Metabolic-inflammatory burden predicts mortality in heart failure across population and ICU cohorts

Systemic metabolic and inflammatory disturbances play a critical role in the pathogenesis and prognosis of heart failure (HF). However, it remains uncertain whether these processes consistently predict mortality across both population-based and critical-care settings.

This study integrated data from the National Health and Nutrition Examination Survey (NHANES) and the eICU Collaborative Research Database to independently derive and examine the construct-level consistency of a Metabolic-Inflammatory Burden (MIB) index. MIB was independently derived in each cohort using principal component analysis (PCA) of available standardized metabolic and inflammatory biomarkers. Cox proportional-hazards regression was applied in the NHANES cohort (outcome: all-cause mortality), and logistic regression with XGBoost in the eICU cohort (outcome: 28-day mortality). To assess incremental prognostic value beyond ICU severity scores, nested model comparisons were performed. Cross-database construct-level integration was conducted to examine the directional consistency of MIB-mortality associations.

In the NHANES cohort (n = 29,400), individuals with HF exhibited less favorable metabolic-inflammatory profiles and higher mortality compared to those without HF (all p < 0.001). Higher MIB was independently associated with increased all-cause mortality. In the eICU cohort (n = 12,946), non-survivors showed similar metabolic-inflammatory elevations. XGBoost achieved the best performance in mortality prediction (AUC = 0.883), with SHAP analysis identifying acute severity scores and metabolic-inflammatory parameters as key contributors. Cross-database integration revealed consistent positive directional associations between MIB and mortality across both settings (NHANES: HR = 1.03, 95% CI: 1.00-1.05; eICU: OR = 1.32, 95% CI: 1.16-1.40), supporting construct-level coherence of the MIB framework.

Elevated metabolic-inflammatory burden (MIB) consistently predicts mortality in both population and critical-care HF cohorts. The consistent directional associations across heterogeneous clinical databases underscore a shared metabolic-inflammatory axis linking chronic metabolic stress with acute deterioration, supporting the prognostic relevance of MIB in diverse HF populations.