INNOVATE.

Methods	Randomised, double‐blind, parallel‐group, multi‐centre, placebo‐controlled trial. Blinding: matched placebo. Methods of allocation not reported. Randomisation stratified by concomitant asthma treatment and country of origin
Participants	N = 482. Mean age: omalizumab: 43.4; placebo: 43.3. FEV1: omalizumab: 61; placebo: 61.6; rescue medication usage: omalizumab: 6.6; placebo: 5.5. Overall AQLQ: 3.9 (both groups); serum total IgE: omalizumab: 197.6; placebo: 189.6; ICS dose (BDP equivalent, mcg/d): omalizumab: 2359; placebo: 2301. All participants receiving high‐dose ICS + LABA. 22% receiving maintenance oral steroids Inclusion criteria: +ve skin prick test to ≥ one aeroallergen; serum IgE: 30 to 700 IU/mL; severe persistent asthma requiring > 1000 BDP or equivalent and LABA treatment; FEV1 40% to 80%; FEV1 reversibility ≥ 12% post SABA; ≥ two exacerbations requiring OCS in previous 12 months or one severe exacerbation resulting in hospitalisation Exclusion criteria: smokers/smoking history of ≥ 10 pack‐years; treatment for exacerbation four weeks before randomisation; use of methotrexate/gold salts/troleandomycin/cyclosporin within three months of first visit; prior omalizumab treatment
Interventions	Subcutaneous omalizumab (0.016 mg/kg per IU/mL) (plus usual care) versus placebo (plus usual care). Study duration: 28 weeks; run‐in phase: seven‐day screening period; eight‐week run‐in phase. Follow‐up: 16‐week (data not presented). During initial four weeks of run‐in phase, medicines adjusted to achieve best control. No further adjustments permitted in last four weeks of run‐in
Outcomes	Exacerbatrions (requiring OCS); hospitalisation; emergency room treatment; lung function; AQLQ; adverse events
Notes	Jadad score: 4. Imbalance between groups at baseline for primary outcome in the trial. Greater instance of exacerbations requiring oral steroids in omalizumab group compared with placebo group. Adjusted data were extracted and entered
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Randomisation stratified by concomitant asthma treatment and country of origin
Allocation concealment (selection bias)	Unclear risk	Methods of allocation not reported
Blinding (performance bias and detection bias) All outcomes	Low risk	Double‐blind
Incomplete outcome data (attrition bias) All outcomes	Low risk	52 participants did not complete the trial, and reasons for withdrawal are included in the trial report
Selective reporting (reporting bias)	Unclear risk	No apparent indication of selective reporting bias