Placebo and expectation

The placebo response in an acupuncture trial is the improvement that comes from everything around the needle: the consultation, the practitioner’s manner, the patient’s expectation of benefit and the ritual of lying still with needles in place. Several trials have set out to measure these parts on their own, by varying how much a practitioner talks to a patient, by asking patients before treatment what they expect, or by telling patients outright that their treatment is a placebo. This page reports what they found. How sham controls are built, and how far real acupuncture beats them, is on Sham acupuncture; this page keeps to expectation, practitioner contact and ritual as quantities a trial can measure.

Practitioner contact as a dose

Ted Kaptchuk and colleagues at Harvard Medical School ran a trial in 2008 that broke the placebo response into three layers and added them one at a time. They randomised 262 adults with irritable bowel syndrome (IBS) to a waiting list, where they were only assessed and observed, or to sham acupuncture with a retracting needle that does not pierce the skin, given in one of two ways. In the “limited” arm the practitioner spent under five minutes at the first visit, said they had been instructed not to converse with patients because this was a scientific study, placed six to eight sham needles at non-acupuncture points and left the patient alone for 20 minutes. In the “augmented” arm the same sham treatment followed a 45-minute first visit scripted for warmth, attention and confidence, with questions about the patient’s symptoms, relationships, lifestyle and what they took the illness to mean. Both sham arms were treated twice a week.

Symptom severity and quality of life rose in the same order, the augmented arm beat the limited arm on all four outcomes, and the results held at six weeks. The authors described the pattern as resembling a graded dose escalation and concluded that the patient-practitioner relationship was the most robust component. A follow-up analysis by John Kelley and colleagues, using videotapes of the visits, found that outcomes differed between the individual practitioners despite the scripted interactions, and that this difference was twice as large as the difference between the arms. AcuiQ’s irritable bowel syndrome page covers the trials of real acupuncture for the condition.

Expectation measured before treatment

A second group of studies asked patients what they expected before randomisation and then followed the outcome. Daniel Kalauokalani and colleagues analysed 135 patients with chronic low back pain from a trial of acupuncture against massage in 2001. Before randomisation each rated how helpful they expected each treatment to be. At 10 weeks, after adjustment for baseline differences, function had improved in 86% of those with higher expectations for the treatment they were given against 68% of those with lower ones. Patients who expected more from massage did better with massage than with acupuncture, and the reverse held for those who expected more from acupuncture.

Klaus Linde and colleagues in Munich pooled four German trials in 2007, covering migraine, tension-type headache, chronic low back pain and knee osteoarthritis, in which 864 patients received 12 sessions over eight weeks of either acupuncture or minimal acupuncture, meaning shallow needling at non-acupuncture points. Patients were asked at the start whether they thought acupuncture worked in general and what they expected for themselves, and after the third session how confident they were that their treatment would help them. A responder was a patient whose main outcome improved by at least half. Patients who rated acupuncture as effective or highly effective had 1.67 times the odds of responding compared with sceptical patients; the odds ratios for high personal expectation and for confidence after the third session were 2.03 and 2.35. The association held after adjusting for condition, treatment arm, age, sex and how long the complaint had lasted, and it held at six months. The authors reported it as an association: patients who expect more may differ in other ways that also predict improvement.

The results are not uniform. Ben Colagiuri and Caroline Smith reviewed nine studies in 2012 that measured or manipulated expectation about acupuncture and related it to a health outcome. Five found a statistically significant effect of expectation on at least one outcome, and three found signs that expectation acted differently with real and sham acupuncture. The designs varied too much to pool.

Believing you had the real treatment

A blinding check asks trial patients at the end which arm they think they were in. R. Barker Bausell and colleagues at the University of Maryland used those answers in 2005 to reanalyse two sham-controlled trials of acupuncture for pain after dental surgery. Neither trial found a statistically significant difference in pain between acupuncture and placebo. In both, patients who believed they had received real acupuncture reported significantly less pain than those who believed they had received the placebo, and the authors found that belief related to pain more strongly than the treatment they were given.

Expectation set by the practitioner

Maria Suarez-Almazor and colleagues in Houston trained acupuncturists in 2010 to deliver treatment in one of two styles, conveying high expectations or neutral ones, and randomised patients with knee osteoarthritis to a waiting list or to one of the two styles, with traditional Chinese acupuncture or sham acupuncture given twice a week for six weeks within each style. Of the patients, 455 received treatment and 72 waited. Traditional acupuncture did not differ from sham, both reduced pain more than waiting, and patients treated in the high-expectation style reported more pain reduction and more satisfaction. About half of each needling arm thought they had received the real treatment, which the authors read as successful blinding. They concluded that part of the pain relief in acupuncture can be mediated by placebo effects tied to the acupuncturist’s behaviour. Knee osteoarthritis covers the wider evidence for that condition.

The ritual also carries side effects

Kaptchuk’s group compared a sham needle with a placebo pill in 2006 in 270 adults with persistent arm pain; its pain results are on the sham page. The same trial recorded what each placebo was said to cause. At consent, patients were warned that acupuncture could temporarily worsen pain, and that amitriptyline, the drug the pill was standing in for, could cause sleepiness, dry mouth, dizziness and restlessness. At two weeks, 15% of the sham needle group reported pain during treatment and 10% more pain after the sham needle was removed, while 20% of the pill group reported drowsiness and 19% dry mouth. Three people withdrew from the pill group because of fatigue or dry mouth, although the placebo capsule held only cornstarch. The authors wrote that the side effects reported mirrored the information patients had been given; an adverse effect that arises from what a patient was told is called a nocebo effect.

How large placebo effects are across medicine

Asbjørn Hróbjartsson and Peter Gøtzsche of the Nordic Cochrane Centre reviewed 234 trials in 2010 that randomised patients to a placebo or to no treatment at all, across 60 conditions. They reported effects as standardised mean differences (SMD), which put different scales on one footing by expressing a difference in units of the spread of scores between patients. Placebo had a small overall effect on continuous outcomes, SMD 0.23, and on pain specifically 0.28, with wide variation even among trials at low risk of bias. Four similarly designed acupuncture trials by an overlapping group of authors reported a large placebo effect on pain, 0.68, while three other pain trials reported little or none, 0.13. Across all trials, physical placebos such as sham acupuncture went with larger effects, as did outcomes the patient reported or took part in, small trials, and trials that did not tell patients a placebo might be used. The authors did not find that placebo interventions have important clinical effects in general; in some settings they can influence patient-reported outcomes, pain and nausea especially, though such effects are hard to tell apart from biased reporting.

Does a placebo need to be hidden?

Two trials told patients openly that they were getting a placebo. Kaptchuk and colleagues randomised 80 patients with IBS in 2010 to no treatment or to pills presented honestly as inert, both with the same quality of contact with the providers. At three weeks the open-label placebo group scored 5.0 on the 1 to 7 global improvement scale against 3.9 without treatment, and also did better on symptom severity and adequate relief. The authors concluded that placebos given without deception may help in IBS and called for further research.

Sujin Lee and colleagues at Kyung Hee University in Seoul tested open-label placebo acupuncture in 2022 in 34 healthy volunteers, of whom 31 were analysed, using heat applied to the hand. Each received an open-label placebo pill, open-label placebo acupuncture and no treatment on three separate days. Both placebos reduced pain, and in the acupuncture condition the effect depended on what participants expected and on who they understood the practitioner to be. The study measured laboratory pain in healthy people, not a clinical condition.

What this means for reading a trial

These studies measure the parts of acupuncture that a sham control is built to reproduce. A trial against waiting list or usual care counts them in the treatment’s effect; a trial against sham cancels most of them, but only as far as the practitioner treats both arms alike, which is hard to hold when the practitioner knows which needle is real. A report that states what patients were told at consent, how long visits lasted, and whether expectation was measured and how it differed between arms lets a reader judge how much of the result could belong to contact and expectation. Reading an acupuncture trial sets out the reporting standard and the bias checks that ask for these details. AcuiQ’s protocols record what each study prescribed, not how its practitioners spoke to patients, so a protocol entry carries none of this information, and the paper behind it is where to look.

What the sources cannot settle

The expectation studies are mostly secondary analyses of trials designed for another question, and patients who expect more may differ in ways the analyses did not adjust for. The trials that set contact or expectation deliberately are few, single-centre and mostly in IBS and knee osteoarthritis, and every outcome they report is one the patient rated. None of these sources establishes how much of the benefit seen in routine acupuncture comes from expectation and contact rather than needling, because a routine visit delivers both together. Mechanism studies in people reports what naloxone, brain imaging and tissue sampling found during needling.