Cut-off points for FeNO for predicting a sputum eosinophil count ≥3% after multiple logistic regression analysis
| Smoking | Atopy | High dose of ICS | FeNO cut-off (ppb) | Specificity (%) | Sensitivity (%) | PPV (%) | NPV (%) |
| No | No | No | 38 | 74 | 66 | 65 | 75 |
| No | No | Yes | 24 | 59 | 78 | 58 | 79 |
| No | Yes | No | 58 | 88 | 48 | 74 | 70 |
| No | Yes | Yes | 35 | 71 | 70 | 64 | 76 |
| Yes | No | No | 23 | 58 | 78 | 58 | 78 |
| Yes | No | Yes | 15 | 37 | 94 | 52 | 89 |
| Yes | Yes | No | 33 | 68 | 70 | 62 | 76 |
| Yes | Yes | Yes | 20 | 54 | 85 | 58 | 83 |
When combining all variables into the logistic model, only FeNO (p<0.0001), smoking (p=0.044) and ICS (p=0.019) were significant predictors of a sputum eosinophil count ≥3%. When atopy was added the p value was not significant although a trend was observed (p=0.086) (for more details on calculation see online supplement).
FeNO, exhaled nitric oxide; ICS, inhaled corticosteroids; NPV, negative predictive value; PPV, positive predictive value.