Ironically, this happens because the medicine is working. FTA:

> When a person’s depression starts to lift, he or she may feel less hopeless and helpless. That sounds like an improvement, but when people feel less helpless but still feel depressed, they may think about suicide as a way out, whereas before they were too immobilized to make a suicide plan. For that reason, a decrease in the symptoms of depression can increase the risk of suicidal thoughts or actions.

While plausible, that paragraph is speculation; it's also plausible that antidepressants push some subgroup of patients toward suicidal ideation.

I'm unaware of any study that answers this question definitively, but I'm interested, so if anyone sees something useful, please chime in.

Not arguing either way, as I don't think I am knowledgeable enough on the subject to do so, but I feel like there is a lot of potential to confuse correlation and causation. It simply could just be that "children and adolescents taking antidepressants" are already in a high risk group that is 3 times as likely "to have suicidal thoughts or to attempt suicide", and antidepressants reduce that number to "twice as likely".

Right and this is where plausible deniability comes in from the point of view of the pharmaceutical companies. Drug studies are problematic because nothing happens in a vacuum.

This is, of course, the very first thing any reputable study tries to control for; presumably the placebo and active populations are both taken from populations with comparable diagnoses.