Comparative prognostic value of nutrition–inflammation indices for 1-year mortality in elderly intensive care unit patients: a retrospective observational study based on MIMIC-IV database

BACKGROUND: Malnutrition and systemic inflammation are frequent in critically ill elderly patients and are strongly linked to poor outcomes. However, the relative prognostic performance of 6 commonly used nutrition–inflammation indices (mNUTRIC, PNI, CONUT, GNRI, HALP, and LAR) in elderly ICU patients has lacked validation in large-scale cohorts. This study aimed to compare the predictive performance of six such indices for 1-year all-cause mortality in elderly ICU patients. METHODS: This retrospective observational study was based on the Medical Information Mart for Intensive Care IV (MIMIC-IV, version 2.2) database. Patients aged ≥ 65 years with an ICU stay ≥ 24 h were included. Six indices (mNUTRIC, PNI, CONUT, GNRI, HALP, and LAR) were calculated using data within 24 h after ICU admission. The primary outcome was 1-year all-cause mortality. Receiver operating characteristic analysis and DeLong tests were used to compare predictive performance among indices. RESULTS: A total of 18 868 patients were included, of whom 5 939 (31.5%) died within one year. All indices were significantly associated with mortality, but mNUTRIC showed the highest discrimination (AUROC = 0.713, 95% CI 0.705–0.720, P < 0.001), outperforming the others. Patients with high mNUTRIC scores (≥ 4) had markedly higher 1-year mortality than those with low scores (47.0% vs. 19.2%, P < 0.001). CONCLUSIONS: Among six nutritional–inflammatory indices, mNUTRIC provided the best prediction of 1-year mortality in elderly ICU patients. Early nutritional risk screening using mNUTRIC may help identify high-risk individuals for targeted interventions.