In a recent editorial for The New York Times, researcher Ted J. Kaptchuk, who directs placebo studies at Harvard, gives his summary of the current state of research (much of it his own) into placebo effects. While much of what he says is true, or at least uncontroversial, in my opinion he shoehorns the facts into his preferred narrative – a popular narrative that can be counterproductive and feeds into unscientific medical treatments.
He lays out one of his premises here:
I’m a researcher who studies the placebo effect, and in some situations, it’s powerful. That said, oral phenylephrine sold over the counter should be removed from the market; despite some people’s love of phenylephrine cold medicines, there’s no evidence that the drug even provides placebo benefits. In clinical trials reviewed by the F.D.A. committee, phenylephrine and a placebo affected patients’ perceptions of nasal congestion equally, but the existing trials do not tell us to what extent people felt better because of placebo effects or because their colds simply resolved on their own.
There is already a lot to unpack here, starting with the premise that the placebo effect can be powerful. But I cannot really comment on a thing without having a working definition of what that thing is. He takes a while to give a definition of placebo effects, which he characterizes as, “Placebo effects are health improvements initiated from the rituals, symbols and behaviors involved with healing.”
That is an inaccurate definition, which I think biases the article and his entire approach to placebo effects. They are apparent or measured “health improvements”, if you include subjective symptoms as health improvements, which is reasonable but needs to be explicitly stated. Apparent improvements may not be real. They may entirely be illusions of how health outcomes are being measured.
For example, there is regression to the mean. Symptoms fluctuate, and people are more likely to seek treatment when symptoms are at their worst, which is statistically likely to be followed by spontaneous improvement. Give someone an inert placebo and they will improve. The improvement was not “initiated” by the placebo in any way. It would have happened without any intervention. It is simply a statistical illusion of observation – but such effects are included in what we measure as placebo effects.
There are other illusory effects in there as well. Clinical trial subjects, for example, may be biased in their perceptions and reporting of their symptoms because of their perceived expectations of the researchers, and their own desires to have the treatment work.
Kaptchuk is focusing on only a subset of placebo effects and treating it as if it is the entire phenomenon. But this fits his narrative, that placebo effects are something to be harnessed and are valuable in and of themselves. He states, correctly, that phenylephrine should have been removed from the market, but not only because it does not work, because it has not been demonstrated that it provides placebo benefit. The implication being that if it did, then it should stay on the market.
The non-specific effects of forming a therapeutic alliance, of the hope and positive human interaction that results, can be psychologically very powerful. For subjective symptoms that appear to have a large psychological component, such as pain and nausea, psychological interventions can be very effective and should be leveraged, especially when treating chronic symptoms. But are such interventions even “placebo” effects? It depends.
In clinical trials placebo effects are everything that happens in the placebo arm of the trial – everything other than a physiological response to an active intervention. So if you are studying the pharmacological effects of a drug, incidental psychological effects are placebo effects. But if you are studying psychological interventions as a treatment, then they are the treatment, not the placebo.
He hints at this dual nature of placebos with his next premise:
In research settings, placebo responses are powerful but a nuisance, as they make detecting a drug’s superiority over a placebo difficult. And in clinical practice they are powerful, but they often require deception, making them unethical. But can placebos ever come out of the shadows and become a legitimate component of health care? My research suggests so.
Again, I would push back against the blanket statement that placebo effects are “powerful”. The research shows they are mostly illusory. For subjective symptoms they can be significant. But as he says, in clinical practice they are not a nuisance, they are part of the treatment. He then sets up a straw man – that the only problem with leveraging placebo effects in clinical practice is that the use of placebos requires deception. But this is false, because it contains a massive and incorrect unstated assumption, that you don’t get placebo (or nonspecific psychological) effects from ethical effective treatments.
If you give a patient an effective painkiller, for example, in addition to a good therapeutic alliance, you will get the full placebo effect plus the pharmacological benefits of the medication. What if you want to avoid the side effects or risk of medication? For most patients you can start with the most benign intervention, such as low dose acetaminophen. You don’t have to give fake medicine to get the placebo effect. You can also measure and then correct any low vitamin levels. And of course, you can also directly give psychological or physical (non-pharmacological) interventions. Most patients have something positive and effective to focus their attention on – sleep hygiene, diet, physical activity, or stress reduction.
But Kaptchuk argues as if placebo effects only come from fake or inert interventions that theoretically require deception. This is why some of his research has focused on that element – what if we remove the deception, do placebo effects from fake medicine still happen? I think this premise is wrong. Kaptchuk summarizes his own research in this area:
As it turns out, placebos can work even when patients know they are getting a placebo. In 2010 my colleagues and I published a provocative study showing that patients with irritable bowel syndrome who were treated with what we call open-label placebos — as in, we gave them dummy pills and told them so — reported more symptom relief compared with patients who didn’t receive placebos.
David Gorski has already critiqued that 2010 study. Patients were told the pills they were receiving were sugar pills, but they were also told, “placebo pills, something like sugar pills, have been shown in rigorous clinical testing to produce significant mind-body self-healing processes”. That is not true (subjective improvement in symptoms is not “self-healing”), and therefore he did not remove deception from the intervention.
In the end I think Kaptchuk is well-meaning but misguided. Levering psychological effects in the treatment of chronic symptoms like pain is a good idea, and there are already pushes to increase this within standard medical treatment. It is likely true that we do not utilize these effect enough. This is mainly an infrastructure problem – it’s a lot easier to prescribe a medication than provide psychological intervention. There simply are not enough providers to give everyone who could benefit such interventions, and insurance needs to cover these treatments. That is where these efforts should be focused.
However, conflating legitimate psychological interventions (whether they are part of the therapeutic alliance or a separate intervention) with “placebo effects” is counterproductive. It also feeds into an entire industry of fake medicine that partly justifies itself with hand-waving statements about how powerful placebos are. Further, since the term “placebo effects” are used to refer to any outcome, not just subjective symptoms, it furthers the confusion. Placebo effects for hard outcomes, like survival and improved physiological parameters, are entirely illusory. But this fact often gets lost in the cheerleading for placebo medicine.
Unambiguous technical definitions are required for clear scientific and clinical thinking. This is critically lacking in popular discussions of placebo effects, and unfortunately Kaptchuk is playing into that confusion.