Blood Eosinophils in Hospitalised Preschool Wheezers Predict School-Age Asthma -- Revisiting Possible Thresholds

Blood Eosinophils in Hospitalised Preschool Wheezers Predict School-Age Asthma -- Revisiting Possible Thresholds | Authorea try { document.documentElement.classList.add('js'); } catch (e) { } var _gaq = _gaq || []; _gaq.push(['_setAccount', 'G-8VDV14Y67G']); _gaq.push(['_trackPageview']); (function() { var ga = document.createElement('script'); ga.type = 'text/javascript'; ga.async = true; ga.src = ('https:' == document.location.protocol ? 'https://ssl' : 'http://www') + '.google-analytics.com/ga.js'; var s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(ga, s); })(); Skip to main content Preprints Collections Wiley Open Research IET Open Research Ecological Society of Japan All Collections About About Authorea FAQs Contact Us Quick Search anywhere Search for preprint articles, keywords, etc. Search Search ADVANCED SEARCH SCROLL This is a preprint and has not been peer reviewed. Data may be preliminary. 24 November 2025 V1 Latest version Share on Blood Eosinophils in Hospitalised Preschool Wheezers Predict School-Age Asthma -- Revisiting Possible Thresholds Authors : Aleksander Adamiec 0000-0001-9407-3419 , Katarzyna Moliszewska , Aleksandra Marchwińska-Pancer , Paweł Kukiełka , Marek Kulus 0000-0002-5360-4372 , and Wojciech Feleszko 0000-0001-6613-2012 [email protected] Authors Info & Affiliations https://doi.org/10.22541/au.176398572.21663998/v1 244 views 129 downloads Contents Abstract Information & Authors Metrics & Citations View Options References Figures Tables Media Share Abstract Background Preschool wheezing disorders are heterogeneous, and a proportion of affected children progress to asthma. Identifying early predictors of persistent disease may enable targeted follow-up and intervention. Eosinophil blood counts (EBCs) are accessible biomarkers, yet their prognostic value during acute infectious wheezing episodes requiring hospitalisation remains unclear. Methods We reviewed records of 323 children aged 1–5 years hospitalised for wheezing at a tertiary centre (2013–2024). EBCs, other laboratory and clinical parameters were retrieved from the episode with the highest eosinophil value. Follow-up beyond age 5 determined respiratory outcomes: no recurrence, transient wheeze, or persistent wheeze/asthma. Logistic regression, ROC analysis, and Youden’s J statistic assessed predictive performance, with bootstrapping for internal validation. Results Persistent wheeze/asthma was diagnosed in 115 children (35.6%). EBCs were linearly associated with asthma risk, with each 100 cell/µl increase corresponding to an 18% increase in asthma odds in the final model. An EBC ≥0.57 ×10 3 /μL identified a high-risk subgroup (>50% asthma risk; OR 5.63, specificity 92.8%). EBC as a continuous variable yielded AUC 0.67, improving to 0.79 when combined with five clinical predictors (aeroallergen sensitisation, food allergy, wheeze outside colds, age <2 years at first episode, hospitalisation at first episode). Eosinophils ≥4% of leukocytes were associated with four-fold higher asthma odds (specificity 87.0%). Conclusion EBCs obtained during acute infectious wheezing independently predict later asthma, especially when integrated with simple clinical features. Incorporating EBC-based risk estimates into routine assessment could guide early counselling, follow-up, and preventive strategies in high-risk preschool wheezers. Blood Eosinophils in Hospitalised Preschool Wheezers Predict School-Age Asthma – Revisiting Possible Thresholds Running title: Eosinophils during wheeze relate linearly to asthma risk Aleksander Adamiec 1, 2 , Katarzyna Moliszewska 1 , Aleksandra Marchwińska-Pancer 1 , Paweł Kukiełka 1 , Marek Kulus 1 , Wojciech Feleszko 1 1 Department of Pneumonology and Childhood Allergy, Medical University of Warsaw, Warsaw, Poland 2 Doctoral School, Medical University of Warsaw, Warsaw, Poland Correspondence: Dr. Wojciech Feleszko, Department of Paediatric Respiratory Diseases and Allergy, The Medical University Children’s Hospital, Medical University of Warsaw, Żwirki i Wigury 63A, Warsaw, Poland. E-mail: [email protected] ORCID ID: Aleksander Adamiec: 0000-0001-9407-3419 Katarzyna Moliszewska: 0009-0009-5459-4338 Aleksandra Marchwińska-Pancer: 0009-0002-3459-281X Paweł Kukiełka: 0009-0007-0303-6999 Marek Kulus: 0000-0002-5360-4372 Wojciech Feleszko: 0000-0001-6613-2012 Acknowledgments : This research received no external funding. Conflict of interest : The authors declare no conflict of interest in relation to this work Author Contributions: Conception, design: AA, WF; Conception and design of the present analysis: AA. Data collection: AA;KM;AMP;PK. Analysis and/or interpretation: AA;WF;MK with AA taking primary responsibility for the statistical analyses. Drafting the manuscript for important intellectual content: AA; KM; AMP; PK wrote the first draft; WF, MK supervised multiple drafts, and all authors contributed to the final draft. This article has no online Supplementary Material. ABSTRACT Background Preschool wheezing disorders are heterogeneous, and a proportion of affected children progress to asthma. Identifying early predictors of persistent disease may enable targeted follow-up and intervention. Eosinophil blood counts (EBCs) are accessible biomarkers, yet their prognostic value during acute infectious wheezing episodes requiring hospitalisation remains unclear. Methods We reviewed records of 323 children aged 1–5 years hospitalised for wheezing at a tertiary centre (2013–2024). EBCs, other laboratory and clinical parameters were retrieved from the episode with the highest eosinophil value. Follow-up beyond age 5 determined respiratory outcomes: no recurrence, transient wheeze, or persistent wheeze/asthma. Logistic regression, ROC analysis, and Youden’s J statistic assessed predictive performance, with bootstrapping for internal validation. Results Persistent wheeze/asthma was diagnosed in 115 children (35.6%). EBCs were linearly associated with asthma risk, with each 100 cell/µl increase corresponding to an 18% increase in asthma odds in the final model. An EBC ≥0.57 ×10³/μL identified a high-risk subgroup (>50% asthma risk; OR 5.63, specificity 92.8%). EBC as a continuous variable yielded AUC 0.67, improving to 0.79 when combined with five clinical predictors (aeroallergen sensitisation, food allergy, wheeze outside colds, age <2 years at first episode, hospitalisation at first episode). Eosinophils ≥4% of leukocytes were associated with four-fold higher asthma odds (specificity 87.0%). Conclusion EBCs obtained during acute infectious wheezing independently predict later asthma, especially when integrated with simple clinical features. Incorporating EBC-based risk estimates into routine assessment could guide early counselling, follow-up, and preventive strategies in high-risk preschool wheezers. Keywords: asthma, eosinophils, prediction, preschool wheeze Introduction Preschool children with wheezing disorders pose significant diagnostic and therapeutic challenges, often requiring extensive medical resources 1 . Wheezing in this age group can encompass a variety of clinical presentations and underlying mechanisms 2 . Viral respiratory infections are a common trigger for wheezing episodes, affecting a considerable proportion of preschool children 3 . In many cases, these episodes are self-limited and resolve without long-term consequences. However, a subset of children experiences recurrent, more severe wheezing episodes, which necessitate hospitalisation and specialist referral for further management 4 . Recurrent wheezing may be an early manifestation of asthma, a chronic condition associated with airway remodelling, hyperresponsiveness, and persistent inflammation. 5 This progression to asthma can significantly affect the long-term quality of life for children, highlighting the need for early identification of individuals at higher risk 6 . Identifying early biomarkers that predict the likelihood of future asthma in children with wheezing disorders remains an important clinical goal 7, 8 . Eosinophil blood counts (EBCs) have been proposed as a simple, accessible tool for predicting the risk of asthma development in wheezing children 9 . Eosinophils are key players in the pathophysiology of allergic asthma, contributing to airway inflammation 10 . In clinical practice, EBCs have been incorporated into predictive tools, such as the Asthma Predictive Index, to help identify children at risk of asthma 11 . Despite this, the evidence supporting the widespread use of EBCs in this context remains limited, and current guidelines are largely based on expert consensus rather than robust data 10, 12 . An EBC threshold for defining eosinophilic inflammation 12, 13 . However, recent evidence suggests that this fixed cutoff may oversimplify the dynamic nature of eosinophil counts during acute wheezing episodes in preschool children, calling for a more nuanced approach to asthma risk stratification 12 . The present retrospective cohort study aimed to explore the relationship between EBCs measured during acute wheezing episodes requiring hospitalisation and the subsequent development of asthma in preschool-aged children. By evaluating the prognostic value of EBCs in predicting future asthma, this study seeks to clarify the potential role of eosinophil counts in guiding early management strategies for children at risk of developing this chronic respiratory condition. 2. Methods Using the hospital’s integrated electronic health record system, we retrospectively extracted clinical and laboratory data on children aged 1 to 5 years who were hospitalised at the Medical University of Warsaw Children’s Hospital between 2013 and 2024 due to wheezing disorders. These cases were identified using the following ICD-10 codes: J20 (Acute bronchitis), J21 (Acute bronchiolitis), and J22 (Unspecified acute lower respiratory infection), B97.4 (Respiratory syncytial virus as the cause of diseases classified elsewhere). Subsequently, the clinical records were reviewed to confirm whether wheezing was documented upon admission. Next, we assessed whether these patients had returned to the hospital or an affiliated outpatient clinic for any reason after the age of 5. Children who fulfilled both criteria were included in the study. For each included subject, we retrieved data from complete blood counts (CBCs) taken during the initial hospitalisation for wheezing. We recorded eosinophil blood count values, both absolute (EOS#) and relative (EOS%, expressed as a percentage of total white blood cells). Additionally, we noted white blood cell count (WBC), platelet count (PLT), and C-reactive protein concentration (CRP). In cases where multiple CBCs were performed during a single episode, all values were recorded for subsequent analysis. We excluded patients with conditions unrelated to acute wheezing which may have affected EBCs at baseline – subjects with parasitic infections, neoplastic disorders, receiving chemotherapy, as well as those who receiver systemic steroids during the 4 weeks preceding the onset of wheezing were excluded from the analysis. In the final step, three researchers (KM, AM, RK) reviewed all follow-up records of patients seen after the age of 5, documenting their respiratory status. These assessments were independently verified by another researcher (AA) to minimise the risk of misclassification. Based on follow-up data, patients were stratified into three groups: – No recurrent wheezing : children with no documented recurrent episodes after the index hospitalisation; – Transient recurrent wheezing : children with recurrent wheezing episodes that resolved before the final follow-up, with or without treatment; – Persistent recurrent wheezing or asthma : children who met the epidemiological definition of asthma: children formally diagnosed with asthma by a physician during a previous visit, who have been assigned the appropriate ICD-10 code (J45, J45.0, J45.1, J45.8, J45.9); those with continued wheezing beyond age 5; or those receiving asthma treatment (e.g., chronic inhaled corticosteroids or leukotriene receptor antagonists) 14 . Where appropriate, these groups were also analysed dichotomously as Asthma vs Non-Asthma (combining the transient wheezing and no recurrent wheezing groups). Statistical analysis All analyses were two-sided with α=0.05. Continuous variables were inspected for distributional shape with Shapiro–Wilk tests and Q–Q plots; most laboratory measures were right-skewed. Accordingly, descriptive statistics for laboratory markers are presented as median [IQR] by group. Categorical variables are shown as n (%). Group comparisons (laboratory markers). For EBC and EOS%, between-group comparisons were made with the Wilcoxon rank–sum test (Mann–Whitney) for the planned two-group contrasts (Asthma vs Transient; Asthma vs No recurrent; Transient vs No recurrent; Asthma vs No asthma). Effects are reported as Hodges–Lehmann median differences with 95% CIs. For WBC, PLT, and CRP, due to skew and occasional low values, outcomes were analysed on the log1p scale (log(1+value)). We used Welch’s ANOVA across the three specific groups (Asthma, Transient wheezing, No recurrent wheezing) and pairwise Welch t-tests vs Asthma (and Asthma vs No asthma). Results are expressed as ratios of geometric means with 95% CIs, obtained by back-transforming mean differences on the log1p scale. For all families of pairwise tests within a biomarker, Holm adjustment controlled the family-wise error rate. Categorical comparisons used χ² tests with multiplicity correction when applicable. Prediction modelling We fitted logistic regression models for asthma (yes/no). Absolute EBC was the primary predictor; in multivariable models EBC was combined with pre-specified clinical covariates (aeroallergen sensitisation, food allergy, wheeze outside colds, atopic dermatitis, parental asthma, age episode). To aid clinical interpretation, the EBC coefficient was scaled per 100 cells/µl and reported as OR per 100 with 95% CI. Model discrimination was summarised by the ROC AUC with 95% CIs derived by bootstrap resampling (B=2000); ROC curves are shown with 95% CI ribbons. We identified thresholds using Youden’s J on the EBC-only ROC and also evaluated pre-specified cut-points from the literature. For each threshold we reported sensitivity, specificity, odds ratio and predicted probability. Because the “wheezing outside of colds” category contained a rare, outcome-enriched level causing quasi-separation, we used Firth bias-reduced estimation as the primary model. Software Data processing and analyses were conducted in R (version 4.5.0) 15 . Ethics The study was acknowledged by the bioethical committee of the Medical University of Warsaw, case number AKBE/206/2025. As the work uses historical data informed consent of the participants was not necessary, as per local regulations. 3. Results Multiple tests In patients who underwent multiple CBC tests over the course of their hospital stay, the association between EBC and future asthma in logistic regression models did not differ significantly when taking into consideration the different achieved results. Hence, in order to maximise signal consistency, all CBC parameters, including EBC, as well as CRP values included in the analyses came from the samples in which the highest overall EBCs were detected. Subject characteristics We identified 1373 children hospitalised with wheezing in the given timeframe. Follow-up data was available for 323 children, who were included in the study. There was no significant difference in median eosinophil counts between children who had follow-up data available and those who did not (median EBC 0.2 vs 0.17, p=0.339, respectively). The study included 323 children, divided into three groups: children with persistent wheezing or asthma (n = 115), children with transient wheezing (n = 72), and children without recurrent wheezing (n = 136). There were no significant differences in age or sex at the index episode or follow-up. A higher proportion of children in the asthma group had eosinophils constituting ≥4% of the total leukocyte count (43, 37.4%) compared to the transient wheezing (8, 11.1%) and no-recurrence (19, 14.0%) groups, suggesting a strong association between early eosinophilia and persistent wheezing or asthma development. The findings were statistically significant with p values detailed in Table 1. Several clinical parameters were also significantly more prevalent in the asthma group. More than four episodes of wheezing were documented in over half of the asthma patients (58, 50.4%) and in an even larger proportion of those with transient wheezing (53, 73.6%), while none of the children without recurrent wheezing met this criterion (0, 0%). Parental asthma was reported in 14 (12.2%) children with asthma, 9 (12.5%) with transient wheezing, and only 2 (1.5%) in the no-recurrence group. From the asthma group, 109 (94.8%) patients had physician-diagnosed asthma, while 6 (5.2%) were included based on the epidemiological diagnosis of asthma – currently receiving treatment for asthma without a formal diagnosis. Due to the large discrepancy between the sizes of the groups it was impractical to compare them statistically, but excluding the 6 patients without a formal diagnosis from further analyses did not significantly impact the results. As such, the asthma group comprises all 115 patients described above. Detailed characteristics and comparative results for all groups are presented in Table 1. Median differences Mean EBCs were not normally distributed among any of the groups. EBC was higher in Asthma vs other phenotypes: Asthma 0.30 [0.10–0.66] vs Transient 0.16 [0.06–0.33] and No recurrent 0.14 [0.02–0.30]; Asthma vs No asthma 0.14 [0.03–0.30]; all Wilcoxon contrasts p<0.01 after Holm. EOS% showed the same pattern (all p<0.001). These findings were visualised in the corresponding box plots and beehive plots, which illustrated the clear upward shift in eosinophil concentration among children with more persistent wheezing phenotypes as well as their distribution among groups (Figure 1, Figure 2). Effects are reported as Hodges–Lehmann median differences (95% CI) in the comparative tables; figures show box/IQR with median dots. Risk Modelling To evaluate the prognostic value of eosinophil blood counts (EBC) and selected clinical variables, several logistic regression models were constructed and compared. The first model assessed the predictive capacity of absolute EBC values alone. The ROC curve for this model yielded an area under the curve (AUC) of 0.673 (95% CI 0.611-0.734, bootstrap), indicating a modest ability to distinguish between children who went on to develop asthma and those who did not. We assessed whether the association between EBC and asthma was linear on the logit scale by comparing a logistic model with a linear EBC term to a model with a natural spline of EBC (df=3). The spline model did not improve fit (likelihood-ratio test p=0.77; AIC 396.5 vs 393.0 for the linear model), so a linear effect was assumed. In the final linear model, each 100 cells/µl increase in EBC was associated with a 24% higher odds of asthma (OR 1.24, 95% CI 1.15–1.36, p<0.01 (Figure 3). While the ROC curve deviates from the diagonal line of random prediction, its performance suggests limited sensitivity and specificity when relying solely on EBC as a predictor (Figure 4). We then examined whether a classic eosinophil percentage threshold (≥4% of total leukocytes) could serve as a useful predictor of asthma. This parameter yielded an odds ratio (OR) of 4.00 (95% CI: 2.30-6.96, p<0.01), with a sensitivity of 37.4% and specificity of 87.0%. Given that EBCs demonstrated a direct relationship with the probability of asthma development, we investigated whether any clinically relevant thresholds could be established. Two such cut-off points were identified: At an EBC of 0.23 ×10³/μL, the analysis using Youden’s J statistic identified the most balanced trade-off between specificity and sensitivity. The sensitivity was 61.7%, specificity was 63.0%, and the corresponding J = 0.25. This cut-off yielded an OR of 2.75 (95% CI: 1.72-4.39, p<0.01) and was associated with a predicted asthma risk of 32.3%. At a higher EBC of 0.57 ×10³/μL, the predicted risk of future asthma exceeded 50%, reaching 50.003%. This threshold yielded an OR of 5.63 (95% CI: 2.91-10.88, p<0.01), with sensitivity of 30.4% and specificity of 92.8%, making it potentially relevant for identifying a smaller subgroup at very high risk. We also tested thresholds commonly used in literature. 12 At 0.15 ×10³/μL the OR was 2.48 (95% CI: 1.53-4.02, sensitivity = 0.70, specificity = 0.51, p<0.01), at 0.3 ×10³/μL the OR was 2.49 (95% CI: 1.55-3.97, sensitivity = 0.53, specificity = 0.69, p<0.01), at 0.45 ×10³/μL the OR was 4.51 (95% CI: 2.51-8.10, sensitivity = 0.35, specificity = 0.89, p<0.01). Three intermediate models were then tested using the framework of the modified Asthma Predictive Index (mAPI), which includes the following parameters: four or more wheezing episodes, parental asthma, atopic dermatitis, sensitisation to aeroallergens, wheezing unrelated to colds, and food allergy. The model using only these mAPI clinical criteria yielded an AUC of 0.732 (95% CI 0.676–0.787, bootstrap). Adding eosinophils ≥4% as a categorical variable improved the model’s performance to an AUC of 0.747 (95% CI 0.691–0.802, bootstrap). Replacing the eosinophil percentage threshold with absolute EBC values as a continuous predictor increase the AUC to 0.762 (95% CI 0.709–0.816, bootstrap). Finally, a comprehensive multivariate logistic regression model was developed. It integrated absolute EBC values with five clinical variables and demonstrated predictive value in univariate analyses: aeroallergen sensitisation, food allergen sensitisation, wheezing unrelated to colds, age under 2 years at the time of the first episode, and the need for hospitalisation during the first wheezing episode. This model demonstrated the highest overall performance, with an AUC of 0.790 (95% CI 0.740–0.840, bootstrap), indicating a stronger balance of sensitivity and specificity. In this final model, EBC was positively associated with asthma (OR per 100 cells/μL increase: 1.18 (95% CI: 1.08-1.30), p < 0.001), confirming that higher eosinophil counts during the initial episode significantly increased the odds of future asthma. Sensitisation to aeroallergens, food, age < 2, and wheezing unrelated to were all significantly associated with increased asthma risk (Table 2). In contrast, hospitalisation during the first ever wheezing episode was associated with reduced odds of asthma development (OR = 0.4, 95% CI: 0.23-0.68, p = 0.001). Substituting the continuous EBC variable in the final multivariate model with dichotomous indicators based on the identified cut-off values (0.23 and 0.57 ×10³/μL) resulted in slightly lower predictive performance, with AUCs of 0.7809 and 0.7881, respectively. As a sensitivity analysis for loss to follow-up, we re-ran the EBC-only logistic model under three extreme assumptions about the children without outcome data: assuming that none of them ever developed asthma (‘nobody’); assuming that all of them did (‘everybody’); and assigning asthma status at random so that the overall asthma prevalence matched that observed among children with follow-up (‘random’). Under these scenarios, the odds ratio for asthma per 0.1×10³ cells/µL of EBC ranged from 1.03 (95% CI 1.01–1.06; AUC 0.536) in the random scenario to 1.07 (95% CI 1.03–1.11; AUC 0.637) and 1.08 (95% CI 1.03–1.15; AUC 0.556) in the ‘nobody’ and ‘everybody’ scenarios, respectively, compared with 1.24 (95% CI 1.15–1.36; AUC 0.673) in the complete-case analysis. Thus, while plausible patterns of loss to follow-up could attenuate the strength of the association, the direction and statistical significance of the EBC–asthma relationship remained robust across this range of assumptions. While hospitalisation during the first wheezing episode decreased future asthma risk, patients who already exhibited traits of eosinophilic inflammation at that staged were significantly more likely to develop asthma (at 0.57 ×10³/μL: OR = 10.57, p<0.01, specificity = 0.32, sensitivity = 0.96). 4. Discussion When parents rush a breathless toddler into hospital in the middle of a winter night, two questions hover over every bedside conversation: Will these frightening attacks keep coming back, and is my child on the road to asthma? Our data suggest that a laboratory value already sitting in the electronic chart—the eosinophil count—can tell a surprisingly nuanced story about that child’s future airways. Our model outperforms existing clinical predictive tools by integrating dynamic eosinophil counts measured during acute episodes with bedside clinical features, offering a more precise and actionable risk stratification for future asthma inception in preschool children. In a retrospective cohort of 323 preschool children admitted for infectious wheezing, we found that higher absolute eosinophil blood counts (EBCs) taken during the acute episode were independently and dose-dependently linked to asthma diagnosed after age five. When treated as a stand-alone dichotomous marker, EBCs offered modest discrimination (AUC 0.67). Modelling the counts as a continuous variable and combined with five bedside clinical features, improved performance (multivariable AUC of 0.79), with every 0.1 × 10³/µL increase in the EBC, associated with 18 % higher odds of subsequent asthma, underscoring the value of reading the marker as a dial rather than a switch. These findings both confirm and extend earlier work from birth cohort studies such as the Tucson Children’s Respiratory Study and CAMP, among others, which linked baseline eosinophilia to later asthma but measured eosinophils during quiescent periods 16-18 . By sampling children in the chaotic midst of viral infection—when neutrophil-predominant inflammation might be expected to drown out type-2 signals—we show that an underlying asthmatic tendency can still shine through 19 . Moreover, we derived pragmatic cut-offs directly from receiver-operating-characteristic optimisation: 0.23 × 10³/µL halved diagnostic uncertainty, while 0.57 × 10³ µL singled out a high-risk minority whose post-school-age asthma probability exceeded 50 %. Older categorical thresholds such as “≥ 4 % of leukocytes” did not outperform these data-driven values and captured barely one-third of future asthma cases, highlighting the cost of relying on legacy cut-points. The highest predictive yield arose when the laboratory signal was analysed together with five easily obtainable clinical cues: aero-allergen sensitisation, food allergy, presence of wheezing that occurs outside colds, the occurrence of a wheezing episode before the second birthday, and whether that inaugural attack demanded admission. An intriguing inverse association emerged for children hospitalised during their very first wheeze—likely reflecting infants who present early with severe but ultimately transient RSV bronchiolitis rather than an entrenched type-2 phenotype 20, 21 . However, when patients hospitalised with their first ever episode of wheezing already presented with evidence for eosinophilic inflammation, they were 10 times more likely to develop asthma than their low-eosinophil counterparts. This highlights the importance of clinical vigilance, as blood counts performed during the first episode of wheezing might hint at the patients future phenotype, despite the obvious lack, at baseline, of a defining asthma feature – the recurrence of wheezing. The developed multifactorial model can be operationalised easily at the bedside or embedded in electronic health-record decision support. This approach transforms a routine blood count into a personalised risk estimate that informs conversations about inhaled-corticosteroid trials, follow-up intensity, or future biologic-prevention studies. Our study has several strengths: it describes a large, well-characterised cohort drawn from a tertiary referral centre; it includes strict confirmation of wheeze at presentation and systematic follow-up beyond the conventional diagnostic age-cut-off of five years; it relies on routinely collected laboratory parameters, enhancing portability; and uses robust statistics, including bootstrapping to mitigate optimism bias. The study includes several limitations. The retrospective design may lead to misclassification and underappreciation of unmeasured confounders, such as environmental tobacco smoke. Asthma diagnoses were extracted from records rather than verified prospectively with spirometry, although nearly 95 % carried a formal physician label. Excluding the remaining cases did not alter results. Eosinophil levels can transiently increase during viral infections, which might weaken the type-2 inflammation signal. This however actually supports the real-world applicability of our findings, as we demonstrate that despite possible interference, eosinophils are still associated with an increased asthma risk during acute wheezing 22 . Our findings may not generalise to community-managed wheezers, but relevant to those children sick enough to require hospitalisation where early prognosis is likely most clinically pertinent 22 . Only 24% of potentially eligible children had follow-up data after age 5, and outcomes were ascertained only for those re-attending our centre. Baseline eosinophil counts did not differ between children with and without follow-up and sensitivity analyses suggest that selection bias was unlikely to fully account for the observed association, although some attenuation of effect size is possible. As such, as well as considering the absence of an external validation cohort means our cut-offs and the model’s weights should be viewed as provisional until replicated prospectively. A prospective, external validation of the model in independent cohorts is warranted and should be planned to strengthen the robustness and generalizability of our findings. Taken together, eosinophil counts emerge from this study as a useful and accessible narrator of preschool wheeze destiny. Read on a continuum and coupled with a handful of bedside facts, they can encourage clinicians to move from “let’s wait and see” to “let’s plan and prevent.” Prospective, multi-centre studies that serially track eosinophil dynamics and test interventions guided by these risk estimates are now needed to determine whether acting on the signal can rewrite the wheezing child’s story before it hardens into lifelong asthma. 1. Brand PLP, Baraldi E, Bisgaard H, Boner AL, Castro-Rodriguez JA, Custovic A, et al. Definition, assessment and treatment of wheezing disorders in preschool children: An evidence-based approach. European Respiratory Journal.32(4):1096-110.2. Jartti T, Smits HH, Bønnelykke K, Bircan O, Elenius V, Konradsen JR, et al. Bronchiolitis needs a revisit: Distinguishing between virus entities and their treatments. Allergy. 2019;74(1):40-52.3. Henderson J, Granell R, Heron J, Sherriff A, Simpson A, Woodcock A, et al. Associations of wheezing phenotypes in the first 6 years of life with atopy, lung function and airway responsiveness in mid-childhood. Thorax. 2008;63(11):974-80.4. Robinson PFM, Fontanella S, Ananth S, Martin Alonso A, Cook J, Kaya-De Vries D, et al. Recurrent severe preschool wheeze: From prespecified diagnostic labels to underlying endotypes. Am J Respir Crit Care Med. 2021;204(5):523-35.5. Boonpiyathad T, Sözener ZC, Satitsuksanoa P, Akdis CA. Immunologic mechanisms in asthma. Semin Immunol. 2019;46:101333.6. Banjari M, Kano Y, Almadani S, Basakran A, Al-Hindi M, Alahmadi T. The relation between asthma control and quality of life in children. International Journal of Pediatrics. 2018;2018(1):6517329.7. Adamiec A, Ambrożej D, Ryczaj K, Ruszczynski M, Elenius V, Cavkaytar O, et al. Preschool wheezing diagnosis and management–survey of physicians’ and caregivers’ perspective. Pediatric Allergy and Immunology. 2020;31(2):206-9.8. Fouzas S, Brand PL. Predicting persistence of asthma in preschool wheezers: Crystal balls or muddy waters? Paediatr Respir Rev. 2013;14(1):46-52.9. Saglani S, Bingham Y, Balfour-Lynn I, Goldring S, Gupta A, Banya W, et al. Blood eosinophils in managing preschool wheeze: Lessons learnt from a proof-of-concept trial. Pediatr Allergy Immunol. 2022;33(1):e13697.10. Rothenberg ME. Eosinophilia. N Engl J Med. 1998;338(22):1592-600.11. Castro-Rodríguez JA, Holberg CJ, Wright AL, Martinez FD. A clinical index to define risk of asthma in young children with recurrent wheezing. American Journal of Respiratory and Critical Care Medicine. 2000;162(4):1403-6.12. Adamiec A, Cieślik M, Mączka K, Tarnoruda J, Jensen S, Chawes B, et al. A systematic review and meta-analysis on absolute eosinophil counts and the risk of asthma in preschool children with wheezing: An eaaci task force report. Pediatric Allergy and Immunology. 2024;35(2):e14078.13. Fitzpatrick AM, Bacharier LB, Guilbert TW, Jackson DJ, Szefler SJ, Beigelman A, et al. Phenotypes of recurrent wheezing in preschool children: Identification by latent class analysis and utility in prediction of future exacerbation. J Allergy Clin Immunol Pract. 2019;7(3):915-24.e7.14. Ghabousian A, Bezerra GN, Zhu Z, Espinola JA, Sullivan AF, Camargo CA, Jr. Impact of varying childhood asthma definitions on incidence and clinical outcomes. Ann Allergy Asthma Immunol. 2025.15. Team RC. _r: A language and environment for statistical computing_. R Foundation for Statistical Computing, Vienna, Austria; 2003.16. Covar RA, Fuhlbrigge AL, Williams P, Kelly HW. The childhood asthma management program (camp): Contributions to the understanding of therapy and the natural history of childhood asthma. Curr Respir Care Rep. 2012;1(4):243-50.17. Wright AL, Taussig LM, Ray CG, Harrison HR, Holberg CJ, Associates TGHM. The tucson children’s respiratory study: Ii. Lower respiratory tract illness in the first year of life. American Journal of Epidemiology. 1989;129(6):1232-46.18. Bleecker ER, Meyers DA, Billheimer D, Li H, Newbold P, Kwiatek J, et al. Clinical implications of longitudinal blood eosinophil counts in patients with severe asthma. J Allergy Clin Immunol Pract. 2023;11(6):1805-13.19. Macchia I, La Sorsa V, Urbani F, Moretti S, Antonucci C, Afferni C, et al. Eosinophils as potential biomarkers in respiratory viral infections. Front Immunol. 2023;14:1170035.20. Ambrożej D, Orzołek I, Makrinioti H, Castro-Rodriguez JA, Camargo CA, Jr., Hasegawa K, et al. Association of respiratory virus types with clinical features in bronchiolitis: Implications for virus testing strategies. A systematic review and meta-analysis. Paediatr Respir Rev. 2024;49:34-42.21. Makrinioti H, Hasegawa K, Lakoumentas J, Xepapadaki P, Tsolia M, Castro-Rodriguez JA, et al. The role of respiratory syncytial virus- and rhinovirus-induced bronchiolitis in recurrent wheeze and asthma-a systematic review and meta-analysis. Pediatr Allergy Immunol. 2022;33(3):e13741.22. Xing Y, Leung AS, Wong GW. From preschool wheezing to asthma: Environmental determinants. Pediatr Allergy Immunol. 2023;34(11):e14049. Baseline participant characteristics by phenotype Demography Male 77 (66.96%) 42 (58.33%) 86 (63.24%) 128 (63.47%) 205 (63.47%) Female 38 (33.04%) 30 (41.67%) 50 (36.76%) 80 (36.53%) 118 (36.53%) Total 115 (35.6%) 72 (22.29%) 136 (42.11%) 323 (64.4%) 323 (100%) Age at test 2.38 (1.68; 3.77) 2.52 (1.78; 3.81) 2.44 (1.68; 3.94) 2.49 (1.70; 3.86) 2.43 (1.68; 3.87) Age at follow-up 9.27 (7.45; 11.25) 9.45 (7.32; 11.61) 9.34 (7.41; 11.29) 9.37 (7.38; 11.33) 9.30 (7.40; 11.34) Laboratory and clinical parameters EBC 0.30 (0.10; 0.66) 0.16 (0.06; 0.33)** 0.14 (0.02; 0.30)*** 0.14 (0.03; 0.30)*** 0.20 (0.06; 0.38) EOS% 2.50 (1.00; 5.15) 1.35 (0.40; 2.92)*** 1.25 (0.20; 2.82)*** 1.30 (0.30; 2.90)*** 1.70 (0.50; 3.60) EOS%>4% 43 (37.39%) 8 (11.11%)** 19 (13.97%)** 27 (12.98%)** 70 (21.67%) WBC 12.69 (9.72; 15.00) 11.86 (9.58; 15.25) 10.44 (7.81; 14.59)* 11.05 (8.44; 15.01) 11.43 (8.76; 15.04) PLT 335.00 (287.50; 404.00) 309.50 (266.00; 370.25) 299.50 (256.75; 352.00)** 304.50 (261.25; 362.00)** 316.00 (266.00; 370.50) CRP 0.90 (0.00; 2.30) 1.50 (0.27; 2.62) 1.30 (0.00; 2.52) 1.40 (0.08; 2.60) 1.30 (0.00; 2.45) Over 4 episodes of wheezing 58 (50.43%) 53 (73.61%)* 0 (0%)** 53 (25.48%)** 111 (34.37%) Parental asthma 9 (12.5%) 2 (1.47%)** 11 (5.29%)* 25 (7.74%) Atopic Dermatitis 34 (29.57%) 20 (27.78%) 12 (8.82%)** 32 (15.38%)* 66 (20.43%) Food allergy 46 (40%) 24 (33.33%) 9 (6.62%)** 33 (15.87%)** 79 (24.46%) Aeroallergen sensitisation 41 (35.65%) 17 (23.61%) 9 (6.6%)** 26 (12.5%)** 67 (20.74%) Age < 2 43 (37.39%) 22 (30.56%) 44 (32.35%) 66 (31.73%) 109 (33.75%) Wheezing unrelated to colds 9 (7.83%) 1 (1.39%) 1 (0.74%)* 2 (0.96%)* 11 (3.41%) Need for hospital care during the first ever episode of wheezing 31 (26.96%) 10 (13.89%)* 106 (77.94%)** 116 (55.77%)** 147 (45.51%) Table 1. Continuous variables: median [IQR]; categorical: n (%). EBC = eosinophil blood counts (10 3 cells/µl), EOS% = eosinophils as a percentage of white blood cells, WBC = white blood cells (10 3 cells/µl), PLT = platelets (10 3 cells/µl), CRP = C-reactive protein (mg/dl). “No asthma” = Transient + No recurrent wheezing. EBC, EOS%: Wilcoxon rank-sum vs Asthma; Hodges–Lehmann difference (95% CI). WBC/PLT/CRP: Welch tests on log1p; effect = ratio of geometric means (95% CI). Holm-adjusted p for pairwise comparisons (*p<0.05; **p<0.01; ** p<0.001 ). Firth bias-reduced logistic regression for asthma EBC (per 100 cells/µl) 0.166 0.080 to 0.259 1.18 1.08 to 1.30 <0.001 Hospitalization with first ever onset of wheezing −0.919 −1.470 to −0.383 0.4 0.23 to 0.68 0.001 Wheezing outside of colds 1.511 0.100 to 3.249 4.53 1.11 to 25.78 0.046 Aeroallergen sensitisation 0.938 0.287 to 1.597 2.55 1.33 to 4.94 0.005 Food allergy 0.746 0.161 to 1.330 2.11 1.17 to 3.78 0.011 Age <2 at index hospitalization 0.724 0.167 to 1.293 2.06 1.18 to 3.64 0.011 Table 2 Odds ratios (OR) with 95% profile-likelihood CIs for EBC (per 100 cells/µl) and covariates hospitalization with first ever episode of wheezing , wheezing outside of colds , aeroallergen sensitisation , food allergy , age <2 at index hospitalization . Firth correction was used to address quasi-separation (rare level in Wheezing outside of colds ). OR1 higher odds. CIs are on the log-odds scale and back-transformed to ORs. Figure 1. Box-and-whisker plots for A: Asthma vs Transient wheezing, B: Asthma vs No recurrent wheezing, C: Transient wheezing vs No recurrent wheezing, D: Asthma vs No asthma. Box = IQR; whiskers = 1.5×IQR; dot = median. P values: Wilcoxon rank-sum, * denotes p<0.01; Hodges–Lehmann (HL) difference reported in text. Figure 2. Predicted probability of asthma from a logistic regression model with EBC modelled using a natural spline (df = 3). Solid line shows the estimated curve; shaded band indicates the 95% confidence interval. A likelihood-ratio test comparing this spline model with a simple linear term for EBC showed no evidence of non-linearity (p = 0.77). Figure 3. Receiver operating characteristic (ROC) curve for absolute eosinophil blood count as a predictor of asthma diagnosis after age five in preschool children hospitalised with infectious wheezing. AUC – area under the curve; Youden’s J optimal threshold marked (230 cells/µl) with orange dot; ribbon represents 95% bootstrap CI, B=2000. Figure 4. Receiver operating characteristic (ROC) curve for the final multivariable logistic regression model predicting asthma after age five in preschool children hospitalised with infectious wheezing. The model integrates absolute eosinophil blood count and five clinical predictors (aeroallergen sensitisation, food allergy, wheezing outside colds, age wheezing episode). AUC – area under the curve; ribbon represents 95% bootstrap CI, B=2000. Information & Authors Information Version history V1 Version 1 24 November 2025 Copyright This work is licensed under a Non Exclusive No Reuse License. Keywords asthma biomarkers eosinophils pediatrics Authors Affiliations Aleksander Adamiec 0000-0001-9407-3419 Warszawski Uniwersytet Medyczny View all articles by this author Katarzyna Moliszewska Warszawski Uniwersytet Medyczny View all articles by this author Aleksandra Marchwińska-Pancer Warszawski Uniwersytet Medyczny View all articles by this author Paweł Kukiełka Warszawski Uniwersytet Medyczny View all articles by this author Marek Kulus 0000-0002-5360-4372 Warszawski Uniwersytet Medyczny View all articles by this author Wojciech Feleszko 0000-0001-6613-2012 [email protected] Warszawski Uniwersytet Medyczny View all articles by this author Metrics & Citations Metrics Article Usage 244 views 129 downloads .FvxKWukQNSOunydq8rnd { width: 100px; } Citations Download citation Aleksander Adamiec, Katarzyna Moliszewska, Aleksandra Marchwińska-Pancer, et al. Blood Eosinophils in Hospitalised Preschool Wheezers Predict School-Age Asthma -- Revisiting Possible Thresholds. Authorea . 24 November 2025. DOI: https://doi.org/10.22541/au.176398572.21663998/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu . Format Please select one from the list RIS (ProCite, Reference Manager) EndNote BibTex Medlars RefWorks Direct import Tips for downloading citations document.getElementById('citMgrHelpLink').addEventListener('click', function() { popupHelp(this.href); return false; }); $(".js__slcInclude").on("change", function(e){ if ($(this).val() == 'refworks') $('#direct').prop("checked", false); $('#direct').prop("disabled", ($(this).val() == 'refworks')); }); View Options View options PDF View PDF Figures Tables Media Share Share Share article link Copy Link Copied! Copying failed. Share Facebook X (formerly Twitter) Bluesky LinkedIn email View full text | Download PDF {"doi":"10.22541/au.176398572.21663998/v1","type":"Article"} Now Reading: Share Figures Tables Close figure viewer Back to article Figure title goes here Change zoom level Go to figure location within the article Download figure Toggle share panel Toggle share panel Share Toggle information panel Toggle information panel Go to previous graphic Go to next graphic Go to previous table Go to next table All figures All tables View all material View all material xrefBack.goTo xrefBack.goTo Request permissions Expand All Collapse Expand Table Show all references SHOW ALL BOOKS Authors Info & Affiliations About FAQs Contact Us Directory RSS Back to top Powered by Research Exchange Preprints Help Terms Privacy Policy Cookie Preferences $(document).ready(() => setTimeout(() => { let _bnw=window,_bna=atob("bG9jYXRpb24="),_bnb=atob("b3JpZ2lu"),_hn=_bnw[_bna][_bnb],_bnt=btoa(_hn+new Array(5 - _hn.length % 4).join(" ")); $.get("/resource/lodash?t="+_bnt); },4000)); (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'9fed111d6c9b0db4',t:'MTc3OTI5ODIwMQ=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();