Tuesday, July 17, 2007

Should placebo be placebo?

I havent been posting for a while. I mentioned the random control trials for Pagaclone to a friend of mine, Gilles (PhD biopsychology), and told him that the trial might not really control for the placebo effect, because Pagaclone is an active compound and does change the brain chemistry. So the patients in the Pagaclone should experience (subtle) changes within themselves, and will know whether they are taking Pagaclone or not, and then they might have a secondary placebo effect: I am taking Pagaclone, so my speech will improve now.

Gilles suggested that the placebo should be a drug similar to Pagaclone, maybe an anti-anxiety drug, because then both groups (the Pagaclone and the control group) will experience a difference, and not know whether they are on Pagaclone or not.


Anonymous said...

A drug similar to pagaclone such as an anti-anxiety drug that also binds to the benzodiazepine (BZ) site (e.g. ativan) may also reduce (or eliminate) the frequency of disfluency. Therefore, the results of such a trial would show a reduced efficacy of pagaclone (i.e., pagaclone would not perform better than the placebo, statistically speaking). Also the benzodiazepine binding anti-anxiety drugs tend to sedate (make the subject sleepy) whereas pagaclone is purported not to sedate.
What struck me when I first heard about pagaclone is that it binds to the BZ sites. Since a lot of other drugs do this, it appears to me that research ought to be reopened on thess classes of drugs and their impacts on fluency. The few previous trial studies that have been conducted re BZ drugs have generally shown negative results. But these could be false negatives. For example, one study used 5 mg doses of valium (diazepam) which is equivalent to about 1 mg of ativan (lorazepam). But my experience has been that 2 mg of ativan works for me.
Also different BZ drugs may work differently on different people re fluency, much like the differential effects of SUI anti-depressants (e.g. prozac, zoloft, etc.) on trial subjects. A patient may have to go through a number of SUI anti-depressants before he (she) finds one that is effective. In short, we may have a fragmentation of the stuttering population into subgroups, affected in different ways by different drugs which makes the research problem much more difficult.

Tom Weidig said...

You are right. So maybe not the same type of drug.

But rather a drug that makes you feel difference, so that you cannot know whether you are on Pagaclone or on this drug.

Anonymous said...

This is one of the standard difficulties in experimental design, I think,

Example: How do you determine the benefits of surgery? There is no true "placebo" in surgery except perhaps for "sham surgery", i.e., cutting someone open, pretending to do something, but in fact doing nothing. But that's unethical.

So how do you fool someone into thinking that they've had (or may have had) a certain sort of treatment without in fact giving it to them, and without damaging them unacceptably in the process? (Doctors learn the Hippocratic principle "Primum non nocere" - "first do no harm".)

Not an easy problem.