Stability of Blood Eosinophil Count in Patients with COPD in the UK Clinical Practice Research Datalink

<p>Blood eosinophil counts may be predictive of corticosteroid response in chronic obstructive pulmonary disease (COPD) patients. However, little is known about measurement stability, which is important for understanding the utility of blood eosinophil counts as a potential biomarker. We evaluated the stability of blood eosinophil counts over 1 year in a population-based cohort of patients with COPD in primary care. Patients were aged ≥ 40 years with forced expiratory volume in 1 second/forced vital capacity < 0.7 and ≥ 1 blood eosinophil measurement taken during a period of stable disease within 6 months of a COPD diagnosis code recorded between January 1, 2010 and December 31, 2012. Generalized linear mixed models were fitted to log-transformed data to estimate the between-(s<sup>2</sup><sub>between</sub>) and within-patient (s<sup>2</sup><sub>within</sub>) variance in eosinophil count; an intra-class correlation coefficient <i>R<sub>i</sub></i> was calculated (s<sup>2</sup><sub>between</sub>/[s<sup>2</sup><sub>between</sub> + s<sup>2</sup><sub>within</sub>]). A sensitivity analysis was performed from which patients who were prescribed systemic corticosteroids or antibiotics at any time during follow-up were excluded. All models were adjusted for age, gender, smoking status, and asthma history. Overall, 27,557 patients were included in the full cohort (51.5% male, mean age [standard deviation] 71.1 [10.6] years) and 54% of patients had ≥ 2 eosinophil measurements (median 2 [interquartile range 1]) during follow-up. For the full cohort, <i>R<sub>i</sub></i> = 0.64, and in the sensitivity analysis subgroup, <i>R<sub>i</sub></i> = 0.70, mainly due to a decrease in s<sup>2</sup><sub>within</sub>. For patients with COPD in primary care, eosinophil measurements demonstrated reasonable repeatability over 1 year, which increased after exclusion of patients who were prescribed systemic corticosteroids or antibiotics.</p>