It seems to me that many of the drugs in this category are effective sometimes for some people. People I have known who were suffering acutely often cycled through several drugs, sometimes in combination, dosages etc. It appears to me to be mostly trial and error.

But for some people the results are miraculous. They solve real problems with few if any side effects. For others, they either don't work at all or have intolerable side effects.

The situation would be a whole lot better if we could predict likely efficacy and side effects in advance, especially cases where treatment is more likely to do harm that to help.

Interestingly, I've read a lot (not so much recently) on the subject and taken several SSRIs myself (with good short term experience, minor sides), but I learned a lot from the discussion here. So many people take these drugs — the data necessary to understand and perhaps to predict ... is out there. Not easy to get at, but really plentiful. Reading this made me think about the feasibility of mining message boards for first person accounts of SSRI experience.