Results
The study population included 123 children (5-18 years) of which 64 (52%) were male. The MCC was positive in 81 (65.9%) children indicating the presence of AHR. In Table 1, the characteristics of the study population are summarized according to the presence of AHR. The median age of children with AHR was significantly younger at the time of diagnosis compared to those without AHR (9.7 vs 12.3 years; P = .001). The prevalence of aeroallergen sensitization in the skin prick test was significantly higher in children with AHR when compared to those without AHR (35.7% vs 58.0%; P = .019). Subclassification according to individual allergen groups revealed for AHR positive children an elevated incidence of allergy to tree pollen (11.9% vs 37.0%; P = .003), animal dander (9.5% vs 27.2%; P = .023) and house dust allergy (21.4% vs 44.4%; P = .012). With regard to admission symptoms,nocturnal cough (9.5% vs 30.9%; P = .008), wheezing not associated with colds (4.8% vs 22.2%; P = .013) and wheezing associated with colds (35.7% vs 58.0%; P = .019) were observed significantly more frequently in AHR positive children compared to children without AHR. Combining wheezing symptoms (ever wheezing ) resulted in an improved significance (40.5% vs 66.7%; P = .005), while significance was reduced for combinations of nocturnal cough withexercise-induced cough . However, dyspnea andexercise-induced dyspnea , even in combination with other parameters, did not reach significance. For the spirometry items, only a baseline forced expiratory flow at 75% of vital capacity (FEF75) <65% was found significantly more frequently in AHR positive children compared to the group without AHR (14.3% vs 30.9%; P = .045). The diagnostic value of significant parameters in the univariate analysis is shown in Table 2.
A comparison of laboratory parameters reveals that the percentage of children with a blood eosinophilia count over 500/ μL was significantly elevated in the AHR positive group compared with children without AHR (22.8% vs 2.4%; P = .002; N = 80). Also, the median of the percentage of blood eosinophilia was significantly higher in AHR positive children compared with the group without AHR (5.6% vs 2.8%; P = .004; N = 93).
Multivariate logistic regression including age, ever wheezing ,nocturnal cough , tree pollen allergy and FEF75<65% confirmed that all parameters chosen are independent predictors of the occurrence of AHR. In Table 3, the results of multivariate analysis are summarized. Multivariate prediction score values, weighted as described in the methods section, were calculated for each child as follows: individual score value = 14 x tree pollen allergy (yes = 1, no = 0) + 11 x ever wheezing (yes = 1, no = 0) + 13 x nocturnal cough (yes = 1, no = 0) + 11 x FEF75<65% (yes = 1, no = 0) – 2 x age (5 – 18 years) + 36. The last summand was added to get only positive score values ranging from 0 to 75 points. The individual score values calculated for our patients ranged from 3 to 61 with a median (interquartile range) of 32 (19-41). The risk of the occurrence of AHR increased with the score value. In Table 4, the test characteristics of the multivariate score at various cutoff points are summarized. At a cutoff point of 35 the presence of AHR is predicted with a specificity of 90.5% and a PPV of 91.5%. The ROC curve displaying sensitivity and specificity of score with ascending cutoff points is shown in Figure 2. The area under the curve (AUC) was 0.813.