Older emergency department (ED) patients are at significant risk of experiencing composite adverse outcomes such as ED revisits, intensive care unit admissions, and all-cause mortality within a short period. Existing predictive tools primarily utilize single scoring systems, showing limited predictive accuracy and clinical applicability. This study evaluated whether combining the identification of seniors at risk (ISAR) and the cumulative illness rating scale (CIRS) with routinely available ED triage information could provide incremental predictive value for 30-day composite adverse outcomes among older ED patients in China, compared with single-scale and clinical-variable models. A prospective, single-center cohort study was conducted involving older patients aged ≥ 65 years attending a tertiary hospital ED in China from March to May 2024. ISAR and CIRS assessments, along with baseline clinical data, were collected upon ED admission. The primary outcome was a composite adverse event within 30 days, including ED revisit, intensive care unit admission, or death. Variable selection was performed using least absolute shrinkage and selection operator regression. Logistic regression models, including single-scale, combined ISAR + CIRS, and baseline clinical models, were constructed and compared based on discrimination, calibration, and clinical net benefit through decision curve analysis. Model interpretability was further evaluated using analyses. A total of 607 older patients were enrolled, with 216 (35.6%) experiencing composite adverse outcomes within 30 days. The combined ISAR + CIRS model showed the highest area under the curve (AUC) among the compared models (AUC = 0.807; 95% confidence interval CI: 0.776–0.839), with a statistically significant improvement over the ISAR-based model (AUC = 0.793; 95% CI: 0.761–0.826). However, the absolute improvement compared with the CIRS-based model was small (AUC = 0.805; 95% CI: 0.773–0.837). The combined model showed favorable calibration and higher net benefit across clinically relevant thresholds. extreme gradient boosting and SHapley Additive exPlanations analyses suggested that ISAR, CIRS, systolic blood pressure, triage level, and pulse rate were important contributors to model output. The combined ISAR + CIRS model provided an interpretable and clinically feasible approach for early risk stratification of older ED patients. Although the improvement in discrimination was modest, integrating acute geriatric risk assessment and chronic comorbidity burden may support more comprehensive risk evaluation alongside routine ED triage and clinical judgment. Further external validation is needed before broader clinical implementation.
Liu et al. (Fri,) studied this question.