First, and most importantly, the answer is no. No single thing can cause a complex behavior such as suicide (or violence), and to say that a drug, or event, or illness made someone do something is simplistic and, well, wrong.
A more refined question would be whether antidepressants (or anything else) can contribute to the suicide process, move someone further along. Cocaine doesn't cause violence, but if you see a violent guy on cocaine, it's best to get out of the way.
The FDA, in 2002 and 2004, issued warnings about suicidal ideations in children who take antidepressants. Here are a couple of points for us to consider.
- These were kids
- The point of discussion is suicidal ideation, not suicide (death) per se.
The authors of a recent review of 29 studies found a similar small increase in risk. Importantly, however, no actual suicides occurred. They also analyzed whether or not the antidepressants were actually helpful to people, and weighted that finding against the risk of suicide. Their results determined the risk/reward ratio to be about 11:1, meaning that patients have 11 times better chance of being helped than hurt by antidepressants.
I should point out that, in my opinion, while there is absolutely no compelling evidence that these drugs promote suicidality, there is also a lack of compelling evidence indicating that they reduce suicidality. Nor would I expect there to be. Again, these drugs may change how someone feels, but there are a million steps between a feeling and an action. Which has the greater impact on suicidality: medications or the availability of a gun?
Asking the question simply, "do antidepressants cause suicide?" misses the meat of the argument. Cause? In whom? When, and under what circumstances? The real question is whether someone who is depressed is more likely to commit suicide off of the medicines than on them.
So while we ponder that question, let me offer a sobering statistic. Only 30% of the people who committed suicide had seen a psychiatrist in the past year. That means 70% hadn't.
Source: JAMA. 2007;297:1683-1696.