Sample size and power calculation
The 6-minute walk test (6MW), which is commonly accepted as the most sensitive outcome measure in pulmonary rehabilitation, was employed for sample size estimation. A change of 54m in 6MW was identified as the minimal significant difference between group means (minDIFF). The
standard deviation (SD) of the measure was reported as 57m according to our past experience. Setting the at 0.05 (twotailed) and the power at 80%, the minimum sample size for each arm was 19 according to the following formula: [1.960.84]2 * 2 * SD2 /minDIFF2. To compensate for the estimated 30% attrition of subjects, the sample size was adjusted to be 27 subjects for each arm. Based on this estimation, 80 patients were eventually recruited.