Table 3

 Comparison of inhaled corticosteroids and placebo for all-cause mortality per 100 patient-years in the various subgroups

Inhaled corticosteroidsPlaceboAdjusted HR (95% CI)p value for interaction
Hazard ratios (HRs) were adjusted for age, sex, baseline post-bronchodilator FEV1, smoking status, body mass index, and individual trial (see Methods section for details).
Overall1.572.100.73 (0.55 to 0.96)
Age (years)
    <600.570.930.60 (0.33 to 1.07)0.443
    ⩾602.913.670.77 (0.56 to 1.06)
Sex
    Women0.771.530.46 (0.24 to 0.91)0.234
    Men1.932.360.80 (0.58 to 1.09)
BMI (kg/m2)
    <251.672.320.68 (0.47 to 0.99)0.686
    ⩾251.431.900.77 (0.50 to 1.17)
Smoking status
    Current1.231.440.87 (0.59 to 1.22)0.231
    Former2.794.520.60 (0.39 to 0.93)
Baseline FEV1
    <60% of predicted2.713.860.67 (0.48 to 0.94)0.707
    ⩾60% of predicted0.830.910.90 (0.54 to 1.53)
Length of study
    ⩽12 months2.434.060.58 (0.31 to 1.10)0.609
    >12 months1.461.860.77 (0.56 to 1.05)
Study medication
    Fluticasone3.615.180.70 (0.48 to 1.01)0.944
    Budesonide1.041.330.73 (0.42 to 1.27)
    Triamcinolone0.730.880.79 (0.38 to 1.63)