Statistical analysis
The significance of differences between the asthmatic and healthy children in their categorical variables was estimated by theχ 2 test and continuous variables (for example, age, BMI, Δ%FEV1) by the t test or ANOVA analyses. Receiver operating characteristic (ROC) curve with analysis of differences in the area under curves (AUC) was used to estimate the diagnostic accuracy. Furthermore, asthmatic subjects with repeated data were further divided into three groups according to the levels of eicosanoid species during exacerbation, i.e. the top 30%, middle 40%, and bottom 30%. A two-way mixed-design analysis of variance (i.e. split-plot ANOVA) was performed for analyzing the effect of stratified eicosanoid levels (top 30% vs. middle 40% vs. bottom 30%) and phases (active exacerbation vs. convalescence).