ACNP Report: Efficacy -- A Closer Look
To begin the series on Antidepressants: Efficacy/Safety, which at this point will focus predominantly on youth (but that can change at any time), let’s begin with the American
College of Neuropsychopharmcology Report, focusing primarily on its conclusions and commentary surrounding the efficacy of antidepressants for depressed youth.
The ACNP Report. As evidence mounted suggesting that SSRI’s were linked to suicidal ideation in youth, the ACNP decided to analyze the data on the issue and provide a report on both the efficacy and safety of these medications in treating depression in children and adolescents. As many of you may have seen, the ACNP issued a preliminary report in 2004 followed by a final report in March 2006, which you can access here if you’d like. They concluded that fluoxetine (Prozac) is a safe and efficacious treatment for youth depression. There is indeed some evidence to support this idea, such as the TADS study (discussed later), though this study has serious methodological shortcomings that weaken its conclusions. However, it is indeed curious that fluoxetine would be the lone medication to show efficacy for youth depression, given its strong chemical similarity to other SSRIs.
A curious conclusion: The ACNP authors believe that the data “…suggest that it is the SSRI action combined with the absence of other pharmacologic effects such as on the noradrenergic system that it crucial for an antidepressant effect in children and adolescents. Tricylcic antidepressants and other new-generation non-SSRI antidepressants are therefore not recommended as a first-line treatment for depressed youth. (p. 478).”
Why this is curious: Let me get this straight. Data show that the SSRIs sertraline (Zoloft), citalopram (Celexa), and paroxetine (Paxil) all yield responses that range from slightly superior to placebo to not superior to a placebo. So somehow it must be the specific “SSRI action” that is doing the trick? WHAT TRICK? The medications are no more efficacious than a placebo, so why are the authors trying to describe WHY SSRI’s are more effective than a placebo?
Conflicts of interest: Well, fortunately, I think I have an answer! The initial ACNP report (2004) was ghostwritten – hat tip to Timothy Scott on this one (Ethical Human Psychology & Psychiatry Summer 2006) – by an agency called GYMR. The final report does not acknowledge the use of the firm (which of course does not mean the firm did not assist in its preparation). Instead it acknowledges the help of Miriam Davis, who is indeed an accomplished medical writer, and like virtually all of the authors of the task force report appears to have received significant funding from pharmaceutical companies at various times. Across the board, the conflicts of interest are mammoth. Ten of eleven authors reported receiving funding from pharmaceutical companies.
Explanation of failed studies: At one point in the ACNP Report, the authors attempt to explain away studies that compared a medication to placebo and found no significant advantage for the medication. “A negative study without an effective comparator [i.e., a study comparing drug to placebo], which fails to prove the effectiveness of the test medication, cannot prove that the drug is not effective… Without an active comparator, it is not possible to determine whether the study failed because (1) the drug is ineffective, or (2) the study sample was treatment resistant, or placebo responsive or the depression transient, or (3) the study design was not optimal in terms of the dose, duration of treatment, the primary outcome measure, or, most commonly, the sample size was too small to detect an effect. (p. 475).”
Why the above arguments are fallacious: The logic is that you can show a drug is effective if it is significantly superior to a placebo, but that you cannot show it is ineffective if it is not significantly superior to a placebo. Sounds like two different sets of rules to me – I’m throwing the bullshit flag! Beside the ridiculous double standard just mentioned, let’s go point by point…
1) Treatment resistant: If a drug is effective, it should effectively treat people with depression who are “treatment resistant.” When the drug is approved, is it marketed for only “non treatment resistant depression?” Of course not. It is true that some samples may be more difficult than others, but one cannot simply claim “Oh, those people were too difficult – it was the patients’ fault it didn’t work!”
2) Placebo responsive: Planet Earth to ACNP… The news is that depression, in general, is quite placebo responsive! Look at the meta-analyses by Kirsch et al (1998, (2002) if you are in need of confirmation that the vast majority of the antidepressant drug effect is replicated by placebo. The small amount of improvement in drug-treatment patients that remains after accounting for the placebo effect may be explained by medication side effects tipping trial participants to the fact that they are on medication, which thus enhances their response (Moncrieff, 2002).
3) The depression was transient: OK, so on one hand, we can excuse a failure of the treatment because the sample was treatment resistant, and on the other hand, we can say that their depression was not serious enough, that it was “transient.” So, if the drug is no more efficacious than a placebo, then the sample was either too chronic/resistant or not depressed seriously enough. Well, that’s certainly having your cake and eating it too!
4) Dose or duration: If there were solid evidence tying dose to response for SSRI’s in depression, then this would be a fair argument. In the absence of such evidence, this is a meager excuse. Granted, a three week study seems a bit short, but all studies mentioned in the ACNP Report were of 8-12 weeks. Shouldn’t we reasonably expect a medication to outshine an inert placebo within two to three months?
5) Primary outcome measure: Fortunately, one of the ACNP’ers, Graham Emslie, would be an expert on this topic! When fluoxetine was submitted for FDA approval for pediatric depression, there was an interesting flip-flop on the primary outcome measure. Emslie, mind you, was the lead author on one of the studies submitted to FDA for approval.
A quote from the FDA’s review: “In the responder analysis based on Dr. Emslie’s original primary efficacy variable (i.e., recovery rate…), the result was not significant (29% for Fluoxetine-treated group and 19% for placebo, p = .339). The analysis based on remission was also not significant (p=.238). But the result was significant based on the percent responder defined by CGI-Improvement scores (p=.040). Note that in Dr. Emslie’s ’97 paper, the percent responders defined by CGI-Improvement was presented as the primary result (p. 8, my emphases in bold).”
So Emslie’s team submitted to the FDA a primary outcome measure for their study, and when fluoxetine showed no superiority on that variable, they simply fished for a variable that reached significance and proclaimed it the primary outcome variable.
I will agree in principle that all measures should be examined and considered, but I find it more than ironic that the ACNP has decided to play the primary outcome measure card when one of their report authors have been caught altering the primary outcome variable.
6) Small sample size. OK, finally, a fair point is made. However, let’s keep in mind that this is the point of meta-analyses and review articles – to examine findings across studies to indicate what a body of literature has to say about a certain topic. Fortunately, a rather excellent meta-analysis has been done on this topic, finding no significant benefit for medication over placebo for depressed youth (full text here of Jureidini et al, 2004). Another good meta-analysis by Whittington and colleagues suggested a moderate benefit for fluoxetine only (e.g., all other SSRI’s were found to be ineffective), though this clearly should be weighed with a possibility of increased suicidality.
More will be forthcoming on this topic. Part 2 should be up in a day or two...
No comments:
Post a Comment