Table 2 Distribution and strength of univariable association between the predictors at baseline and sensitisation to LA allergens at the 32nd month of the training programme
Questionnaire only (model 1)Questionnaire + SPT (model 2)Questionnaire + BR (model 3)Questionnaire + SPT + BR (model 4)
βOR (95% CI)βOR (95% CI)βOR (95% CI)βOR (95% CI)
Intercept−2.86−3.16−3.44−3.55
Symptoms indicative for asthma0.702.2 (1.1 to 4.3)0.621.9 (1.0 to 3.9)0.611.9 (1.0 to 3.8)0.501.7 (0.8 to 3.5)
Allergic symptoms1.384.6 (2.2 to 9.5)1.103.2 (1.5 to 7.0)1.394.3 (2.1 to 9.1)1.073.1 (1.4 to 6.9)
Positive SPT to common allergens0.992.8 (1.4 to 5.8)0.842.4 (1.2 to 5.1)
Measurable PC20 (⩽32 mg/ml)1.324.0 (2.0 to 7.8)1.213.6 (1.8 to 7.2)
ROC area (95% CI) after bootstrapping0.731 (0.663 to 0.799)0.754 (0.682 to 0.826)0.777 (0.710 to 0.844)0.785 (0.717 to 0.853)
Hosmer-Lemeshow test (p value)0.9990.9850.9910.991
  • ROC area SPT to common allergen 0.69 (95% CI 0.61 to 0.77).

  • ROC area measurable PC20 0.69 (95% CI 0.60 to 0.77).

  • From the questionnaire model, the individual probability of developing LA sensitisation can be estimated using the following formula:

  • P(sensitisation) = 1/(1+ exp(−(−2.86 + symptoms indicative for asthma × 0.70 + allergic symptoms × 1.38))).

  • Predictor is valued as 1 when present and 0 when absent.

  • β, regression coefficient after multiplication by correction factor from bootstrapping procedure; BR, bronchial responsiveness to methacholine; 95% CI, 95% confidence interval; OR, odds ratio; ROC, receiver operating characteristic; SPT, skin-prick testing.