The red cell distribution width to albumin ratio as a novel predictor of 180-day mortality in lung cancer patients

The red cell distribution width-to-albumin ratio (RAR) integrates inflammation and nutritional compromise - two hallmarks of lung cancer progression. This study validates RAR as a predictor of long-term mortality in critically ill lung cancer patients, addressing gaps in conventional ICU prognostication. In this MIMIC-IV-based retrospective cohort, 973 lung cancer patients meeting ICU criteria were stratified by RAR quartiles (calculated as RDW-CV (%) / albumin (g/dL)). Primary and secondary endpoints were 180-day and 365-day all-cause mortality. Multivariable Cox regression (adjusted for demographics, severity scores, and interventions), restricted cubic splines (RCS), and ROC analysis (at the specified endpoint) evaluated RAR's predictive utility. Subgroup analysis tested interactions with disease severity. Among 973 critically ill lung cancer patients, those in the highest RAR quartile (6 ≤ RAR < 14.29) exhibited substantially elevated mortality compared to the lowest quartile: 66.4% versus 31.1% at 180 days (P < 0.001) and 73.4% versus 43.2% at 365 days (P < 0.001). Multivariable Cox regression confirmed RAR as an independent mortality predictor, with each unit increase in continuous RAR associated with 20% higher 180-day risk (HR = 1.20, 95%CI 1.14-1.27) and 19% higher 365-day risk (HR = 1.19, 95%CI 1.13-1.25), while patients in the highest quartile faced more than doubled mortality hazard versus the lowest quartile (180-day HR = 2.32, 95%CI 1.74-3.09; 365-day HR = 2.07, 95%CI 1.60-2.67). Restricted cubic spline analysis demonstrated a linear dose-response relationship between rising RAR and mortality risk (P < 0.001). RAR significantly outperformed SOFA in discriminative accuracy for both endpoints (180-day AUC 0.65 (95%CI 0.62-0.69) vs. 0.56 (95%CI 0.53-0.60); 365-day AUC 0.64 (95%CI 0.60-0.67) vs. 0.55 (95%CI 0.51-0.59)). Subgroup analyses revealed diminished predictive utility in high-severity patients (SOFA ≥ 5; P = 0.006), though robustness persisted across other clinical strata. RAR is a robust, accessible predictor of long-term mortality in lung cancer ICU patients. Its strong discriminative ability (> 65% mortality at 6 ≤ RAR < 14.29) supports clinical utility for early risk stratification, particularly before multi-organ failure develops.