A recent article in the major, prestigious medical journal Lancet concluded that
"Clinically important differences exist between commonly prescribed antidepressants for both efficacy and acceptability in favour of escitalopram and sertraline. Sertraline might be the best choice when starting treatment for moderate to severe major depression in adults because it has the most favourable balance between benefits, acceptability, and acquisition cost."
Here is a link to the abstract: http://www.thelancet.com/journals/lancet/article/PIIS0140-6736(09)60046-5/fulltext
Predictably the news headlines about this article read something like this: "Zoloft and Cipralex better than other drugs"
Taking a closer look at the data, as presented in the abstract, reveals the following:
1) mirtazapine (Remeron) was actually the most "efficacious" drug of all the drugs studied. But its side-effect profile/tolerability was less favourable than some of the others.
2) mirtazapine, escitalopram (Cipralex), venlafaxine (Effexor), and sertraline (Zoloft) were all quite similar in terms of "efficacy", and were all significantly superior to duloxetine (Cymbalta), fluoxetine (Prozac), fluvoxamine (Luvox), paroxetine (Paxil), and reboxetine.
The study is impressive, in that it was a meta-analysis including the data from over 25 000 patients.
But the study is substantially weakened by the fact that it does not look at long-term outcomes (over a year or more).
As I've written before, I feel that the best assessments of effectiveness for conditions such as depression, which recur over a period of years, require data that also cover a period of years, rather than just months.
Oddly, the findings about bupropion are not mentioned in the abstract.
The paper is further weakened by not looking at tricyclics at all.
I think the results of the study should not be overvalued. The study may reasonably guide a first choice of antidepressant, though. A few very particular points to take from this study are that venlafaxine was not shown to be dramatically superior to all other antidepressants (despite what their advertising has inferred quite often), also that the new antidepressant duloxetine is clearly not dramatically superior either (which encourages us to be wary of the marketing hype behind it -- see my previous entry on Cymbalta).
Another result from this study confirms an observation I've had in my practice, that mirtazapine (Remeron) can be a very good antidepressant, provided its side effects can be tolerated (sedation and weight gain).
The authors wisely note that the study ought not to prompt someone to change a medication that is working well for them. The study measures differences between groups; for a given individual sometimes one particular medication can work best (e.g. fluvoxamine, duloxetine, or fluoxetine), even if it is not the most effective for a group.