Posted by Larry Hoover on February 9, 2003, at 18:48:49
In reply to Re: Anti_depressants little more than placebos? » Larry Hoover, posted by mattdds on February 9, 2003, at 17:41:30
> Larry,
>
> "Placebo groups in clinical trials are not untreated"
>
> First, I don't wish to engage in a heated argument about this, and I am aware you are probably the most pharmacologically knowlegdable person who posts here.Well, I don't know about that, but thanks for the compliment.
> Just one thing. I agree with you that placebo treatment is still treatment. The alliance with professionals in hopes of getting an effective treatment can affect one's mood. However, this is also true of people getting antidepressant treatment. So don't all these nonspecific treatment factors (e.g. therapeutic alliance, participating in a study, the simple act of taking a pill to try to help yourself) figure in to the AD group's response as well?
Yes, and that's why both groups must be treated identically, in all respects, except for the active treatment.
>How do we know that ALL of the benefit we are seeing is from the pharmacological effect?
We don't try and interpret the results that way. What is important is the difference in response between the groups. It is inferred that the difference is the pharmacological effect.
>And if these "other" things figure in, doesn't that make the actual drug effect quite small?
Yes, in absolute terms, a 15% difference is pretty typical. E.g. 35% placebo response, vs. 50% pharmacological response. But, focussing solely on that one characteristic overlooks the fact that clinical practice offers a good deal of the other aspects of treatment as a matter of course. In other words, one would expect much greater than 15% response to active antidepressant treatment in the general population.
>If they are somehow exempt from that effect, I'd like to know why.
>
> But antidepressants do somewhat better, so isn't that something? Maybe.That's the nature of a clinical trial, though.
>But it could also be an "unblinding effect", which I'm sure you've heard of. In other words the side effects are fairly obvious to most people, and it quickly becomes apparent who is taking what. This might very well skew the results.
Yes, it might unblind the treatments, but the outcome would still be valid. I took part in a couple clinical trials, and I knew if I was treated or not. In one case I could not tolerate the med, and I dropped out (which removes me from the statistical analysis, preventing skew). In the other case, I knew that I was treated and continued treatment. The treatment worked, so even though I was "unblinded", my outcome was a valid measurement for inclusion in the statistical analysis.
>Why not use active placebos, like diphenhydramine or something with obvious anticholinergic effects.
That's a valid consideration, and more and more, active comparators are being used. An active comparator can also give some insight into methodological flaws. A recent study into the effectiveness of St. John's wort comes to mind (published in JAMA). The placebo response in this study was only 4.3%, whereas SJW response was over 14%. Despite a finding of a significant difference between the two groups (p > 0.02), the authors declared SJW ineffective. They did not, however, consider the low placebo response to be of any importance. An active comparator would have been a very important validator, in this case. There's something seriously wrong with an antidepressant trial obtaining only 4.3% placebo response.
> My opinion is not totally formed about this topic, and I try to keep an open mind (really!), despite my bad luck with them. But I am pretty convinced there are some major flaws in AD trials.
>
> Respectfully,
>
> MattThe biggest flaw is something you probably don't even consider: subject selection. More than 90% of the depressed population would be excluded from consideration as subjects in clinical studies, most often because of comorbid health conditions. How on Earth is one to extrapolate from an unrepresentative sample (simple uncomplicated depression) to the population at large (multiple health problems) if the latter group is excluded from study?
poster:Larry Hoover
thread:140316
URL: http://www.dr-bob.org/babble/20030208/msgs/140345.html