Dear Doctor Werner:

Since you are not allowed to prescribe placebos in your practice (either single or double blindedly), you have no way of knowing what is causing the apparent "improvement" in some depressed patients who are prescribed antidepressants zero blindedly. So you attribute the entire improvement to the drug. The whole point of a clinical trial is to determine if there is a difference between active drug and placebo, and the Kirsch study shows the difference, except for the severely depressed (many of whom have lost hope and believe nothing will help, thus the lower placebo effect) is minimal. The history of medicine is filled with treatments that based on clinical ( i.e. "physician") experience, were thought to work, and only later found to be worthless after rigorous scientific study. Your comment about "statistics for the sake of statistics" suggests a lack of understanding of the scientific method, something I have found rampant among practicing physicians. Also, your comment that antidepressants are "safe" flies in the face of tens of thousands of adverse event reports filed with the FDA concerning antidepressants. And, no, I am not a scientologist, I am a professional statistician, and I will believe an analysis made by qualified researchers with no financial ties to the pharmaceutical industry any day over studies selectively published by persons who have a direct financial interest in the outcome.(I am not implying you are in this group). But I suppose that if, for instance, fluoxetine apparently makes some of your patients feel better despite ruining their sex lives, I say keep writing the scripts. At least it is generic and can be had by the uninsured for $4 a month at Wal-Mart, as opposed to drugs still under patent, whose real effects (if any) are no better documented, and cost upwards of $4 per pill (and in some cases much much more).

I wonder what the results would be if patients (outside of a clinical trial) could be prescribed (double blind) placebos that looked identical to the real medications. Obviously this is unethical, since neither the physician nor the patient would know this was being done. Which is why we have a clinical trial, which is a miniaturized, controlled version of just this I actually believe that statistical studies, when properly designed and conducted, do in fact produce results indicative of what would happen in the target population; i.e. I believe that the science of statistics "works". Therefore, if my idea in the first sentence could actually be carried out on a large scale, I believe there would be many many physicians and patients unwittingly singing the praises of placebos they thought were real drugs.