Szafranski 2003.

Methods	Design: randomised, double‐blind, placebo‐controlled, parallel‐group, multi‐centre study Duration: 12 months (+ 2 weeks run‐in period) Location: 89 centres from 11 countries
Participants	Population: 812 participants were randomised to formoterol (201), budesonide (198), formoterol/budesonide combination (208), and placebo (205) Baseline characteristics Age (mean years): form 63, bud 64, form/bud 64, pbo 65 % Male: form 76, bud 80, form/bud 76, pbo 83 % FEV1 predicted: form 36, bud 37, form/bud 36, pbo 36 Pack‐years (mean): form 45, bud 44, form/bud 44, pbo 45 Inclusion criteria: males and females aged 40 and older; symptoms for 2+ years; history of at least 10 pack‐years Exclusion criteria: history of asthma or seasonal rhinitis before 40 years of age; relevant cardiovascular disorders; use of beta‐blockers; current respiratory tract disorders other than COPD or any other significant diseases or disorders; requiring regular use of oxygen therapy; exacerbation during run‐in
Interventions	1. Formoterol 12 bid (LABA) 2. Budesonide 400 bid (ICS) 3. Formoterol/budesonide 9/320 bid (LABA/ICS) 4. Placebo (PBO) Inhaler device: Dry powder turbuhaler Allowed co‐medications: terbutaline (0.5 mg) as reliever. Disallowed medication included parenteral steroids, oral steroids, antibiotics and nebulised treatment from 4 weeks before; inhaled steroids from 2 weeks before; inhaled long‐acting beta2‐agonists from 48 hours before; inhaled short‐acting beta2‐agonists from 6 hours before; other bronchodilators from 6 to 48 hours before
Outcomes	St George's Respiratory Questionnaire (SGRQ), COPD exacerbations, FEV1, vital capacity, morning and evening PEF, diary card data
Notes	Funding: AstraZeneca Identifier(s): AZ SD‐039‐CR‐0629
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A total of 812 patients were randomised [no other details, industry sponsored]
Allocation concealment (selection bias)	Unclear risk	No details provided
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Double‐blind [presumed subject and investigator]
Blinding of outcome assessment (detection bias) All outcomes	Low risk	No details provided. Inlcuded outcomes unlikely to be affected by detection bias
Incomplete outcome data (attrition bias) All outcomes	High risk	Withdrawal high and uneven between groups (formoterol 31.8%, placebo 43.9%). An intention‐to‐treat analysis was used.
Selective reporting (reporting bias)	High risk	Quality of life [primary] stated as outcome but not reported in enough detail to include in meta‐analysis.