Table 3

Multiple linear regression models of the form y = α + β1FEV1 + β2log(eosinophil) + β3inhaled steroids where (1) y = log PD20 and (2) y = bronchodilator reversibility (absolute value)

Unstandardised regression coefficientConfidence intervalp value
Regression coefficient is the change in the dependent variable (log PD20 or bronchodilator reversibility) per unit change in the independent variable (FEV1 % predicted, log eosinophil count).
Log PD20, R2=0.45
    FEV1 % predicted0.020.01 to 0.030.001
    Log eosinophil count–0.013–0.32 to 0.290.93
    Inhaled steroids (yes v no)–0.42–0.89 to 0.060.083
    Constant1.470.45 to 2.480.006
Bronchodilator reversibility, R2=0.25
    FEV1 % predicted0–0.005 to 0.0040.80
    Log eosinophil count0.1080.002 to 0.210.045
    Inhaled steroids (yes v no)0.101–0.063 to 0.2660.22
    Constant0.22–0.13 to 0.580.21