Murray 2004

Methods	A randomised, double‐blind, active‐controlled, multi‐centre, parallel‐group study over 12 weeks from November 1999 to September 2000, at 33 centres in the USA. Run‐in 2 weeks, single‐blind placebo
Participants	Population: 267 adolescents and adults (12 to 73 years) with persistent asthma  Baseline characteristics: mean age 34 years; FEV₁ 66% predicted Concomitant ICS used by 0% of participants   Inclusion criteria: 12 years of age or older with a 6‐month history of asthma; must have been treated with as‐needed SABA alone during the previous month with no oral or ICS use within 1 month, or LABA within 72 hours of study entry; FEV₁ % predicted between 40% and 85%; bronchodilator reversibility by an increase ≥ 15% in FEV₁ over baseline within 30 minutes of inhalation of 2 puffs (180 μg) of albuterol  Exclusion criteria: pregnancy and/or lactation, life‐threatening asthma, hospitalisation attributable to asthma twice or more in the last year, current smoker or > 10‐pack‐year history of smoking, significant concurrent disease including a recent upper or lower respiratory tract infection. Medications prohibited before and throughout the study included inhaled, oral, or parenteral corticosteroids, theophylline or other bronchodilators, anticholinergics, leukotriene modifiers, cromolyn, and nedocromil
Interventions	• Fluticasone propionate and salmeterol 100/50 μg twice daily • Fluticasone propionate 100 μg twice daily Delivery Diskus
Outcomes	Primary efficacy variables: mean change from baseline in AM predose FEV₁ at endpoint for fluticasone propionate/salmeterol 100/50 compared to salmeterol 50; area under the serial FEV₁ curve at treatment week 12 relative to treatment day 1; baseline for fluticasone propionate/salmeterol 100/50 compared to fluticasone propionate 100
Notes	Sponsored by GSK
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Treatment assignments were generated in blocks of 6 by a computer‐based random codes system
Allocation concealment (selection bias)	Unclear risk	Not stated
Blinding (performance bias and detection bias) All outcomes	Low risk	Double‐blind
Independent Assessment of causation (detection bias) Asthma‐related events	High risk	Causation of SAEs not independently assessed
Incomplete outcome data (attrition bias) All outcomes	Low risk	228/267 (85%) completed the study
Selective reporting (reporting bias)	Low risk	Full data on GSK website