Purpose: Structural progression of medial knee osteoarthritis (OA) is associated with an increased knee adduction moment (KAM), a surrogate measure of the compartmental knee loads. The KAM is thus a target for mechanical intervention. Based on real-time pressure data and individualized training thresholds, the use of plantar pressure-based auditory feedback to shift weight medially was found to reduce the KAM in individuals with medial knee OA. However, its potential mechanisms remain unclear. Thus, this study aimed to elucidate movement strategies adopted by individuals with medial knee OA in response to pressure-based feedback gait retraining and compare movement strategies of substantial and partial responders of the intervention.
Methods: Subjects with radiographic and symptomatic medial knee OA (n = 26; 85% female; 63±9 yr; 30.4±5.5 kg/m2; 11/15 KL-2/3) participated in this IRB-approved study. A wireless insole (Moticon) was used to supply real-time pressure data to a smartphone app for generating auditory feedback. All subjects completed a gait retraining session between the baseline and immediate post-training (IMD) gait test in the lab. After being instructed on subtly modifying gait to stop feedback, subjects self-perceived and adopted movement strategies by following feedback. They continued the training at home for three weeks and returned to the lab for another (W3) gait test. Subjects were categorized as substantial responders if their KAM changes at W3 were ≥6%, which was shown achievable in medial knee OA via medial weight shift. All walking trials were acquired using 3D motion capture (Qualisys & Bertec). Kinematics and kinetics (Visual 3D) were calculated relative to proximal segment. External joint moments were normalized by percent body weight and height (%BW×Ht). Mixed linear models were used to evaluate the Time, Group, and Interaction effect. Walking trials were nested within Subject, and within-subject variability over Time was modeled by an unstructured covariance matrix. Significant results were further analyzed based on the means of least square predictions. Multiple comparisons were adjusted by the Bonferroni’s method. Estimated differences [95%CI] and p-values are reported.
Results: There were 13 substantial and 13 partial responders, and their baseline characteristics and gait were not different. Ten out of the 13 substantial responders achieved ≥6% KAM1 reduction at both IMD and W3. Five out of the 13 partial responders achieved ≥6% KAM1 reduction at IMD but not W3. Their speed-adjusted KAM1 reduction were 0.32 [0.11, 0.52], 0.49 [0.27, 0.70] and 0.49 [0.28, 0.71] %BW×Ht (all p < 0.001), equivalent to 10 and 16% at IMD and W3, respectively. Four subjects discontinued after the first visit; although one subject reduced KAM1 by 7% at IMD, they were considered partial responders for this analysis. The speed-adjusted KAM1 reduction of the partial responders was 0.20 [0.07, 0.34] %BW×Ht (p < 0.001; 6%) at IMD. They showed an increased KFM by 0.31 [0.03, 0.58] %BW×Ht (p = 0.018), making their KFM at IMD 0.46 [0.00, 0.92] %BW×Ht (p = 0.048) higher than that of the substantial responders.
Substantial responders reduced speed via reducing strides at IMD and W3, whereas the partial responders reduced speed via reducing both stride and cadence at IMD. The groups employed different mechanisms to medialize their COP that shortened the lever arm to reduce KAM. For example, at IMD, the substantial responders medialized COP, which translated proximally to less rearfoot inversion, ankle inversion, and shank lateral tilt (Fig. 1). Decreased hip adduction, which suggests trunk lean, cannot be verified due to the lack of a difference in pelvis lateral tilt. This series of kinematic changes moved the knee joint center closer to the frontal-plane GRF that translated medially regardless. At W3, the substantial responders maintained the described kinematic changes, except decreased hip adduction. Meanwhile, partial responders widened their step, showed less shank lateral tilt, knee adduction, and thigh lateral tilt and an increased mediolateral GRF. These findings suggested that COP medialization was likely a consequence of wider step, and the lever arm was shortened by the altered GRF vector orientation. In the sagittal plane, the partial responders shifted their COP more posteriorly and reduced the anteroposterior GRF less than the substantial responders. Consequently, their sagittal-plane GRF acted further away from the knee joint center, resulting in an increased KFM.
Conclusions: Substantial and partial responders adopted different movement strategies in response to pressure-based auditory feedback to reduce the KAM. These findings will aid researchers to suggest strategies to subjects to effectively reduce the KAM and avoid increasing KFM that may offset the reduced KAM, improving this intervention to benefit more individuals.
