Blood eosinophil count as a predictor of hospital length of stay in COPD exacerbations.

Background and Objective: Airway inflammation accompanying exacerbations varies among individuals with some having neutrophilic, while others showing eosinophilic inflammation. This study assessed the cut-off values of blood eosinophil count for identifying subjects with longer hospital length of stay (LOS) with acute exacerbations of chronic obstructive pulmonary disease (AECOPD).
Methods: Patients were recruited at presentation to the hospital with an AECOPD. Complete blood picture with differential count was taken on admission. Patients were treated with a standard course of systemic corticosteroid and antibiotic and evaluated at 8 weeks post-exacerbation for lung function measurement and 6-min walk. They were followed up in 1 year for any readmissions or mortality. Cut-off values of eosinophils for assessment of longer LOS were calculated using receiver operating characteristic (ROC) curve analysis.
Results: A total of 346 patients with admission eosinophil count were included in the analysis (333 (96.2%) were males; mean ± SD age: 74.9 ± 7.8 years; mean forced expiratory volume in 1 s (FEV ₁ ): 43.4 ± 16.3% predicted). The median (interquartile range (IQR)) of the absolute peripheral eosinophil count, percent eosinophil count and LOS were 0.11 (0.25) × 10 ⁹ /L, 1 (3) % and 5 (7) days, respectively. Using the median LOS of ≥5 days as the cut-off, ROC analysis of the cut-off value of eosinophil count associated with longer LOS was at <2% (area under the curve (AUC): 0.666, P < 0.001) while absolute eosinophil count was at <0.144 × 10 ⁹ /L (AUC: 0.645, P < 0.001). These eosinophil cut-off values could predict longer LOS independent of age, lung function and previous hospital admissions, but had no association with readmissions for AECOPD and mortality at 12 months.
Conclusion: An eosinophil value of <0.144 × 10 ⁹ /L on admission or <2% was associated with longer hospital LOS for AECOPD.
(© 2019 Asian Pacific Society of Respirology.)