Extremely disappointing analysis in the article and also the cited paper's abstract [0]. The only way this data could possibly hold any value is if they found a way to control for how depression might influence an individual's likelihood of choosing to participate in the trial in the first place, as well as trial completion rates.

Even an amateur could read the headline and instantly understand this critical point the experiment's design, and yet it's not even acknowledged under the "Risk of bias" section.

[0] https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD...