Association between immuno-nutritional biomarkers and mortality in hospitalized geriatric population

This study aimed to identify the most sensitive immuno-nutritional and systemic inflammation biomarkers for predicting in-hospital all-cause mortality in older adults.

A retrospective observational study was conducted in 2,067 hospitalized geriatric patients aged ≥60 years in the Department of Geriatrics, Lodz, Poland, from 2017 to 2024. Blood-based immuno-nutritional indices were calculated from routine laboratory tests at admission, including NLR (Neutrophil-to-Lymphocyte Ratio), LMR (Lymphocyte-to-Monocyte Ratio), PNI (Prognostic Nutritional Index), PLR (Platelet-to-Lymphocyte Ratio), LCR (Lymphocyte-to-C-Reactive Protein Ratio), DLR (D-dimer-to-Lymphocyte Ratio), MWR (Monocyte-to-White Blood Cell Ratio), SII (Systemic Immune-Inflammation Index), SIRI (Systemic Inflammation Response Index), CAR (C-Reactive Protein-to-Albumin Ratio), DAR (D-dimer to Albumin Ratio), PAR (Platelet-to-Albumin Ratio), NAR (Neutrophil-to-Albumin Ratio), PIV (Pan-Immune-Inflammation Value), C-Reactive Protein (CRP) and White Blood Cell (WBC) count. Differences between survivors and non-survivors were analyzed using Mann-Whitney U and Chi-square tests. Prognostic accuracy was assessed via Receiver Operating Characteristic (ROC) curve analysis, with statistical significance set at p ≤ 0.05. Additionally, multivariable logistic regression, calibration assessment, and 10-fold cross-validation were used to confirm the robustness and internal validity of prognostic models.

The mean age was 80.88 ± 8.33 years for men and 82.92 ± 7.72 years for women. Men had higher levels of inflammatory biomarkers (NLR, SIRI, CAR), while women exhibited better nutritional and immune profiles (higher PNI, LMR). Non-survivors of both sexes showed significantly higher NLR, PLR, DLR, SII, SIRI, CAR, DAR, PAR, NAR and PIV, and significantly lower levels of LMR, PNI, LCR and MWR compared to survivors (p < 0.001). The Prognostic Nutritional Index (PNI) demonstrated the highest predictive value for in-hospital mortality (AUC = 0.837; sensitivity = 0.88, specificity = 0.64), followed by CAR and LCR. Other indices, including DLR, DAR, and NAR, also showed significant but comparatively lower predictive accuracy. In multivariable analysis, age, PNI, LCR, and NAR remained independent predictors of mortality (AUC for final model = 0.852).

This study highlights PNI as the most sensitive and reliable biomarker for predicting in-hospital mortality among older adults. These results support using PNI and inflammatory markers in clinical assessments to better identify high-risk geriatric patients and reduce mortality.