Tuesday, October 24, 2006

Depakote for Mania: Small Beans

In the latest Journal of Clinical Psychiatry, there is a trial of divalproex (Depakote ER) for mania. These patients were all hospitalized, so we’re talking about a relatively severe group of patients. The 21-day study used the following measures: Mania Rating Scale (MRS), Manic Syndrome Scale (MSS), and Behavior and Ideations Scale (BIS). The authors noted that there was a statistically significant difference found between Depakote and placebo on all three measures.

But wait, how much of an effect are we talking about? Did Depakote outshine placebo by a large margin? One is left wondering about this point, as the authors did not report the effect size of the difference between groups. I calculated the effect sizes (d) and here is what I found for each measure: MRS: .229; MSS: .233; BIS: .235. Remember that the generally used criteria are d = .2 (small), d = .5 (medium), d = .8 (large). According to conventional standards, then, we’re talking small beans here. An advantage for Depakote that was very small in size. Depakote also resulted in a significant increase in body weight (1.8 kg) compared to placebo (.5 kg). 1.8 kg in three weeks? Keep gaining weight at that rate over a year and you’re looking at Zyprexa!

The authors acknowledged no ghostwriters, but considering that four of the authors work for Abbott (the maker of Depakote), I suppose they didn’t need to hire out to find a friendly writer. How else do you explain that there is no discussion of the wimpy gains in comparison to placebo? Ahhhh, when industry and science collide...

4 comments:

a retired scientist/practicioner said...

If you are truly "an academic with a respectable amount of clinical experience," then you would know that this trial was an FDA Stage III Trial initiated for the purpose of earning the drug a new indication, and as such BY DEFINITION includes some of the company's research and development scientists as authors. These trials costs several millions of dollars to conduct, and the company which owns the drug patent has to fund the study (unless you would like the NIMH to fund drug companies' clinical trials).

Also, as an academic, you should know that effect sizes are simply a metric of change scores divided by standard deviation, and are thus not a reliable standard of "clinical significance" of the observed effect, as the value is dependent upon the degree of variability among subjects. Also, effect sizes are not a standard by which the FDA grants new indications for medications. Over 35 years of international clinical experience have substantiated the fact that valproate products are extremely effective in treating bipolar mania. The only question being addressed in this trial seems to be one of whether or not this "extended release" delivery system of valproate has an impact on the efficacy of the active ingredient (the efficacy of which has been established for decades now).

There is plenty of room to scrutinize, question, and debate the impact of commercial interests on science and the practice of medicine. However, your complaints regarding this particular study are the "small beans." It would be nice if every outcome study addressed the issue which is important to me as an individual, but that is no basis upon which to judge the merit of the study. The basis for judgment, in my opinion is this: what did the study intend to demonstrate? Did it state this objective clearly? Did it accomplish this objective? For the current study, it passes the test.

What I have stated is my educated opinion. I do not mean it as a personal slight against you, the blogger.

CL Psych said...

Thanks for your post. I'm glad to reply to your concerns. First, yes it was an FDA Phase III trial. There is no doubt that corporate employees were authors, which in and of itself is not necessarily a problem. But having such a small effect size and not disucssing it is, in my opinion, a problem. Indeed, most journals these days claim that they would like authors to report effect sizes. This could have been done in the present study, but it was not. My suspicion is that the corporate authorship had something to do with the lack of effect size reporting, though I suppose any author could have made the same decision.

I am aware that the FDA does not consider size of effect in approving applications. I didn't say the medication should not have been FDA approved. I said that the effect size was very small, and I stand by that assertion.

When you say that valproate is "extremely effective," I do not see the data from this study to support that assertion. You seem to be saying that effect size is a poor measure of clinical importance, but what else do you propose to use? Surely you are not suggesting that we just rely on "international clinical experience" alone?

In a clinical trial, I venture to guess that the opinion of most scientists would be that the size of the effect should be reported. In this case, it was not. I thus calculated the effect and put it in the context of standard interpretations of effect size. It is up to readers to decide if my interpretation is adequate.

I greatly appreciate your comment, as informed public debate is often sorely lacking on such topics. Should you have more to add, please feel free to do so.

Alone said...

Hi. I'd like to offer a slightly different criticism of this same paper.

On the face of it, this is a paper about the efficacy of Depakote ER in mania. But what is the context? Look at it like a movie, or a novel. Huckleberry Finn wasn't a story about kids, and it wasn't written in 1945; it had a context, and that context was that slavery is for idiots. So what's this paper's context?

"Medications are or are not good for mania, and we have to study them to find out which are. We can't assume that just because Depakote is good, Depakote ER will be. That's not rigorous. That's not science. We have to do a trial."

That's a magician's distraction, it's smoke. The real context is this:

"Currently approved treatments of the acute manic phase of bipolar disorder can be categorized primarily as mood stabilizers (e.g. divalproex sodium, lithium, and carbamazepine) or as atypical antipsychotics."

That's a quote from the paper, BTW.

"Deconstruct" the paper. It's not about the efficacy of Depakote ER, it is really about 1) the idea that seizure drugs are mood stabilizers (and other things aren't); 2)certain people are in a better position to tell us what is true, and so we should defer to them.

Everyone "knows" you need a therapeutic Depakote level. Everyone "knows" higher doses have better efficacy vs. increased side effects.

But these things turn out not to be true. But papers like this one add volume to the papers "for" seizure drugs as mood stabilizers-- and the important consequence is that no one ever has to go back and look at the first principles. If Bowden said it, it must be true.

Self-promotion: do me a favor, read my post on the same paper. If my point is not clear by the end of it, I'll go away quietly.

http://thelastpsychiatrist.com/2006/11/because_i_said_so.html

CL Psych said...

Alone,

I read your comment and the post on your site. You did an excellent job picking apart the article and its references. I was focused on a different topic in my original post, but I see your point that the authors were, at one point in the paper, citing research indicating that dose-pushing was a way to increase efficacy, when the data did not really support the assertion. The whole "mood stabilizer" idea has generally been overblown, especially for the anticonvulsants -- shouldn't they show some sort of prophylactic value before we get all excited about them?

I've read a few other posts on your site. Very thoughtful and insightful. I certainly agree that psychiatry is very much into labeling, to the point where the labels become more important than the data. I’m nearly certain that these labels come straight from the marketing departments of drug companies. “Mood stabilizer” and “broad spectrum psychotropic agent” didn’t just show up in someone’s breakfast cereal one morning!