Low sample size, high drop out rate, and mostly women. While this is interesting, each of those is reason for concern.

Only 94 started, then divided into 3 groups, only 56 completed. Now their study design required 54 to complete so this isn't fatal, but it is still concerning that the sample size is so low. It limits the conclusions that can be drawn.

High drop out is very common in all exercise programs. To the point where we should not ignore them - odds are high you will drop out to and so we need to ask what happens to drop outs (but how to ask - good luck)

Of the 56 that completed, 38 were women. The female body is different from the male body in ways that we would expect to affect low back pain (expect does not mean it does, only that we should check for it in studies), and should be noted.

This is still interesting. If you only read this study then you didn't learn enough, but fortunately there are a lot of other studies and the trend that any exercise is good seems to hold. I'm sure someone can find a meta-analysis that shows that.

That also worked for my grad school :)

A randomized trial assigned 94 adults with chronic low back pain to 12 weeks of exercise, live by videoconference with a kinesiologist or alone with pre recorded videos, or to a waitlist. Among the 57 finishers, pain fell versus waitlist by roughly 2.5 points on a 0 to 10 scale with live sessions and just over 2 with videos, each a clinically worthwhile drop of over 20%. Formats did not differ, though the trial was not designed to show equivalence. Pain at rest barely moved; what fell was movement evoked pain: pain on standing from a chair or a short walk. The sample was small, a third dropped out, the physical test assessor was unblinded, and video attendance was self reported (about 14 of 36 sessions versus 19 live). Why women reported less relief than men remains open.