Exploding the Antidepressant Myth
Irving Kirsch
Irving Kirsch is a clinical psychologist and Associate Director at Harvard Medical School. He is an internationally respected clinician and researcher who holds professorships at the University of Connecticut and at Hull University. His work is published in some of the most respected medical journals. And his findings have influenced regulatory bodies to change their prescribing guidelines for depression. Throughout his clinical practice he has referred depressed patients to psychiatrists for drug treatment. Sometimes the medication seemed to work, sometimes it didn’t. On those occasions that the drugs worked he assumed the active ingredient in the medication had effected a cure, perhaps by altering the brain chemistry of the patient. He was seemingly on solid ground in this assumption. Antidepressants have been prescribed since the 1950s, and in the 1960s groundbreaking research suggested depression was caused by a deficiency of different brain chemicals, including serotonin and norepinephrine. The theory was eloquent and simple: depressed people have a chemical imbalance in their brain, antidepressants correct this imbalance. Pharmaceutical companies were quick to see the significance of this research, leading to the development of different classes of antidepressants. First, tricyclics, then MAO inhibiters, followed by SSRIs, then SNRIs, and now SSREs. Each class of drug promised a new, more effective approach to treatment, leading to $19 billion per year in sales for various pharmaceutical companies. The same drug companies claimed 80% of depressed patients could be treated by antidepressants, company representatives visited doctors and attended conferences, advertisements ran in prestigious journals, and sponsored clinical trials were published. The evidence for the efficacy of antidepressants seemed overwhelming. Yet the author retained some doubts as not all his patients were helped by antidepressants.
The doubt was partly caused by his own research over many years into the placebo effect. This is when a patient experiences relief just by knowing he is being treated, even if the treatment is counterfeit. The placebo effect should not be underestimated, it is an extremely powerful phenomenon. Placebos can affect the body and mind; they can reverse the effects of genuine medication; and, as a team at Imperial College, London, discovered, they can be up to 20% more effective than real drugs. Moreover, the bigger and more dramatic a patient perceives the intervention to be, the larger the placebo effect is. To this end, placebo treatment has moved far beyond giving a sugar pill to patients suffering from an undetermined illness. Researchers have found giving a placebo with a well-known brand name is more effective than a generic pill, taking more placebo pills is more effective than less pills, and placebo injections are more beneficial than tablets. Expensive placebos also work better than cheaper ones. But the most dramatic example of the power of placebos can be found in placebo surgery. In the 1990s Dr. Bruce Moseley, a physician to the Houston Rockets basketball team, pretended to perform arthroscopic surgery on athletes with osteoarthritis of the knee. The patient was admitted into theater, anesthetized, the skin cut on the knee, then sewn back up. No surgery was performed, but the stitched skin made the unaware patient think that surgery had been carried out. Osteoarthritis is a debilitating condition, it does not just disappear, yet immediately after surgery, and for two years of follow-up, the athletes were symptom free. Similarly, surgeons told Parkinson’s disease patients of a procedure that would relieve their condition. The patients were duly anesthetized, taken to theater, cut open, then sewn back up. An astonishing 45% of those operated on got better.
To determine the placebo effect in depression, Professor Kirsch and his team conducted a meta-analysis of all the studies in which a placebo was used to treat depression. He gathered the results from thirty-eight clinical trials involving over 3,000 depressed patients that had been published in leading medical journals. Four types of treatment were analyzed: antidepressant medication, psychotherapy, placebo, and no treatment. As was expected, there was a substantial improvement in patients treated with drugs or psychotherapy, a large improvement in the placebo group, and little improvement in those who did not receive treatment. The surprise was that antidepressants only offered a 25% therapeutic benefit over taking a placebo. This seemed small considering every new generation of antidepressants were advertised as a revolutionary breakthrough. As the meta-analysis was of all types of antidepressants, perhaps some classes of antidepressants were more effective than others. Further analysis quashed this notion: newer antidepressants were not more effective than the first antidepressants developed in the 1950s. However, the meta-analysis offered further revelations. Some of the medications used to treat depression were not antidepressants, nor were they known to act as antidepressants. These medications included sleeping pills, sedatives, thyroid hormones, stimulants, and opiates. How could it be that these drugs worked as effectively as antidepressants? The answer lay in the side effects of these drugs.
Inactive placebos do not cause side effects. In contrast, taking an antidepressant, a sleeping pill, or an opiate is likely to cause some sort of side effect. In clinical trials a patient is not told whether they are taking an active drug or an inactive placebo, neither does the doctor administering treatment know. This is known as a double-blind trial, and it is considered the gold standard of clinical testing. Yet, due to medical ethics, the patient is told that the active drug might cause side effects, and they are told what these side effects might be. They are also told that the side effects might appear before any therapeutic benefit. Picture the scene: you are told you are trialling a drug for depression, though you might be given a placebo; you are also told that you might experience specific side effects which may appear before any therapeutic benefit. At this point, most people would wonder whether they have been given the drug or the placebo. As the trial progresses a person might experience unexpected fatigue or nausea. He knows that this is a side effect of the drug, then reasonably suspects he has been given the active drug. In fact, 80% of patients in clinical trials guess correctly which group they are in, mainly due to the presence of side effects.
When a person guesses they are taking the active drug their expectations change. They expect the drug to have a therapeutic effect. This was confirmed in a clinical trial carried out by Columbia University. All participants were given an antidepressant, and all were told they received the active drug. In this case 60% of patients responded positively to treatment. The trial was then repeated with the same antidepressant. This time half the patients received the active drug and half received an inert placebo. Neither group knew whether they received the drug or the placebo. This time only 46% of patients noted a therapeutic benefit. In other words, patients are 14% more likely to respond positively to treatment if they know they are receiving a drug rather than a placebo. This figure of 14% is an important one. Previously it was noted that a meta-analysis of clinical trials for antidepressants yielded a 25% therapeutic benefit over placebos. It was noted that the majority of patients in these trials correctly guess they are taking the active drug due to the side effects. It has also been established that when patients know they are taking the active drug they receive a 14% therapeutic boost. Therefore, the clinical benefits of taking an antidepressant over a placebo are in the region of 11%. In effect, the primary worth of antidepressants lies in an enhanced placebo effect. To test this hypothesis, Professor Kirsch and his team conducted a further meta-analysis of trials involving the popular drug Seroxat also known as Paxil. Once they had accounted for side effects, they found no meaningful difference between taking Seroxat versus taking a placebo. The question is, how do these drugs get approved in the first place? The answer lies in the regulatory process of the Federal Drug Agency, which licenses all drugs in the United States.
The meta-analysis by Professor Kirsch of clinical trials involving antidepressants was published in a highly respected medical journal. Controversy soon ensued. In the midst of the dispute, Professor Kirsch received a letter from a senior fellow in health policy at George Washington University suggesting a Freedom of Information request to the FDA. The suggestion was made as the FDA has a more complete data set then those found in published studies, as it requires pharmaceutical companies to submit the results of all trials, whether published or not. As much as 40% of data on antidepressants is never published, nor made available to researchers and doctors. In addition, there is a tendency for only successful clinical trials to be published. This means that any doctor scanning the medical literature is only likely to find positive results on antidepressants. As the FDA data contains every trial conducted, and as the FDA also excludes all inadequate or poorly controlled studies, it enables a researcher to obtain a rigorous, complete picture regarding the efficacy of a drug. Professor Kirsch was successful in obtaining comprehensive data for the six most widely prescribed antidepressants. These are Prozac, Paxil, Zoloft, Effexor, Serzone, and Clexexa, though they are known by different names in different countries. Using this complete data, he discovered that there was even less of a drug effect for antidepressants than that suggested by the published literature. The drug effect of antidepressants stood at less than 11%. He also discovered that there was no difference in results for people with severe depression compared to mild depression, and no difference between low dosing and high dosing, which was interesting as most drugs are dose dependent. After analyzing the FDA data, Professor Kirsch and a multinational team published these findings accompanied by nine expert commentaries. This time there was no controversy, no doubt over the data or the accuracy of the analysis. After publication, the surprise was of a different nature.
A group of researchers actively engaged in antidepressant clinical trials wrote that the findings were not new. They referred to the data as a “dirty little secret.” The pharmaceutical companies knew it, the regulatory bodies knew it, and specialist academic researchers knew it. Only doctors not engaged in clinical research and the general public were unaware. For the most part, even practicing psychiatrists were unacquainted with the lack of evidence supporting the use of antidepressants. However, the FDA data threw up further “dirty little secrets.” Pharmaceutical companies were withholding negative studies from publication. They were publishing the same data numerous times but under different authors and in different journals. They published only some of the data from multisite studies. And they published data different to that submitted to the FDA. They also published incomplete data to skew the results. For instance, one study of Prozac was of 245 patients, yet the published paper only reported on 27 patients, making the drug seem more effective. Another such example was to be found in the clinical trials of Paroxatine—also known as Paxil or Seroxat. GlaxoSmithKline manufactured this drug and conducted three trials focussing on major depression in children. One trial showed mixed results, the second trial showed no significant difference between the drug and a placebo, while the third trial showed the placebo might be more effective than the drug. Only the first favorable trial was published. The other two were buried until a confidential internal memo fell into the hands of the Canadian Medical Association. The memo stated that GlaxoSmithKline was to manage dissemination of clinical data to minimize any negative commercial impact. A second team of researchers at Oregon University followed up Professor Kirsch’s work. They analyzed the data submitted to the FDA, then compared it with the published data for the same drug. They found that of thirty-eight positive trials submitted to the FDA all but one were published. In contrast, almost all thirty-six trials demonstrating negative results were suppressed. The drug companies have not contested this data. Two major pharmaceutical companies even hired Professor Kirsch and his team to help design more effective trials, as they admitted they could not get their drugs to work better than a placebo.
So why didn’t the FDA step in if they had all the data? Internal FDA documents show that this data was deliberately withheld from doctors and the public. They also reveal that the FDA actively urged antidepressant manufacturers not to disclose the failed trials as it might scare physicians and the public away from the drugs. This flirtation between the FDA and the pharmaceutical industry began in 1992 when the United States government stopped being the sole funder of the FDA. Instead, the principal clients of the FDA became the pharmaceutical companies who funded the very regulatory body that was supposed to oversee their trials. This relationship, coupled with the disclosure of manipulated data, eventually led to officials from the regulatory agencies of the European Union to ask how some antidepressants were ever approved.
In 2008, following the publication of Professor Kirsch’s second meta-analysis of antidepressants and the accompanying nine expert commentaries, every major newspaper and news channel in the U.K. carried the study as their lead story. This was repeated through much of Europe, the United States, Canada, Australia, and China. Critics within the medical community were lined up to offer a different opinion. Interestingly, these critics did not contest the data, nor did they question the validity of the meta-analysis. The main criticism came not from pharmaceutical representatives, academics, or researchers. It came primarily from clinicians and physicians who were not involved in medical research. Their common refrain was, “but it works in clinical practice.” The problem with this argument is that so do placebos, and Professor Kirsch’s meta-analysis demonstrated this. Clinicians generally do not give placebos when treating patients, they only give the active drug. If clinicians gave placebos to patients then they would have seen the positive effect of placebos upon depression. Secondly, in clinical practice, as opposed to clinical trials, patients know they are getting the active drug. As we have seen, when a patient knows they are receiving a prescribed drug for a specific condition, the efficacy of that drug significantly increases. Finally, continuation studies of antidepressants over a long period of time have shown there to be little difference between a placebo and an antidepressant. In the long term, placebos are 95% as effective as medication. This led the U.K. regulatory body—the National Institute for Health and Clinical Excellence—to conclude that there is little difference between a drug and placebo in prolonged studies.
The therapeutic benefit of antidepressants over placebos is not backed up by evidence-based medicine. However, the work of Professor Kirsch has more far-reaching conclusions. It also questions the belief that depression is a physical brain disease caused by a chemical imbalance. The theory of chemical imbalance has become well known over recent years. Even nonspecialists and many members of the general public know that a brain lacking in serotonin or dopamine results in apathy, poor motivation, or depression. It is a persuasive theory and attractive to pharmaceutical companies, for if the problem is chemical, then it can be rectified by other chemicals. But there is a difficulty: the evidence does not support the theory.
The chemical imbalance theory was formed in the 1960s, based on weak and contradictory data. It remains a popular explanation of depression. The first two antidepressants were iproniazid—made from leftover rocket fuel—and imipramine. In primitive trials without a placebo they appeared to relieve depression, though their mechanism of action was unknown. In 1965 a groundbreaking paper argued that depression was caused by a deficiency of the neurotransmitter norepinephrine. A second paper in 1967 emphasized a deficiency of serotonin. Against this research it was argued that the two new antidepressants of iproniazid and imipramine must work by increasing either norepinephrine or serotonin in the brain. However, this was a theory not supported by any data. Researchers have tried to find evidence for the chemical imbalance theory, largely with contradictory results. If the chemical imbalance theory is correct then increasing serotonin in the brain should relieve depression. As we have seen from the meta-analysis of antidepressant treatment, increasing serotonin in the brain through drug intervention does not relieve depression any more than an inert placebo. Conversely, when drugs are administered to deplete serotonin and norepinephrine, it should induce depression. This simply does not happen. Neurotransmitter depletion has been attempted in at least ninety studies with conclusive results: in healthy volunteers reducing serotonin and other neurotransmitters does not effect mood. The authors of the most complete meta-analysis of serotonin depletion conclude: “there is no direct correlation between serotonin in the brain and mood.” Fifty years of research has consistently rejected the chemical imbalance theory promoted by pharmaceutical companies. This does not mean that depression does not exist. It resides in the brain in some form as shown by neuroimaging scans. People who are sad or depressed have altered brain states, and the affected areas of the brain change when depressed people get better. But finding a representation of depression in the brain does not mean it is caused by a chemical imbalance or a physical illness. Life circumstance and emotions also alter neural networks, as do traumatic events. As depressed people feel more worthlessness, a sense of guilt, or experience hopelessness, this altered brain state is strengthened and reinforced, leading to further depression. It may well be that life events and the way we interpret those events make us feel depressed. In contrast, leading researchers at Oxford and Cambridge have claimed that effective treatments for depression work by altering this sense of hopelessness. In part, this is how antidepressants and placebos also seem to work: they give hope to the patient.