AIR 2.

Methods	Design: randomised double‐blind controlled trial (NCT00231114) Number of participating centres: 30 from USA, Canada, Brazil, United Kingdom, the Netherlands and Australia
Participants	Inclusion criteria: Participants on stable maintenance asthma medications for at least four weeks before entry Baseline Asthma Quality of Life Questionnaire (AQLQ) score 6.25 or lower Pre‐bronchodilator FEV1 > 60% of predicted Airway hyperresponsiveness (methacholine PC20 8 mg/mL) At least two days of asthma symptoms during the four‐week baseline period Being a non‐smoker for at least one year with less than 10 pack‐year smoking history Exclusion criteria: Life‐threatening asthma Chronic sinus disease Respiratory diseases such as emphysema Use of immunosuppressants, beta‐adrenergic blocking agents or anticoagulants History in the previous year of three or more hospitalisations for asthma History in the previous year of three or more lower respiratory tract infections History in the previous year of four or more pulses of OCS use for asthma No. of randomly assigned participants: 297 (196 BT vs 101 placebo) Age (mean (SD)): BT group: 40.7 (11.9) versus control group: 40.6 (11.9) Sex (% men): BT group: 42.6 versus control group: 38.8 PC20 mg/mL (geometric mean (95% CI)): BT group: 0.27 (0.22 to 0.34) versus control group: 0.31 (0.22 to 0.43) Pre‐bronchodilator FEV1 (% predicted: mean (SD)): BT group: 77.8 (15.65) versus control group: 79.7 (15.14) Dose OCS (mg/d): BT group: 6.4 (1.97) versus control group: 5 Dose ICS (µg/d): BT group: 1960.7 versus control group: 1834.8 Dose LABA (µg/d): BT group: 116.8 (34.39) versus control group: 110.3 (26.7) Asthma severity (percentage of participants with severe refractory asthma according to criteria of the American Thoracic Society) BT group: 86% versus control group: 88% AQLQ score: BT group: 4.3 (1.17) versus control group: 4.32 (1.21)
Interventions	Intervention: bronchial thermoplasty plus medical management Control: sham bronchoscopies plus medical management
Outcomes	Main outcome: Asthma Quality of Life Questionnaire (AQLQ) score (change from baseline) Secondary outcomes: Percentage symptom‐free days (change from baseline) Total symptom score (change from baseline) Number of puffs of rescue medication used (change from baseline) Percentage days rescue medication used (change from baseline) Asthma Control Questionnaire (ACQ) score (change from baseline) Morning peak expiratory flow (change from baseline) Pre‐ and post‐bronchodilator FEV1 (predicted) Number of severe exacerbations (per participant per year) Percentage of participants experiencing severe exacerbations Respiratory‐related unscheduled physician office visits Emergency department (ED) visits (per participant per year) Hospitalisations in the post‐treatment period (percentage of participants) Days lost from work/school or other activities due to asthma (days/y)
Notes	Funded by Asthmatx Inc
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomisation sequence computer generated using minimisation
Allocation concealment (selection bias)	Unclear risk	The publication of the trial results did not describe the efforts made to ensure concealment of the randomisation sequence
Blinding of participants and personnel (performance bias) All outcomes	Low risk	The trial ensured blinding of participants by comparing bronchial thermoplasty versus a sham procedure. The control group received three bronchoscopies that involved sedation and a mimic procedure to that performed in the bronchial thermoplasty group. A catheter was deployed into the airways through a bronchoscope, an electrode array was expanded and a radiofrequency controller was activated, simulating indistinguishable audio and visual signals, but no radiofrequency energy was delivered. Duration and time intervals between groups coincided Participants in both groups were unable to guess their treatment after the first bronchoscopy, but in the second procedure, a larger proportion of participants in the bronchial thermoplasty group guessed their assigned treatment (the trial publication reported within‐group comparison P values at second bronchoscopy of 0.011 for bronchial thermoplasty group vs 0.342 for sham group) The bronchoscopy team was unblinded to participant assignments
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "All follow‐up and assessment visits were conducted by a blinded assessment team"
Incomplete outcome data (attrition bias) All outcomes	Low risk	Minimum recruitment goal of 225 participants (150 in the bronchial thermoplasty group and 75 in the sham group) described Intention‐to‐treat analysis described as participants who were randomly assigned and received at least a bronchoscopy 288 participants randomly assigned (190 in the bronchial thermoplasty group and 98 in the sham group), 278 completed the 12 months of follow‐up visits (93% of the sample) Participants lost to follow‐up: nine participants in the bronchial thermoplasty group versus one participant in the control group No participants were withdrawn from the study for asthma worsening Six participants in the bronchial thermoplasty group were withdrawn by investigators (none in the control group), and two participants in each group were withdrawn for medical reasons
Selective reporting (reporting bias)	Low risk	All outcomes described in the study protocol (NCT00231114) were reported in the main publication (Castro 2010)