C-reactive protein-to-albumin ratio adds independent prognostic value to the clinical frailty scale in older emergency department patients: a multicenter cohort study

Abstract Background Older patients account for a growing share of emergency department (ED) visits with heterogeneous risk that conventional triage does not capture. The Clinical Frailty Scale (CFS) reflects baseline reserve and the C-reactive protein-to-albumin ratio (CAR) reflects acute inflammatory–nutritional stress. Whether combining these complementary domains improves risk stratification, particularly in intermediate-acuity older patients, is unestablished. Methods Multicenter retrospective cohort of 5,928 patients aged ≥ 65 years presenting to three Korean tertiary EDs between August and October 2023. CFS was assessed at triage; CAR was calculated from C-reactive protein (CRP) and albumin obtained within 1 h of arrival. The primary outcome was in-hospital mortality; secondary outcomes were hospital and intensive care unit (ICU) admission. Multivariable logistic regression models with and without CFS and log(1 + CAR) were compared, adjusting for age, sex, study site, Korean Triage and Acuity Scale (KTAS) category, vital signs, and mental status (AVPU). Discrimination (area under the receiver operating characteristic curve [AUROC]; DeLong test), reclassification (net reclassification improvement [NRI], integrated discrimination improvement [IDI]), and calibration (Brier score) were assessed following the TRIPOD statement. Results In-hospital mortality was 4.0%; 49.2% were admitted and 12.8% directly to ICU. After full multivariable adjustment, CFS (adjusted odds ratio [aOR] 1.12; 95% CI 1.03–1.22; p = 0.007) and log(1 + CAR) (aOR 1.28; 95% CI 1.15–1.42; p < 0.001) remained independent mortality predictors. Adding CFS + CAR to the adjusted baseline improved discrimination (AUROC 0.870 → 0.882; ΔAUC + 0.013, p = 0.002), reclassification (NRI + 34.4%; 95% CI 22.3–46.5%), and calibration. The gain was largest within KTAS level 3 ( n = 4,280; AUROC 0.785 → 0.816; NRI + 51.4%). A 2 × 2 risk matrix showed an 8-fold mortality gradient across CFS × CAR strata in KTAS 3 (0.7% to 5.9%). The CFS × CAR interaction was non-significant ( p = 0.46), supporting an additive rather than synergistic relationship. Conclusions Combining a bedside frailty score with an inflammatory–nutritional biomarker from routine laboratory testing provides modest but statistically significant incremental prognostic information beyond triage, vital signs, and mental status, with greatest benefit in the intermediate-acuity subgroup. Both inputs are obtainable at no additional cost where these labs are routine for older ED patients.