There is a battle underway in the genteel circles of academic psychiatry. The disputed question is: Are antidepressants (ADs) good or bad for patients with bipolar disorder?

The major figureheads in this drama are respected psychiatrists on opposite coasts. In the pro-AD corner, weighing in with an endowed chair and full professorship at UCLA, we have Lori Altshuler, M.D.; and in the anti-AD corner, weighing in with an assistant professorship at Harvard Medical School, we have Nassir Ghaemi, M.D.

Even as the contenders are lacing up their gloves, the crowd remains pretty confused.

There are two issues here:

1. Do antidepressants lead to manic switching, and if so, do they differ in switching risk?


2. Do antidepressants in some way worsen the overall course of bipolar disorder, by causing rapid cycling or in some other way?

The influential APA Treatment Guidelines for Bipolar Disorder are in the Ghaemi corner, being antidepressant-shy. They recommend avoiding antidepressants as first line treatment for bipolar depression, and prefer either Lamictal or lithium. If you must use an antidepressant, the APA endorses a very short list of relatively safe ones: Paxil and Wellbutrin.

But Team Altshuler has strong momentum since the recent publication of a definitive review of the issue arguing that ADs are just fine for bipolar depression. Let’s go through its findings in some detail.

Researchers from the U.K. and the Netherlands teamed up with the statistical superstars at the Cochrane Collaborative to conduct a “systematic review” of randomized controlled trials of treatments of bipolar depression (Am J Psychiatry 2004; 161:1537 - 1547). One of the advantages of a systematic review is that the results of different studies are crunched in a specially designed statistical program that gives more weight to studies with larger numbers of subjects.

The researchers identified 12 studies in the world psychiatric literature that met criteria of being well-designed, randomized controlled trials focusing on the use of antidepressants in patients with bipolar depression. After having done more statistics than is truly healthy, they reported four major findings:

• Finding #1: Patients with bipolar depression who are put on antidepressants do better over a four to 12 week time frame than patients put on placebo, in terms of both response rates and remission rates.


• Finding #2: Patients on SSRIs have the same switch rate to mania as patients on placebo.


• Finding #3: Tricyclics seem to cause somewhat more manic switching than SSRIs.


• Finding #4: Wellbutrin’s “lower” risk of manic switching is a myth: The two studies upon which the APA based this recommendation did not involve bipolar patients at all, but rather patients coded as “manic-depressive, depressed” in DSM-II, which translates to “unipolar depression” in DSM-IV language!

So, is the case closed? Hardly. Here are some of the problems with interpreting these findings.

First of all, the trials endorsing the effectiveness of ADs were pretty short--no longer than 12 weeks. Ghaemi’s position is that antidepressants tend to worsen the course of bipolar disorder over the long term, perhaps by eventually leading to rapid cycling (see TCR Vol. 1, No. 8, for our interview with Ghaemi on this topic).

Secondly, these were all welldesigned randomized controlled trials. Wait--that’s supposed to be an advantage, right? Not always. RCTs are typically designed to measure one specific thing, efficacy, and are not statistically powered to actually compare rates of side effects, one of which is the rate of switching to mania. Thus, the fact that these well-designed studies show no SSRI vs. placebo differences in switch rates does not carry the statistical legitimacy that you might imagine (see Ghaemi et al, Int J Neuropsychopharm 2003; 6:303-308). On the other hand, the lack of switch difference certainly doesn’t lend any support to the anti-AD camp.

At the end of the day, it certainly appears that the anti-AD camp is increasingly under siege. An oft-cited report (Altshuler et al, Am J Psychiatry 2003; 160:1252-1262) was an observational study in which 1000 patients with bipolar disorder in community treatment were simply followed over time in order to see what meds they end up on and how they did.

Of the 1000 or so enrolled, 189 patients had received an adequate trial of antidepressants (defined here as at least two months), and 84 (or 44%) of these had a good antidepressant response. This very select group of 84 was observed further – 43 of them stopped their ADs within six months, and their relapse rate was 70%. The other 41 continued their ADs for at least six months, and they did much better, with a relapse rate of only 36%. At the end of the study, a total of 15 patients became manic, but the majority of these (nine) were patients who had been taken off antidepressants, again casting doubt on the idea that ADs lead to manic switching.

While this study is usually spun as endorsing the long term use of ADs in bipolar depression, the quality of the data is not compelling. Patients in the two groups may have differed from each other in important ways, but since these differences cannot be controlled for in an observational study, we’ll never know.

So what’s the bottom-line? Unclear, but the tide of opinion is certainly shifting away from current APA guidelines. Antidepressants, especially SSRIs, are less verboten in bipolar depression than formerly thought.

TCR VERDICT: AD’s in bipolar disorder:Time to be less squeamish.