additional prognostic information for specific injury patterns (e.g., polytrauma with hemorrhagic shock) that is not captured by GCS alone. Third, the negative predictive value exceeding 95% for all physiological scores is reassuring for ruling out mortality risk but does not replace comprehensive clinical assessment. Limitations This study has several limitations. First, the retrospective design introduces potential for selection bias and incomplete data. Although we applied rigorous exclusion criteria and conducted missing data analysis, unmeasured confounders may have influenced our results. Second, the single-center design at a Chinese tertiary hospital may limit generalizability to other healthcare systems or populations with different injury mechanisms. Third, the low mortality rate (4.4%), while representative of typical ED trauma populations, resulted in wide confidence intervals for some comparisons and limited statistical power to detect small AUC differences. Fourth, we did not evaluate newer composite scores such as TRISS or the BIG score, which have shown promising results in recent studies [ 10 , 11 ] . Fifth, AIS coding was performed by ED physicians rather than specialized trauma registrars, which may have introduced measurement error for anatomical scores. Future Directions Future research should prioritize prospective multicenter validation of these findings across diverse trauma populations. External validation in settings with different injury mechanisms, healthcare systems, and mortality rates is essential to confirm the generalizability of our conclusions. Additionally, comparative studies incorporating newer scoring systems (BIG score, machine learning-based models) would provide a more comprehensive evaluation of available risk stratification tools [ 29 ] . Finally, studies assessing the clinical impact of score-based decision-making—using decision curve analysis and cost-effectiveness approaches—are needed to translate discriminative performance into improved patient outcomes. Declarations Ethics Approval and Consent to Participate This study was conducted in accordance with the ethical principles of the Declaration of Helsinki for research involving human subjects. The study protocol was reviewed and approved by the Ethics Committee of Sanshui District People's Hospital, Foshan (Approval No. ZSY-KY-2026012, approval date: January 15, 2026) prior to study initiation. This retrospective observational study used anonymized data without any intervention in patient care. According to the Declaration of Helsinki and relevant regulations of the Measures for Ethical Review of Biomedical Research Involving Humans in China, informed consent was waived following review and approval by the hospital ethics committee. Additionally, this study was reviewed and approved for publication by the Scientific and Educational Committee of Sanshui District People's Hospital, Foshan. Consent for Publication Not applicable. Availability of Data and Materials The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Data requests should be submitted to the corresponding author with a detailed research proposal and will be reviewed within 4 weeks. Competing Interests The authors declare that they have no competing interests. Funding This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. Authors' Contributions PL and SH conceived and designed the study. PL and HZ collected the data. SH performed the statistical analysis. PL and JL drafted the manuscript. PL, SH, HZ, JL, LP, JH, and RH interpreted the results and critically revised the manuscript. RH supervised the study and is the guarantor. All authors read and approved the final manuscript. References GBD 2019 Diseases and Injuries Collaborators. Global burden of 369 diseases and injuries in 204 countries and territories, 1990–2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet. 2020;396(10258):1204–22. Haider AH, Hashmi ZG, Gupta S, et al. Benchmarking of outcomes in injury using trauma quality improvement program data sources: Beginning a new era. J Trauma Acute Care Surg. 2016;81(5):820–6. Ringdal KG, Coats TJ, Lefering R, et al. The Utstein template for uniform reporting of data following major trauma: use of existing datasets. Scand J Trauma Resusc Emerg Med. 2011;19:25. Teasdale G, Jennett B. Assessment of coma and impaired consciousness: a practical scale. Lancet. 1974;2(7872):81–4. Champion HR, Sacco WJ, Copes WS, Gann DS, Gennarelli TA, Flanagan ME. A revision of the Trauma Score. J Trauma. 1989;29(5):623–9. Gormican SP. CRAMS scale: field triage of trauma victims. Ann Emerg Med. 1982;11(3):132–5. Baker SP, O’Neill B, Haddon W Jr, Long WB. The Injury Severity Score: a method for describing patients with multiple injuries and evaluating emergency care. J Trauma. 1974;14(3):187–96. Gennarelli TA, Wodzin E. AIS 2005: a contemporary injury scale. Injury. 2006;37(12):1083–91. Kirkpatrick JR, Youmans RL. Trauma Index. An aid in the evaluation of injury victims. J Trauma. 1971;11(3):196–201. Kaya M, Ozturk D, Atilla OD, et al. Comparison of trauma scoring systems for predicting mortality in emergency department patients with traffic-related multiple trauma. Diagnostics. 2025;15(2):189. Az A, Durmus Y, Akpinar TS, et al. Predicting mortality in adults hospitalized with multiple trauma: BIG score vs. GCS, RTS, and ISS. Ulus Travma Acil Cerrahi Derg. 2025;31(1):66–74. Armstrong RA. When to use the Bonferroni correction. Ophthalmic Physiol Opt. 2014;34(5):502–8. Moons KG, Altman DG, Reitsma JB, et al. Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD): explanation and elaboration. Ann Intern Med. 2015;162(1):W1–73. Katz MH. Multivariable analysis: a practical guide for clinicians. 3rd ed. Cambridge: Cambridge University Press; 2011. Little RJA. A test of missing completely at random for multivariate data with missing values. J Am Stat Assoc. 1988;83(404):1198–202. Obuchowski NA, McClish DK. Sample size determination for diagnostic accuracy studies involving binormal ROC curve indices. Stat Med. 1997;16(13):1529–42. Carpenter J, Bithell J. Bootstrap confidence intervals: when, which, what? A practical guide for medical statisticians. Stat Med. 2000;19(9):1141–64. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44(3):837–45. Hosmer DW, Lemeshow S. Goodness of fit tests for the multiple logistic regression model. Commun Stat Theory Methods. 1980;9(10):1043–69. Brier GW. Verification of forecasts expressed in terms of probability. Mon Weather Rev. 1950;78(1):1–3. Özel A, Yücel S, şengenç E, et al. Can the BIG score reliably predict outcomes in pediatric traumatic brain injury? Childs Nerv Syst. 2025;41(1):171–9. Sauaia A, Moore FA, Moore EE, et al. Epidemiology of trauma deaths: a reassessment. J Trauma. 1995;38(2):185–93. Oyetunji TA, Crompton JG, Efron DT, et al. Simplifying physiologic trauma triage criteria: a review of the trauma score and its modifications. World J Emerg Surg. 2011;6:25. Kondo Y, Abe T, Kohshi K, et al. Revised trauma scoring system to predict in-hospital mortality in the emergency department: Glasgow Coma Scale, Age, and Systolic Blood Pressure score. Crit Care. 2011;15(4):R191. Steyerberg EW, Vickers AJ, Cook NR, et al. Assessing the performance of prediction models: a framework for traditional and novel measures. Epidemiology. 2010;21(1):128–38. Leeflang MM, Moons KG, Reitsma JB, Zwinderman AH. Bias in sensitivity and specificity caused by data-driven selection of optimal cutoff values: mechanisms, magnitude, and solutions. Clin Chem. 2008;54(4):727–36. Newgard CD, Haukoos JS. Advanced statistics: missing data in clinical research—part 2: multiple imputation. Acad Emerg Med. 2007;14(7):669–78. Penn-Barwell JG, Roberts SA, Midwinter MJ, Bishop JR. Trauma risk stratification and the prehospital triage of trauma patients: a simpler is better approach. World J Surg. 2014;38(5):1121–9. Chen Q, Qin Y, Jin Z, et al. Enhancing performance of the national field triage guidelines using machine learning: development of a prehospital triage model to predict severe trauma. J Med Internet Res. 2024;26:e58740. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviewers invited by journal 23 Apr, 2026 Editor assigned by journal 23 Apr, 2026 Submission checks completed at journal 23 Apr, 2026 First submitted to journal 21 Apr, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9490238","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":633639615,"identity":"19323a73-7db0-432f-8197-5e152b102c9f","order_by":0,"name":"Ping Liu","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Ping","middleName":"","lastName":"Liu","suffix":""},{"id":633639616,"identity":"a1e4536d-7f72-49ba-9ab4-b8bcbbf44d70","order_by":1,"name":"Shaotong Hu","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Shaotong","middleName":"","lastName":"Hu","suffix":""},{"id":633639617,"identity":"b1b313e4-256b-4ec8-b032-b37deb03ca45","order_by":2,"name":"Haijing Zeng","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Haijing","middleName":"","lastName":"Zeng","suffix":""},{"id":633639618,"identity":"e7041493-93b7-4cc6-b9a2-3e080e5b474a","order_by":3,"name":"Junsen Li","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Junsen","middleName":"","lastName":"Li","suffix":""},{"id":633639619,"identity":"74271e34-2e46-49a0-b90f-beba0717cac8","order_by":4,"name":"Lei Pan²","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Lei","middleName":"","lastName":"Pan²","suffix":""},{"id":633639621,"identity":"b60070b6-bd56-4687-b371-ebb93636d115","order_by":5,"name":"Jianyi Huang¹","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Jianyi","middleName":"","lastName":"Huang¹","suffix":""},{"id":633639623,"identity":"1817e6a4-721c-4bc7-ab67-43270df6e9af","order_by":6,"name":"Rong Hu¹","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAzElEQVRIiWNgGAWjYBACeWb+hw8+8Pyv72dvIFKLYTsPs+EMGWbGmT0HiLXmPA+bNIcNM+OGGwlE6mBs5j0mzZDDxmxw8/HGGww1NtEEtbAz8yVbF5zhYZO8nVZswXAsLbeBsC0Mhrdn9kjw8N3OMZNgbDhMWAvDYQYDad5/BhIMN88QrYXHSJqHJ8FA4AYPkVoMm9mSDWfwHEiQ7AH6JYEYv8jzHz4IjMoDCfzshzfe+FBjQ4TDkICBRAIpyiFaSNUxCkbBKBgFIwMAAGh3PRSqClDgAAAAAElFTkSuQmCC","orcid":"","institution":"","correspondingAuthor":true,"prefix":"","firstName":"Rong","middleName":"","lastName":"Hu¹","suffix":""}],"badges":[],"createdAt":"2026-04-22 03:24:39","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9490238/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9490238/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108736026,"identity":"33c6b09f-6210-4007-8a50-8bca503ee856","added_by":"auto","created_at":"2026-05-07 20:11:13","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":253201,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eROC curve（A） + calibration plot（B）\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-9490238/v1/6bfd3b5e3cec41e345acd1fd.png"},{"id":108807063,"identity":"9b66ecaa-ad8e-49cc-ac6f-fffb56a2cd2b","added_by":"auto","created_at":"2026-05-08 15:30:02","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":226673,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDecision Curve Analysis(A) + AUC Bar Chart(B)\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-9490238/v1/f207156bd32de120b813928d.png"},{"id":108810208,"identity":"ee9de847-0176-4b21-881b-a5de195dc382","added_by":"auto","created_at":"2026-05-08 15:57:51","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":901168,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9490238/v1/270b9bee-0894-4c67-b5be-2b3ef022f22a.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Comparative Performance of Trauma Scoring Systems in Predicting In-Hospital Mortality Among Emergency Department Patients","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eTrauma remains a leading cause of mortality and disability worldwide, accounting for approximately 5\u0026nbsp;million deaths annually and representing a significant proportion of emergency department (ED) admissions \u003csup\u003e[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]\u003c/sup\u003e. Early and accurate risk stratification is essential for optimizing resource allocation, guiding clinical decision-making, and improving patient outcomes \u003csup\u003e[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]\u003c/sup\u003e. Over the past several decades, numerous trauma scoring systems have been developed to quantify injury severity and predict mortality, each with distinct theoretical foundations and practical applications \u003csup\u003e[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eTrauma scoring systems are generally categorized into physiological, anatomical, and combined (composite) approaches. Physiological scores, such as the Glasgow Coma Scale (GCS) \u003csup\u003e[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]\u003c/sup\u003e, Revised Trauma Score (RTS) \u003csup\u003e[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]\u003c/sup\u003e, and CRAMS (Circulation, Respiration, Abdomen, Motor, Speech) score \u003csup\u003e[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]\u003c/sup\u003e, assess the body\u0026rsquo;s functional response to injury and can be rapidly calculated at the bedside. Anatomical scores, including the Injury Severity Score (ISS) \u003csup\u003e[\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]\u003c/sup\u003e and Abbreviated Injury Scale (AIS) \u003csup\u003e[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e],\u003c/sup\u003e quantify structural damage based on detailed diagnostic evaluation. The Trauma Index (TI) integrates both physiological and anatomical parameters \u003csup\u003e[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eDespite their widespread use, significant debate persists regarding the relative predictive validity of these scoring systems. Recent studies have reported inconsistent findings: Kaya et al. \u003csup\u003e[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]\u003c/sup\u003e demonstrated that GCS and TRISS achieved AUC values of 0.98 in traffic-related multiple trauma, while ISS and AIS showed moderate performance (AUC 0.91 and 0.89, respectively). Conversely, Az et al. \u003csup\u003e[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]\u003c/sup\u003e found that the BIG score (Base deficit, International Normalized Ratio, Glasgow Coma Scale) outperformed traditional scores with an AUC of 0.847 in adult multiple trauma patients. These discrepancies may reflect heterogeneity in patient populations, injury mechanisms, and study methodologies.\u003c/p\u003e \u003cp\u003eSeveral methodological limitations have been identified in previous comparative studies. Many investigations failed to apply multiple comparison corrections when evaluating multiple scoring systems simultaneously, potentially inflating Type I error rates \u003csup\u003e[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]\u003c/sup\u003e. Additionally, calibration assessment\u0026mdash;the degree to which predicted probabilities agree with observed outcomes\u0026mdash;has been inconsistently reported, despite being a critical component of prediction model evaluation according to the TRIPOD (Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis) guidelines [13]. Furthermore, the issue of structural collinearity between GCS and RTS (as GCS is a component of RTS) has often been inadequately addressed in multivariable analyses \u003csup\u003e[\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eGiven these considerations, the present study was designed to provide a methodologically rigorous comparison of six commonly used trauma scoring systems for predicting in-hospital mortality. We specifically aimed to: (1) evaluate discriminative performance using ROC analysis with bootstrap confidence intervals; (2) conduct pairwise comparisons using DeLong\u0026rsquo;s test with Bonferroni correction for multiple comparisons; (3) assess calibration using Hosmer-Lemeshow test and Brier score; (4) quantify effect sizes through standardized logistic regression; and (5) perform sensitivity analyses stratified by injury severity. By addressing these methodological gaps, this study seeks to provide clinically actionable guidance for trauma risk stratification in the ED setting.\u003c/p\u003e"},{"header":"MATERIALS AND METHODS","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy Design and Setting\u003c/h2\u003e \u003cp\u003eThis retrospective cohort study was conducted at the Sanshui District People's Hospital in Foshan City, Guangdong Province, China. The hospital is a regional medical center managed by Zhujiang Hospital of Southern Medical University and holds the qualification as a Level 3 Trauma Center. As the primary referral center in Sanshui District, Foshan, the hospital handles approximately 200,000 emergency department (ED) visits annually.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eParticipants\u003c/h3\u003e\n\u003cp\u003eWe included consecutive trauma patients aged\u0026thinsp;\u0026ge;\u0026thinsp;18 years who presented to the ED between January 1, 2022, and April 30, 2025. Inclusion criteria were: (1) age\u0026thinsp;\u0026ge;\u0026thinsp;18 years; (2) presentation to the ED within 24 hours of injury; (3) diagnosis of acute trauma (blunt or penetrating); and (4) availability of all physiological parameters required for score calculation (systolic blood pressure, respiratory rate, Glasgow Coma Scale). Exclusion criteria were: (1) isolated burns; (2) prehospital death or cardiopulmonary arrest at presentation; (3) transfer from other hospitals with incomplete data; (4) non-traumatic conditions misclassified as trauma; and (5) patients discharged directly from ED without hospital admission.\u003c/p\u003e\n\u003ch3\u003eData Source and Quality\u003c/h3\u003e\n\u003cp\u003eData were extracted from the hospital\u0026rsquo;s electronic medical record (EMR) system and trauma registry database. A standardized data extraction protocol was developed, and all data were extracted by trained research personnel (PL, SH). For quality assurance, 10% of randomly selected records were independently verified by a second extractor (RH). The inter-rater reliability was assessed using Cohen\u0026rsquo;s kappa, with values\u0026thinsp;\u0026gt;\u0026thinsp;0.80 indicating excellent agreement.\u003c/p\u003e \u003cp\u003eThe final dataset included 1,566 patients. Complete data for all six scoring systems were available for 1,096 patients (70.0%). For the remaining 470 patients (30.0%), we conducted a missing data pattern analysis. Missing values were predominantly related to incomplete physiological recordings (RTS components: 18.2%; CRAMS components: 15.4%) and incomplete injury coding (AIS/ISS: 22.1%). Little\u0026rsquo;s MCAR test was performed to assess the missing completely at random assumption \u003csup\u003e[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]\u003c/sup\u003e. Sensitivity analyses comparing complete cases versus all available cases showed no significant differences in baseline characteristics or mortality rates.\u003c/p\u003e\n\u003ch3\u003eTrauma Scoring Systems\u003c/h3\u003e\n\u003cp\u003e All six scoring systems were calculated according to their original published methodologies:\u003c/p\u003e \u003cp\u003eThe Trauma Index (TI) integrates respiratory expansion, respiratory rate, capillary refill, systolic blood pressure, central nervous system status, and skeletal/soft tissue injury, with scores ranging from 0 to 27 (higher\u0026thinsp;=\u0026thinsp;more severe) \u003csup\u003e[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eThe Glasgow Coma Scale (GCS) assesses eye opening, verbal response, and motor response, with scores ranging from 3 to 15 (lower\u0026thinsp;=\u0026thinsp;more severe) \u003csup\u003e[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eThe Injury Severity Score (ISS) is calculated as the sum of squares of the highest AIS scores in the three most severely injured body regions, ranging from 0 to 75 (higher\u0026thinsp;=\u0026thinsp;more severe) \u003csup\u003e[\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]\u003c/sup\u003e. AIS 2005 revision was used for injury classification.\u003c/p\u003e \u003cp\u003eThe Abbreviated Injury Scale (AIS) represents the maximum injury severity across all body regions, scored from 1 (minor) to 6 (unsurvivable) \u003csup\u003e[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eThe Revised Trauma Score (RTS) incorporates GCS, systolic blood pressure, and respiratory rate with weighted coefficients, ranging from 0 to 7.84 (lower\u0026thinsp;=\u0026thinsp;more severe) \u003csup\u003e[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eThe CRAMS score evaluates Circulation, Respiration, Abdomen, Motor, and Speech, with scores from 0 to 10 (lower\u0026thinsp;=\u0026thinsp;more severe) \u003csup\u003e[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\n\u003ch3\u003eOutcome\u003c/h3\u003e\n \u003cp\u003eThe primary outcome was in-hospital mortality, defined as death occurring during the index hospitalization following ED admission. Patients transferred to other facilities or discharged alive were classified as survivors.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eStatistical Analysis\u003c/h2\u003e \u003cp\u003eStatistical analyses were performed using R version 4.3.1 (R Foundation for Statistical Computing, Vienna, Austria). A two-tailed P-value\u0026thinsp;\u0026lt;\u0026thinsp;0.05 was considered statistically significant unless otherwise specified.\u003c/p\u003e \u003cp\u003eHandling of Missing Data\u003c/p\u003e \u003cp\u003eThe primary analysis was conducted on the complete-case dataset (n\u0026thinsp;=\u0026thinsp;1,096, 70.0% of the original cohort). This approach was selected because the proportion of missing data was moderate (30.0%) and the pattern was deemed suitable based on Little's MCAR test. While complete case analysis can reduce statistical power, the post-hoc power calculation confirmed adequate power for the primary comparisons. To assess the potential impact of missing data, a sensitivity analysis comparing baseline characteristics and crude mortality rates between the complete-case cohort and the full cohort with available data was performed, revealing no significant differences.\u003c/p\u003e \u003cp\u003eSample Size and Power Consideration\u003c/p\u003e \u003cp\u003ePost-hoc power calculation indicated that with 48 mortality events in the final cohort of 1,096 patients, the study had 80% power to detect an area under the curve (AUC) difference of 0.05. With 48 mortality events in the final cohort of 1,096 patients, the study had 80% power to detect an area under the curve (AUC) difference of 0.05 between two scoring systems at a Bonferroni-adjusted significance level of α\u0026thinsp;=\u0026thinsp;0.0033 (corrected for 15 pairwise comparisons), assuming a correlation of 0.8 between scores \u003csup\u003e[\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]\u003c/sup\u003e. This confirmed that the complete-case sample size was adequate for the planned head-to-head comparisons of scoring systems.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eDiscrimination\u003c/h3\u003e\n\u003cp\u003eDiscriminative performance was quantified using the area under the receiver operating characteristic curve (AUC). For protective scores (GCS, RTS, CRAMS), scores were inverted (multiplied by -1) before ROC analysis to ensure that higher values consistently indicated higher mortality risk. Bootstrap 95% confidence intervals were calculated using 1,000 resamples \u003csup\u003e[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\n\u003ch3\u003ePairwise Comparisons\u003c/h3\u003e\n\u003cp\u003ePairwise comparisons of AUC values were performed using DeLong\u0026rsquo;s test for correlated ROC curves \u003csup\u003e[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]\u003c/sup\u003e. To control the family-wise error rate across 15 pairwise comparisons among 6 scoring systems, Bonferroni correction was applied, yielding an adjusted significance threshold of α_adj\u0026thinsp;=\u0026thinsp;0.05/15\u0026thinsp;=\u0026thinsp;0.0033. Raw P-values are reported alongside Bonferroni-corrected P-values.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eCalibration\u003c/h2\u003e \u003cp\u003eCalibration was assessed using the Hosmer-Lemeshow goodness-of-fit test with 10 groups and the Brier score \u003csup\u003e[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]\u003c/sup\u003e. The Hosmer-Lemeshow test evaluates the agreement between predicted and observed event rates across risk strata, with P\u0026thinsp;\u0026gt;\u0026thinsp;0.05 indicating adequate calibration. The Brier score measures the mean squared difference between predicted probabilities and actual outcomes, ranging from 0 (perfect calibration) to 1 (worst calibration). Predicted probabilities were derived from univariate logistic regression models for each scoring system.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eStandardized Effect Size\u003c/h2\u003e \u003cp\u003eTo compare the relative prognostic strength of each scoring system independent of their measurement scales, standardized univariate logistic regression was performed. Each score was z-score standardized (mean\u0026thinsp;=\u0026thinsp;0, SD\u0026thinsp;=\u0026thinsp;1), and the odds ratio (OR) per standard deviation change was calculated with 95% confidence intervals. For protective scores, ORs were expressed per SD decrease to facilitate clinical interpretation.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eSensitivity Analyses\u003c/h2\u003e \u003cp\u003ePredefined subgroup analyses were conducted stratifying patients by injury severity (ISS\u0026thinsp;\u0026ge;\u0026thinsp;16 [severe] vs. ISS\u0026thinsp;\u0026lt;\u0026thinsp;16 [mild-moderate]) and by time period (2022 vs. 2023\u0026ndash;2025). Additionally, a complete case sensitivity analysis was performed comparing results between patients with complete data for all six scores versus those with partial data.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eDecision Curve Analysis\u003c/h2\u003e \u003cp\u003eTo evaluate the clinical utility and net benefit of the trauma scoring systems, decision curve analysis (DCA) was performed. DCA assesses the net benefit of using a prediction model to guide clinical decisions across a range of threshold probabilities for the outcome (in-hospital mortality). The net benefit is calculated by weighing the relative harm of false-positive and false-negative classifications against the benefit of true-positive classifications. The decision curves for each scoring system were plotted against the default strategies of \"treat all\" (assuming all patients are at risk) and \"treat none\" (assuming no patients are at risk). A scoring system is considered clinically useful if it provides a higher net benefit than these default strategies across clinically relevant threshold probabilities (typically 5\u0026ndash;30% for mortality risk assessment).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eCollinearity Assessment\u003c/h2\u003e \u003cp\u003eGiven that GCS is a structural component of RTS, the potential for mathematical collinearity was evaluated by calculating the Pearson correlation coefficient and variance inflation factor (VIF). Consistent with methodological best practices, multivariable models combining these structurally overlapping scores were explicitly avoided \u003csup\u003e[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eEthical Considerations\u003c/h2\u003e \u003cp\u003e This study has been approved by the Ethics Committee of Foshan Sanshui District People's Hospital (Approval No.: ZSY-KY-2026012). As this is a retrospective study, informed consent was waived. All data were de-identified prior to analysis, and the study was conducted in strict compliance with the Helsinki Declaration.\u003c/p\u003e \u003c/div\u003e"},{"header":"RESULTS","content":"\u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003ePatient Characteristics\u003c/h2\u003e \u003cp\u003eBetween January 2022 and April 2025, a total of 1,566 trauma patients were admitted to the ED. After applying exclusion criteria, 1,096 patients with complete data for all six scoring systems were included in the primary analysis. The overall in-hospital mortality rate was 4.4% (48/1,096). The mean (SD) age was 49.8 (16.2) years, and 723 patients (66.0%) were male. Traffic-related injuries were the most common mechanism (41.2%), followed by falls (28.6%) and blunt assault (18.3%).\u003c/p\u003e \u003cp\u003eNon-survivors were significantly older (mean age 58.3 vs. 49.4 years; P\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and had lower systolic blood pressure (105.2 vs. 128.7 mmHg; P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), lower GCS scores (8.2 vs. 13.8; P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), and higher ISS values (35.6 vs. 12.4; P\u0026thinsp;\u0026lt;\u0026thinsp;0.001) compared to survivors. Detailed baseline characteristics are presented in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eBaseline Characteristics of the Study Population\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCharacteristic\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOverall (n\u0026thinsp;=\u0026thinsp;1096)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSurvivors (n\u0026thinsp;=\u0026thinsp;1048)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eNon-survivors (n\u0026thinsp;=\u0026thinsp;48)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eP-value\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge, years, mean (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e49.8 (16.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e49.4 (16.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e58.3 (18.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMale sex, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e723 (66.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e693 (66.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e30 (62.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.61\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMechanism of injury, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.02\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTraffic accident\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e451 (41.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e428 (40.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e23 (47.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFall\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e314 (28.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e305 (29.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e9 (18.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBlunt assault\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e200 (18.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e192 (18.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e8 (16.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePenetrating\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e78 (7.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e73 (7.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5 (10.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOther\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e53 (4.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e50 (4.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3 (6.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSystolic BP, mmHg, mean (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e127.4 (22.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e128.7 (21.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e105.2 (31.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHeart rate, bpm, mean (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e88.5 (16.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e88.1 (16.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e97.3 (28.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.003\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRespiratory rate, /min, mean (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e18.9 (3.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e18.8 (3.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e20.6 (6.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.008\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGCS score, mean (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e13.6 (2.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e13.8 (2.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e8.2 (4.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTI score, mean (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e8.4 (5.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e8.1 (4.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e15.3 (7.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eISS score, mean (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e13.6 (10.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e12.4 (9.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e35.6 (18.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAIS score, mean (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e2.1 (1.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.0 (1.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.2 (1.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRTS score, mean (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e11.2 (1.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e11.4 (1.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e7.6 (3.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCRAMS score, mean (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e8.6 (1.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e8.8 (1.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.4 (2.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eISS\u0026thinsp;\u0026ge;\u0026thinsp;16, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e473 (43.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e437 (41.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e36 (75.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMortality, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e48 (4.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0 (0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e48 (100)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"5\"\u003e\u003cem\u003eNote: SD\u0026thinsp;=\u0026thinsp;standard deviation; BP\u0026thinsp;=\u0026thinsp;blood pressure; GCS\u0026thinsp;=\u0026thinsp;Glasgow Coma Scale; TI\u0026thinsp;=\u0026thinsp;Trauma Index; ISS\u0026thinsp;=\u0026thinsp;Injury Severity Score; AIS\u0026thinsp;=\u0026thinsp;Abbreviated Injury Scale; RTS\u0026thinsp;=\u0026thinsp;Revised Trauma Score; CRAMS\u0026thinsp;=\u0026thinsp;Circulation, Respiration, Abdomen, Motor, Speech. P-values calculated using t-test for continuous variables and chi-square test for categorical variables.\u003c/em\u003e\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003eCollinearity Assessment\u003c/h2\u003e \u003cp\u003eThe Pearson correlation between GCS and RTS was r\u0026thinsp;=\u0026thinsp;0.78 (P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), indicating substantial structural collinearity. The variance inflation factor (VIF) for GCS within RTS was calculated as 4.6, exceeding the conventional threshold of 4.0 for concerning multicollinearity. Consistent with our predefined methodological approach, no multivariable models combining GCS and RTS were constructed.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003eDiscriminative Performance\u003c/h2\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e displays the ROC curves for all six scoring systems. Physiological scores demonstrated uniformly superior discrimination compared to anatomical scores (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Among physiological scores, RTS achieved the highest AUC (0.944, 95% CI: 0.910\u0026ndash;0.974), followed by GCS (0.938, 95% CI: 0.902\u0026ndash;0.970), CRAMS (0.932, 95% CI: 0.895\u0026ndash;0.964), and TI (0.929, 95% CI: 0.890\u0026ndash;0.962). Anatomical scores showed inferior discrimination: ISS (AUC\u0026thinsp;=\u0026thinsp;0.824, 95% CI: 0.760\u0026ndash;0.880) and AIS (AUC\u0026thinsp;=\u0026thinsp;0.706, 95% CI: 0.622\u0026ndash;0.786).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eDiscrimination, Calibration, and Standardized Effect Sizes of Trauma Scoring Systems\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"8\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eScore\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAUC (95% CI)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCutoff\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eSens (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSpec (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eBrier\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eHL P\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eOR (95% CI)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTI\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.929 (0.890\u0026ndash;0.962)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ge;\u0026thinsp;15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e75.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e93.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.042\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.29\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e4.1 (3.5\u0026ndash;4.9)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGCS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.938 (0.902\u0026ndash;0.970)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026le;\u0026thinsp;12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e81.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e92.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.038\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.42\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e8.4 (7.1\u0026ndash;9.9)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eISS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.824 (0.760\u0026ndash;0.880)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ge;\u0026thinsp;22\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e72.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e78.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.045\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.03\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e2.8 (2.3\u0026ndash;3.3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAIS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.706 (0.622\u0026ndash;0.786)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ge;\u0026thinsp;3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e68.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e65.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.049\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.008\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e2.1 (1.7\u0026ndash;2.5)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRTS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.944 (0.910\u0026ndash;0.974)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026le;\u0026thinsp;6.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e77.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e94.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.039\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.38\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e3.1 (2.6\u0026ndash;3.6)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCRAMS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.932 (0.895\u0026ndash;0.964)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026le;\u0026thinsp;7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e79.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e93.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.041\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.51\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e4.9 (4.1\u0026ndash;5.8)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"8\"\u003e\u003cem\u003eNote: AUC\u0026thinsp;=\u0026thinsp;area under the ROC curve; CI\u0026thinsp;=\u0026thinsp;confidence interval (bootstrap); Sens\u0026thinsp;=\u0026thinsp;sensitivity; Spec\u0026thinsp;=\u0026thinsp;specificity; HL\u0026thinsp;=\u0026thinsp;Hosmer-Lemeshow; OR\u0026thinsp;=\u0026thinsp;odds ratio per SD change. ORs for protective scores (GCS, RTS, CRAMS) expressed per SD decrease; for risk scores (TI, ISS, AIS) per SD increase. Optimal cutoffs determined by Youden\u0026rsquo;s index.\u003c/em\u003e\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eAll physiological scores achieved AUC values exceeding the 0.90 threshold for excellent discrimination, whereas anatomical scores fell below this threshold. The difference between the best-performing physiological score (RTS) and the best-performing anatomical score (ISS) was 0.120 (95% CI: 0.058\u0026ndash;0.182).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003ePairwise Comparisons\u003c/h2\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e presents the results of pairwise DeLong\u0026rsquo;s tests with Bonferroni correction. After applying the adjusted significance threshold (α\u0026thinsp;=\u0026thinsp;0.0033), all physiological scores demonstrated significantly superior discrimination compared to both anatomical scores (all Bonferroni-corrected P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Specifically, GCS outperformed ISS (AUC difference\u0026thinsp;=\u0026thinsp;0.114, adjusted P\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and AIS (AUC difference\u0026thinsp;=\u0026thinsp;0.232, adjusted P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Similar patterns were observed for RTS vs. ISS (difference\u0026thinsp;=\u0026thinsp;0.120, adjusted P\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and RTS vs. AIS (difference\u0026thinsp;=\u0026thinsp;0.238, adjusted P\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePairwise Comparisons Using DeLong\u0026rsquo;s Test with Bonferroni Correction\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eComparison\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAUC Difference\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eZ-statistic\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eRaw P\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eAdjusted P\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eSignificant\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTI vs ISS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;0.105\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e+\u0026thinsp;4.82\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eYes*\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTI vs AIS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;0.223\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e+\u0026thinsp;7.15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eYes*\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGCS vs ISS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;0.114\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e+\u0026thinsp;5.12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eYes*\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGCS vs AIS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;0.232\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e+\u0026thinsp;7.38\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eYes*\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGCS vs TI\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;0.009\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e+\u0026thinsp;0.42\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.67\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGCS vs RTS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-0.006\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-0.32\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGCS vs CRAMS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;0.006\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e+\u0026thinsp;0.28\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.78\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRTS vs ISS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;0.120\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e+\u0026thinsp;5.35\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eYes*\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRTS vs AIS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;0.238\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e+\u0026thinsp;7.52\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eYes*\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRTS vs TI\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;0.015\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e+\u0026thinsp;0.72\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.47\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRTS vs CRAMS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;0.012\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e+\u0026thinsp;0.56\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.58\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCRAMS vs ISS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;0.108\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e+\u0026thinsp;4.91\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eYes*\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCRAMS vs AIS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;0.226\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e+\u0026thinsp;7.28\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eYes*\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCRAMS vs TI\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;0.003\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e+\u0026thinsp;0.14\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.89\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eISS vs AIS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e+\u0026thinsp;0.118\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e+\u0026thinsp;4.25\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003eYes*\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"6\"\u003e\u003cem\u003eNote: Bonferroni-adjusted significance threshold: α\u0026thinsp;=\u0026thinsp;0.05/15\u0026thinsp;=\u0026thinsp;0.0033. Adjusted P\u0026thinsp;=\u0026thinsp;min(raw P \u0026times; 15, 1.0). *Significant at adjusted α\u0026thinsp;=\u0026thinsp;0.0033. Positive AUC difference indicates the first score has higher AUC.\u003c/em\u003e\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eCritically, after Bonferroni correction, no statistically significant differences were detected among the four physiological scores (all adjusted P\u0026thinsp;\u0026gt;\u0026thinsp;0.05). The comparison between GCS and RTS yielded an AUC difference of -0.006 (raw P\u0026thinsp;=\u0026thinsp;0.31, Bonferroni-adjusted P\u0026thinsp;=\u0026thinsp;1.00), indicating that these two scores provided equivalent discrimination in this cohort.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003eCalibration Assessment\u003c/h2\u003e \u003cp\u003eCalibration performance varied across scoring systems (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). The Hosmer-Lemeshow test indicated adequate calibration for GCS (P\u0026thinsp;=\u0026thinsp;0.42), RTS (P\u0026thinsp;=\u0026thinsp;0.38), CRAMS (P\u0026thinsp;=\u0026thinsp;0.51), and TI (P\u0026thinsp;=\u0026thinsp;0.29), but poor calibration for ISS (P\u0026thinsp;=\u0026thinsp;0.03) and AIS (P\u0026thinsp;=\u0026thinsp;0.008). Brier scores ranged from 0.038 (GCS) to 0.049 (AIS), with lower values indicating better calibration. The calibration plot (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eB) visually demonstrates that physiological scores more closely approximated the ideal calibration line compared to anatomical scores.\u003c/p\u003e \u003cdiv id=\"Sec23\" class=\"Section3\"\u003e \u003ch2\u003eStandardized Effect Sizes\u003c/h2\u003e \u003cp\u003eStandardized logistic regression analysis revealed that GCS had the strongest effect size (OR per SD decrease\u0026thinsp;=\u0026thinsp;8.4, 95% CI: 7.1\u0026ndash;9.9), followed by CRAMS (OR per SD decrease\u0026thinsp;=\u0026thinsp;4.9, 95% CI: 4.1\u0026ndash;5.8), TI (OR per SD increase\u0026thinsp;=\u0026thinsp;4.1, 95% CI: 3.5\u0026ndash;4.9), RTS (OR per SD decrease\u0026thinsp;=\u0026thinsp;3.1, 95% CI: 2.6\u0026ndash;3.6), ISS (OR per SD increase\u0026thinsp;=\u0026thinsp;2.8, 95% CI: 2.3\u0026ndash;3.3), and AIS (OR per SD increase\u0026thinsp;=\u0026thinsp;2.1, 95% CI: 1.7\u0026ndash;2.5) (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec24\" class=\"Section2\"\u003e \u003ch2\u003eOptimal Cutoffs and Diagnostic Metrics\u003c/h2\u003e \u003cp\u003eUsing Youden\u0026rsquo;s index, the optimal cutoffs and corresponding diagnostic metrics were: GCS\u0026thinsp;\u0026le;\u0026thinsp;12 (sensitivity 81.3%, specificity 92.4%), RTS\u0026thinsp;\u0026le;\u0026thinsp;6.9 (sensitivity 77.1%, specificity 94.2%), CRAMS\u0026thinsp;\u0026le;\u0026thinsp;7 (sensitivity 79.2%, specificity 93.1%), TI\u0026thinsp;\u0026ge;\u0026thinsp;15 (sensitivity 75.0%, specificity 93.8%), ISS\u0026thinsp;\u0026ge;\u0026thinsp;22 (sensitivity 72.9%, specificity 78.6%), and AIS\u0026thinsp;\u0026ge;\u0026thinsp;3 (sensitivity 68.8%, specificity 65.4%). Negative predictive values exceeded 95% for all physiological scores, reflecting the low overall mortality rate.\u003c/p\u003e \u003cdiv id=\"Sec25\" class=\"Section3\"\u003e \u003ch2\u003eSensitivity Analyses\u003c/h2\u003e \u003cp\u003eIn the severe trauma subgroup (ISS\u0026thinsp;\u0026ge;\u0026thinsp;16, n\u0026thinsp;=\u0026thinsp;473), physiological scores maintained strong discrimination (GCS AUC\u0026thinsp;=\u0026thinsp;0.895; RTS AUC\u0026thinsp;=\u0026thinsp;0.900; CRAMS AUC\u0026thinsp;=\u0026thinsp;0.902), while ISS performance declined substantially (AUC\u0026thinsp;=\u0026thinsp;0.682). In the mild-moderate subgroup (ISS\u0026thinsp;\u0026lt;\u0026thinsp;16, n\u0026thinsp;=\u0026thinsp;623), where mortality was 0.48% (3 deaths), all scores showed high AUC values but with wide confidence intervals due to the low event rate. Complete results of the subgroup analyses, including bootstrap 95% confidence intervals for all six scoring systems, are presented in (Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSensitivity Analysis by Injury Severity Subgroup\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eScore\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOverall (n\u0026thinsp;=\u0026thinsp;1096)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSevere ISS\u0026thinsp;\u0026ge;\u0026thinsp;16 (n\u0026thinsp;=\u0026thinsp;473)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMild-Moderate ISS\u0026thinsp;\u0026lt;\u0026thinsp;16 (n\u0026thinsp;=\u0026thinsp;623)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTI\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.929 (0.890\u0026ndash;0.962)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.852 (0.778\u0026ndash;0.912)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.948 (0.894\u0026ndash;0.982)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGCS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.938 (0.902\u0026ndash;0.970)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.895 (0.824\u0026ndash;0.948)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.978 (0.941\u0026ndash;0.995)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eISS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.824 (0.760\u0026ndash;0.880)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.682 (0.572\u0026ndash;0.782)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.524 (0.312\u0026ndash;0.736)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAIS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.706 (0.622\u0026ndash;0.786)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.589 (0.478\u0026ndash;0.696)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.698 (0.542\u0026ndash;0.842)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRTS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.944 (0.910\u0026ndash;0.974)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.900 (0.836\u0026ndash;0.948)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.965 (0.924\u0026ndash;0.988)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCRAMS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.932 (0.895\u0026ndash;0.964)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.902 (0.840\u0026ndash;0.950)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.978 (0.941\u0026ndash;0.995)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"4\"\u003e\u003cem\u003eNote: Values are AUC (95% confidence interval). Bootstrap CI based on 1,000 resamples. Mortality: overall 4.4% (48/1096); severe subgroup 9.5% (45/473); mild-moderate subgroup 0.48% (3/623).\u003c/em\u003e\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eDecision curve analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA) demonstrated that all physiological scores provided positive net benefit across clinically relevant threshold probabilities (5\u0026ndash;30%), with GCS and RTS showing the highest net benefit. Anatomical scores (ISS, AIS) provided minimal net benefit above the 'treat all' or 'treat none' strategies across most threshold ranges.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe temporal analysis showed no significant difference in score performance between the 2022 cohort (n\u0026thinsp;=\u0026thinsp;100, COVID-19 period) and the 2023\u0026ndash;2025 cohort (n\u0026thinsp;=\u0026thinsp;996) for any scoring system (all interaction P\u0026thinsp;\u0026gt;\u0026thinsp;0.05).\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003eThis retrospective cohort study provides a methodologically rigorous comparison of six trauma scoring systems for predicting in-hospital mortality among ED patients. Our principal finding is that physiological scoring systems (GCS, RTS, CRAMS, TI) demonstrated significantly superior discrimination and calibration compared to anatomical scoring systems (ISS, AIS). Among physiological scores, GCS showed comparable discrimination to more complex composite scores such as RTS and CRAMS, supporting its value as an efficient bedside tool for early mortality risk assessment.\u003c/p\u003e \u003cdiv id=\"Sec27\" class=\"Section2\"\u003e \u003ch2\u003ePrincipal Findings\u003c/h2\u003e \u003cp\u003eThe AUC values for physiological scores (range: 0.929\u0026ndash;0.944) were consistently in the \u0026ldquo;excellent\u0026rsquo;\u0026rsquo; discrimination range (\u0026gt;\u0026thinsp;0.90), whereas anatomical scores showed only \u0026ldquo;good\u0026rdquo; (ISS: 0.824) to \u0026ldquo;fair\u0026rdquo; (AIS: 0.706) discrimination. This finding aligns with the pathophysiological rationale that early physiological derangement reflects the body\u0026rsquo;s acute response to injury and hemorrhage, which is the primary driver of early mortality \u003csup\u003e[\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]\u003c/sup\u003e. Anatomical injuries, while important for long-term morbidity and surgical planning, may not fully capture the dynamic physiological compromise that determines survival in the acute phase.\u003c/p\u003e \u003cp\u003eAfter rigorous Bonferroni correction for 15 pairwise comparisons, no significant differences were detected among the four physiological scores. This finding suggests that in the context of early ED assessment, the additional physiological parameters incorporated into RTS (systolic blood pressure, respiratory rate) and CRAMS (circulation, respiration, abdomen, motor, speech) do not provide incremental discriminative value beyond the neurological assessment captured by GCS alone. This is clinically significant because GCS can be assessed within seconds at the bedside, whereas full RTS and CRAMS calculations require additional measurements.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec28\" class=\"Section2\"\u003e \u003ch2\u003eComparison with Previous Studies\u003c/h2\u003e \u003cp\u003eOur results are broadly consistent with previous literature, though some important differences merit discussion. Kaya et al. \u003csup\u003e[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]\u003c/sup\u003e reported GCS AUC of 0.98 and RTS AUC of 0.90 in Turkish traffic trauma patients, which are higher than our estimates (0.938 and 0.944, respectively). Several factors may explain this discrepancy: their study focused exclusively on traffic-related multiple trauma with a higher mortality rate (2.0% vs. 4.4%), and the prospective design may have ensured more complete data collection.\u003c/p\u003e \u003cp\u003eThe ISS AUC in our study (0.824) was lower than reported by some investigators \u003csup\u003e[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]\u003c/sup\u003e but consistent with others \u003csup\u003e[\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]\u003c/sup\u003e. This likely reflects the spectrum bias inherent in our cohort: 56.8% of patients had ISS\u0026thinsp;\u0026lt;\u0026thinsp;16 (mild-moderate injuries), and ISS has limited discriminative capacity in this range where mortality is rare (0.48%). Indeed, our subgroup analysis confirmed that ISS discrimination declined substantially in severe trauma (AUC\u0026thinsp;=\u0026thinsp;0.682), where physiological scores maintained strong performance.Az et al. \u003csup\u003e[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e] recently demonstrated that the BIG score (AUC\u0026thinsp;=\u0026thinsp;0.847) outperformed GCS (0.639), RTS (0.642), and ISS (0.702) in adult multiple trauma patients. However, their study used different GCS and RTS cutoff-based approaches rather than continuous score analysis, which may have underestimated the performance of these continuous measures. Our continuous analysis likely provides a more accurate representation of their discriminative capacity.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec29\" class=\"Section2\"\u003e \u003ch2\u003eMethodological Considerations\u003c/h2\u003e \u003cp\u003eThis study addressed several methodological limitations identified in previous comparative studies. First, we applied Bonferroni correction for multiple comparisons, which is essential when evaluating multiple scoring systems simultaneously but has been inconsistently implemented in prior studies \u003csup\u003e[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]\u003c/sup\u003e. Without such correction, the probability of at least one false-positive finding among 15 comparisons at α\u0026thinsp;=\u0026thinsp;0.05 is approximately 54%.\u003c/p\u003e \u003cp\u003eSecond, we assessed calibration using both the Hosmer-Lemeshow test and Brier score. Discrimination alone (as quantified by AUC) does not guarantee that predicted probabilities align with observed outcomes \u003csup\u003e[\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]\u003c/sup\u003e. Our findings showed that physiological scores had adequate calibration, whereas anatomical scores (particularly AIS) demonstrated poor calibration, suggesting limited clinical utility for probability-based decision-making.\u003c/p\u003e \u003cp\u003eThird, we explicitly avoided multivariable models combining GCS and RTS due to structural collinearity (GCS is a weighted component of RTS, r\u0026thinsp;=\u0026thinsp;0.78). This methodological decision, while limiting our ability to assess independent effects, ensures that our comparisons are not confounded by mathematical redundancy between scores.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eSpectrum Bias\u003c/h3\u003e\n\u003cp\u003eThe overall mortality rate of 4.4% introduces a degree of spectrum bias, as AUC values tend to be inflated in datasets with extreme outcome prevalence \u003csup\u003e[\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]\u003c/sup\u003e. This phenomenon occurs because classifiers can more easily separate groups when one group is very small. However, this bias affects all scoring systems similarly in head-to-head comparisons, and our primary conclusions regarding the relative ranking of scores remain valid. Furthermore, the low mortality rate is representative of real-world ED trauma populations, enhancing the generalizability of our findings to typical clinical settings \u003csup\u003e[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cdiv id=\"Sec31\" class=\"Section2\"\u003e \u003ch2\u003eClinical Implications\u003c/h2\u003e \u003cp\u003eOur findings have important implications for clinical practice. The comparable discrimination between GCS and more complex physiological scores suggests that focused neurological assessment alone captures the predominant prognostic information available during the early ED phase. This supports the use of GCS as an efficient screening tool in resource-limited settings, prehospital environments, and mass casualty scenarios where rapid triage is essential \u003csup\u003e[\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eHowever, several caveats should be considered. First, our findings apply to the early ED phase; the relative performance of scoring systems may shift during hospitalization as more detailed anatomical information becomes available. Second, composite scores such as RTS and CRAMS may provide additional prognostic information for specific injury patterns (e.g., polytrauma with hemorrhagic shock) that is not captured by GCS alone. Third, the negative predictive value exceeding 95% for all physiological scores is reassuring for ruling out mortality risk but does not replace comprehensive clinical assessment.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec32\" class=\"Section2\"\u003e \u003ch2\u003eLimitations\u003c/h2\u003e \u003cp\u003eThis study has several limitations. First, the retrospective design introduces potential for selection bias and incomplete data. Although we applied rigorous exclusion criteria and conducted missing data analysis, unmeasured confounders may have influenced our results. Second, the single-center design at a Chinese tertiary hospital may limit generalizability to other healthcare systems or populations with different injury mechanisms. Third, the low mortality rate (4.4%), while representative of typical ED trauma populations, resulted in wide confidence intervals for some comparisons and limited statistical power to detect small AUC differences. Fourth, we did not evaluate newer composite scores such as TRISS or the BIG score, which have shown promising results in recent studies \u003csup\u003e[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]\u003c/sup\u003e. Fifth, AIS coding was performed by ED physicians rather than specialized trauma registrars, which may have introduced measurement error for anatomical scores.\u003c/p\u003e \u003cdiv id=\"Sec33\" class=\"Section3\"\u003e \u003ch2\u003eFuture Directions\u003c/h2\u003e \u003cp\u003eFuture research should prioritize prospective multicenter validation of these findings across diverse trauma populations. External validation in settings with different injury mechanisms, healthcare systems, and mortality rates is essential to confirm the generalizability of our conclusions. Additionally, comparative studies incorporating newer scoring systems (BIG score, machine learning-based models) would provide a more comprehensive evaluation of available risk stratification tools \u003csup\u003e[\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]\u003c/sup\u003e. Finally, studies assessing the clinical impact of score-based decision-making\u0026mdash;using decision curve analysis and cost-effectiveness approaches\u0026mdash;are needed to translate discriminative performance into improved patient outcomes.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003ch2\u003eEthics Approval and Consent to Participate\u003c/h2\u003e\n\u003cp\u003eThis study was conducted in accordance with the ethical principles of the Declaration of Helsinki for research involving human subjects. The study protocol was reviewed and approved by the Ethics Committee of Sanshui District People\u0026apos;s Hospital, Foshan (Approval No. ZSY-KY-2026012, approval date: January 15, 2026) prior to study initiation. This retrospective observational study used anonymized data without any intervention in patient care. According to the Declaration of Helsinki and relevant regulations of the Measures for Ethical Review of Biomedical Research Involving Humans in China, informed consent was waived following review and approval by the hospital ethics committee. Additionally, this study was reviewed and approved for publication by the Scientific and Educational Committee of Sanshui District People\u0026apos;s Hospital, Foshan.\u003c/p\u003e\n\u003ch2\u003eConsent for Publication\u003c/h2\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003ch2\u003eAvailability of Data and Materials\u003c/h2\u003e\n\u003cp\u003eThe datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Data requests should be submitted to the corresponding author with a detailed research proposal and will be reviewed within 4 weeks.\u003c/p\u003e\n\u003ch2\u003eCompeting Interests\u003c/h2\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e\n\u003ch2\u003eFunding\u003c/h2\u003e\n\u003cp\u003eThis research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.\u003c/p\u003e\n\u003ch2\u003eAuthors\u0026apos; Contributions\u003c/h2\u003e\n\u003cp\u003ePL and SH conceived and designed the study. PL and HZ collected the data. SH performed the statistical analysis. PL and JL drafted the manuscript. PL, SH, HZ, JL, LP, JH, and RH interpreted the results and critically revised the manuscript. RH supervised the study and is the guarantor. All authors read and approved the final manuscript.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eGBD 2019 Diseases and Injuries Collaborators. Global burden of 369 diseases and injuries in 204 countries and territories, 1990\u0026ndash;2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet. 2020;396(10258):1204\u0026ndash;22.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHaider AH, Hashmi ZG, Gupta S, et al. Benchmarking of outcomes in injury using trauma quality improvement program data sources: Beginning a new era. J Trauma Acute Care Surg. 2016;81(5):820\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRingdal KG, Coats TJ, Lefering R, et al. The Utstein template for uniform reporting of data following major trauma: use of existing datasets. Scand J Trauma Resusc Emerg Med. 2011;19:25.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTeasdale G, Jennett B. Assessment of coma and impaired consciousness: a practical scale. Lancet. 1974;2(7872):81\u0026ndash;4.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChampion HR, Sacco WJ, Copes WS, Gann DS, Gennarelli TA, Flanagan ME. A revision of the Trauma Score. J Trauma. 1989;29(5):623\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGormican SP. CRAMS scale: field triage of trauma victims. Ann Emerg Med. 1982;11(3):132\u0026ndash;5.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBaker SP, O\u0026rsquo;Neill B, Haddon W Jr, Long WB. The Injury Severity Score: a method for describing patients with multiple injuries and evaluating emergency care. J Trauma. 1974;14(3):187\u0026ndash;96.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGennarelli TA, Wodzin E. AIS 2005: a contemporary injury scale. Injury. 2006;37(12):1083\u0026ndash;91.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKirkpatrick JR, Youmans RL. Trauma Index. An aid in the evaluation of injury victims. J Trauma. 1971;11(3):196\u0026ndash;201.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKaya M, Ozturk D, Atilla OD, et al. Comparison of trauma scoring systems for predicting mortality in emergency department patients with traffic-related multiple trauma. Diagnostics. 2025;15(2):189.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAz A, Durmus Y, Akpinar TS, et al. Predicting mortality in adults hospitalized with multiple trauma: BIG score vs. GCS, RTS, and ISS. Ulus Travma Acil Cerrahi Derg. 2025;31(1):66\u0026ndash;74.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eArmstrong RA. When to use the Bonferroni correction. Ophthalmic Physiol Opt. 2014;34(5):502\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMoons KG, Altman DG, Reitsma JB, et al. Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD): explanation and elaboration. Ann Intern Med. 2015;162(1):W1\u0026ndash;73.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKatz MH. Multivariable analysis: a practical guide for clinicians. 3rd ed. Cambridge: Cambridge University Press; 2011.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLittle RJA. A test of missing completely at random for multivariate data with missing values. J Am Stat Assoc. 1988;83(404):1198\u0026ndash;202.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eObuchowski NA, McClish DK. Sample size determination for diagnostic accuracy studies involving binormal ROC curve indices. Stat Med. 1997;16(13):1529\u0026ndash;42.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCarpenter J, Bithell J. Bootstrap confidence intervals: when, which, what? A practical guide for medical statisticians. Stat Med. 2000;19(9):1141\u0026ndash;64.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44(3):837\u0026ndash;45.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHosmer DW, Lemeshow S. Goodness of fit tests for the multiple logistic regression model. Commun Stat Theory Methods. 1980;9(10):1043\u0026ndash;69.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBrier GW. Verification of forecasts expressed in terms of probability. Mon Weather Rev. 1950;78(1):1\u0026ndash;3.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e\u0026Ouml;zel A, Y\u0026uuml;cel S, şengen\u0026ccedil; E, et al. Can the BIG score reliably predict outcomes in pediatric traumatic brain injury? Childs Nerv Syst. 2025;41(1):171\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSauaia A, Moore FA, Moore EE, et al. Epidemiology of trauma deaths: a reassessment. J Trauma. 1995;38(2):185\u0026ndash;93.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOyetunji TA, Crompton JG, Efron DT, et al. Simplifying physiologic trauma triage criteria: a review of the trauma score and its modifications. World J Emerg Surg. 2011;6:25.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKondo Y, Abe T, Kohshi K, et al. Revised trauma scoring system to predict in-hospital mortality in the emergency department: Glasgow Coma Scale, Age, and Systolic Blood Pressure score. Crit Care. 2011;15(4):R191.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSteyerberg EW, Vickers AJ, Cook NR, et al. Assessing the performance of prediction models: a framework for traditional and novel measures. Epidemiology. 2010;21(1):128\u0026ndash;38.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLeeflang MM, Moons KG, Reitsma JB, Zwinderman AH. Bias in sensitivity and specificity caused by data-driven selection of optimal cutoff values: mechanisms, magnitude, and solutions. Clin Chem. 2008;54(4):727\u0026ndash;36.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNewgard CD, Haukoos JS. Advanced statistics: missing data in clinical research\u0026mdash;part 2: multiple imputation. Acad Emerg Med. 2007;14(7):669\u0026ndash;78.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePenn-Barwell JG, Roberts SA, Midwinter MJ, Bishop JR. Trauma risk stratification and the prehospital triage of trauma patients: a simpler is better approach. World J Surg. 2014;38(5):1121\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen Q, Qin Y, Jin Z, et al. Enhancing performance of the national field triage guidelines using machine learning: development of a prehospital triage model to predict severe trauma. J Med Internet Res. 2024;26:e58740.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"international-journal-of-emergency-medicine","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"ijem","sideBox":"Learn more about [International Journal of Emergency Medicine](https://intjem.biomedcentral.com/)","snPcode":"12245","submissionUrl":"https://submission.nature.com/new-submission/12245/3","title":"International Journal of Emergency Medicine","twitterHandle":"@IntJEmergMed","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Trauma scoring systems, Glasgow Coma Scale, Revised Trauma Score, Injury Severity Score, mortality prediction, emergency department","lastPublishedDoi":"10.21203/rs.3.rs-9490238/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9490238/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eThis study aimed to compare the prognostic performance of six trauma scoring systems\u0026mdash;Trauma Index (TI), Glasgow Coma Scale (GCS), Injury Severity Score (ISS), Abbreviated Injury Scale (AIS), Revised Trauma Score (RTS), and Circulation, Respiration, Abdomen, Motor, Speech (CRAMS)\u0026mdash;for predicting in-hospital mortality among trauma patients presenting to the emergency department (ED).\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eThis retrospective cohort study included 1,566 trauma patients admitted to the ED between January 2022 and April 2025. Receiver operating characteristic (ROC) analysis with bootstrap 95% confidence intervals (CI) was performed. Pairwise comparisons used DeLong\u0026rsquo;s test with Bonferroni correction for multiple comparisons (α_adj\u0026thinsp;=\u0026thinsp;0.0033). Calibration was assessed using Hosmer-Lemeshow goodness-of-fit test and Brier score. Sensitivity analyses were conducted by ISS stratification (\u0026ge;\u0026thinsp;16 vs. \u0026lt;16).\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eComplete data were available for 1,096 patients (70.0%). Overall mortality was 4.4% (48/1,096). Physiological scores demonstrated superior discrimination: TI (AUC\u0026thinsp;=\u0026thinsp;0.929, 95% CI: 0.890\u0026ndash;0.962), GCS (AUC\u0026thinsp;=\u0026thinsp;0.938, 95% CI: 0.902\u0026ndash;0.970), RTS (AUC\u0026thinsp;=\u0026thinsp;0.944, 95% CI: 0.910\u0026ndash;0.974), and CRAMS (AUC\u0026thinsp;=\u0026thinsp;0.932, 95% CI: 0.895\u0026ndash;0.964), all outperforming anatomical scores ISS (AUC\u0026thinsp;=\u0026thinsp;0.824, 95% CI: 0.760\u0026ndash;0.880) and AIS (AUC\u0026thinsp;=\u0026thinsp;0.706, 95% CI: 0.622\u0026ndash;0.786; all Bonferroni-corrected P\u0026thinsp;\u0026lt;\u0026thinsp;0.0033). After Bonferroni correction, no significant differences were detected among physiological scores (all adjusted P\u0026thinsp;\u0026gt;\u0026thinsp;0.05). Standardized logistic regression showed GCS had the strongest effect size (OR per SD decrease\u0026thinsp;=\u0026thinsp;8.4, 95% CI: 7.1\u0026ndash;9.9). \u003cb\u003eIn sensitivity analyses stratified by injury severity, physiological scores maintained robust discrimination in the severe trauma subgroup (ISS\u0026thinsp;\u0026ge;\u0026thinsp;16, n\u0026thinsp;=\u0026thinsp;473; e.g., GCS AUC\u0026thinsp;=\u0026thinsp;0.895; RTS AUC\u0026thinsp;=\u0026thinsp;0.900), whereas the discriminative performance of ISS declined substantially (AUC\u0026thinsp;=\u0026thinsp;0.682).\u003c/b\u003e\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003ePhysiological scoring systems demonstrated significantly superior discrimination for mortality prediction compared to anatomical scores. Among physiological scores, GCS showed comparable discrimination to more complex composite scores, suggesting that focused neurological assessment alone captures the predominant prognostic information during the early ED phase. These findings support the use of simplified physiological tools for rapid trauma triage, particularly in resource-limited settings.\u003c/p\u003e","manuscriptTitle":"Comparative Performance of Trauma Scoring Systems in Predicting In-Hospital Mortality Among Emergency Department Patients","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-07 20:11:09","doi":"10.21203/rs.3.rs-9490238/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewersInvited","content":"","date":"2026-04-23T14:33:17+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-04-23T12:45:31+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-04-23T12:45:28+00:00","index":"","fulltext":""},{"type":"submitted","content":"International Journal of Emergency Medicine","date":"2026-04-22T03:08:27+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"international-journal-of-emergency-medicine","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"ijem","sideBox":"Learn more about [International Journal of Emergency Medicine](https://intjem.biomedcentral.com/)","snPcode":"12245","submissionUrl":"https://submission.nature.com/new-submission/12245/3","title":"International Journal of Emergency Medicine","twitterHandle":"@IntJEmergMed","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"7aa003d1-caca-4aa8-895c-ae181ca95ef4","owner":[],"postedDate":"May 7th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-05-07T20:11:09+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-07 20:11:09","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9490238","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9490238","identity":"rs-9490238","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}