Integrated Risk Stratification Using Comorbidity Burden, Frailty, and Systemic Inflammation to Predict 30-Day Mortality in Older Emergency Department Patients: A Prospective Cohort Study

Abstract

Background/Objectives: Estimating mortality risk in older adults presenting to the emergency department (ED) is challenging because risk is shaped by comorbidity burden, frailty, and systemic inflammation. We examined the individual and combined prognostic contributions of the Age-adjusted Charlson Comorbidity Index (CCI), the Clinical Frailty Scale (CFS), and the Systemic Immune-Inflammation Index (SII) to 30-day mortality. Methods: This prospective observational cohort included 326 adults aged ≥65 years evaluated in a tertiary ED from September 2025 through April 2026. CCI, CFS, and SII were assessed at admission, with SII analyzed after logarithmic transformation [ln(SII)]. Associations with 30-day mortality were examined using logistic regression. Discriminative performance was evaluated by receiver operating characteristic analysis, and the integrated model underwent bootstrap internal validation and calibration assessment. Results: Thirty-day mortality occurred in 60 patients (18.4%). CCI, CFS, and ln(SII) were each independently associated with mortality in the integrated multivariable model (all p < 0.05). Individual AUCs were 0.741 for CCI, 0.703 for CFS, and 0.662 for ln(SII), with CCI showing the highest value. The integrated model achieved the highest observed discrimination (AUC 0.793; optimism-corrected AUC 0.783) and showed acceptable internal calibration (Brier score 0.127; bootstrap-corrected calibration intercept 0.003; calibration slope 0.957). Conclusions: Comorbidity burden, frailty, and systemic inflammation each provided independent prognostic information for 30-day mortality in older ED patients. Although CCI was the strongest individual predictor, the integrated model achieved the highest observed discrimination. However, its incremental benefit over some two-predictor models was modest and not statistically significant.