Nonlinear association between blood eosinophil counts and clinical outcomes in ICU-Admitted patients with acute exacerbation of COPD: a retrospective observational study.

Background: This retrospective cohort study aimed to examine the association between blood eosinophil counts and in-hospital mortality among patients with acute exacerbation of chronic obstructive pulmonary disease (AECOPD) admitted to the intensive care unit (ICU).

Methods: Data were retrieved from the MIMIC-IV 2.2 database. AECOPD patients with blood eosinophil counts measured within the first 24 h following ICU admission were included. Kaplan-Meier (KM) survival analysis compared in-hospital mortality across eosinophil thresholds (0.10, 0.15, 0.20, and 0.30 × 10⁹/L). Four Cox regression models adjusted for confounders, with patients categorized into three eosinophil groups: eosinophil counts < 0.1 × 10⁹/L (Group I), 0.1 × 10⁹/L ≤ eosinophil counts < 0.3 × 10⁹/L (Group II), and eosinophil counts ≥ 0.3 × 10⁹/L (Group III), with the Group I serving as the reference group. Restricted cubic spline (RCS) analysis based on a Cox proportional hazards model was used to assess the non-linear relationship between eosinophil counts and in-hospital mortality.

Results: A total of 1,855 patients were analyzed. KM survival curves showed significant mortality differences at eosinophil thresholds of 0.10, 0.15, and 0.20 × 10⁹/L. Compared with Group I, Group II consistently exhibited a significantly lower risk of in-hospital mortality across all Cox models. Group III showed a mortality reduction only in the unadjusted model, which lost statistical significance after adjustment. RCS analysis revealed an "L"-shaped association, with an inflection point around 0.042 × 10⁹/L.

Conclusion: Lower eosinophil counts were associated with increased in-hospital mortality among critically ill AECOPD patients. Absolute eosinophil counts may serve as a prognostic biomarker to support personalized ICU management. Prospective studies are needed to validate these findings and inform clinical decision-making.