Potentially this is quite a positive result. A couple of issues...

Potentially this is quite a positive result. A couple of issues that I can see.

1. The trial design talks about a 10 point drop in the KOOS pain component from baseline. This scale ranges from 0 to 36. I assume NRS 4-6 are people with a KOOS pain threshold from 14 to 21. I am not sure why they then changed to a 50% drop in KOOS pain in the analysis. According to the KOOS website 8-10 is the minimum important clinically important change for pain is 8-10. A 50% drop for those in NRS 4 & 5 would be less than 8-10 points. I would really like to see the raw KOOS numbers.

2. There were only 112 people in this study and 126 were recruited. What happened to the 14 missing and what effect did this have on the statical power of the study.

3. Knee pain has a large placebo response and the response p-value was only 0.026. This is statically significant, but only weak. There is a reasonable chance this could be due to chance alone.