Feasibility of multidomain cardiovascular–kidney–metabolic biomarker assessment for in-hospital mortality in older heart failure inpatients: an exploratory retrospective cohort study

Cardiovascular-kidney-metabolic (CKM) dysfunction integrates cardiac, renal, metabolic, and inflammatory risk and is particularly relevant for older patients hospitalized with heart failure. Whether standardized multidomain CKM-related biomarker assessment is feasible using real-world hospital data in older heart-failure inpatients and what limitations arise from such data remains poorly characterized. Retrospective cohort study using hospital records from Zhongnan Hospital (Jan 2022–Dec 2024). Patients with missing in-hospital outcome status were excluded. Primary outcome was in-hospital mortality. Associations between age, sex, ECG major abnormality, BNP, BUN, IL-6, and mortality were evaluated using Firth penalized logistic regression in complete-case patients. Nested ROC models described apparent discrimination (DeLong tests), bootstrap resampling estimated optimism-corrected discrimination, and an exploratory CKM biomarker burden analysis was performed. Among 1,645 records, 1,131 patients had interpretable in-hospital outcomes (64 deaths, 1,067 survivors). A complete CKM biomarker panel (BNP, BUN, and IL-6) was available in only 186 patients (16.4%; 17 deaths, 169 survivors; median non-survivor age 82 years), who were older and had higher observed mortality than patients excluded for incomplete data. In this selected complete-case cohort, only older age (adjusted OR per year, 1.10; 95% CI, 1.04–1.16; P < 0.001) and BNP per doubling (OR, 1.40; 95% CI, 1.05–1.86; P = 0.021) were independently associated with mortality; sex, ECG major abnormality, BUN, and IL-6 were not. Apparent discrimination was numerically higher for the biomarker-enriched than the clinical-ECG model (AUC 0.841 vs. 0.762), but nested DeLong comparisons were not significant ( P ≥ 0.060) and the optimism-corrected C-index was 0.79. Observed mortality rose across exploratory CKM-burden groups (8.3%, 12.0%, 20.0%), but these subgroups were very small. The incremental biomarker, ECG, discrimination, and CKM-burden findings remained exploratory and hypothesis-generating. In this selected complete-case cohort of older heart-failure inpatients, age and BNP were the most consistent markers of in-hospital mortality. A complete CKM biomarker panel was available in only a minority of patients, indicating that standardized multidomain CKM-related assessment was not systematically implemented in routine care. The feasibility, incremental prognostic value, and clinical utility of broader CKM-related multidomain biomarker assessment remain unproven and require standardized prospective validation; these findings should not be interpreted as a prognostic model ready for clinical use.