I was involved in this sort of stuff for years (writing about it, not taking the stupid things). I launched two antidepressants, using enough “clinical data” to sink a ship, but I also had to ignore a number of studies where placebo groups did better than those taking SSRIs. A study done at the University of Toronto showed CBT (cognitive behavioural therapy) lit up the same areas of the brain as antidepressants. One clinician threw in an interesting wrench when he suggested that raising serotonin might be working—not because it’s raising serotonin—but because it’s lowering dopamine. Conclusion? We know diddly about the brain, and the only reason doctors prescribe SSRIs for anxiety is because a few controlled trials showed some relief (totally sideline to the study itself). Then again, that’s how a lousy blood pressure medication became the darling of the erectile disfunction category (I worked on that, too!). Spend enough time around clinical trials, and you’ll realize antidepressants shouldn’t be used for more than a year. Basically, they haven’t been studied on patients for longer (not in large groups). By the way, meta-analysis is useless. It combines small sample studies with different criteria, then concludes that the number is significant. It isn’t. It’s like sticking parts from four different model cars together and saying you’ve got a hybrid.