Author's response to reviews Title: Laterally wedged insoles in knee osteoarthritis: do biomechanical effects decline after one month of wear? Authors:

conclusions. This sentence has been reworded as suggested. Background: add Barrios et al 2009 The paper by Barrios et al has been added to the list of clinical trials on wedged devices. Background: state the hypothesis The hypothesis has now been stated at the end of the introduction section. State why excluding Grade 1 and 4 participants Grade 1 and 4 participants were excluded in keeping with the selection criteria for the larger RCT in which these participants were enrolled. This has now been stated in the methods. Clarify directionality of knee alignment exclusion Valgus knee alignment >185 degrees was the exclusion criterionthis has now been clarified in the methods section. As stated above, this was enforced as it was a selection criteria for the larger RCT in order to ensure that participants did not have predominant lateral compartment disease (in which cases lateral wedges are unlikely to be beneficial). Please provide actual durometer value for insoles. Please add a picture of the device as a figure. The durometer reading has been added to the manuscript as requested. A photograph of the insoles has also been added. What was done with the tibial torsion measure? A further sentence has been added to the methods section describing what was done with the tibial torsion measure. Methods, gait analysis: 2 nd -to-last sentence needs “peak” added The word “peak” has been added as suggested. Discussion: Why were the reductions only 4.2-5.1%? These are smaller..... Our reductions in the adduction moment of 4.2-5.1% are only marginally smaller than the studies cited by the reviewer. However we have added a short paragraph to the discussion highlighting and discussing this as requested. Why were pain and disease severity not used as covariates? As this was a within-subject design and radiographic disease severity would remain stable amongst individuals over the one month time frame, disease severity was not used as a covariate. Similarly, any immediate effect of the wedges on pain was expected to be similar within subjects at each time point, thus we also did not include pain as a covariate. Given that our results show similar reductions in the knee adduction moment at each time point, our data support this premise that any effect of pain on the moment is systematic. Had our data shown differential effects on the knee adduction moment over time, it is possible that results could have been due to changes in pain rather than wedge degradation, and inclusion of pain as a covariate would have been warranted. Add error bars to the interaction plots Error bars have been added to the interaction plots as suggested. We hope you find these revisions satisfactory and look forward to your response soon.