The sample size of 28 for each group was determined from a power analysis assuming a 20% attrition rate. According to
Cohen’s calculation, an ! level of .05, low effect size of 0.40, and power of 0.70 were achieved through repeated-measures analysis of variance, yielding 20 subjects. After considering the attrition rate, 28 samples for each group were needed, and these subjects were alternately assigned to the control (n = 24) and experimental (n = 24) groups in order of registration. According to patients’ daily diaries, within the foot bath group, 3 subjects withdrew from the study with leukocytopenia and high fevers, and 1 withdrew because of dizziness. Within the foot massage group, 3 subjects withdrew because of leukocytopenia and high fevers, and 1 from noncompliance (G75%). The mean compliance rate at home was 91.5% in the foot bath group and 95.4% in the foot massage group.