METHODS: Twenty stroke patients from conventional rehabilitation (CR) (n = 10) and RR (n = 10) groups were recruited through a purposive sampling method. Patients in the CR group received a two-hour session of a five-day-a-week home-based CR program for 4 weeks. Patients in the RR group received a five-day-a-week of an hour combined physiotherapy and occupational therapy session and a one-hour robotic therapy session using the HAL® Cyberdyne lower-limb, for 4 weeks. The mid-thigh circumferences of both limbs were measured on day 1 (baseline), week 2 and week 4 of rehabilitation program.

RESULTS: The results revealed no statistically significant difference in the mid-thigh circumferences between the paretic (F1.05,9.44 = 1.96, p = 0.20), and the normal (F1.05,9.44 = 1.96, p = 0.20) sides in the CR group (n = 10). For the comparison between the paretic and normal sides in the RR group (n = 10), the paretic mid-thigh circumferences revealed significant time effect results (F2,18 = 11.91, p = 0.001), which were due to changes between baseline and week 2, and baseline and week 4 measurements. Interestingly, the normal mid-thigh circumferences also revealed a significant time effect (F2,18 = 6.56, p = 0.007), which is due to changes between baseline and week 4. One-way analysis of variance was employed to compare the mean average between groups due to the difference in the baseline measurements of the mid-thigh circumferences between the paretic side of the CR and the RR groups. With this adjustment, the average means mid-thigh circumferences after 4 weeks of therapy were shown to be significantly different between the CR and RR groups (F1,18 = 12.49, p = 0.02).