SALTO.

Methods	This was a randomised, open‐label, phase IIIB, multicentre study with a parallel group design. Patients were treated with either SMART i.e. Symbicort® Turbohaler® 160/4.5μg/inhalation (delivered dose), 1 inhalation b.i.d. plus as‐needed (in response to symptoms), or conventional best practice. The study consisted of the following periods: 2‐week run‐in period followed by a 26‐week randomised treatment period. Usual therapy used in run‐in period. A total of 194 centres in Belgium and Luxembourg participated in this study, between December 2004 and June 2006
Participants	Population: 908 adults and adolescents were randomised. All were analysed for efficacy and safety and 867 completed the study. 38% classified as moderate persistent asthma, 36% severe persistent asthma and 27% mild persistent asthma. Mean ICS daily dose during run‐in 579 µg/day (range 100 to 2000). Inclusion criteria: Male and female, adolescent (≥ 12 years of age) and adult patients with persistent asthma, currently treated with IGCS or IGCS and LABA.
Interventions	Investigation medication was Symbicort® Turbohaler® 160/4.5μg/inhalation (delivered dose), 1 inhalation b.i.d. + as‐needed in response to symptoms. Comparators were conventional best practice, active stepwise individualised treatment according to the GINA treatment guidelines.
Outcomes	Primary variable Time to first severe asthma exacerbation Secondary variables Number of severe asthma exacerbations Mean use of as‐needed medication Prescribed asthma medication Peak expiratory flow (PEF) Patient‐reported outcomes (PROs) Asthma Control Questionnaire (ACQ) score Satisfaction with Asthma Treatment Questionnaire (SATQ) score Safety SAEs Discontinuation due to AE(s) Definition of severe exacerbation Not specified
Notes	Twenty patients reported a total of 20 SAEs during treatment (9 in the SMART group and 11 in the current best practice group). Six patients discontinued treatment due to an SAE/AE [4 in the SMART group (including two patients who died : one suicide and one myocardial infarction with no relation with the treatment) and 2 in the current best practice group].
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomization code assigned from a computer generated randomisation schedule" Demoly 2009
Allocation concealment (selection bias)	Low risk	"Patients were randomised, strictly sequentially...using coded envelopes. When a patient had been randomised, the envelope was opened and the code was revealed." Demoly 2009
Blinding (performance bias and detection bias) All outcomes	High risk	Open‐label study
Incomplete outcome data (attrition bias) All outcomes	High risk	27/450 (6%) on SiT and 14/458 (3%) discontinued treatment
Selective reporting (reporting bias)	Low risk	Data have been obtained for all primary outcome measures