[ Pobierz całość w formacie PDF ]
.Now this is a rather complicated figure,so let me walk you through it.The triangles represent the drugresponse on each of the clinical trials; the circles indicate theplacebo response.The size of the triangle or circle reflects thenumber of subjects in the trial.The larger the shape, the largerthe trial.This is important because data from larger trials aremore reliable than data from smaller trials.So when doing a meta-analysis, more weight is given to large trials than to small trials.Figure 2.1.The response to drug and placebo at different levels of initialseverity of depression.9The  Dirty Little Secret 33The most important things to look at in Figure 2.1 are the solidhorizontal line, representing the average drug response, and thedashed diagonal line, representing the average placebo response.The difference between them is the drug effect.That differencegets greater and greater as baseline severity increases, until itfinally reaches clinical significance (the shaded area) for the mostextremely depressed patients  those with Hamilton scale scoresof about 28 or more at the beginning of the study they were in.The average drug placebo difference in this small group of rela-tively small studies was just over four points on the Hamiltonscale.A four-point difference is clinically significant, according toNICE, but it is still rather small.Differences in sleep patterns, forexample, can produce a six-point difference in depression scores,without any other differences in symptoms of depression.Still,this relationship seems to be reliable.The worse the depression,the greater the drug effect.If you look at the figure again, you will see that there issomething a bit strange about it.The response to the drug doesnot become greater as depression increases.Instead, the placeboresponse gets smaller, and that is what makes the drug placebodifference larger.Now this seems very curious.Why is there lessof a placebo response among extremely depressed patients,without much change in the drug response? I can think of twofactors that might account for this.First, these patients tend tohave been chronically depressed.They are much more likely tohave been on antidepressant medication before, and they knowwhat it feels like.Second, physicians are likely to prescribe higherdoses to patients who are more severely depressed.As I showlater in this chapter, the dose-response studies that have beendone tell us that this does not make much difference in the effectof the antidepressant.Low doses of SSRIs are just as effective ashigher doses.But unlike the therapeutic effects, the side effectsof SSRIs are dose-dependent.The higher the dose of the medica-tion, the more side effects you get.Putting these two factorstogether suggests that more extremely depressed patients areparticularly likely to recognize whether they have been put on34 The Emperor s New Drugsplacebo or on the real drug.When they don t experience theside effects they are used to, even on high doses of the new medi-cation, they may conclude that they have been placed in theplacebo group, and this recognition may dampen the placeboeffect.If this is the case, then even the relatively small but clin-ically significant drug effect seen in extremely depressedpatients may be a placebo effect in disguise.Now I have to admit that my speculation about the severityeffect being due to breaking blind, and guessing correctly whetheror not one has been given the real drug, is just conjecture.Theremay be other explanations.I have not been able to think of any,nor has anyone suggested to me another plausible explanation.Still, I consider my proposed explanation to be no more than ahypothesis, which might very well turn out to have been mistaken.But it might also be correct.And if it is, then there may not be areal drug effect, even amongst the most severely depressed patients.A LITTLE GOES A LONG WAYPrior to submitting our analysis of the published data to a journal,Sapirstein and I were concerned that we might have underesti-mated the drug effect by lumping together effective drugs withineffective drugs  a concern that proved to be unfounded, asthere turned out not to be any meaningful differences betweenone type of drug and another, even when looking at drugs thatare not antidepressants [ Pobierz całość w formacie PDF ]