de Boeck 2017.

Study characteristics
Methods	Double‐blind, randomised, phase 2, cross‐over study. Duration (total): 24 weeks (2x 8‐week cross‐over periods, separated by an 8‐week washout period). Location: multicentre (39 centres) in UK, France, Belgium, Canada, Germany, Switzerland, Italy, Netherlands.
Participants	95 children or adolescents aged 6 ‐ 17 years with CF were randomised and 84 completed the study. Inclusion criteria: aged between 6 and 18 years with a confirmed diagnosis of CF and FEV1 (% predicted) between 30% and 90%. Mean age (SD): 12.0 (3.0) years (reported for 92 children). Gender split: 55 females (60%); 37 males (40%) (reported for 92 children). Mean (SD) FEV1 (% predicted) at screening: 72.23 (11.6), range (38.3 to 89.9). Use of dornase alfa and maintenance antibiotics (if established 3 months before screening) were permitted during the study. 63 participants (68.5%) had previously used or were using dornase alfa at screening. Other CF therapies except nebulised hypertonic saline were also permitted during the study. Exclusion criteria: failing the mannitol tolerance test, design to identify participants with bronchial hyper‐reactivity.
Interventions	Treatment: inhaled mannitol 400 mg 2x daily. Control: 10 mg of non‐spray dried, non‐respirable raw material mannitol. Interventions administered 2x daily in a cross‐over fashion. Treatment duration: 8 weeks for each treatment arm, separated by an 8‐week wash‐out period.
Outcomes	Primary outcome: FEV1 % predicted (absolute change from each treatment period baseline). Relative change from each treatment period baseline of FEV1 % predicted was also measured. Secondary outcomes: FVC % predicted (absolute and relative change from each treatment period baseline); FEF25-75% predicted (absolute and relative change from each treatment period baseline); safety (physical examination and adverse event data); sputum weight.
Funding Source	The study was funded by Pharmaxis Limited and an employee of Pharmaxis was among the authors. Pharmaxis participated in the study design, data collection, data interpretation and and writing of the study report.
Conflicts of Interest	The authors declare that Pharmaxis provided travel support for investigator meetings, but that no investigator received any personal funding to participate in the study.
Notes	The original planned sample size was 160 participants, calculated to have 90% power of detecting a difference of 3% absolute change in FEV1 % predicted. Recruitment was halted early after 95 participants were recruited (centres exhausted potential candidate pools). Post hoc power calculations showed that the recruited sample size had 80% power of detecting a difference of 3.5% absolute change in FEV1 % predicted.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Randomisation was conducted in a 1:1 ratio and stratified according to age (6 ‐ 11 years or 12 ‐ 17 years) and dornase alfa use (user or non‐user). No information given of how random sequence was generated.
Allocation concealment (selection bias)	Unclear risk	No information provided.
Blinding (performance bias and detection bias) Participants	Low risk	Double‐blind achieved with a placebo also containing a low dose of mannitol.
Blinding (performance bias and detection bias) Clinicians	Low risk	Double‐blind achieved with a placebo also containing a low dose of mannitol.
Blinding (performance bias and detection bias) Outcome assessors	Unclear risk	No information provided regarding blinding of outcome assessors.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	95 participants were randomised and 84 completed the study. Reasons for withdrawal were clearly documented in a flow diagram and primary and secondary efficacy endpoints were analysed on an intention‐to‐treat basis. Missing data were imputed using 'last observation carried forward;' such a method does not capture change over time so may have introduced bias into results. However, it is not stated how much data has been imputed and for how many participants, therefore the risk of bias is unclear.
Selective reporting (reporting bias)	Low risk	All important outcomes of the phase 2 efficacy and safety trial seem to be reported.
Other bias	High risk	Participants underwent a mannitol tolerance test at screening; those who failed the test or in whom the test was incomplete were not entered into the study and thus, study population included only those with CF who passed the tolerance test and not all potential participants with CF. Sponsored by Pharmaxis; authors or study staff worked for Pharmaxis or had financial interest. The study may be underpowered; a priori and post‐hoc power calculations are based on different sizes of treatment difference (3% or 3.5% change in FEV1 % predicted).