It's my own belief that the data relating SSRI's to suicide risk represents two phenomena:
1) pt's given antidepressants w/o proper instruction often experience a rather immediate lifting of mood in the first couple days which represents a placebo effect based on the possibility that something will now get better. Then, when this wears off a few days later, pt's are crestfallen and are likely to think, "See, even this didn't work. Nothing will make me feel better. This will go on forever," and this may lead to increased suicide risk. Explaining this to pt's may be why I've never had a pt initiated on antidepressants c/o new or increased SI in the first weeks.
2) the data used for the FDA warning basically shows that pts given SSRI's had greater chance of SI and suicidal behavior - which probably represents the fact that those with more severe depression were more likely to receive antidepressants.
At least, that's how I fit it into my little world.
I do tell this to pt's and I explain that carefully done research published after the warning shows antidepressants LOWERING suicide risk.