The sample size was determined based on an analysis of
test power before the study began using the Minitab 15
software program (Statistical and Process Management Software for Quality Improvement). The two-sample t-test
was used for the power analysis. The parameters used were assumed mean differences (2), assumed standard deviation (3•2), alpha (0•05) and power level (80%). The results showed that the sample size (n) must be 42 for each group. Based on the expected drop-out rate, inclusion of 45 patients in each group was planned. However, we closed the present study once we had recruited 65 patients (32 in the intervention group and 33 in the control group) at the end of 14 months due to the small number of available patients with no disease diagnosis additional to COPD.