First, antidepressants work. The study found antidepressants had odds ratios from about 1.5 to 2, which means people on antidepressants were more likely to recover from depression than people who were not on antidepressants.

Second, the effect size was generally small. In terms of a statistic called “Cohen’s d”, it only reached 0.32 for the average antidepressant. What does that mean in real life? It depends on the exact distribution being looked at; but for comparison, a diet pill with an effect size of 0.32 would equate to about 9 pounds. It’s not zero, but it’s not miraculous either.

Third, some antidepressants did better than others. In terms of efficacy (how well the medicine worked), the older antidepressant amitriptyline seemed most effective, followed by newer antidepressant mirtazapine. In terms of tolerability (how few side effects there were), the leaders were agomelatine (a European antidepressant not available in this country), and fluoxetine (better known as Prozac). Few medications did well in both categories simultaneously, although a few of the SSRIs like escitalopram (Lexapro) and paroxetine (Paxil) had generally respectable performances. But the error bars (the uncertainty on each result) are so wide that it’s hard to really draw any firm conclusions about one being better than another.

How does this study change how we treat depression? The relatively small effect size should be disappointing to people who expect antidepressants to work miracles. But it’s important to remember that these are averages. Each drug worked very well for some people, didn’t work at all for others, and on average had a small positive effect. We know that people respond to medications differently. An antidepressant trial is always just an educated guess on what will work for any particular patient. If one doesn’t work, another one might. So the effect size for a rational regimen of trying drugs-- rejecting the ones that don’t work, and continuing the ones that do-- will be higher than the effect size for any individual drug. So the first lesson from this study is to be willing to try different things.

The second lesson is to pay attention to each individual patient’s needs when looking for a tradeoff between medication strength and side effects. Amitriptyline, the antidepressant which this study found to be most effective, isn’t for everyone. While it’s undeniably a powerful medication, it can also cause weight gain, sleepiness, dry mouth, and a host of other side effects. And even though this study found fluoxetine was well-tolerated and rarely caused major problems, it wasn’t very optimistic about its ability to treat the toughest depression cases. The point is less “here’s the best antidepressant” and more “here are the tradeoffs that each antidepressant involves”. This study helps us quantify those tradeoffs better, but it’s still a question of how many (and which) side effects any given patient is willing to accept.

People who are trying their first antidepressant medication and worried about side effects may want something more tolerable, like fluoxetine. And people who haven’t done well on antidepressants before, aren’t worried about side effects, and just want whatever’s most likely to work may want something more like amitriptyline. If you have a preference like this, talk to your doctor about your concerns and what you want out of medication to see if you can make a treatment plan that balances your needs.