Patients with a previous diagnosis of cancer, helminth parasite infections and allergic diseases were excluded. A total of 2353 patients met the inclusion criteria, 277 children without infections were comprised the control group. The demographic and baseline characteristics of all subjects are summarized in Table 1. The median age of bacterial group was younger than other groups (P<0.001). To eliminate confounding effect age-induced, we analyzed data matched for age.
Peripheral eosinophils count from each group
The mean of eosinophil in the neonates afflicted with bacterial infections particularly G- infections was significantly higher than age-matched controls (0.63±0.60 vs 0.44±0.20, P= 0.015). Blood eosinophil numbers in neonates was very high for virus group compared with the other groups, however, this did not result in an exploration as there were no statistically significant differences among them because of a little sample size of 22 subjects, the same as MP group(n=6) (Table 2). Of the 2353 patients enrolled in the final analysis, 255 patients with eosinophilia defined as the count ≥0.6 *109/L, respectively, were identified during the study period. In addition, results from patients with eosinophil levels and 24 pathogens were further compared. The supplemental figure 2A lists the etiological distribution of 255 patients with different eosinophil levels and pathogens. Among them, patients with eosinophils ≥0.6 *109/L make up respectively 43.5% and 42.0% of the total patients afflicted with CMV DNA+ and Kpn (Supplemental Figure 2B).
Correlation of peripheral eosinophils count with multiple clinical characteristics
A correlation was found between eosinophils with the age and pathogens by using the Spearman correlation coefficient test, whether 2 of 3 factors were controlled for (R = -0.27, 0.28; P<0.001), whereas no significance in eosinophils and diseases stage (R = 0.14; P = 0.062). In addition, significant differences were seen among pathogens with different age (R= -0.38; P<0.001). Patients admitted in ICU had higher eosinophils than outpatients (0.46±0.60 vs 0.16±0.24, P <0.001), but had no significant difference compared with control group (0.45±0.20, P>0.99). We found that blood eosinophil numbers were high for neonates (<0.1 year) with infection indicative of infectious status in neonates (Figure 1A). The number of subjects with eosinophils ≥ 0.6 cells/µL decreased with ages increased in virus and G- group (Figure 1B-C).
We collected the settings where the patients admitted to, 1631 of the total 2353 children were outpatients, 542 from the general internal medicine department and 180 from the intensive care unit (ICU). The mean eosinophils count at ICU was higher than those of the other two settings. Patients who were bacteria positive possessed a higher number of eosinophils in ICU as well as the general internal medicine department than outpatients (0.46±0.60, 0.35±0.45 vs 0.16±0.24, P <0.001, <0.001), whereas no significant differences were seen with eosinophil count in the patients from ICU and control group (0.46±0.60 vs 0.45±0.20, P>0.99).
The results as shown in the table 2 demonstrated that patients infected with virus, MP and bacteria particularly G- organisms had higher eosinophil numbers compared to age-matched controls in neonates, however, there were no statistically significant differences between patients aged >0.1 year and the controls.
We did not always obtain paired samples for 3 stages of diseases, we lost samples for analyses. For the specific analyses reported here, we only obtained complete data sets from 168 patients (25 on group 1, 47 on group 2, 87 on group 3, and 9 on group 4).Using repeated measures, as is shown in the supplemental table 1, eosinophil numbers of group 1 decreased over hospitalization time, whereas those of group 2-4 reached their peaks on the acute stage, however there were no statistical significant differences in terms of eosinophil count with infection progression (P = 0.101), whether combined with the pathogen. In other words, there was no interaction in terms of eosinophils between stage and pathogen (P = 0.067).
Diagnostic value of eosinophil for infection
Areas under the receiver operating characteristic curve (AUC) was found to have a value of 0.64 (95% CI: 0.56–0.71) for eosinophil alone, which had no significant differences with the value of 0.68 (95% CI 0.61–0.75) for CRP alone in neonates with infections (consist of virus, bacteria and MP). The cutoff value for eosinophil was 0.60 *109/L, with a sensitivity and specificity, PPV, and NPV of 52.6%, 82.4%, 71.7% and 64.5%, respectively. When eosinophil was combined with CRP, the AUC was 0.81 (95% CI 0.75–0.86), with sensitivity, specificity, PPV, and NPV values of 60.6%, 99.3%, 97.1% and 67.1%, respectively. There were significant differences when compared with CRP alone in neonates with infections (P = 0.021). (Table 3, Figure 2A). Furthermore, we also analyzed the diagnostic value of eosinophil in neonatal patients with G- bacteria and virus (Figure 2B-C).