Vermeersch 2019.

Study characteristics
Methods	Design: randomised, double‐blind, placebo‐controlled, parallel‐group design Duration: 13 weeks Location: 5 centres across Italy
Participants	Population: 301 adults with COPD (GOLD stages B to D) were randomly assigned to azithromycin (n = 147) or placebo (n = 154) Baseline characteristics: age (mean years): 66.5 (SD 9.5); % male (mean): 51; % FEV₁ (mean, pre‐bronchodilator): 0.925 L; pack‐years (mean): 44; exacerbation history: in the previous 12 months; 89 participants = 1 AECOPD exacerbation, 78 = 2 AECOPD exacerbations, 50 = 3 AECOPD exacerbations, 84 = ≥ 3 exacerbations Inclusion criteria: history of severe COPD diagnosed with a pulmonary function test and tracheostomy Exclusion criteria: allergy to macrolides, life expectancy < 1 year
Interventions	Azithromycin: loading dose 500 mg once daily for 3 days, followed by 250 mg every 2 days for 13 weeks Placebo Allowed co‐medications: not reported
Outcomes	Reduction in number of exacerbations, time to first exacerbation, reductions in number of hospitalisations, time to first hospitalisation, quality of life, SAEs, AEs
Notes	Funding: not reported Identifier: not reported
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	This was not reported in the actual pre‐publication document but in the protocol. Randomisation was performed using an online randomisation schedule; a unique randomisation code was obtained through a secured web‐based programme, as reported on clinicaltrials.gov. We assume that they carried out what they intended
Allocation concealment (selection bias)	Low risk	This was not reported in the pre‐publication document but in the protocol. It was reported on clinicaltrials.gov that they intended to use identical packaging, labelling, schedule of administration, and appearance. We assume that they carried out what they intended to do as stated in their protocol
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Study was reported as double‐blind in the pre‐publication document; at clinicaltrials.gov, this study was reported as triple‐blind, but this was not mentioned anywhere else
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Study was reported as triple‐blind at clinicaltrials.gov, but it is unclear whether or not outcome assessors were investigators. This is not mentioned anywhere in the pre‐publication document, but we assume that they carried out what they intended to do as stated in their protocol
Incomplete outcome data (attrition bias) All outcomes	Low risk	Overall attrition was 13%. Azithromycin group had 11% attrition, placebo group had 16% attrition. In the PRISMA diagram at 90 days, dropout, withdrawal, and mortality were similar, so we considered attrition as low. We requested further information about how attrition was handled in the ITT analyses
Selective reporting (reporting bias)	Low risk	All outcomes were reported according to the protocol
Other bias	Low risk	This study was stopped early due to slow participant recruitment, but this would not affect the overall analysis