Given the nasty side effects, does the reliance of antidepressants on the placebo effect present an ethical dilemma?
Given the often nasty side effects, how do we deal with the ethical dilemma presented by evidence that suggests that near half the effectiveness of antidepressants is placebo effect?
There’s a famous advertising idiom which goes: “I know only 50% of my advertising works, I just don’t know which 50%.” The same might be said of antidepressant medications.
Hitting the press in the past few weeks, the article in the Medical Journal of Australia is the latest in a number warning about the potential harms of antidepressants.1 This is not a new story, but it is a controversial one.
The debate really got started in 1998. That year, a Harvard University psychology professor, Irving Kirsch, published a meta-analysis with one his graduate students, Guy Sapirstein. The paper, Listening to Prozac but hearing placebo: A meta-analysis of antidepressant medication, caused a storm.
Fascinated by the placebo effect, Kirsch wanted to know whether the belief that one has taken an antidepressant can produce some of the effects of that medication. And this is exactly what he seemed to find. In that first meta-analysis of randomised controlled trials on antidepressants versus placebo, there appeared to be a substantial placebo effect on depression.
“What surprised us was how small the drug effect was. 75% of the improvement in the drug group also occurred when people were give dummy pills with no active ingredient in them.”2
Based on the response-expectancy theory, this finding supported the idea that what people experience depended partly on what they expected to experience. People’s responses can be both subjective and physiological, and can be altered by changing their expectations.
Famously reported by a World War II US army medic, Henry Beecher, the placebo effect can be powerful. In Beecher’s case he had run out of morphine to treat the wounded soldiers so he decided to replace it with a saline solution and told the soldiers it was morphine to keep them calm. To his surprise, almost half of the soldiers reported that the inert saline solution actually reduced or erased their pain.
From these early observations, the placebo effect was considered to be a purely psychological phenomenon. However, more recent research has identified some of the complex physiological responses embedded within it.
This complexity is not really surprising when we think about the biochemical and neurophysiological interdependence of the human system. In Western medicine, separating mind and body responses has a long history. Linking them together was more in the realm of non-Western healing traditions but the recent upsurge in research involving the physical benefits of practices such as mindfulness, yoga, and prayer has confirmed this fundamental interdependence. We don’t yet fully understand it, but the physiological effects are there and measurable.
So back to Kirsch and his apple-cart upsetting antidepressant paper. At first, his meta-analysis methods were slammed. At the time of this initial paper, meta-analysis was still considered a bit fringe.
Four years later, Kirsch once again raised the ire of the medical community by publishing a further paper on the subject.3 This time he and his colleagues had sought access to not just the published empirical research on antidepressants but also the unpublished data held by the US Food and Drug Administration (FDA).
Provision of both published and unpublished material to the FDA is a requirement of pharmaceutical companies when they apply for licensing approval of a new drug. The requested data, obtained under the US Freedom of Information Act, was for six new-generation antidepressants that accounted for the bulk of antidepressant prescriptions being written at the time.
Kirsch found that almost half of the clinical trials that had been sponsored by the drug companies had not been (and still haven’t been) published. The results of these unpublished trials were known only to the drug companies and the FDA, and most of them failed to find a significant benefit of drug over placebo. All of the published and unpublished trials used the same primary measure of depression – the Hamilton Depression Rating Scale (HAM-D) (see box, page 35) – which enabled ready comparison of results, including the clinical significance of the drug-placebo differences.
When taken as a whole, only 43% of the trials showed a statistically significant benefit of antidepressant over placebo. The remaining 57% were failed or negative trials. Other meta-analyses have reported similar percentages.4, 5
Given that the control in all of these studies was a placebo – a standard trial practice in this context – the meta-analysis showed that, for all six of these new antidepressants, the placebo response accounted for 82% of the response found.
So the people in these studies were not responding to receiving an antidepressant per se, but rather to the thought that they were taking an antidepressant. Put another way, the meta-analysis demonstrated that there was a strong therapeutic response to antidepressant medication but the response to placebo was almost as strong.
As most clinicians know, severity of depression is one of the few predictors of response to treatment. What is perhaps less well known is that the type of antidepressant has little if any impact on treatment response. In a 2011 meta-analysis that compared the different types of second-generation antidepressant medications to each other, no clinically relevant differences in effectiveness were detected for the treatment of acute, continuation, and maintenance phases of major depressive disorder6. And this applied regardless of sex, age, ethnicity, co-morbid conditions or accompanying symptoms.
The bottom line appears to be that although the type of medication does not make a clinically significant difference in outcome, response to placebo does.7
And there is an interesting twist. Although the difference between active drug and placebo was not clinically significant – study participants got equal benefit whether they were taking the active drug or the placebo in each of the studies – the difference was significant statistically. This raises the question of what do antidepressants have in common that make their effect on depression slightly statistically significantly better than placebo?
The answer is side effects. If a patient knows they are in a clinical trial where they could receive either an antidepressant or a placebo, and that the active drug could cause side effects, research suggests subjects are very good at successfully guessing which arm of the study they are in, and this influences their response.8
Where to from here?
We could advocate for antidepressants to be used as active placebos but they often have significant side effects, some of which are serious, reduce quality of life, and are common reasons for patients wanting to stop their antidepressant medication. Knowing this, is it ethical to prescribe something where a safer alternative – a placebo – is available? Or we could go the other way and prescribe placebos as antidepressants. The practice of placebo prescribing is something clinicians are known to do. This is, however, another ethically fraught option. For a placebo to work a person must believe they are receiving active medication. Deceiving patients in this way is not great for building practitioner-patient trust.
So what are the options for managing depression? If the meta-analyses are anything to go by, there are no significant differences between antidepressants, psychotherapy (of any sort), the combination of antidepressants and psychotherapy, and other treatments such as physical exercise and acupuncture.6, 11 They all offer similar benefits to patients with depression.
So if there are no differences in treatment outcome between the various available treatment modalities, the ethical response is to choose those interventions with the least risk. From all the available evidence, antidepressants are the riskiest and most potentially harmful. These medications then seem best reserved as a last resort for the most refractory depression, where other options have been tried and failed.
This statement is controversial, but if the aim is to practice evidence-based medicine then the results of these international meta-analyses are worthy of serious consideration. The issue of how antidepressants have come to be approved by organisations such as the FDA – given the weight of the published and unpublished clinical trial data – is a whole other question.
Dr Ursula King is Medical Editor, The Medical Republic, and a rural emergency practitioner
1 Davey, C.G., and Chanen, A.M. The unfilled promise of the antidepressant medications, Med J Aust 2016; 204 (9): 348-350 doi: 10.5694/mja16.00194
2 Kirsch I., & Sapirstein G. Listening to Prozac but hearing placebo: A meta-analysis of antidepressant medication. Prevention and Treatment, 1998; 1, 2a doi: 10.1037/1522-3722.214.171.124a
3 Kirsch I., Moore T. J., Scoboria A., & Nicholls S. S. The emperor’s new drugs: An analysis of antidepressant medication data submitted to the U.S. Food and Drug Administration. Prevention and Treatment, 2002; 5, 23 doi: 10.1037/1522-37126.96.36.1993a
4 Turner E. H., Matthews A. M., Linardatos E., Tell R. A., & Rosenthal R. Selective publication of antidepressant trials and its influence on apparent efficacy. New England Journal of Medicine, 2008; 358, 252–260.
5 Khin N. A., Chen Y. F., Yang Y., Yang P., & Laughren T. P. Exploratory analyses of efficacy data from major depressive disorder trials submitted to the US Food and Drug Administration in support of new drug applications. Journal of Clinical Psychiatry, 2011; 72, 464–472. doi: 10.4088/JCP.10m06191
6 Gartlehner G., Hansen R. A., Morgan L. C., Thaler K., Lux L., Van Noord M., et al. Comparative benefits and harms of second-generation antidepressants for treating major depressive disorder. Annals of Internal Medicine, 2011; 155, 772–785. doi: 10.1059/0003-4819-155-11-201112060-00009
7 Kirsch, I. Antidepressants and the Placebo Effect, Z Psychol. 2014; 222(3): 128–134. doi: 10.1027/2151-2604/a000176
8 Rutherford B. R., Sneed J. R., & Roose S. P. Does study
design influence outcome? Psychotherapy and Psychosomatics, 2009; 78, 172–181.
9 Raz A., Campbell N., Guindi D., Holcroft C., Déry C., & Cukier O. Placebos in clinical practice: A Pan-Canadian Review of attitudes and patterns of use between academic psychiatrists and non-psychiatrists. Canadian Journal of Psychiatry, 2011; 56, 198–208.
10 Tilburt J. C., Emanuel E. J., Kaptchuk T. J., Curlin F. A., & Miller F. G. Prescribing “placebo treatments”: Results of national survey of US internists and rheumatologists. British Medical Journal, 2008; 337, 1097–1100. doi: 10.1136/bmj.a1938
11 Khan A., Faucett J., Lichtenberg P., Kirsch I., & Brown W. A. A systematic review of comparative efficacy of treatments and controls for depression. PLoS One, 2012; 7, e41778 doi: 10.1371/journal.pone.0041778