Seemungal 2008.

Study characteristics
Methods	Design: randomised, double‐blind, single‐centre, placebo‐controlled, parallel‐group design Duration: 52 weeks Location: 2 outpatient chest clinics at 2 hospitals in the UK (London)
Participants	Population: 109 adults with moderate to severe COPD were randomly assigned to erythromycin (n = 53) or placebo (n = 56) Baseline characteristics: age (mean years, SD): 67.1 (8.5); % male (mean): 63; % FEV₁ predicted (mean, SD): 49.9 (18); pack‐years (mean): 51.6; exacerbation history: 35% of participants had ≥ 3 exacerbations in the previous 12 months Inclusion criteria: severity of COPD was moderate to severe; FEV₁ between 30% and 70% predicted; mean FEV₁ (L): 1.27 (erythromycin) and 1.36 (placebo) Exclusion criteria: history of asthma; bronchiectasis; neoplasia; unstable cardiac status (including prolonged QTc and arrhythmias); macrolide allergy or history of abnormal liver functions
Interventions	Erythromycin (250 mg 2 times a day) Placebo Allowed co‐medications: inhaled steroids, no changes were made unless there was a clinical need, which excluded the participant from the study. No antibiotics or oral steroids during 1‐month run‐in period of exacerbation‐free symptoms
Outcomes	Number of people with an exacerbation, exacerbation frequency, time to first exacerbation, spirometry and inflammatory markers, bacteriology, adverse events
Notes	Funding: British Lung Foundation Identifier: NCT00147667
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer‐generated permutated block random sequence generation was carried out
Allocation concealment (selection bias)	Low risk	Randomisation numbers were stored in sealed envelopes
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Placebo and erythromycin were concealed in identical capsules
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Unblinding occurred only after data entry
Incomplete outcome data (attrition bias) All outcomes	Low risk	All outcomes/dropouts were explained in a CONSORT diagram
Selective reporting (reporting bias)	Low risk	All pre‐specified outcomes were reported
Other bias	Low risk	No other bias was identified