I have avoided this topic for some time because of the complexity emanating from the heterogeneity of the depressive syndrome, complex statistics, changes in study design over time, use of meta-analysis, changing FDA requirements for the studies, issues with regard to publication bias or withholding publication of negative results, issues with regard to patient selection and representativeness of those patients, questionable assumptions that changes in scale ratings adequately measure changes in severity, data that shows an increase in placebo effect over time which is likely the result of recruiting patients that either the most severe or the least severe (least responsive to placebo and most responsive to placebo), financial incentives in the medical and pharmaceutical industries etc. However, recently this issue has been raised often by patients in my practice and I am thus going to engage it.
This current debate was kicked off by an article in January 6, 2010 Journal of the American Medical Association (JAMA) that concluded that the results of the published meta-analysis showed that the magnitude of benefit from antidepressant therapy when compared to placebo increased with the severity of depression and that in mild or moderate depression the benefit was minimal or nonexistent. It does seem intuitive that less depressed people would more likely respond to an antidepressant and placebo similarly. Two other points need to be made about the study. First, it is a meta-analysis and by definition it includes many studies and thus the heterogeneous condition of depression is magnified because of different research protocols. Second, the idea of placebo does not mean that nothing is done. It simply means that the active agent being evaluated is not in the placebo group. However, the placebo group is still given attention, they go through the mechanics of taking a pill, they are evaluated and asked questions etc. This is something and possibly something that is meaningful. It is not nothing. These ideas made the leap to the popular press in a January 29, 2010 article in Newsweek entitled, "The Depressing News About Antidepressants." Their conclusion that the therapeutic effects of antidepressants are generally trivial is a distortion of the placebo effect.
In the June 23, 2011 and the July 14, 2011 issues of New York Review of Books Marcia Angell, M.D., the former Editor-in-Chief of The New England Journal of Medicine, joined the discussion by saying that clinical trials failed to demonstrate antidepressant effectiveness in mild to moderate depression and further that they have serious adverse effects and "are greatly overused because of the pharmaceutical industry's influence over the psychiatric profession." Peter Kramer, the author of Listening to Prozac, responded in the July 9 2011 Sunday Magazine New York Times with a rebuttal to Angel. Both refer to Irving Kirsch's book, The Emperor's New Drugs:Exploding the Antidepressant Myth, in which he examined the FDA reviews of the clinical trials leading to the approval of 6 antidepressants. He found that placebos were 82% as effective as active drug as measured by the Hamilton Depression Rating Scale (HAM-D). The difference in ratings of 1.8, Kirsch believes, supports the view that antidepressants are no more effective than placebo. Kramer responds that FDA trials do not answer questions regarding mild depression because of recruitment of patients who are not depressed. He also notes that antidepressants have been found effective in people with co-morbid conditions and in long term maintenance prevention studies.
What are we to make of this data and analysis? First and foremost consider the terminological inexactitude of the word depression. There are dozens of different variants. Are we then comparing apples to apples or not? Second, the recruitment process tend to enroll patients who are very seriously ill or barely ill at all. Are they really comparable to patients seen in a private office or clinic? This recruitment (and changes in study design) is what probably underpins the finding that the placebo effect has grown over time which in turn narrows the field to show a therapeutic effect. Third, the criticisms of antidepressant efficacy all emanate from the clinical trials that are part of the FDA approval process. These studies are designed to demonstrate to the FDA that the drug is safe and that it produces a greater reduction in a standard set of symptoms than does the placebo under double blind conditions. The two issues should be noted regarding these studies are that they are long enough to show or not show an effect and that a standard set of symptoms are appraised by using a standard rating instrument such as the HAM-D. This appraisal is not identical to treatment in the community. Changes in symptom rating is not the same as monitoring the changes in personal anxieties, concerns, worries, suffering and performance in a persons actual life experience.
How does this inform my opinion about the use of antidepressants? It is definitely one factor I take into account and I do think that very mild depressions often respond to antidepressants in a manner similar to placebo; but, in depressions that are more severe or that have other co-morbidities such as panic states, obsessive compulsive disorder, social anxiety, eating disorders or others I know that antidepressants are clearly much more effective than placebo. Depressions of the atypical type or melancholic type also respond to antidepressants much more than to placebo. Also, depressions that are a part of bipolar disorder presentations clearly respond more robustly to antidepressants than to placebo. In the final analysis, the best judgment about the usage of antidepressants is formed by the experienced clinician who is well trained and who evaluates his or her experience continuously and discriminates between cases carefully taking into account a patient's presentation as well as their evolution with medication in their social environment.