Lipson 2018.

Study characteristics
Methods	Study design: RCT, parallel group Follow‐up: 52 weeks Dates conducted: June 2014 to July 2017 Location, no. of centres: multicentre trial performed in 37 countries
Participants	Inclusion criteria: Aged 40 years of age or older Symptomatic COPD (CAT score ≥ 10) FEV1 < 50% predicted normal and a history of at least 1 moderate or severe exacerbation in the previous year; OR FEV1 50% to 80% predicted normal and at least 2 moderate exacerbations or 1 severe exacerbation in the previous year Exclusion criteria: No specified exclusion criteria Baseline imbalances: none reported. Treatment group: age (65.3 ± 8.2 years); female (1385 (33%)) Control group: age (65.2 ± 8.3 years); female (714 (34%)) N Screened: 13,906 N Randomised: 10,355 N Triple therapy: 4151 N LABA/LAMA: 2070 N Completed: 9087 (88%)
Interventions	Treatment group: triple therapy (fluticasone furoate (100 µg) + umeclidinium (62.5 µg) + vilanterol (25 µg)), once daily, administered in a single dry‐powder inhaler (Ellipta, GSK) Control group: LABA/LAMA (umeclidinium (62.5 µg) + vilanterol (25 µg)), once daily, administered in a single dry‐powder inhaler (Ellipta, GSK) Run‐in period: 2 weeks, participants continued to take their own medications. Co‐interventions: none reported.
Outcomes	Primary: Annual rate of moderate to severe exacerbations, stratified effect estimates provided by peripheral eosinophil levels (< 150 cells/µL vs ≥ 150 cells/µL) Secondary: Time to first moderate‐to‐severe exacerbation Annual rate of severe exacerbations Change in trough FEV1 Change in HRQoL: SGRQ All‐cause mortality Adjudicated cause‐specific mortality: cardiovascular, respiratory causes Annual rate of all exacerbations Change in dyspnoea: TDI Serious adverse event: composite measure Serious adverse event: pneumonia
Notes	IMPACT (NCT02164513) Study did not exclude patients with a history of asthma. Study run‐in period continued previous inhaled medications. Participants taking ICS who were randomised to LABA/LAMA therapy would have had an abrupt cessation of ICS therapy.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: “patients who underwent randomization” Comment: Probably done
Allocation concealment (selection bias)	Low risk	Quote: “randomization code will be generated by GSK using a validated computerized system. Subjects will be randomized using an Interactive Voice Response System (IVRS‐RAMOS)” Comment: Probably done.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: “double‐blind, parallel‐group”; “an independent data monitoring committee reviewed unblinded safety information at regular intervals” Comment: Probably done.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: “double‐blind, parallel‐group”; “an independent data monitoring committee reviewed unblinded safety information at regular intervals” Comment: Probably done.
Incomplete outcome data (attrition bias) All outcomes	High risk	52 weeks: 758 (18%) prematurely discontinued in intervention groups, 163 (4%) due to ‘lack of efficacy’; 566 (27%) prematurely discontinued in control group, 172 (8%) due to ‘lack of efficacy’
Selective reporting (reporting bias)	Low risk	Protocol available. All outcomes reported.