CBC-Derived Inflammatory Indices and ICU Outcomes in Mechanically Ventilated Patients: A Two-Database Multicenter Validation Study

Abstract Background Complete blood count (CBC)-derived inflammatory indices represent accessible, cost-effective prognostic biomarkers in critical care. However, their predictive value for ventilator-associated pneumonia (VAP) and mortality in mechanically ventilated patients remains inadequately characterized. This study assessed correlations between CBC-derived inflammatory indices and ICU outcomes through multicenter validation. Methods This retrospective study analyzed two independent cohorts: PhysioNet Colombia (n = 141) and CARPEDIEM Northwestern United States (n = 686). Six indices—neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), systemic immune-inflammation index (SII), aggregate index of systemic inflammation (AISI), prognostic nutritional index (PNI), and hemoglobin-albumin-lymphocyte-platelet (HALP) score—were calculated from baseline values. Primary outcomes included ICU length of stay and VAP. Analyses comprised Spearman correlations, Mann-Whitney U tests, ROC analysis, and logistic regression. Results Mean age was 50.2 ± 20.5 and 60.3 ± 15.3 years in primary and validation cohorts, respectively, with VAP incidence of 33.3% and 41.2%. In the validation cohort, NLR (ρ = 0.101), PLR (ρ = 0.093), SII (ρ = 0.092), and HALP (ρ=-0.103) correlated significantly with ICU length of stay (all p < 0.05). VAP patients demonstrated higher PLR (250.0 vs. 195.0, p = 0.004), higher SII (2.27 vs. 1.62×10³, p = 0.001), and lower HALP (13.21 vs. 16.70, p = 0.006). ROC analysis yielded AUC values of 0.585 for SII (cut-off ≥ 1.48×10³), 0.575 for HALP (cut-off ≤ 25.4), and 0.574 for PLR (cut-off ≥ 158.6). Quartile analysis demonstrated dose-response relationships for VAP across SII, PLR, and HALP quartiles. Significant associations were observed exclusively in high-severity patients (SOFA ≥ 8), indicating effect modification by illness severity. Non-survivors exhibited higher NLR (12.86 vs. 9.57, p < 0.001) and lower PNI (34.5 vs. 37.5, p < 0.001). Multivariate analysis confirmed PNI as an independent mortality predictor (OR = 0.617, 95% CI: 0.474–0.778, p < 0.001). Conclusion CBC-derived inflammatory indices demonstrate significant associations with ICU outcomes in mechanically ventilated patients. SII, PLR, and HALP show promise for VAP risk stratification, while PNI emerges as an independent mortality predictor. Effect modification by severity underscores immunological dysregulation in critical illness.