Associated diagnostic spectrum and stratified characteristics of patients with eosinophilia: a retrospective study

Peripheral blood eosinophilia occurs in diverse clinical conditions, including allergic diseases, infections, autoimmune disorders, hematologic malignancies, and solid malignancies. Although parasitic infection and allergic disease have traditionally been regarded as common settings, the distribution of associated diagnoses may differ in real-world tertiary-care practice. Describing the associated diagnostic spectrum of persistent eosinophilia and its relationship with eosinophil burden and demographics may improve clinical recognition and stratified evaluation.

To characterize the associated diagnostic spectrum and stratified features of persistent eosinophilia, compare persistent with transient eosinophilia, and provide evidence for standardized clinical evaluation of elevated eosinophil counts.

This single-center retrospective study used the clinical database of West China Hospital, Sichuan University, including outpatients and inpatients with eosinophilia between 2024 and 2025. Patients with a peak absolute eosinophil count (AEC) > 0.52 × 10^9/L were eligible. Patients with only one AEC measurement or <30 days between their earliest and latest available measurements were excluded. Persistent eosinophilia required at least two AEC values ≥0.50 × 10^9/L separated by ≥30 days; the remaining eligible patients were classified as transient. Baseline demographics and peak AEC were compared between groups. Persistent eosinophilia without a diagnosis generally considered an established explanation for eosinophilia was classified as no established eosinophilia-related diagnosis, and associated diagnostic distributions were assessed across study-specific peak AEC bands, age, sex, and ethnicity. Multivariable analyses were performed.

Of 23,069 patients included, 19,745 had transient and 3,324 had persistent eosinophilia. Patients with persistent eosinophilia were older, more often male, and had higher peak eosinophil counts with greater proportions in the two higher study-specific AEC bands. The mean age difference was 4.03 years (95% CI 3.35-4.71; Hedges g = 0.19). Solid malignancy was the most frequent associated diagnostic category (34.2%), followed by no established eosinophilia-related diagnosis (32.1%). Across the study-specific AEC bands, the proportion without an established related diagnosis decreased, whereas allergic disease increased. Associated diagnostic distribution varied by age and sex, and peak AEC did not differ significantly across solid malignancy subtypes. Among patients without an established related diagnosis, CKD was recorded in 565/1,067 (53.0%); adjusted CKD associations were weak and exploratory.

In this tertiary-care cohort, persistent eosinophilia occurred across a broad associated diagnostic spectrum, with solid malignancy and no established eosinophilia-related diagnosis as the two most frequent categories. These descriptive findings do not establish causation. Differential repeat testing, referral selection, unmeasured medication exposure, and residual confounding limit clinical interpretation.