Lactate to albumin ratio predicts one year mortality in ICU patients receiving invasive mechanical ventilation

The lactate-to-albumin ratio (LAR), a composite biomarker reflecting both inflammatory burden and nutritional status, has been associated with adverse outcomes in various critically ill populations. However, its prognostic value in patients receiving invasive mechanical ventilation (IMV) remains unclear. A retrospective cohort study was conducted using the MIMIC-IV database. Adult ICU patients who received IMV and had lactate and albumin measurements within 24 h of admission were included. The primary outcome was 1-year all-cause mortality. Multivariable Cox proportional hazards and modified Poisson regression models were used to assess associations between LAR and mortality. Restricted cubic spline (RCS) analysis evaluated dose-response relationships. Subgroup, interaction, and sensitivity analyses were also performed. The Boruta algorithm was applied to compare LAR's predictive importance with traditional severity scores. A total of 9,195 IMV patients were included, with a mean age of 63.8 years; 40.3% were female. Patients were stratified into high and low LAR groups based on an optimal cutoff of 1.48. Higher LAR was independently associated with increased 1-year mortality (adjusted HR: 1.31, 95% CI: 1.20-1.43) and ICU mortality (adjusted RR: 1.27, 95% CI: 1.17-1.38). RCS analysis showed a linear positive association between LAR and 1-year mortality. Subgroup analyses demonstrated stronger associations in younger patients and those with lower RDW. The Boruta feature selection algorithm confirmed LAR as an important predictor, ranking above the SOFA score. Overlap weighting analysis further validated the robustness of the findings. Elevated LAR is independently associated with increased short- and long-term mortality in ICU patients receiving IMV. While not a replacement for comprehensive scoring systems, LAR may serve as a convenient and valuable supplementary biomarker, especially in patients with indeterminate risk profiles.