At this point this is all about semantics, but correlation does in fact work both ways. But anyway, I see where the whole thing's coming from now, they looked only among suicide cases, so obviously they can't draw any conclusion about those who did not commit suicide.