No, most studies are not so highly confounded and the results are more clear-cut. Antidepressants are tough to study - they require very careful protocols, lots of repeat studies, and different types of studies as well. Sometimes just longer-term studies to tease out the differences. Sometimes longitudinal studies, where large populations of patients are studied for very long times vs cohort populations which have different drugs or protocols.
Psychiatry is a bit of an art as a consequence. Many patients require combinations of approved drugs, tailoring of therapies to symptoms, etc. But psychiatrists see real-world patients getting real drugs over time. They see how their patients respond. We also know what happens when depressed patients stop taking their drugs if they become noncompliant, or are put on a drug challenge - where they are carefully observed after the doctor takes them off the drugs. The results are compelling: patients with major depression who quit their drugs relapse and commit suicide with increasing frequency after each challenge. By the time you have a patient who has relapsed five times, if they are left untreated as many as 95% of them will die by suicide. (!)
So, it is obviously true that 1) most antidepressants do help about 70% of patients they are tried in, at least for while and 2) antidepressant studies are highly confounded by placebos. So, medicine progresses, just not easily.
