Aitken 2012.

Study characteristics
Methods	Double‐blind, randomised parallel trial. Duration: intervention for 26 weeks, followed by 26‐week open‐label treatment with mannitol. Multicentre: 53 sites. Location: USA, Canada, Argentina, Europe.
Participants	305 participants with CF: mannitol group (n = 184) and control group (n = 121). Mean age: 20 years, range: 6 ‐ 53 years, 154 aged < 18 years. Mannitol group: 19% aged 6 ‐ 11 years; 30.4% aged 12 ‐ 17 years. Control group: 19.8% aged 6 ‐ 11 years; 32.2% aged 12 ‐ 17 years. Gender split: 51.5% male, 48.5% female. All participants clinically stable at start of study, mean (SD) baseline FEV1 % predicted 63.8 (15.9). 75.1% received concomitant dornase alfa. Participants receiving hypertonic saline excluded.
Interventions	Treatment: inhaled dry powder mannitol, 400 mg 2x daily, 10 x 40 mg capsules. Administered using RS01 inhaler model 7, Plastiape, Italy. Control: subtherapeutic mannitol 50 mg 2x daily, 10 x 50 mg capsules. Administered using RS01 inhaler model 7, Plastiape, Italy (same inhaler as for active intervention).
Outcomes	Primary outcome: FEV1 (absolute value over the 26‐week study). Secondary outcomes: FEV1 % predicted at 26 weeks; FVC; FEF25-75; PDPE; sputum weight; HRQoL (as assessed by CFQ‐R, including burden of treatment and assessment of symptoms provided by Pharmaxis). Other outcome measures: adverse events and sputum microbiology (also additional non‐routine antibiotic use and hospitalisations provided by Pharmaxis)
Funding Source	Supported by Pharmaxis Limited.
Conflicts of Interest	No specific conflicts of interest of the authors of the manuscript reported. Stated that the study sponsor participated in the study design, data collection, data analysis, data interpretation, and writing of the report. After completion of the trial, the data were held and analysed by the sponsor.
Notes	Additional data supplied by manufacturer of mannitol (Pharmaxis) for the majority of outcomes reported in this review. Pharmaxis stated that the lung function results in the paper were presented as per the primary analysis method also described in the publication. The FEV1 and FVC data were presented in full, however they acknowledged that in this study the outcome FEF25-75 was mentioned in the methodology and not presented in the results section.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Further details were provided by Pharmaxis: the master randomisation list, stratified by country and dornase alfa user (yes/no) for a parallel design was prepared using SAS Version 8.1 by an external company. 300 randomisation numbers (180 active and 120 control) were generated for each country and each dornase alfa user/non‐user group. Randomization blocking by country was done in paired blocks of 5, 1 block for dornase alfa users and 1 block for non‐users.
Allocation concealment (selection bias)	Low risk	Pharmaxis confirmed randomisation managed via an IVRS, therefore the investigator was unaware prior to randomising the participant which specific blinded pack of treatment they would be allocated to. This provided an extra level of security (over and above the blinded nature of the study) against selection bias.
Blinding (performance bias and detection bias) Participants	Low risk	Described as double blind ‐ mannitol and low‐dose mannitol control administered as capsules identical in taste and appearance with identical methods of administration.
Blinding (performance bias and detection bias) Clinicians	Low risk	Described as double blind ‐ mannitol and low‐dose mannitol control administered as capsules identical in taste and appearance and with identical methods of administration. Pharmaxis confirmed investigators and study staff, including statisticians and all outcome assessors at investigator sites e.g. spirometry technicians were blinded. Both Bilton 2011 and Aitken 2012 used the same low‐dose mannitol control that was identical in taste and appearance to the 400 mg mannitol active intervention.
Blinding (performance bias and detection bias) Outcome assessors	Low risk	Described as double blind ‐ mannitol and low‐dose mannitol control administered as capsules identical in taste and appearance and with identical methods of administration. Pharmaxis confirmed blinding of investigators and study staff including statisticians and all outcome assessors.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Higher dropout rate with mannitol, 17% versus 12% for control, for adverse events and other reasons e.g. withdrawal of consent. However, paper provides flow diagram with timing and reasons for drop out and which group the participants were in. Sensitivity analyses conducted by Pharmaxis (methods of imputation of missing data for withdrawals) showed a consistent treatment effect in favour of mannitol and no change to conclusions.
Selective reporting (reporting bias)	Low risk	Limited information was reported in the study publication; particularly HRQoL and lung function. Additional data were provided by Pharmaxis on request for all primary outcomes of this review and many secondary outcomes.
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, the participant population included only those with CF who passed the tolerance test and not all potential participants with CF. Sponsored by manufacturer of mannitol (Pharmaxis); authors or study staff worked for Pharmaxis or had financial interest.