A Simple Risk Assessment Tool for Older Adult Trauma Patients-Silver Trauma Score.

INTRODUCTION The recent American College of Surgeons Best Practice Guidelines emphasize greater sensitivity in geriatric trauma triage, which has led to a steep increase in trauma team activations. This underscores the need for accurate, easy-to-use prediction tools to guide early decision-making and triage.

Methods

We conducted a retrospective cohort study of 3897 hospitalized blunt trauma patients aged ≥ 65 y presenting directly to a Level 1 trauma center (2017-2023). Efficacy of the Injury Severity Score, Revised Trauma Score, Trauma and Injury Severity Score, quick sequential organ failure assessment, and early warning systems (national early warning score, modified early warning score, and rapid emergency medicine score) was assessed for predicting in-hospital mortality and emergency department (ED)-to-intensive care unit (ICU) disposition. Univariate and multivariable logistic regression analyses were performed with internal cross-validation. Based on these comparisons, two geriatric-specific models-the Silver Trauma Score (STS): STS-Mort and STS-ED-to-ICU-were derived from clinical variables available during the initial ED evaluation.

Results

STS-Mort demonstrated the highest sensitivity and discrimination for mortality (sensitivity 0.85; area under receiver operating characteristic curve (AUC-ROC) 0.77, 95% confidence interval [CI] 0.76-0.78), outperforming quick sequential organ failure assessment (sensitivity 0.57; AUC-ROC 0.70, 95% CI 0.68-0.71). For ED-to-ICU triage, STS-ED-to-ICU achieved the highest sensitivity: 0.83 and discrimination (AUC-ROC 0.80, 95% CI 0.80-0.80), exceeding Injury Severity Score (sensitivity 0.60; AUC-ROC 0.73, 95% CI 0.72-0.73). Univariate analysis showed mortality and ED-to-ICU rates increased with STS score.

Conclusions

STS-Mort and STS-ED-to-ICU improved sensitivity for geriatric risk stratification. STS-Mort may serve as a general prognostic tool, whereas STS-ED-to-ICU reflects institutional geriatric triage practices and may require site-specific calibration. These models may support early resource allocation and reduce undertriage, with performance influenced by variability in ICU admission practices.