Primary, secondary and additional efficacy endpoints (modified intent-to-treat population; efficacy estimand unless stated otherwise)

BFF MDI 320/10 μgBFF MDI 160/10μgFF MDI 10 µg
Patients, n619617607
Primary endpoint
 Change from baseline in morning pre-dose trough FEV1 (mL) at week 12#
  LSM difference versus FF MDI (95% CI)34 (9–60)32 (7–57)
Secondary endpoints
 Change from baseline in morning pre-dose trough FEV1 (mL) at week 12 (attributable estimand)#
  LSM difference versus FF MDI (95% CI)50 (24–76)44 (18–70)
 Time to first moderate/severe COPD exacerbation
  Patients with exacerbation n (%)220 (35.5)223 (36.1)241 (39.7)
  HR versus FF MDI (95% CI)0.827 (0.688–0.995)0.803 (0.668–0.966)
 SGRQ responder rate at week 12+
  Responders n (%)315 (51.98)331 (54.00)262 (43.96)
  Difference versus FF MDI % (95% CI)7.55 (1.67–13.43)11.05 (5.18–16.91)
 Change from baseline in mean daily rescue medication use (puffs·day-1) over 12 weeks#
  LSM difference versus FF MDI (95% CI)−0.32 (−0.56 to −0.08)−0.29 (−0.53 to −0.05)
Additional endpoints
 Rate of moderate/severe exacerbations per year§
  Adjusted rate of exacerbations+/−SE0.93±0.0730.98±0.0761.39±0.106
  Rate ratio versus FF MDI (95% CI)0.67 (0.54–0.82)0.71 (0.58–0.87)
 Time to first moderate/severe COPD exacerbation in patients with a history of ≥2 COPD exacerbations in the last 12 months§
  Patients with exacerbation, n (%)93 (39.4)97 (40.2)106 (45.7)
  HR versus FF MDI (95% CI)0.783 (0.590–1.039)0.799 (0.603–1.059)

BFF: budesonide/formoterol fumarate dihydrate; MDI: metered dose inhaler; FF: formoterol fumarate dihydrate; FEV1: forced expiratory volume in 1 s; LSM: least squares mean; HR: hazard ratio; SGRQ: St George's Respiratory Questionnaire; COPD: chronic obstructive pulmonary disease. #: analysed by repeated measures linear mixed model; : analysed by Cox regression; +: analysed by logistic regression model; §: analysed by negative binomial regression. Rate of exacerbations per year=total number of exacerbations / total years of exposure across all patients for the treatment. ƒ: p-values for this endpoint are considered nominal since these analyses were not under Type I error control.