Backgroundand Objectives: Reliable prognostic biomarkers in sepsis that are inexpensive and universally available remain limited. Eosinophil counts decrease in systemic illness, but the prognostic value of static versus dynamic eosinophil measures in sepsis is uncertain. We investigated the association between eosinophil measurements, including early trajectories, and in-hospital mortality in a large sepsis cohort. Materials and Methods: We conducted a retrospective observational study of adult patients hospitalized with sepsis between 2021 and 2024 at a tertiary referral center. Absolute eosinophil counts were assessed at admission, at eosinophil nadir, and dynamically at approximately 48 and 72 h. Eosinophil trajectories were categorized as persistently undetectable, decrease, non-decrease, or rise from undetectable. Multivariable logistic regression models adjusted for demographic factors, comorbidities, and markers of disease severity were used to evaluate associations with in-hospital mortality. Results: Among 3932 patients, in-hospital mortality was 48.5%. Static eosinophil measures at admission and at eosinophil nadir were lower in non-survivors but were not independently associated with mortality after adjustment. Pancytopenia at eosinophil nadir was independently associated with increased mortality (OR 1.65, 95% CI 1.37–1.99). In contrast, eosinophil dynamics provided additional prognostic information. A decrease in eosinophil counts at 72 h was independently associated with higher mortality (OR 1.43, 95% CI 1.08–1.91), while 48 h changes were not. Persistently undetectable eosinophil trajectories were associated with the highest mortality, whereas recovery from undetectable levels was associated with lower risk. Conclusions: In sepsis, static eosinophil counts lack independent prognostic value, whereas delayed eosinophil decline and persistently suppressed trajectories are associated with increased in-hospital mortality. Eosinophil dynamics were associated with in-hospital mortality and may provide additional information on the host response during sepsis.
Mușat et al. (Fri,) studied this question.