Data are presented as n, least squares mean±se or n (%), unless otherwise stated. Differences and odd ratios are umeclidinium (UMEC) versus glycopyrronium (GLYCO). ^#: analysis performed using a mixed-model repeated measures with covariates of treatment, baseline dyspnoea index score or baseline St George's Respiratory Questionnaire (SGRQ) or chronic obstructive pulmonary disease assessment test (CAT) total score (scores recorded on day 1), centre group, 24H subset flag, day, day by baseline and day by treatment interactions. ^¶: analysis performed using a separate logistic regression model at each visit with covariates of treatment, baseline dyspnoea index score or SGRQ or CAT score at baseline, centre group and 24H subset flag. Transition dyspnoea index (TDI) response was defined as patients who had a ≥1 unit TDI focal score. SGRQ response was defined as an SGRQ total score of 4 units below baseline (scores recorded on day 1) or lower. CAT response within a patient was defined as a CAT score 2 units below baseline or lower. ⁺: inference of statistical significance was drawn from p-values for treatment comparisons on these non-lung-function related endpoints and they were termed statistically significant if p<0.05. 24H: subset of patients who completed 24-h spirometry on day 1 and day 84.