Table 4 Criterion validity: performance of the 8‐item predictor and 5‐item discriminant function predictor* (DFP) versus bronchial hyperresponsiveness (PC20) and a prior physician diagnosis of asthma (MD asthma) (n = 118 subjects).
| Predictor | Test positive | Sensitivity | Specificity | Correctly classified (%) |
|---|---|---|---|---|
| 8‐item predictor† | ||||
| PC20 ⩽8 mg/ml | 62 (52.5%) | 71% | 70% | 70% |
| PC20 ⩽4 mg/ml | 44 (37.2%) | 61% | 85% | 74% |
| MD asthma | 21 (17.8%) | 79% | 98% | 94% |
| 5‐item DFP | ||||
| PC20 ⩽8 mg/ml | 44 (37.2%) | 52% | 81% | 65% |
| PC20 ⩽4 mg/ml | 33 (30.0%) | 47% | 90% | 69% |
| MD asthma | 18 (15.3%) | 71% | 99% | 93% |
*From Burney et al.15
†For each dichotomised outcome variable, separate 8‐item logistic regression models were developed, based on the best combination of sensitivity and specificity. In each case, the 8 items included ever experiencing trouble breathing (continuously or repeatedly), wheezing in the previous 12 months, an attack of shortness of breath in the previous 12 months, having been awakened by nocturnal cough and/or chest tightness, and allergic respiratory symptoms when around animals, feathers, a dusty part of the house, or outdoor environmental allergens and pollens.