Read the actual study...just because there was a small, statistically significant numeric improvement in pain score and functionality, the changes did not exceed the minimally important difference (MID) in either scale.
To recap the results:
In case you didn't notice, .69 is less than 1, and 2.04 is less than 5. The authors even go one step further to calculate a second probability that the results were clinically relevant, and the results are terrible: 11.9% and 8.5% to achieve the MID, respectively.
So the improvements are clinically inconsequential IMO.
And I didn't chose just one study. It's the largest, most recent meta analysis on the topic, comprising data from ninety-six RCTs including 26 169 participants. It's the best data we have, despite its limitations.
I don't care to dissect each of the smaller trials you mentioned, as I know better data doesn't exist. I'm sorry the results contradict your worldview, but there was no misrepresentation of the data I linked.