The placebo effect and the meaning response

What it is

The measurable change produced by the context of a treatment rather than by its active ingredient — the ritual of taking something, the expectation that it will help, the attention of whoever provides it, and the act of doing something about your own suffering.

It is the single most useful idea on this site, and it is almost always described wrongly. Two corrections do most of the work.

It does not mean "nothing happened" or "you imagined it". Placebo analgesia can be blocked by naloxone, a drug that blocks opioid receptors, which means the brain is releasing its own opioids in response to expectation. In Parkinson's disease, placebo administration produces measurable dopamine release in the striatum. These are physiological events, not politeness.

But it is also much smaller than folklore claims. The widely repeated figure that "a third of patients respond to placebo" comes from a 1955 paper by Henry Beecher that could not distinguish the placebo effect from people simply getting better on their own. Most illnesses fluctuate; people seek help when they feel worst, and then regress toward their average. A great deal of what looks like placebo response is that, plus a tendency to report improvement to someone who is being kind to you.

Some researchers prefer the meaning response, a term from the anthropologist Daniel Moerman, because what is doing the work is not the sugar pill but what the whole encounter means to the person in it. That framing travels better to practices that involve no pill at all.

Where it came from

The word is Latin — placebo, "I shall please" — and entered English through the medieval Office of the Dead, where it acquired an unflattering sense of flattery. It appears in medicine by the late eighteenth century, meaning a treatment given to satisfy rather than to cure.

Beecher's "The Powerful Placebo" (1955) is the founding modern document and is also the source of the overstatement. He reviewed trials in which patients improved on inert treatment and concluded the placebo itself was responsible. The trials had no untreated comparison group, so there was no way to separate the placebo from time passing.

Hróbjartsson and Gøtzsche corrected this in 2001, in the New England Journal of Medicine, by doing something obvious that nobody had done: they collected trials containing both a placebo arm and a genuine no-treatment arm, so the two could be compared directly. Across those trials, placebo had little effect on objective or binary outcomes. It had a modest but real effect on continuous subjective ones, and the effect was clearest for pain.

Ted Kaptchuk's group at Harvard then produced the result that unsettled everyone's assumptions. In a 2010 trial in irritable bowel syndrome, patients were given inert pills in a bottle openly labelled "placebo", and told plainly that the pills contained no active medication. They improved anyway, more than the no-treatment group. Open-label placebo should not work if deception is the mechanism, and it does, which means the mechanism is something else.

How it is used

Three ways, and they pull in different directions.

As a control. This is why trials are blinded at all. If you want to know what a substance does, you have to subtract everything that happens around it, and the placebo arm is how that subtraction is performed.

As a defence. "It might be placebo, but it works for me" is offered constantly in support of practices with no evidence, and it is doing more work than the speaker usually realises — it concedes the substance and defends the experience.

As a dismissal. "That's just placebo" is used to end conversations, and it is unkind and imprecise. A response mediated by expectation is still a response, and telling someone their relief was not real is both wrong and needlessly cruel.

What we can and cannot say

We can say exactly where the effect is strong and where it disappears, and one study shows it better than any argument.

In 2011 Michael Wechsler and colleagues published an asthma trial in the New England Journal of Medicine with four arms: albuterol inhaler, placebo inhaler, sham acupuncture, and no treatment. Patients rated their improvement after each. By self-report, the real inhaler, the fake inhaler and the sham acupuncture all performed about equally, and all three beat doing nothing.

Then the researchers measured FEV1 — how much air the patient could actually force out of their lungs. Only albuterol improved it. The placebo arms were flat.

That is the whole subject in one experiment. The feeling of improvement and the improvement came apart, and only one of them was the drug. For a symptom like pain, where the feeling is the outcome, that gap does not matter much. For asthma, where you can feel fine while your airways close, it can kill you.

We can say the effect has components, and they can be separated. A 2008 study by Kaptchuk's group varied the encounter itself: some IBS patients received sham acupuncture briskly, others received it with a warm, attentive, unhurried practitioner. The warm condition produced substantially more improvement. The ritual, the attention and the relationship are doing identifiable work, independent of anything in the treatment.

We can say the same mechanism runs in reverse. The nocebo effect — expectation producing harm — is well documented, and it is why listing side effects can generate them, why patients told a drug is a cheaper generic report more problems, and why negative expectation is a genuine clinical difficulty rather than a curiosity.

We can say what this does and does not license, which is the reason this article exists.

It does not license inferring anything about a substance from how a person felt after taking it. A real response to the context of a treatment is not evidence about the content of it. Every honest testimonial on earth is compatible with the material doing nothing at all, and that is not an insult to the person giving it.

It also does not license contempt. The things that generate the meaning response — being attended to, taking your own distress seriously, marking a bad period with an act, having a practice at all — are not frauds. They are among the more reasonable things a person can do, and much of conventional medicine benefits from them too.

What follows is a question about cost rather than about mechanism. An inert practice that is cheap, safe, honest about itself, and displaces nothing is in a different category from one that is expensive, hazardous, sold on a false claim, or standing between someone and a treatment that would work. The first is a ritual. The second is covered under medical harm.

We cannot say how much of any individual experience was which. No one can, from the inside. A single person, unblinded, who wanted it to work, cannot separate the substance from the ritual from the ordinary course of getting better — and this is the whole reason the trial apparatus exists, rather than a slight on anyone's judgement.

Further reading

• Wechsler et al., "Active Albuterol or Placebo, Sham Acupuncture, or No Intervention in Asthma" (NEJM, 2011) — short, readable, and the most clarifying paper on this list.
• Hróbjartsson and Gøtzsche, "Is the Placebo Powerless?" (NEJM, 2001).
• Daniel Moerman, Meaning, Medicine and the "Placebo Effect".
• Ted Kaptchuk's open-label placebo papers, which are freely available and genuinely strange.