Looking at the study, randomization and blinding both are good. I don't see a predicted difference, which would determine the number of subjects needed to 'power' the study. Lacking that, this looks to be potentially underpowered. Yes, they got published, but many studies that are just hypothesis-generating are as well. There's also a potential confounder: self-generated urolithin from ingested polyphenols.

Also, it's JAMA, so a clinical (in the sense of 'find what works') focus rather than a research one. OTOH, they're human-focused.

I see they included subjects with controlled T2DM - which, without getting into details, tends to lower mitochondrial flux. But there are only five, so not significant.

Disclaimer: I've got a stat background, but am very much not a bio stat guy - I probably missed something important (and would love getting 'schooled' on this by someone who is)