Tuesday, February 26, 2008

Prozac Attack, Take 58

Today's news offers a study from Professor Irving Kirsh, Psychologist in the field of placebo effect. Prof. Kirsh, et al. have done a meta-analysis on 47 clinical trials for depression and have determined that the effect of the most popular medications (always referred to as "drugs" in such studies) for depression provide no benefit, except in severe cases. Of course, much of the media has been reporting with "unbiased" reports, with headlines such as "Antidepressant may not work" "Do we need antidepressants. and "Antidepressants no better than placebos." Even Fox (fair and balanced) fails to disclose some very basic facts regarding this article- the original which can be found here.

Some of those facts are:

1. This article, (published by the way on an open access journal-a sort of wikipedia for scientists) is a meta-analysis. It is a reconstruction of statistics determined after the fact. Given that one can choose the statistical analysis afterwards can allow one to manipulate the data.

2. There is a selection bias regarding studies chosen for the meta-analysis. Per Glaxo Smith Kline's representative, only studies that were submitted prior to approval, rather than a comprehensive (or at least randomly selected comprehensive) set of studies.

3. As we learn in statistics, failure to prove separation (the null hypothesis) does not prove equality. Many things are involved in a study, such as sample size, completion rates, and statistical methods. It is as if I take two animals, a cat and a dog, and try to prove that they are different based on certain criteria (e.g. type of hair, teeth, claws, number of legs, and weight). If I do this with one hundred cats and dogs, I can prove a difference. If I do it with two, I cannot. This failure to separate statistically does not allow me to state that cat=dog!

4. There is a likely bias of the conductor of the study. Prof. Kirsch has been singing this placebo song for years, and is not just limited to antidepressants. He also has written similar opinions regarding Asthma, and Irritable Bowel Syndrome. So what's his angle? Research money? Perhaps. Personal Bias? Attention? I do not know. He is a supporter of cognitive behavioral therapy, in lieu of medication. Furthermore, he espouses the usage of a certain medication for treatment of depression: can you guess which one? How about Placebo! Dr. Kirsch notes in an interview to NPR that regarding the effect of placebos:
"For years and years and years it has been treated as just noise, as trash. Now we're finding out that it's not just trash, it's really treasure. It's something to be mined, something to be understood, something to be made use of."
Fascinating. I'm not sure how he's going to sell that idea to the FDA: Hey! Don't use a drug that might not work, use one that we know won't work.

5. The UK may have a financial agenda. The nationalized system recently has done this with Alzheimer's medications (the NICE study declaring them to be not cost effective). Perhaps we are seeing the UK starting to put up barriers to obtaining treatment for depression.

Well. Any doc worth his salt will tell you that we are quite aware of the "placebo effect." I used to be quite annoyed and amused with a neurologist in my residency days who used to declare that a person was not having a particular neurological condition just because it improved when he gave a shot of saline to the patient. He would smile and state, "must be one of yours." So of course, we should understand that a "placebo effect" can do some interesting things- that is the amazing part of the mind.

I do not fault Prof. Kirsch his opinion or his data. I do not fault the news media for reporting. They both have their rights. I just strongly disagree with the findings as horribly biased, and the reporting as merely sensationalistic. It is, in short, bad science.

The choice for antidepressant treatment should be made after consultation with a qualified health professional. I do not recommend merely "just trying" any medication, psychotropic or otherwise. It would be a shame, though, if doctors or patients allowed such bad science to bias their decision to offer or consider these medications.

No comments: