Those meta-analyses compare CBT to other common forms of psychotherapy, including the psychoanalytic kind (these days usually called “psychodynamic” therapy). The general picture is this: the effects of CBT are essentially the same as any other kind of psychotherapy. They all reduce depression symptoms, and they all still seem to work up to a year later (this particular kind of meta-analysis has to assume all the trials are comparable, though — and that’s often quite a big assumption).

It’s a very similar story for drugs: the meta-analyses show that essentially all forms of antidepressant work better than placebo. But these effects are likely a bit overblown by all the dodgy practices in the scientific literature. And the evidence for one antidepressant being substantially better than another is, to use the kind of language one often sees in the review studies, “limited” (which means researchers only have the vaguest clue).

Although this is good news in one sense, it’s worrying (even depressing) in another. There is evidently a gaping hole in our evidence base on treating depression. If essentially all the major therapy types work to the same degree, despite being based on entirely different — often opposing — principles, it’s pretty difficult to pin down exactly why they work. What exactly are the therapists doing in their sessions that makes the treatment effective? Can we really say that Beck was correct about CBT being the best treatment if other forms of therapy, which take an entirely different approach, can do the same job?