Figure 3.

Incidence rate ratios (IRRs) with 95% CIs for the association between increasing airway responsiveness and the development of COPD_LLN, stratified by history of smoking (top panels), and history of asthma/asthma-like symptoms (bottom panels) (Airway responsiveness according to the slope at the first examination, with slope <0.44% × µmol^-1 as the reference category. Adjusted for study, sex, education, FEV₁ predicted, age, body mass index, second-hand smoking and occupational exposures to vapours, gas, dusts or fumes. Also adjusted for history of asthma/asthma-like symptoms, history of active smoking (when these are not used for stratification) and for lifetime pack-years (only analysis on smokers, top right panel; data on pack-years not available for 133 subjects). Number of subjects: never smokers, 1670; smokers, 1944; without asthma/asthma-like symptoms, 2890; with asthma/asthma-like symptoms, 852. P_int is the P value for the interaction between slope group and stratification variable. The y-axis is on log₂ scale).