Trunk lean over the stance limb during gait has been linked to a reduction in the knee adduction moment, which is associated with joint loading. Differences were examined in knee adduction moments and frontal plane trunk lean during gait between subjects with knee osteoarthritis and a control group of healthy adults. Additionally, subject variability in human motion data presents a challenge to researchers when trying to detect differences between subject groups. The individual differences in neutral posture between subjects is a source of variation in joint angles. A method was developed using principal component analysis (PCA) to objectively reduce this inter subject variability.
Gait analysis was performed on 80 subjects (40 osteoarthritis). Models were developed to define lateral thoracic tilt, as well as pelvic tilt. The trunk and pelvis frontal plane angles were used to describe trunk lean and pelvic tilt. Angles were calculated across the stance phase of gait. We analyzed the data, (i) by extracting discrete parameters (mean and peak) waveform values, and (ii) using principal component analysis (PCA) to extract shape and magnitude differences between the waveforms.
Osteoarthritis (OA) subjects had a higher knee adduction moment than the control group (α=0.05). Although the discrete parameters for trunk lean did not show differences between groups, PCA did detect characteristic waveform differences between the control and osteoarthritis groups. The data show that subjects display similar waveform shapes, however waveforms vary in magnitude, suggesting a variation in posture between subjects. The results from the PCA reveal that the first PC, which captures the most variation in the data, represents this variation in magnitude. The second PC describes a significant difference in range of motion between the subject groups.
Subjects with knee OA were found to have a different range of motion of their pelvis and trunk than control subjects. These changes are consistent with a strategy to lower the knee adduction moment. As an alternative to conventional subjective methods, PCA should be employed to reduce inter subject variability in order to ensure objective analysis in human motion waveform data.