Two force plates (Bertec, USA) were used to record the ground reaction forces (GRF). Seven optical tracking rigid plates with each consisting inhibitor of four markers and one shell were designed according to the body characters and were attached to the lateral aspects of the feet (bilateral instep), shanks (bilateral surface of tibia), thighs (bilateral surface of the thigh), and pelvis (over the center point between both posterior superior iliac spines), respectively. With each foot on one force plate, the subjects performed the squat exercises under three conditions: (1) both feet pointing straight ahead (neutral squat), (2) hip adduction and 30o of feet adduction (squeeze squat), and (3) hip abduction and 30o of feet abduction (outward squat).
The subjects were required to perform each activity from an initial upright position with the feet shoulders width apart, the arms in 90o of shoulder flexion and elbows extension. They were also instructed to maintain the feet in the initial position during the exercise. At a low descending speed, the subjects squatted down until the thighs were parallel with the ground and then in a continuous motion ascended back to the upright position. For each subject, six successful trials were recorded. Local coordinate systems were defined for the foot, shank, thigh and pelvis segments through digitized palpated bony landmarks. The bony landmarks included left/right ilium anterior superior, left/right prominence of the greater trochanter external surface, left/right femur lateral/medial epicondyle, left/right fibula apex of lateral malleolus, left/right tibia apex of medial malleolus, left/right dorsal aspect of first metatarsal head, and left/right dorsal aspect of fifth metatarsal head.
These local coordinate systems enabled the calculation of the floating axis angles at the knee joint (Grood and Suntay, 1983). The raw kinematic data were smoothed using a fourth-order zero lag digital Butterworth low pass filter with cut-off frequency at 6 Hz. Three-dimensional (3D) joint angles, moments and forces were calculated in the Visual 3D software (C-Motion Inc., Rockville, MD, USA) based on the subjects�� lower limbs length, body mass and ground reaction force. The forces and moments were normalized to body weight (BW) and percent of body weight times height (% BW��Ht), respectively.
Statistical Analysis To generate ensemble graphs, data throughout a squat cycle were normalized to 101 points GSK-3 (0%�C100%). The average measures of every subject were obtained from six trials, and then these individual data were averaged for all subjects. Nonparametric Wilcoxon sign-rank tests were performed using SPSS. Statistical significance was set at p<0.05. Results Compared to the neutral squat, the outward squat demonstrated an offset towards varus at the knee, whereas the squeeze squat displayed a valgus offset during approximately 10�C90% of the squat cycle (Figure 1).