It seems to be a very mixed situation, without one single correct answer to give. It isn't as black and white as 'antidepressants work' or 'antidepressants don't work'
Rather, it seems highly dependent on how depressed the subject is. And whether it is clinical severe depression, or great sorrow. There is a difference. People who are unhappy for a life-related reason, due to circumstances, don't seem to respond much more than placebo, if at all, whereas for the very severely depressed patient, who will be depressed even if somebody came over and waved a magic wand over their heads and set their life to be easy and in every way perfect in circumstance, such people can still be deeply depressed, even if not necessarily unhappy, at least bar due to the depression itself there can be much more likelihood of success.
When there is a biochemical basis to the depression is when people seem to respond.
I was reading some meta-analyses (analyses of not studies, but the results averaged across groups of studies) yesterday, and here is what I came up with:
http://ahrp.org/jama-antidepressant-meta-analysis-reveals-22-years-of-deception/https://www.nimh.nih.gov/about/directors/thomas-insel/blog/2011/antidepressants-a-complicated-picture.shtml This one is interesting, it makes a mention of the relatively new ketamine therapy, although their description of the effects only lasting days, contradicts every experience report from a firsthand patient and most other resources that I have ever read about it. Most say effects last months, at the longer end, and a good few weeks to a bit over a month at the short end. Although at least they got it right about the way it can help those with severe and treatment-resistant depression as a maintenance therapy administered as single doses spaced far apart and do so dramatically, in a matter of hours, not the many weeks expected for SSRIs, IF the latter work, or the likes of tricyclics (in the case of tricyclics, BTW it seems like clomipramine might be the best of them), and lacks the great danger associated with MAOIs, especially those of the irreversible type, which I have seen reports of categorizing them as some of the most effective of the antidepressants. Unfortunately there are severe dietary restrictions, some of them are hydrazines [hydrazine compounds, based off the parent compound hydrazine itself, N2H4, a highly toxic relative of ammonia, often used in rocket fuels, are often as not pretty toxic, and a hydrazine linkage in a drug usually makes for a covalent-binding, irreversible attachment of the drug to its target. Often pretty nasty little fuckers. Its a group usually avoided in drug design for that reason.]
And for someone who has an accidental violation of the dietary and medicinal restrictions with the older irreversible MAOIs especially, the results can be fatal, and even if not, then a really, really seriously unpleasant, traumatic way to get fucked over. And if the likes of serotonin syndrome or an adrenergic hypertensive crisis does kill someone, its one of just about the most unpleasant ways to go that I can think of.