Dahl 2010.

Methods	Design: double‐blind, parallel, randomised, controlled trial. 52 weeks' duration. Additional inhaled bronchodilators other than albuterol discontinued. Modified intention‐to‐treat analysis performed. Origin of participants not stated Run‐in: 2 weeks
Participants	Population: 1732 participants with a diagnosis of moderate to severe COPD were randomly assigned. Mean age was 63 years. Recruitment was predominantly from Europe, Russia and the UK. In the indacaterol 300 mcg, indacaterol 600 mcg, formoterol, and placebo arms, 55.6%, 53.2%, 50.9% and 51.9% of participants were taking concomitant ICS, respectively. Mean FEV1 was between 50% and 52% predicted in all arms of the study Inclusion criteria: age > 40, smoking history > 20 pack‐years, FEV1/FVC < 0.7, postbronchodilator FEV1 30% to 80% predicted Exclusion criteria: respiratory tract infection or hospitalisation in previous 6 weeks, oral corticosteroids or change in ICS in previous month, diagnosis of asthma
Interventions	1. Indacaterol 300 mcg 2. Indacaterol 600 mcg 3. Formoterol 12 mg 4. Placebo
Outcomes	Primary endpoint: 24‐hour postdose trough FEV1 after 12 weeks, active medication compared with placebo. Other outcomes: Transitional Dyspnoea Index (TDI), use of as needed salbutamol, St George Respiratory Questionnaire (SGRQ), BODE index (body mass index, obstruction, dyspnoea, exercise), safety and tolerability Follow‐up on days 1, 2, 15, 29, 84, 85, 113, 168, 197, 253, 364 and 365 Values reported at baseline and at weeks 12 and 52
Notes	Study was supported by Novartis, and some authors were Novartis employees. Novartis directly supplied data for: Indacaterol versus placebo (trough FEV1, quality of life, dyspnoea, peak FEV1, number of participants experiencing at least one exacerbation, and mortality); Indacaterol versus LABA (trough FEV1, quality of life, dyspnoea, peak FEV1, number of participants experiencing at least one exacerbation).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Participants were randomly allocated to treatment using an automated interactive system
Allocation concealment (selection bias)	Low risk	Allocation via automated interactive system, with both participants and investigators blinded to allocation; double‐dummy technique
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blinded trial
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Outcome assessors were blinded to treatment allocation
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Reasons for dropout across treatment and control arms reported. Higher dropout rate in placebo arm. Efficacy data from 6 sites excluded on the basis of "non‐conformance with good clinical practice"
Selective reporting (reporting bias)	Low risk	All prespecified outcomes reported, not just statistically significant outcomes