We’ve all seen the commercials for antidepressants; you can hardly turn on a TV and miss them. And we’ve all heard jokes about the side effects listed off by a rushed voice while some joyful ovoid creature ironically bounces happily across the screen. But there’s something much deeper about these commercials worth criticizing which fewer people even think to notice: their presentation of the idea of a “chemical imbalance.” (For instance, take this commercial for Zoloft, and pay attention to the bit from 0:21 to 0:30 seconds.) The kind of marketing represented by these commercials has ingrained this notion into popular question so deeply that it is hardly ever questioned at all, but there is surprisingly little evidence to support the idea.
In fact, the evidence is so weak that in Ireland in 2003, GlaxoSmithKline (the maker of Paxil, Wellbutrin, and other antidepressant drugs) published a letter explaining that, in response to demands from the Irish Medical Board, they were retracting their previous advertisements claiming that Paxil “works by bringing serotonin levels back to normal.” This claim amongst others, they explained, was “not consistent with the scientific literature.” An article published in the Public Library of Science in November 8, 2005 by Jeffrey R. Lacasse and Jonathan Leo, titled ‘Serotonin and Depression: A Disconnect between the Advertisements and the Scientific Literature,’ noted this event: “The Irish equivalent of the FDA, the Irish Medical Board, recently banned GlaxoSmithKline from claiming that paroxetine corrects a chemical imbalance even in their patient information leaflets.” The authors then go on to describe correspondence with various media outlets, none of which could supply any study to support their claims when questioned on their references to the chemical imbalance hypothesis. They conclude by rightly observing the problem with this promotion: “If a patient comes into the office believing the serotonin theory and the doctor doesn't take the time to correct them, we wonder where that leaves the issue of informed consent … they might have chosen an alternative approach to their distress if they were fully informed.”
If these drugs don’t work by correcting a ‘chemical imbalance,’ how do they work? Clearly, they do ‘work’ to produce altered states of consciousness and a whole host of possible side effects. And clearly, they do inhibit the reuptake of neurotransmitters. But quite simply, we don’t have to postulate any ‘chemical imbalance’ in order to explain or understand this. A drug like marijuana can at least temporarily ‘improve’ depression as well, but we don’t have to postulate any ‘imbalance’ of THC in order to understand this either. Dr. Steven Hyman, professor of neurobiology at Harvard Medical School and head of the National Institute of Mental Health from 1996 to 2001, noted in a paper titled Initiation and Adaptation: A Paradigm for Understanding Psychotropic Drugs that antidepressants work to create a state of mind that is “both qualitatively and quantitatively different than normal.”
To be completely clear, I don’t consider this necessarily to be a bad thing at all. The issue is not that people have some moral obligation never to experience drug–altered states, but that they should not be lied to or misled. If someone finds that marijuana, antidepressants, or anything else truly does work to relieve their depression, and they can use them in a responsible way that does not negatively impact their life, then by all means they should continue to do so. Whether either drug works by ‘correcting’ something or simply by creating drug-induced states which allow them to enjoy life is, in the end, totally irrelevant.
But while it is obvious that antidepressants do ‘work’ to produce altered states of consciousness, it is actually not clear that they ‘work’ to reduce depression due to their chemical action at all. To put this more directly, it is possible that the effect that antidepressants do have on depression is actually no more than a placebo effect. At first glance, this suggestion might sound absurd: don’t these drugs have to be proven more efficient than placebo to the FDA before they can be approved for popular use at all? Unfortunately, the situation just isn’t nearly so simple.
Marcia Angell, M.D. is an American physician currently working in the Department of Social Medicine at Harvard Medical School. She is also a member of the Association of American Physicians, the Institute of Medicine of the National Academy of Sciences, the Alpha Omega Alpha National Honor Medical Society, a Master of the American College of Physicians, and the first woman ever to serve as the editor-in-chief for the New England Journal of Medicine. In an article published to the New York Times’ Review of Books in January of 2009, she spoke out: “It is simply no longer possible to believe much of the clinical research that is published, or to rely on the judgment of trusted physicians or authoritative medical guidelines. I take no pleasure in this conclusion, which I reached slowly and reluctantly over my two decades as an editor of The New England Journal of Medicine. … [Drug] company employees or their agents often design the studies, perform the analysis, write the papers, and decide whether and in what form to publish the results. Sometimes the medical faculty who serve as investigators are little more than hired hands, supplying patients and collecting data according to instructions from the company.” She goes on to note part of the answer to the previous question: “The results of all the trials (there may be many) are submitted to the FDA, and if one or two trials are positive—that is, they show effectiveness without serious risk—the drug is usually approved, even if all the other trials are negative.” That is, the FDA doesn’t actually require proof that any drug is on the whole better than placebo; it only requires one or two studies; and negative results simply don’t count, meaning it doesn’t matter how many tries it might take for a manufacturer to find those two positive outcomes.
Furthermore, the results of all of these trials do not even have to be published.
In a landmark 1998 study (Listening to Prozac but Hearing Placebo: A Meta-Analysis of Antidepressant Medication), Irving Kirsch had performed a meta-analysis incorporating all the published studies on antidepressants into one large picture. He found that a whopping 75% of antidepressants’ effect could be attributed to placebo, which was surprising enough. But susequently, he was criticized by defenders of the benefits of antidepressants for only incorporating the published studies into his analysis without knowing what information the unpublished studies might hold—and so Kirsch and colleagues proceeded to file a Freedom of Information request against the FDA in order to gain access to these trials. What he discovered from this request alone is shocking: while roughly 94% of all published trials found a benefit for the antidepressants beyond placebo, forty percent of all studies were not published, because most of them did not find a positive result. Accounting for all the trials together, only about 51% of all trials found any benefit for antidepressants beyond placebo! Marcia Angell’s article, linked above, noted this phenomena as well: “A review of seventy–four clinical trials of antidepressants, for example, found that thirty–seven of thirty–eight positive studies were published. But of the thirty–six negative studies, thirty–three were either not published or published in a form that conveyed a positive outcome. It is not unusual for a published paper to shift the focus from the drug’s intended effect to a secondary effect that seems more favorable.” Clearly, once Kirsch’s analysis accounted for all these additional unpublished studies, the result for antidepressants would be even more dismal.
In the updated, more comprehensive analysis (Initial Severity and Antidepressant Benefits: A Meta-Analysis of Data Submitted to the Food and Drug Administration), the placebo effect now accounted for 82% of the antidepressants’ effect . So what does this number actually mean in practical terms? The studies in question ranked depression on what is known as the Hamilton Rating Scale (or HAM–D). An example HAM-D questionnaire can be seen here. When the statistic that placebo accounted for 82% of the antidepressant effect is translated into this scale, the result is that the net benefit of taking an antidepressant over taking a placebo is only 1.8 points! As the example questionnaire shows, a total of six points on this scale relate entirely to patterns of sleep. A 1.8 point difference could thus be explained in full if either patients in the drug group begin falling asleep more easily, or if they tend to stay asleep more easily, or if they find themselves falling asleep more easily if and when they do wake up during the night. And since many trials administer benzodiazepines like Xanax—well known for their sedative and hypnotic properties—along with the antidepressant in question, even this incidental detail alone might explain the small difference between the drug and placebo groups completely.
It is also crucial to note that even placebo-controlled studies of antidepressants are not always truly “double–blind.” In RCTs (random controlled trials) today, patients are not considered legally to have given true consent unless they consent with full knowledge of both the design of the study and the possible effects and side effects of the drug; and thus, they go into the trial knowing in detail what side effects the active drug is likely to produce. Then, if they experience those side effects, they can easily infer that they have been given the real thing; and if they do not, they can infer that they are on the placebo. This means that since expectations are lessened in those experiencing no side effects and raised in those who are, the placebo effect is actually weakened for the placebo control group and enhanced for the drug group. One study, indeed (Inert placebo versus active medication: Patient blindability in clinical pharmacological trials) found that 75% of all patients were able to correctly guess whether they were in the drug or placebo group, and that virtually 100% of those experiencing side effects were able to correctly guess that they were on the active drug. In another study (Side Effects and the ‘Blindability’ of Clinical Drug Trials), even an independent evaluator was able to distinguish whether patients were taking a placebo or the active drug on the basis of reported side effects alone. It is plausible that this unblinding effect might even explain the small difference between drug and placebo groups.
Psychologist David Antonuccio, a professor at the University of Nevada, points out another possible problem with clinical study design: many trials involve a ‘placebo wash–out’ period in which all participants are placed on a placebo for some specified period of time, so that anyone who responds positively to the placebo can then be removed from the study altogether. He says: “if you put everybody on an antidepressant and washed out everyone who responds, people would say, ‘That’s a very biased strategy against the drugs.’ Well, I believe we have a strategy here that’s biased against the placebo condition.” Another example is pointed out by Joanna Moncrieff, M.D. Senior Lecturer in Social and Community Psychiatry at the University College of London: in some trials, both the drug and placebo group are taken from groups of people who are already on the active drug prior to the trial. These individuals are then assigned either to continue using the active drug, or to be transitioned onto the placebo; so it is possible that the “control group” in some trials is actually not just on a placebo, but in a state of withdrawal—making it very easy to find that someone who is not in a state of withdrawal would rank better on measures of depression than someone who is. These are just two of many ways in which the study design of most clinical trials may be biased or corrupted, even on top of the inability to prevent unblinding, but it is not my purpose in this essay to survey every possible way that clinical studies can be corrupted.
In any case, one thing is clear: either the data sent to the FDA for the approval of these drugs was accurate, or it was not. If it was, then the conclusions of Kirsch’s meta-analysis hold, and these drugs should never have been approved. If it was not, then there was never any solid basis for concluding that these drugs are effective in the first place—and still they should never have been approved. Either we can safely conclude that these drugs have no real efficacy beyond placebo, or else we must admit that present data gives us no grounds for any conclusion at all and that efficacy has yet to be demonstrated. Certainly, many people do report and believe that antidepressants have helped them; for a long time, I was one of these people myself. The problem is that many people believe that inert placebos have helped them as well; and personal experience simply cannot determine whether a given response was placebo or due to a direct chemical effect—only rigorous statistical analysis can do so. And given the results of these studies, it is overwhelmingly likely that anyone who believes that an antidepressant has helped them would in fact have improved just as much on any inert substance.
If anything, this should be taken as good news for those who believe antidepressants have helped them: however much you believe your antidepressant has helped you, you probably in fact achieved that improvement yourself—and you should have little fear of relapse should you decide to go off of these drugs and save yourself the time, money, and risk of possible side effects (particularly over long-term use, where the possible effects of antidepressant use are still unknown) associated with continued use. However, there is one important word of caution: withdrawal symptoms in the days and sometimes weeks following going off of a medication can be severe, and these tend to be more or less dramatic in proportion to how quickly the drug was ended and how much of it was ended at once. If you decide to try going off of your antidepressant, you should taper off as slowly and gradually as possible, and try to find qualified supervision to guide you through the process. On the other hand, it is important to keep in mind that even dramatically lowered mood in the weeks following cessation of an antidepressant is not evidence of ‘relapse,’ but is overwhelmingly likely to be symptom of withdrawal which should pass quickly in time for almost everyone.
Still, there is one last important point to make before closing: if the conclusion you are beginning to draw from this is to think: “hey, if these drugs are just placebos, then I really don’t need them,” then this is perhaps not entirely accurate. In clinical trials, patients given placebos actually do improve substantially faster and more fully than patients who are given nothing at all. The benefit produced by a placebo actually is ‘real,’ and it is not all attributable to the improvement in depression that tends to happen naturally over time. Thus, if you believe that antidepressants have helped you and you are doing nothing else in particular to fight your depression, going off of them without replacing them with anything is probably not a good idea. To achieve the same level of improvement or more that antidepressants seem to have granted you, one would generally have to replace them with either an exercise regiment, or a vitamin regiment perhaps including neurotransmitter precursors and amino acids (I have personally found l–theanine, l–tyrosine, and 5–HTP beneficial), or psychotherapy, or something else. Like antidepressants, it is unlikely that any of these things is working to ‘correct’ any biochemical cause of depression on any direct physical level; but unlike antidepressants, all of these things are much cheaper, and none have any serious risk of negative side effects. In many cases they have positive side effects, as should be obvious enough in the case of exercise. In any case, the primary benefit of all of these things seems to come from mainly the mere fact of doing something while having faith in a positive outcome.
Our expectations of the future, whether positive or negative, tend to create loops of experience which feed on themselves. If we expect the future to be bleak, then we are likely to emotionally disengage, in which case this very act of disengagement tends to bring about an emotionally bleak future. Likewise, we are more likely to pay closer attention to ‘negative’ mood swings and consider them evidence of an approaching depression, thus pushing us into an even more negative mood. Yet, if we have positive expectations of the future, then this very attitude too tends to help create an emotionally engaging future—and we are more likely to focus on, and appreciate, our ‘positive’ mood swings instead. This is not to say that the experience of depression is ‘all in one’s mind’ or that it can simply be easily ‘willed away,’ however, as these psychological tendencies can take serious and difficult work to overcome. And not everyone comes preset with the same tendencies and expectations, either, meaning the job is objectively much harder for some people than others. But those who, for whatever reason, are set with a more difficult job should not be duped into believing that their depression has an actual biochemical ‘cause,’ as if dangerous and expensive drugs which aren’t even pleasurable are the only way to fight and overcome depression—even its most severe form.