Methods	Parallel-group, multicentre study (90 centres in US). Study had 2 treatment periods: fixed dosing during period 1 (up to week 12), and variable dosing in period 2 (weeks 12 to 24) Jadad quality score = 3
Participants	Stable asthmatic patients 12 years and over % ELIGIBLE OF SCREENED POPULATION: Not reported % RUN-IN PATIENTS RANDOMISED: 558/1596 = 35% RANDOMISED: 558 (Salmeterol 50 bid + ICS: 281; ICS + placebo: 277) WITHDRAWALS: S + ICS: 35; ICS: 42 AGE mean (range): 39 (12 to 77) GENDER (% male): 42 BASELINE % PREDICTED FEV1 mean: 80.5 BASELINE DOSE OF ICS (mean): Not reported by treatment groups BDP 400 to 800 mcgs/day; triamcinolone acetate 1200 to 1600 mcgs/day; flunisolide 1000 to 1500 mcgs/day; FP 440 to 660 mcgs/day ASTHMA DURATION: Not reported ATOPY (%): Not described SMOKING STATUS: Not reported ELIGIBILITY CRITERIA: Asthma for at least 6 months and treated with medium dose ICS for at least 30 days before screening; baseline FEV1 of 65% to 95% normal; >= 12% improvement from baseline in lung function following inhaled bronchodilator; best FEV1 within +/− 15% of the best predose FEV1 obtained at screening; no more than 1 nighttime awakening requiring albuterol and fewer than 18 puffs albuterol during the previous week EXCLUSION CRITERIA: Pregnancy and/or lactating; life-threatening asthma; asthma hospitalisation within 3 months of screening; change in asthma regimen 30 days before screening; significant concurrent diseases including recent URTI; systemic corticosteroids in 30 days before screening CRITERIA TO ENTER RUN-IN PERIOD 2: 20% decrease from the screening visit pre-dose FEV1; >= 20% decrease from the mean morning baseline PEF on any one of the 7 days immediately preceding the visit; total symptom score of >= 8 during any week before run-in visit; 2 or more nighttime awakenings due to asthma requiring treatment with albuterol during any week period before run-in visit CRITERIA FOR RANDOMISATION FOLLOWING RUN-IN PERIODS: Patients who regained asthma control following step-up treatment during run-in period 2; best FEV1 >= 65% of predicted and >= 15% of the best predose FEV1 obtained at visit IA
Interventions	LABA + ICS vs INCREASED dose of ICS OUTCOMES: Reported at 1, 4, 8 and 12 and 24 weeks (only those at 12 weeks used) DOSEOPTIMISATION PERIOD: 10 weeks; patients commenced on FP 250 mcg bid or equivalent for 2 weeks. Stable patients had their dose dropped to FP 100 bid. Patients who became unstable on this were eligible to continue. FP 250 bid was commenced and patients stable on this were eligible to continue INTERVENTION PERIOD: 12 to 24 weeks (data at 12 weeks used) TEST GROUP: (FP 100 bid + S 50 bid) salmeterol 50 mcg bid + fluticasone propionate 100 bid CONTROL GROUP: Fluticasone propionate 250 bid DEVICE: Diskhaler NUMBER OF DEVICES: 1 COMPLIANCE: Monitored during study CO-TREATMENT: prn SABA
Outcomes	PULMONARY FUNCTION TEST: FEV1; am PEF; pm PEF SYMPTOMSCORE: Mean change in daily symptom score (score of 0 to 5 Likert scale) FUNCTIONAL STATUS: Exacerbations (defined as any worsening of asthma that required asthma medication beyond blinded study drugs or albuterol); rescue medication use; nocturnal awakening; % rescue-free days; % symptom-free days INFLAMMATORY MARKERS: Not assessed ADVERSE EFFECTS: Reported WITHDRAWALS: Reported Primary outcome measure: proportion of patients who remained in study after 12 weeks and who did not withdraw due to lack of efficacy
Notes	Full-text publication. Additional data available from GSK trials register Funded by Glaxo Wellcome Confirmation of methodology and data obtained User-defined number: 600 (1000-400)
Risk of bias
Item	Authors’ judgement	Description
Adequate sequence generation?	Yes	See Appendix 3
Allocation concealment?	Yes	See Appendix 3
Blinding? All outcomes	Yes	Identical inhaler devices
Incomplete outcome data addressed? All outcomes	No	Last observation carried forward: “All analyses for efficacy and safety were conducted through use of the intent-to-treat population (all randomised patients). All data from patients who were withdrawn from the study early were included in the analyses, data available up to the time of study discontinuation being used. No interpolation was used for missing data.”
Free of selective reporting?	Yes	OCS-treated exacerbations available on request from GSK