Association of eosinophil levels with 30-day overall mortality in hospitalized patients with aspiration pneumonia Authors: Xia Jing, Li Li, Junyan Zhang, Haihong Zhao, Jing Li, Jixiu Fan, Jin Huo, Yueqin Li, Junkang Zhao Published: 2026 ABSTRACT AbstractAspiration pneumonia (AP) is expected to have an increasing contribution to morbidity and mortality, particularly in the elderly population over the next coming decades. However, there are no reliable disease evaluation methods and prognostic indicators for patients with AP. A total of 841 patients with AP were included in this single-center prospective cohort study, among which 406 patients were in the eosinopenia group (an absolute eosinophil count < 0.05 × 109/L), and 435 patients were in the non-eosinopenia group (an absolute eosinophil count ≥ 0.05 × 109/L). The primary outcome measure was 30-day overall mortality. The Cox proportional-hazards regression models with overlap weighting according to the propensity score were employed to estimate the associations between eosinopenia level and 30-day overall mortality. 841 AP patients (406 in eosinopenia, 435 in non-eosinopenia), mean ages (75.15 ± 16.76 years in eosinopenia, 78.23 ± 15.28 years in non-eosinopenia), 71 deaths within 30 days (62 in eosinopenia, 9 in non-eosinopenia), 30-day death rates were 15.27% in eosinopenia and 2.07% in non-eosinopenia), 30-day overall mortality hazard ratio by multivariate Cox regression (HR 3.884. for eosinopenia vs. non-eosinopenia, 95% CI 1.772–8.514), after propensity score weighting-overlap weighting (HR 2.337, 95% CI 1.052–5.194), sensitivity analyses ratio (HR 2.951, 95% CI 1.163–7.484) by propensity score matching and (HR 2.574, 95% CI 1.143–5.796) by propensity score regression. The Aspiration Pneumonia Patients with Low Eosinophil counts (APPLE) prospective cohort study suggested that eosinopenia in patients with AP was associated with an increased risk of 30-day overall mortality.Trial registration: Chinese Clinical Trial Registry. URL https://www.chictr.org.cn/. Registration number ChiCTR2300076254. Registration date September 28, 2023.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1038/s41598-026-57569-0.Keywords: Aspiration pneumonia, 30-day overall mortality, Eosinopenia, Overlap weighting, Propensity-score methodsSubject terms: Diseases, Medical research, Risk factors Introduction Aspiration pneumonia (AP) is best regarded not as a distinct entity but as part of a continuum that also encompasses community- and hospital- acquired pneumonias1. It is estimated that AP accounts for 5 to 15% of cases of community-acquired pneumonia, but figures for hospital-acquired pneumonia are unavailable2. It is expected to contribute increasingly to mortality and morbidity, particularly in the elderly population over the next coming decades3. Among patients hospitalized for community-acquired pneumonia (CAP), an aspiration event causes pneumonia in 5% of patients under 80 years of age and 10% of patients aged 80 years or older4. Up to 10% of patients who normally live at home and are admitted to a general hospital with pneumonia are diagnosed with AP, and this prevalence increases by over 30% for patients living in nursing homes5. The mortality of nursing home-acquired pneumonia ranges from 13 to 41%6. AP has unique clinical features, including older-age prevalence than other CAP. Assessing the severity of the disease is crucial for planning and stratifying the care of patients with AP, as identifying patients who are likely to have significant adverse outcomes is the first step in reducing mortality. Assessing the severity of AP can also guide treatment choices, such as the need for hospitalization or intensive care, suitability for discharge, choice of antibacterial agents, course of treatment, and route of administration. Therefore, there is an urgent need to find reliable disease evaluation methods and good prognostic indicators for patients with AP. Eosinophils have been evolutionarily conserved across organisms and are present in the blood circulation of all vertebrates, including reptiles, amphibians, mammals, and fish7,8. Even invertebrates such as annelids, insects, and crabs contain eosinophil-like cells8. Eosinophils are an integral component of the innate immune system. They are phagocytes that have similar chemotaxis, phagocytosis, and antibacterial and antifungal killing rates as monocytes and polymorphonuclear leukocytes (PMNLs)7. Eosinophils even possess antiviral abilities9. In addition, in clinical practice, we have observed a significant decrease in peripheral blood eosinophil counts in some patients with AP. However, few prospective cohort studies have been conducted to evaluate the effect of peripheral blood eosinophil count on mortality in patients with AP. We conducted the Aspiration Pneumonia Patients with Low Eosinophil counts (APPLE) study to observe the 30-day overall mortality in patients with AP and low eosinophil counts compared to the non-eosinopenia group. Methods Study design The APPLE study was a prospective cohort study conducted from January 2016 to December 2023 in Shanxi Bethune Hospital in Shanxi Province, China. The patients were divided into two groups: the group with AP and low absolute eosinophil count (eosinopenia group, with an absolute eosinophil count < 0.05 × 109/L) and the group with AP without low absolute eosinophil count (non-eosinopenia group, with an absolute eosinophil count ≥ 0.05 × 109/L)10,11. The study design of the APPLE study is shown in Fig. 1. This clinical study is registered at Chinese Clinical Trial Registry (Registration number: ChiCTR2300076254) (e-Appendix 1). This study was conducted in accordance with the Declaration of Helsinki and approved by the ethics committee of Shanxi Bethune Hospital (approval YXLL-2015-148). Written informed consent was obtained from each subject. The full trial protocol is available in the Online Resource. Flowchart of patients with aspiration pneumonia enrolled in this study. Inclusion criteria The inclusion criteria for the study were as follows: (1) an aspiration event witnessed by a person nearby, such as family members, caregivers, and medical staff, or factors indicating an aspiration event; (2) acute respiratory symptoms and the presence of respiratory system findings suggestive of pneumonia, such as cough, sputum, fever, tachypnea, and pathological respiratory sounds; (3) compatible radiologic findings of aspiration pneumonia on chest Computed Tomography (CT). The appropriate chest CT imaging findings for the diagnosis of AP were determined by two different experienced chest radiologists, with clinical information on AP as an infiltrative lesion in the dependent position (the posterior segment of the upper lobes and the superior segment of the lower lobes in a supine position and the bilateral basal segments, middle lobes, and lingular region in an erect position)12; (4) over the age of 18 years old; and (5) signed informed consent by patients or their immediate family members. In case of no diagnostic agreement between the reviewers, the patient was excluded from the study. The exclusion criteria were as follows: (1) infiltrative lung lesion in the dependent position without signs of bacterial infection; (2) confirmation of a different diagnosis (e.g., lung cancer, tuberculosis, asthma, Chronic Obstructive Pulmonary Disease (COPD), Corona Virus Disease-2019 (COVID-19), bronchiectasis); (3) patients with a history of solid tumors or hematological malignancy, autoimmune and endocrine disorders, parasitic and fungal diseases, allergic disorders, patients who received steroids within the previous six months, and those suffering from a disease or receiving therapy that suppresses bone marrow activity. All these clinical conditions can affect peripheral eosinophil counts; (4) pregnant or lactating women; (5) participants who are currently enrolled in other clinical studies; and (6) patients with severe mental illness. We obtained demographic information, laboratory findings, comorbidities, and hospital course using the electronic medical record system (EMRS). Arterial blood gas analysis was performed when the patient was not on oxygen therapy upon admission. Microbiological samples were collected from the respiratory tract (sputum, pleural fluid, endotracheal aspirate, or bronchoalveolar lavage) and blood within 24 h of hospitalization. Diagnostic testing was determined by the attending physician caring for the patient, and the culture results determined whether there was bacterial infection. All baseline demographic data, clinical comorbidities, and laboratory indicators were extracted directly and completely from the hospital’s unified EMRS within 24 h of each patient’s admission. The EMRS of our institution has a mandatory entry system for key clinical and laboratory data of inpatients, which eliminates the possibility of missing recording of core variables for patients with aspiration pneumonia. For the primary outcome indicator (30-day overall mortality), we adopted a 100% follow-up strategy combined with EMRS review and telephone follow-up for dual verification. A dedicated follow-up team was assigned to conduct standardized telephone follow-up for all included patients at the 30th day after admission, and the survival status was cross-checked with the hospital’s death registration system and the local household registration department. Death events were strictly confirmed by official death certificates, ensuring no missing or inaccurate outcome data. The outcome assessment work for this study was undertaken by an independent research team that was completely separated from the patient recruitment and laboratory testing teams. This team was blinded to all eosinophil-related data throughout the study, including the absolute eosinophil count results of all patients and the grouping status (eosinopenia group vs. non-eosinopenia group). The independent outcome assessors only collected and adjudicated the 30-day survival status of patients, with no access to any of the patients’ peripheral blood test results during the entire process of data collection and outcome verification. Study outcomes The primary outcome measure was 30-day overall mortality, which included AP deaths from admission to the hospital up to 30 days after admission. Confirmation of death was based on the death certificate. Sample size estimation As an exploratory study, the sample size of this study was based on patients who met the inclusion criteria in the preliminary trial (100 patients, 50 patients in the eosinopenia group and 50 patients in the non-eosinopenia group), and the primary clinical endpoint 30-day overall mortality was 16% and 8%, respectively. The estimates were made using PASS software, requiring 345 patients per group (α = 0.05, power = 0.8, 2-tailed test). Considering there would be approximately 20% missing data in follow-ups, we set the preliminary estimated sample size at 414 cases per group. Statistical analysis Continuous variables were summarized using the median (interquartile range) for all samples. Categorical variables were reported as frequencies and percentages for the crude and matched samples and as percentages only for the weighted samples. Cox proportional-hazards regression models were employed to estimate the association between eosinopenia and the 30-day overall mortality. An initial multivariable Cox regression model was constructed, which included demographic factors, clinical factors, laboratory tests, and medications. Additionally, propensity-score methods were utilized to minimize the effects of confounding factors. The propensity score for the predicted probability of eosinopenia was estimated using a multivariable logistic-regression model, which incorporated the same covariates as the Cox regression model. The primary analysis utilized propensity-score methods with overlap weighting. Overlap weighting is a propensity-score technique that aims to emulate key features of randomized clinical trials. Weights were assigned to each patient, proportional to the probability of that patient being in the opposite group, ensuring the inclusion of all available patients and exact balance for the mean of all covariates in the model. This method generates a weighted pseudo-sample of patients in the reference group with a distribution of measured covariates that mirrors that of the exposed group. Baseline characteristics were compared between groups using standardized differences in both the unweighted and weighted samples (differences > 10% were considered meaningful). In secondary analysis, we used propensity-score matching13 and another analysis included the propensity score as an additional covariate. For the propensity-score matching, the nearest-neighbor method was adopted to create matched samples based on the logit of the propensity score using a caliper width equal to 0.2 of the standard deviation of the propensity score logit. A significance level of 0.05 for two-sided comparisons was considered statistically significant. Hazard ratios (HRs) with 95% confidence intervals are reported. Statistical analyses were conducted using R software, version 4.2.3 (R Project for Statistical Computing). Results Study patients and follow-up From January 2016 to December 2023, a total of 1041 patients with AP who signed informed consent were initially enrolled. Among these patients, 200 patients were excluded due to recurrent AP during the follow-up period. Ultimately, a total of 841 patients were included in this study, among which 406 patients were in the eosinopenia group and 435 patients were in the non-eosinopenia group. There were no lost patients in the study. The patient flow of the study is shown in Fig. 1. Patients’ characteristics Table 1 presents a comparison of baseline characteristics before and after propensity score weighting. The eosinopenia group had a younger age (mean 75.15 ± 16.76) but a higher BMI (mean 20.19 ± 2.96) compared to the non-eosinopenia group (mean 78.23 ± 15.28 and mean 19.79 ± 2.38, respectively). The proportion of patients with a history of trauma was higher (14.3%), while the proportions with a history of hypertension, stroke, and congestive heart failure were lower (62.8%, 72.9%, and 21.9%, respectively) in the eosinopenia group. Additionally, the proportion of patients without bacterial infection was lower (39.7%) in the eosinopenia group. Moreover, the levels of lymphocytes (LYM), platelets (PLT), albumin (ALB), and arterial partial pressure of oxygen (PaO2) were lower in the eosinopenia group, whereas the levels of neutrophils (NEUT), CRP, procalcitonin (PCT), blood urea nitrogen (BUN), D-dimer, and lactate (Lac) were higher. After propensity score weighting, all covariates between the two groups were balanced and comparable (SMD < 0.001). Table 1Baseline characteristics of eosinopenia and non-eosinopenia groups before and after propensity-score overlap weighting.UnweightedAfter overlap weightingNon-eosinopenia groupN = 435Eosinopenia groupN = 406Standardized mean differenceNon-eosinopenia groupN = 435Eosinopenia groupN = 406Standardized mean differenceDemographicsFemale (%)29.028.30.0129.229.2< 0.001Age, mean (SD), y78.23 (15.28)75.15 (16.76)0.1976.93 (16.89)76.93 (15.39)< 0.001BMI, mean (SD), kg/m219.79 (2.38)20.19 (2.96)0.1519.98 (2.50)19.98 (3.13)< 0.001Chronic medical comorbiditiesTrauma (%)8.314.30.1911.611.6< 0.001Stroke (%)82.372.90.2376.276.2< 0.001Hypertension (%)68.062.80.1167.167.1< 0.001Diabetes (%)30.627.80.0630.530.5< 0.001Congestive heart failure (%)26.721.90.1125.025.0< 0.001Chronic medicationsβRB (%)21.623.90.0523.123.1< 0.001BDZ (%)7.68.10.028.18.1< 0.001Laboratory testsNEUT, median (IQR), ×109/L5.48 (3.80, 7.93)8.44 (5.85, 11.70)0.696.79 (4.37, 9.55)6.89 (4.56, 9.67)< 0.001LYM, median (IQR), ×109/L1.20 (0.86, 1.69)0.78 (0.52, 1.10)0.671.04 (0.72, 1.42)0.99 (0.62, 1.40)< 0.001HB, median (IQR), g/L111.00 (95.00, 125.00)111.00 (91.25, 128.00)0.02109.00 (94.00, 125.00)111.00 (91.00, 128.00)< 0.001PLT, median (IQR), ×1012/L212.00 (156.50, 262.00)179.00 (127.00, 253.25)0.13213.10 (150.00, 256.00)183.00 (139.00, 246.00)< 0.001CRP, median (IQR), mg/L32.62 (10.04, 71.96)52.63 (17.32, 98.04)0.3740.47 (11.80, 87.10)44.24 (12.54, 84.06)< 0.001PCT, median (IQR), ng/mL0.10 (0.03, 0.52)0.24 (0.09, 1.11)0.290.14 (0.04, 0.66)0.17 (0.04, 0.81)< 0.001ALB, median (IQR), g/L33.30 (30.10, 36.60)31.10 (27.30, 34.35)0.4932.40 (28.90, 35.60)32.20 (28.89, 35.50)< 0.001BUN, median (IQR), mmol/L6.10 (4.45, 9.45)7.20 (4.80, 12.38)0.346.40 (4.70, 10.60)6.50 (4.40, 10.43)< 0.001D-dimer, median (IQR), ng/mL422.00 (252.50, 773.00)759.50 (385.75, 1807.50)0.42534.59 (297.66, 956.03)539.69 (302.44, 1088.90)< 0.001PaCO2, median (IQR), mmHg41.20 (36.90, 45.60)40.35 (35.10, 47.80)0.1141.50 (35.65, 46.53)40.60 (35.53, 46.96)< 0.001PaO2, median (IQR), mmHg77.00 (66.75, 88.50)64.00 (57.80, 71.07)0.8870.60 (61.36, 79.50)68.70 (61.30, 76.16)< 0.001Lactate, median (IQR), mmol/L1.80 (1.40, 2.30)2.00 (1.40, 2.90)0.352.00 (1.50, 2.50)1.70 (1.30, 2.50)< 0.001Without bacterial infection (%)57.239.70.3649.649.6< 0.001Other non-medical conditionsTracheotomy (%)18.420.40.0519.019.0< 0.001MedicationsWithout glucocorticoids (%)94.090.90.1295.995.9< 0.001Drinking_history (%)4.611.60.266.66.6< 0.001Bedridden (%)74.793.10.5286.486.4< 0.001Low systolic_BP (%)0.93.90.201.41.4< 0.001Respiration rate > 30 times/min (%)0.710.80.452.42.4< 0.001βRB, β receptor blocker; ACEI, angiotensin converting enzyme inhibitors; BDZ, benzodiazepines; NEUT, Neutrophil; LYM, Lymphocyte; HB, Hemoglobin; PLT, Platelet; CRP, C-reactive protein; PCT, Procalcitonin; ALB, Albumin; BUN, Blood Urea Nitrogen; PaCO2, Arterial partial pressure of carbon dioxide; PaO2, Arterial partial pressure of oxygen; BP, Blood pressure. Baseline characteristics of eosinopenia and non-eosinopenia groups before and after propensity-score overlap weighting. βRB, β receptor blocker; ACEI, angiotensin converting enzyme inhibitors; BDZ, benzodiazepines; NEUT, Neutrophil; LYM, Lymphocyte; HB, Hemoglobin; PLT, Platelet; CRP, C-reactive protein; PCT, Procalcitonin; ALB, Albumin; BUN, Blood Urea Nitrogen; PaCO2, Arterial partial pressure of carbon dioxide; PaO2, Arterial partial pressure of oxygen; BP, Blood pressure. Primary clinical endpoints Among the 841 patients included in the analysis, a total of 71 deaths occurred. There were 62 cases (15.27%) in the eosinopenia group and 9 cases (2.07%) in the non-eosinopenia group. In multivariate Cox regression analysis, patients in the eosinopenia group were more likely to experience the primary endpoint event compared to those in the non-eosinopenia group (hazard ratio, 3.884; 95% confidence interval [CI] 1.771–8.514). After propensity score weighting-overlap weighting, patients in the eosinopenia group remained more likely to experience death (hazard ratio, 2.337; 95% CI 1.052–5.194). In sensitivity analyses using propensity score matching and propensity score regression, the same conclusion was reached (hazard ratio, 2.951; 95% CI 1.163–7.484; hazard ratio, 2.574; 95% CI 1.143–5.796) (Table 2; Fig. 2) Table 2Associations between eosinopenia and 30-day overall mortality in AP patients in the multivariable analysis and propensity-score analyses.AnalysisHazard ratio (95% CI)Multivariable analysis3.884 (1.772–8.514)Propensity-score analysesWith overlap weighting2.337 (1.052–5.194)Adjusted for propensity score2.574 (1.143–5.796)With matching2.951 (1.163–7.484) Associations between eosinopenia and 30-day overall mortality in AP patients in the multivariable analysis and propensity-score analyses. Kaplan–Meier plot of 30-day overall mortality in patients with AP. Incremental prognostic value of eosinopenia To evaluate whether eosinopenia provides additional prognostic information beyond established risk factors, we assessed the incremental predictive value using the C-index, Integrated Discrimination Improvement (IDI), and Net Reclassification Improvement (NRI). The inclusion of eosinopenia resulted in a significant improvement in the model’s discriminative ability, with the C-index increasing from 0.867 to 0.894 (ΔC = 0.027; 95% CI 0.005–0.050; P = 0.018). Reclassification analysis further supported the additive value: the IDI was 0.024 (95% CI 0.003–0.061; P = 0.022). Notably, the NRI for events (mortality) was 0.338 (95% CI 0.125–0.420; P = 0.022), indicating that the model with eosinopenia correctly reclassified a substantial proportion of deceased patients into higher risk categories. These findings suggest that eosinopenia contributes unique prognostic information independent of other clinical and laboratory parameters. Discussion The APPLE single-center prospective cohort study prospectively evaluated peripheral blood eosinophil levels in patients with AP. Through Cox regression analysis, statistical adjustments after propensity score with overlap weighting, sensitivity analyses by using propensity score matching and propensity score regression, it was suggested that eosinopenia is associated with a higher risk of 30-day overall mortality in patients with AP. These results support that eosinopenia may be used as a prognostic indicator to help identify high-risk patients and guide clinical decision-making. Our results show that the 30-day overall mortality of patients with AP and eosinopenia is 15.27%, which is consistent with other findings14,15. Reported mortality rates for AP range from 12.9% to 59.1% (with an average of 17.8%)14. Within the context of this single‑center cohort, eosinopenia appears to be a potentially useful prognostic indicator for mortality in patients with AP, though it should not be regarded as a definitive or standalone predictive marker. Traditionally, eosinophils were viewed as key effector cells in parasitic infections, allergic reactions16 and tissue damage17, but emerging evidence—including the present study—suggests that eosinopenia is associated with adverse clinical outcomes, supporting a protective role of eosinophils in infection18–22 and inflammation7,23. Other studies have shown that eosinopenia is a strong predictor of 18-month mortality in hospitalized patients with acute exacerbation of COPD combined with CAP11. CAP patients with eosinopenia had significantly higher in-hospital mortality, mechanical ventilation, risk of sepsis, significantly longer median length of stay among eosinopenic survivors, and significantly reduced median time to in-hospital death10. This study supports and expands previous research, confirming the association between eosinopenia and adverse outcomes in pneumonia patients, and extends this conclusion to the AP patients for the first time, thereby supplementing the evidence of prognostic markers in the AP field. Because eosinophil counting is rapid, inexpensive, and widely available in routine clinical settings, it can help clinicians early identify high-risk AP patients for timely intervention. A number of previous studies have reported prognostic factors for AP, including lymphocyte count, serum creatinine, lactate dehydrogenase, and the blood urea nitrogen-to-albumin ratio24–26, all of which are easily measured in peripheral blood, similar to eosinophils. However, most prior studies were retrospective in design and did not employ rigorous statistical methods such as propensity score weighting or multiple sensitivity analyses to reduce confounding. In contrast, our study used multiple complementary statistical approaches and sensitivity analyses to confirm the independent association between eosinopenia and 30-day mortality, supporting the robustness and reliability of our findings. Patients with AP and eosinopenia have a reduced ability to clear the aforementioned pathogens, and the infection is less easily controlled, leading to increased death. However, the underlying immunological mechanisms require further validation through more in-depth and targeted experimental studies. In clinical practice, Pseudomonas aeruginosa, Klebsiella pneumoniae, Escherichia coli, and Staphylococcus aureus were the main pathogenic bacteria detected in patients with AP27,28. Ravin et al. reported that eosinophils can engulf bacterial organisms such as Staphylococcus aureus, Escherichia coli, and Pseudomonas aeruginosa via phagocytosis29. In vivo, eosinophil cationic protein (ECP) had antibacterial effects against Pseudomonas aeruginosa and Staphylococcus aureus30,31. Both ECP and major basic protein-1 (MBP-1) had bactericidal activity against Staphylococcus aureus and Escherichia coli in vitro32. Eosinophile-derived reactive oxygen species can destroy Escherichia coli by biding to eosinophilic peroxidase (EPO)29. Yousefi et al. indicated that eosinophils respond to bacteria such as Escherichia coli by catapult-like ejecting mitochondrial DNA33,34. A critical question in prognostic marker research is whether a new marker offers incremental value over existing parameters. Our analysis confirmed that eosinopenia independently contributed to risk prediction, significantly improving model discrimination and reclassification. This finding supports the clinical utility of routinely monitoring eosinophil counts in AP patients, as it aids in identifying high-risk individuals who might be overlooked by traditional scoring systems. Our trial has some potential limitations: (1) This study was conducted at a single center, which restricts the external validity and generalizability of our findings. Caution is needed when extrapolating these results to other populations and settings, and multicenter studies are required to validate our conclusions. As this study was performed in a large tertiary general hospital, the enrolled cohort included a high proportion of critically ill patients, which may have introduced selection bias and may not fully represent the spectrum of patients with aspiration pneumonia in other clinical settings. Future multicenter, prospective studies with larger and more geographically diverse samples are warranted to validate the present findings and improve the external validity and generalizability of the results. (2) Consistent with other studies on AP, our study used clinical determinants to define AP. Unfortunately, to date, there is still no standard definition of AP. (3) The possibility of unobserved confounding factors will affect the results of the study. (4) All covariates in this study were measured only at the time of admission, and repeated measurements during follow-up were not available. Therefore, we were unable to account for potential time-dependent confounding arising from changes in disease severity or treatment response over time. (5) Although the PaO₂ value at admission was included in the analysis, baseline FiO₂ and specific oxygen delivery methods were not recorded. Consequently, comprehensive oxygenation status and precise disease severity could not be adequately adjusted, and disease severity may thus have become an unmeasured confounding factor in this study. Future studies are warranted to establish and validate a dedicated prognostic scoring system for aspiration pneumonia by incorporating eosinophil count with established clinical scores and routine laboratory parameters. Direct comparisons of predictive performance using ROC curve analysis and external validation cohorts should be performed to verify and strengthen the clinical utility of such a composite scoring tool. Conclusions In summary, our study indicates that eosinopenia is associated with increased 30-day mortality in patients with aspiration pneumonia. The primary clinical implication of these findings is that eosinophil count may serve as a simple, readily available prognostic marker. Future research should prioritize the development and validation of a composite prognostic scoring system for aspiration pneumonia that incorporates eosinopenia together with established severity scores and routine laboratory parameters. Further studies exploring the underlying immunological mechanisms will also help deepen our understanding of the role of eosinophils in the pathogenesis of aspiration pneumonia. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 Abbreviations Aspiration pneumonia patients with low eosinophil Aspiration pneumonia Community-acquired pneumonia Confidence interval C-reactive protein Lymphocytes Platelets Albumin Arterial partial pressure of oxygen Neutrophils Procalcitonin Blood urea nitrogen Eosinophil cationic protein Major basic protein-1 Eosinophilic peroxidase Author contributions J.Z. had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis, including and especially any adverse effects. J.Z., X.J., and L.L. were responsible for the study design. J.L. and J.H. included all participants and performed the baseline evaluation. L.L., J.F. and Y.L. completed the follow-up and were responsible for data collection. Statistical analyses were performed by J.Z. with contributions from H.Z. The first draft of the manuscript was written by X.J. and J.Z. J.Z. commented on previous versions of the manuscript. All authors read and approved the final manuscript. Funding This work was supported by the Science and Technology Achievement Transformation Guidance Project of Shanxi Province (202404021301059), the Scientific Research Project of Shanxi Provincial Health Commission (2023127), the Scientific Research Foundation for Talent Introduction of Shanxi Bethune Hospital (2023RC27), and the Science and Technology Innovation Project of Higher Education of Shanxi Province (2024L100). Data availability The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate This study was conducted in accordance with the Declaration of Helsinki and approved by the ethics committee of Shanxi Bethune Hospital (approval YXLL-2015-148). Written informed consent was obtained from all of the participants in the study. Consent for publication Written informed consent for publication was obtained from each subject. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Xia Jing and Li Li contributed equally to this work. References Associated Data Data Availability Statement The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.