Research published this week based on multiple studies conducted over the past several years concluded that placebos can be effective even when those taking them are aware that they’re being given inactive pills.1 This was a surprising finding since most of us have been led to believe that placebos work because people think they’re being given active medication, suggesting that the “placebo effect” is dependent on an element of deception. Because of an ethical proscription against lying to patients in clinical practice, the use of placebos has therefore been historically limited to research although calls to integrate them into medical care as safer alternatives to medications are increasingly common.
With this week’s published research, the UK-based newspaper The Guardian ran an article with the headline, “I knew they were sugar pills but I felt fantastic’ – the rise of open-label placebos.” It told of a woman named Linda Buonanno who was enrolled in a clinical trial for the treatment of irritable bowel syndrome and despite being aware that she would be receiving an “inactive substance like a sugar pill” nonetheless felt better. “I know it’s crazy,” she said, “I felt fantastic… I was able to go out dancing and see my friends again.” When the study concluded, the article said, she was “unable to obtain further effective placebos and her symptoms returned.”
For those familiar with the science of placebos, this is something of a ridiculous statement, reflecting a widespread misunderstanding of placebos and how they might work.
In recent years, other research attempting to link placebo response to observable neurochemical changes in the body and brain has led to a flurry of media headlines touting the increasingly popular belief that placebos can "help people heal themselves." And yet, a more accurate view of placebos is that “they” aren’t really working at all.
To understand the details of why this is so, we need to start by defining a placebo as an inactive or inert pill. Although placebos are often referred to as “sugar pills,” a sugar pill isn’t an inert substance—if it was, no one would ever buy Skittles or any other pill-sized candy for that matter. So, modern day placebos aren’t sugar pills—they’re pills that look like medications, but contain no actual ingredients, sugar or otherwise, that cause any physiologic effects in the human body.
Next, we have to discuss the traditional role of placebos in research where they have served as comparisons to active medications in “randomized controlled trials” (RCTs). By randomly assigning half the subjects in a study to active medication and the other half to placebo, researchers can be confident that any observed differences in outcome between the two groups are due to medication since all of the other conditions that might affect an outcome are on average the same. Research subjects are usually “blinded” to whether they’ve been given active medication or placebo, as are the investigators rating their treatment response (this is known as “double-blinding”).
The importance of blinding was discussed in a previous Psych Unseen blogpost called, “The Blind Leading the Blind: Medications, Gluten, & Violins,” but the basic idea is that blinding prevents a study subject’s expectations from influencing the outcome of the study. It's well-known that the expectation of a treatment response can lead to the subjective perception of that response, whether positive (e.g. a subject feels better or reports a reduction in symptoms) or negative (e.g. side effects; the so-called “nocebo" response). For placebos, this mechanism of action has been called the “expectancy hypothesis,” though psychologically speaking, it’s simply an illustration of the power of confirmation bias. Believe an effect will happen and you’ll tend to look for evidence that it did. When blinding is well-done in a study, this confirmation bias is distributed equally across both treatment conditions and therefore cancelled out, so that it can’t be used to explain any observed differences in outcome between comparison groups.
What most people writing about placebos in the media forget is that the responses reported by research subjects taking placebos aren’t only due to expectation effects or confirmation bias. In an RCT, both positive and negative outcomes may also be a result of all of the other things that are occurring during study enrollment. For example, research subjects participate in assessments consisting of many hours of weekly face time with research staff who take thorough histories and closely monitor how a subject is feeling over the course of the study. This often amounts to much more time and attention than a patient would spend with their usual clinician in routine care.
This explains why placebo response in conditions like depression can often be as high as 50 percent—having someone listen to you and watch over you closely can make people feel better. Therefore, while research subjects treated with placebos are receiving an inert pill, that’s not the same as saying that they’re receiving “no treatment”—far from it.
Beyond the therapeutic effects of contact with caring individuals, research subjects are usually are paid for their time – usually modestly, but sometimes more significantly. They also often want to please the researchers and in the case of those with treatment refractory conditions, can feel thankful to have been enrolled in a study. These factors can lead research subjects to want to report positive effects, claiming that they feel better when they don’t necessarily feel any different.
It’s just as important to keep in mind that the reported outcomes in clinical trials might have nothing to do with the experimental conditions at all. Again taking the example of depression, symptoms like mood or insomnia often ebb and flow in response to other things happening in one's life or for no detectable reason at all. Likewise, side effects like headaches might occur from any of a number of reasons, independent of being in a research study or taking a medication or a placebo.
With this background in place, we can now discuss what new research adds to our understanding about placebo response. The paper that made headlines this week was a meta-analysis of five different studies that each intended to examine the therapeutic potential of placebos in various medical and psychiatric conditions. Accordingly, placebos were used not as controls, but as the active treatment arm of the studies in which they were administered in “open-label” fashion without blinding. In other words, unlike in a traditional RCT, placebo-treated subjects were aware that they were being given placebos. Since in aggregate the studies found that the placebo-treated patients did better than patients randomized to control treatments, the meta-analysis concluded that “open-label placebos appear to have positive clinical effects compared to no treatment.”
To examine whether or not that conclusion is sound, let’s pick a representative study from the five included in the meta-analysis—a 3-week clinical trial involving 80 subjects with irritable bowel syndrome published in 2010.2 This study compared placebo—an inert pill—to “no treatment” and found that placebo-treated patients had significantly greater improvements in reported symptoms. As with the meta-analysis, the published study concluded that “placebos administered without deception may be an effective treatment for irritable bowel syndrome.”
First of all, it should be noted because the study was not blinded, the outcomes are biased against the no treatment condition. That said, the patients who received "no treatment" did improve, highlighting the potential therapeutic value of simply being in a research study. But the critical issue here is how to explain the greater response to placebo and what to make of the investigators’ claim that the placebo in the study was administered “without deception.” In the actual study, while the placebo-treated patients were given pills that were “truthfully described as inert or inactive pills, like sugar pills, without any medication in it,” they were also told that “placebo pills, something like sugar pills, have been shown in rigorous clinical testing to produce significant mind-body self-healing processes.” This instruction was part of a “fifteen minute script” read to patients at the start of the study explaining that “the placebo effect is powerful” and “the body can automatically respond to taking placebo pills like Pavlov’s dogs who salivated when they heard a bell,” with another reminder of these discussion points repeated half-way through the study. The no-treatment subjects, meanwhile, were presumably told nothing about what to expect.
Although the researchers were no doubt genuinely confident in the ability of placebos to produce “powerful mind-body self-healing processes,” that claim is at best debatable and at worst misleading and deceptive, despite claims to the contrary. For one thing, placebos by definition have no physiologic effects such that the term “placebo effect” itself isn’t really accurate. Rather, the observed benefits of placebo over no treatment in an open-label study like this are best attributed to confirmation bias, not the actual pills (the investigators could have addressed this by using a better control condition that involved administering inert pills without telling subjects that they were “powerful”). In other words, the benefit of placebo in this study was most likely due to subjects being told that they were being given a powerful treatment that promoted self-healing, not what they were actually given. With a similar claim, the placebo pill could have just as effectively been replaced by having the subjects wear a charm, receive a prayer on their behalf, or get smacked on the forehead while the researchers shouted "you are healed!"
In the end then, the claim that "placebo pills have been shown in rigorous clinical testing to produce significant mind-body self-healing processes" is, strictly speaking, a lie. More to the point, Linda Buonanno’s placebo response didn’t depend on access to “effective placebo pills” at all.
When researchers use placebos as an active treatment condition, they’re not so much studying the healing power of placebos as they’re studying the therapeutic potential of expectation, confirmation bias, and the power of suggestion. Stated in a more positive way, they’re studying hope. Dr. Ted Kaptchuck, Professor of Medicine at Harvard Medical School and lead author of the irritable bowel syndrome study discussed above, has written that “placebo effects are improvements in patients’ symptoms that are attributable to their participation in the therapeutic encounter, with its rituals, symbols, and interactions.”3
And yet, while hope has the potential to lift one’s spirit in the face of illness, even Kaptchuck tempers his enthusiasm for placebos with realistic clarifications about their limits:
“…though placebos provide relief, they rarely cure. Although research has revealed that objective neurobiological pathways and correlates of placebo responses, the evidence to date suggests that the therapeutic benefits associated with placebo effects do not alter the pathophysiology of disease beyond their symptomatic manifestations; they primarily address subjective and self-appraised symptoms.”
In the final analysis then, research has revealed that placebo response mostly consists of the ability to feel better in the setting of hope and expectation. The importance of that hope, expressed within the “therapeutic alliance,” is something that all clinicians should attempt to maximize in their relationships with patients struggling with illness. But a good clinician doesn’t depend on deception to do that, any more than they need a placebo.
1. Charlesworth JEG, Petkovic G, Kelly JM, et al. Effects of placebo without deception compared with no treatment: a systematic review and metanalysis. Journal of Evidence-Based Medicine 2017 DOI: 10.1111/jebm.12251
2. Kaptchuk TJ, Friedlander E, Kelley JM, et al. Placebos without deception: A randomized controlled trial in irritable bowel syndrome. PLoS One 2015; 5:e15591.
3. Kaptchuk TJ, Miller FG. Placebo effects in medicine. New England Journal of Medicine 2015; 373:8-9.