West 2010.

Study characteristics
Methods	RCT. Parallel design. Location: single centre in Australia. Duration: at least 10 days.
Participants	23 children and adolescents with CF admitted to hospital for IV antibiotics for a respiratory exacerbation (as defined by Wood 2002). Needed previous experience at home with any PEP device. 1 from acapella group was discharged early on Day 6, so only 10 analysed in that group. Age mean (SD) range: PEP 13.5 (3.3) 7 ‐ 18 years; acapella 10.4 (2.2) 7 ‐ 13 years. Sex: PEP 9 females, 3 males; acapella 8 females, 2 males, the gender of the one participant from the acapella group who was discharged early was not identified in the paper. FEV1 % predicted mean (SD) range: PEP 74.67 (19.8)%, 56% ‐ 114%; acapella 58.9 (23)%, 29% ‐ 95%. Exercise performance (m) mean (SD) range; PEP 798.3 (233.6), 390 ‐ 1100 m; acapella 576 (293.7), 290 ‐ 1200 m.
Interventions	PEP mask (n = 12) versus acapella (n = 11). 2 supervised treatment sessions each day for a 10‐day period. Treatment was standardised to consist of 10 sets with the allocated device in a sitting position. Each set consisted of 10 breaths through the device followed by one or two huffs and cough. The pressure settings for the device were standardised for the study to provide between 15 and 20 cm H₂O of positive pressure. All participants received concurrent IV antibiotics; any other treatment was in accordance with direction from a respiratory physician who was not aware of the treatment allocation of participants.
Outcomes	Lung function, exercise performance (modified shuttle walk test), wet weight of sputum and satisfaction questionnaire. Outcomes measured prior to randomisation and after 10 days.
Notes	Sample size calculation undertaken (18 participants per treatment arm needed to detect a 10% change in FEV1).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	36 pieces of paper (18 PEP and 18 acapella) were put in double‐sealed envelopes and a research assistant (who was not involved with recruitment, assessment, or treatment) withdrew 1 envelope, determined group allocation, and then discarded the envelope.
Allocation concealment (selection bias)	Low risk	36 pieces of paper (18 PEP mask and 18 acapella) were placed in double‐sealed envelopes and for each participant a research assistant (who was not involved with recruitment, assessment, or treatment) withdrew one envelope to determine group allocation
Blinding (performance bias and detection bias) All outcomes	Low risk	Outcome assessors were blinded for lung function and modified 10‐metre shuttle test.
Incomplete outcome data (attrition bias) All outcomes	Low risk	22 out of 23 participants completed the study; 1 participant was discharged home on day 6 for home IV treatment and was not available to complete the 10 days of treatment.
Selective reporting (reporting bias)	Low risk	Limitations of the study were identified ‐ specifically smaller than planned sample size due to changes in clinical practice. This impacts the power of the study to detect an effect if one exists and may have contributed to lack of statistical differences between the intervention groups.
Other bias	Unclear risk	There appeared to be differences at baseline for age, FEV1, and exercise performance. The PEP mask group was older, had a higher FEV1, and could cover more distance in the 10‐metre shuttle test. It was noted that parents were allowed to assist their child in completing the satisfaction questionnaire.