Results
Patients with higher CONUT scores or lower GNRI scores (indicating worse nutritional status) had significantly longer hospital stays, lower levels of nutrition-related indices (albumin, total cholesterol, low-density lipoprotein cholesterol (LDL-C), serum ferritin, transferrin saturation), lower lymphocyte counts, and higher NT-pro-B-type natriuretic peptide (NT-proBNP) and inflammatory markers (C-reactive protein (CRP), procalcitonin (PCT), interleukin-6 (IL-6)) (all P < 0.05). Over a median follow-up of 11.5 months, 83 patients (42.3%) died. In unadjusted analyses, patients with worse nutritional status had significantly higher rates of all-cause death, all-cause readmission, and heart failure-related readmission (all P < 0.05). Multivariable Cox regression revealed that CONUT (hazard ratio (HR) = 1.366, 95% confidence interval (CI): 1.181–1.579, P < 0.001) was an independent risk factor for all-cause mortality, while GNRI (HR = 0.962, 95% CI: 0.927–0.999, P = 0.043) was a protective factor. The CONUT and GNRI showed poor consistency (κ = 0.235, P < 0.01). Using receiver operating characteristic (ROC) analysis based on vital status at the end of follow-up, the area under the curve (AUC) for predicting all-cause mortality was 0.86 (95% CI: 0.80–0.91) for CONUT, 0.79 (95% CI: 0.73–0.86) for GNRI, and 0.87 (95% CI: 0.82–0.92) for a combined model incorporating both scores via logistic regression. Net reclassification improvement (NRI) analysis showed that compared with the GNRI score, the NRI of the CONUT score was 0.580 (95% CI: 0.309–0.852), and the NRI of the combined score was 0.760 (95% CI: 0.499–1.022), both reaching statistical significance (P < 0.01). By contrast, the NRI was 0.083 (95% CI: −0.197 to 0.363, P = 0.560) for the combined score versus the CONUT score.
