This is what a $150 subscription to the NEJM gets you: 

From the abstract:

Conclusions Augmentation of citalopram with either sustained-release bupropion or buspirone appears to be useful in actual clinical settings.

I can't be the only person who actually reads the articles and not just the titles, can I?  There has to be at least one other person?

565 Celexa failures (i.e. did not achieve remission) from the previous STAR*D trial were then randomized to  Celexa (avg dose 54mg) + Wellbutrin or Celexa + Buspar.  30 percent of the augmented patients (either Wellbutrin or Buspar) achieved remission. 

From this it is concluded "These findings show that augmentation of SSRIs with either agent will result in symptom remission."

How the hell do you conclude that?  Is it a mere coincidence that the remission rates of Celexa+Wellbutrin in this group were the same as Celexa alone in the other study (30%)-- and the same as almost every other monotherapy trial for every other antidepressant?

In other words, how can you be sure it was the combination of Celexa+Wellbutrin that got the patients better, and not the Wellbutrin alone?  What would have happened if you had given these patients Wellbutrin but taken them off Celexa?  They would have done half as well?  Are you sure?

I'm not saying that it might not be true that two drugs are better than one, I'm saying that this study doesn't show that.  If anything, this study actually supports switching as a strategy (i.e. fail Celexa, so switch to Wellbutrin)-- because two drugs are not proven here to be twice as good as one alone, but I can certainly prove they carry twice as many side effects and are twice as expensive. 

Here we have a massive expenditure of tax dollars that will undoubtedly lead to treatment guidelines that will be clinically misleading and economically wasteful.  How much did the NIMH pay for this?  And for CATIE?  I'm not a Pharma apologist, but what  was wrong with forcing Pharma to pay for their own studies which we get to pick apart?  These government sponsored studies are no better.  Gee-- the generic came out on top?
 

I don't even know what to make of this:

4041 patients show up and consent to be in a massive antidepressant trial,  and almost 25% can't even score a HAM-D of 14? (7=complete cure.)  Who are these people?  What were they thinking?

And then of the ones who actually stay to participate (N=2876), their average HAM-D is 21? For two years?

And Celexa cures a third of these patients?  Half of them in less than 6 weeks? After two years walking around HAM-D =21?  Cures?  Celexa?  40mg?  Hello? 

Remember, this is open label.  These people, who presumably have been in psychiatric treatment for a long time (mean length of illness 15 years), know that they are taking 40mg of Celexa.   Not a new experimental drug with a new mechanism of action.  Celexa.  1/3rd get cured.  After all this time.

BTW, the people who failed this Celexa study  get moved into Star-D II.  What is the relevance of this?  Well, in this study 63% were female, 75% were white, 40% were married, 87% were high school grads or greater, 56% had jobs.  It is the opposite of this demographic that is most likely not to have gotten better. 

Evaluation of Outcomes with Citalopram for Depression Using Measurement-Based Care in STAR*D.