Modi 2010a.

Study characteristics
Methods	RCT. Parallel design (3 arms). Location: multicentre (20 centres) in USA. Duration: 3 years, but terminated early, duration of study participation ranged from 1.3 to 2.8 years.
Participants	166 participants with CF enrolled initially. 15 dropped out (11 from the CPT group and 4 from the flutter group) in the initial 60 days of the study with a further 41 withdrawing due to lost to follow up; lack of time; treatment preference and decrease in health. Data missing from 5 participants. Randomised 58 (31%) to PD&P, 51 (30%) to flutter and 57 (39%) to HFCWO. Gender split: 54% male. Mean (range) age: 14.2 (7 ‐ 44) years. Participants split into 86 children (7 ‐ 12 years), 44 adolescents (13 ‐ 17 years) and 36 adults (over 18 years). Mean FEV1 88.2%.
Interventions	Flutter versus HFCWO versus PD&P. Flutter: self‐administered in 3 stages ‐ (1) loosening and mobilisation breaths (2) mucus mobilisation and (3) expectoration. PD&P: treatment administered by caregiver using a wedge and consisted of positioning, percussion (vibration ) and forced expiratory technique with coughing between 6 positions; after each position participants instructed to do 3 forced expiratory technique and cough. Each treatment was 2x daily
Outcomes	Rate of decline in FEV1, time to need for antibiotics for pulmonary exacerbations, use of other pulmonary therapies, participant satisfaction, adherence, quality of life. Measurements of satisfaction were recorded before and after study and every 4 months with phone diary, but no identification of type of activities outlined in this abstract. ‐ details in online supplement
Notes	This study ID refers to the flutter versus PD&P section of the study. 166 participants enrolled and a total of 56 withdrew (15 before Day 60, 41 after Day 60). 110 left in at early study termination and those who withdrew after Day 60 included in ITT analysis (n = 151). Funded by Hill‐Rom. Sample size calculation undertaken (60 participants per group) to detect a difference between annual rates of decline in FEV1 of 2% predicted.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Electronic randomisation stratified by age.
Allocation concealment (selection bias)	Unclear risk	Not discussed.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Not discussed.
Incomplete outcome data (attrition bias) All outcomes	High risk	Drop outs apparent over the 3 abstracts but reasons not discussed. 15 dropped out and data missing from 5 participants. 130 provided adherence data in the 2006 abstract, but other abstracts and main papers describe166 participants. Uneven drop outs across treatment arms and age groups led to early termination.
Selective reporting (reporting bias)	Low risk	Measurements of satisfaction were recorded before and after study and every 4 months with phone diary, these were identified as effectiveness, convenience, comfort, and overall satisfaction. Satisfaction with the therapy was an independent predictor of withdrawing.
Other bias	High risk	Study supported by Hill‐Rom, Inc and the CF Foundation.