A sham control is a procedure built to look and feel like acupuncture to the person receiving it while leaving out the part the trial is testing: the needle going through the skin, or going in at the chosen points. A 2003 review by Dincer and Linde at the Technical University of Munich looked at 47 randomised trials that compared real and sham acupuncture and found five different kinds of sham among them. Every kind in use touches, presses or punctures the skin, takes the same time and attention as the real treatment, or both, and the trials that measured it found people improve on it. A result reported “versus sham” is therefore the difference between two procedures that each do something. This page sets out the kinds of sham, what was measured about each, and what the comparison can and cannot tell a reader.
What a sham is meant to separate
Acupuncture comes with a ritual: a consultation, a practitioner’s attention, lying still with the needles in place, the expectation of benefit. A trial that compares acupuncture with no treatment measures all of that together. A sham arm is meant to keep the ritual and remove the needling, so that any extra improvement in the acupuncture arm can be credited to the needling itself, which trial reports call the specific effect. Everything else is the nonspecific effect. The sham also keeps patients from knowing which arm they are in, which matters because most outcomes in these trials, such as pain, are rated by the patients themselves. Reading an acupuncture trial covers blinding and the other design questions in full.
The kinds in use
Dincer and Linde sorted the shams in their 47 trials as follows.
| Kind of sham | Trials | What the patient receives |
|---|---|---|
| Superficial needling at the real points | 2 | Needles at the points the real treatment uses, inserted only a little way |
| Needling at real points not indicated for the condition | 4 | Full needling, at points the tradition does not use for that complaint |
| Needling outside acupuncture points | 27 | Needles at sites that are not acupuncture points, usually shallow; the German GERAC trials used this kind |
| Placebo needles | 5 | A needle that retracts into its handle or rests on the skin without going through it, such as the Streitberger needle and the Park sham device |
| Pseudo-interventions | 9 | A device that looks active but is switched off, such as a laser |
The review also found that real and sham treatments in the same trial often differed in more than the sham itself: in how the needles were manipulated, how deep they went, and whether the practitioner sought de qi, the aching or spreading sensation described on Needling and de qi. It found no clear link between the kind of sham a trial used and its result, and concluded that grouping all these procedures together as “placebo” controls was misleading. The US Department of Veterans Affairs evidence map of 2014 lists the same range, adding deactivated electrical stimulation, and notes that no standard defines what an appropriate sham is. A 2001 German trial of 177 people with chronic neck pain used an inactivated laser as its sham: one week after five treatments, acupuncture had reduced pain on movement more than massage did, but not significantly more than the sham laser.
Needles that do not go in
Streitberger and Kleinhenz at the University of Heidelberg described the first placebo needle in 1998. Its blunt tip slides back into the handle when pressed against the skin, so the shaft appears to shorten as a real needle does when it goes in. In a crossover test with 60 volunteers, 54 felt the real needle penetrate and 47 felt the placebo penetrate; 34 reported de qi with the real needle and 13 with the placebo. None suspected that the placebo had not gone through the skin.
Park and colleagues at the University of Exeter validated a sham device in 2002 that holds a similar telescoping needle in a guide tube on an adhesive base. In one trial, none of 58 stroke patients believed they had received the sham. In a second, of 63 volunteers who had never had acupuncture, the 40 on whom a panel of acupuncture experts agreed were about 15 times as likely to have felt de qi with the real needle as with the sham. The authors took that as evidence the device was inactive, using de qi as the stand-in for activity.
Both devices hide the needle from the patient but not from the practitioner, who knows which needle is in the hand. Takakura and Yajima in Tokyo built a pair of needles in 2007 meant to mask practitioners as well: an opaque guide tube holding either a penetrating needle or one whose tip only presses on the skin. Ten acupuncturists each placed 40 of them at LI04 (Hegu) and, on average, judged 17 correctly, 16.6 incorrectly and 6.4 as unidentifiable: about as many right as wrong. In 60 volunteers who each received one of each at TB05 (Waiguan), the needle that did not penetrate still produced a sensation of penetration in 42% of applications and de qi in 33%.
Why none is inert
A sugar pill is inert in the sense that it has no pharmacological action. No sham acupuncture procedure is inert in that sense, and the sources give three reasons.
First, touch acts on the nervous system. Lund and Lundeberg at the Karolinska Institute in Stockholm argued in 2006 that light touch on the skin stimulates a class of slow nerve fibres, the C tactile afferents, which signal to a brain region, the insula, involved in emotion rather than to the areas that map touch. They proposed that shams which press or prick the skin trigger this response and relieve the emotional component of pain, which would explain why sham did as well as acupuncture for migraine and low back pain in some trials. The paper states this as a hypothesis; it reports no new experiment.
Second, a needle that goes in is a needle. Hugh MacPherson and colleagues in the Acupuncture Trialists’ Collaboration reanalysed 20 sham-controlled trials of chronic pain in 2014. Acupuncture beat every kind of sham, but by less where the sham needles penetrated the skin than where they did not or where the sham used no needles at all. The authors concluded that penetrating needles appear to have “important physiologic activity” and recommended that trials stop using them as shams.
Third, the ritual itself acts, and a needle ritual more than a pill. Kaptchuk and colleagues at Harvard Medical School compared two placebos in 270 adults with persistent arm pain: a validated sham needle with a blunt retracting tip, twice a week for six weeks, or a placebo pill daily for eight weeks. Over the whole course, pain fell faster on the sham needle, by 0.33 points a week on a 10-point scale against 0.15 on the pill, though the two did not differ during the first two weeks. The authors concluded that placebo effects depend on the behaviour built into a medical ritual.
How large the sham effect is
Three pooled analyses measured the gaps between acupuncture, sham and no acupuncture across many trials. All report standardised mean differences (SMD), which put different pain scales on one footing by expressing a difference in units of the spread of scores between patients.
| Analysis | Trials and patients | Acupuncture against sham | Sham, or acupuncture, against no acupuncture |
|---|---|---|---|
| Madsen, Gøtzsche and Hróbjartsson, 2009; three-armed pain trials | 13 trials, 3,025 patients | SMD 0.17, about 4 mm on a 100 mm pain scale | Sham against none: SMD 0.42 |
| Linde and colleagues, 2010; trials with both a sham and a no-acupuncture arm, any condition | 37 trials, 5,754 patients | Not pooled; trials with larger sham effects reported smaller acupuncture-over-sham effects | Sham against none: SMD 0.45 |
| Vickers and colleagues, 2018; individual patient data, four chronic pain conditions | 39 trials, 20,827 patients | Close to 0.2 SD | Acupuncture against none: close to 0.5 SD |
The three groups read similar numbers differently. Madsen and colleagues, in Copenhagen, judged the difference between acupuncture and sham small enough to lack clinical relevance and impossible to separate from bias, noting that the practitioners were unblinded in all 13 trials. Linde and colleagues concluded that sham procedures carry moderately large nonspecific effects, larger perhaps than inert placebos, which would make a small specific effect hard to detect. Vickers and colleagues concluded that acupuncture is effective for chronic pain and that its effect cannot be explained by placebo alone, while finding that the size of the effect in a trial depended mainly on what the control group received.
The German trials: sham and standard care in the same study
The German Acupuncture Trials (GERAC), published in 2006 and 2007, compared real acupuncture with sham acupuncture made of shallow needling at non-acupuncture points and, in three of them, with standard medical care, in more than 3,000 patients treated in hundreds of practices. The tension-type headache trial, run by many of the same investigators, had no standard-care arm. Each acupuncture arm received about ten sessions over roughly six weeks.
| Condition | Patients | Real acupuncture | Sham acupuncture | Standard care |
|---|---|---|---|---|
| Chronic low back pain (Haake, 2007) | 1,162 | 47.6% responded | 44.2% responded | 27.4% responded (drugs, physiotherapy and exercise) |
| Knee osteoarthritis (Scharf, 2006) | 1,007 | 53.1% success | 51.0% success | 29.1% success (physiotherapy and anti-inflammatory drugs as needed) |
| Migraine prevention (Diener, 2006) | 960 randomised, 794 analysed | 2.3 fewer migraine days a month | 1.5 fewer | 2.1 fewer (preventive drugs) |
| Tension-type headache (Endres, 2007) | 409 | 33% responded | 27% responded | No standard-care arm |
For back pain and knee osteoarthritis, both acupuncture arms did better than standard care and did not differ from each other. For migraine, none of the three groups differed. For tension-type headache, the primary measure did not separate real from sham, but real acupuncture came out ahead on most secondary measures, including 1.8 fewer headache days a month. The knee trial’s authors listed three explanations for real and sham matching: placebo effects, the greater contact time patients in both needling arms had with a practitioner, or a physical effect of needling regardless of where the needles go. Their trial could not choose between them, and they noted that patients could not be blinded to whether they received acupuncture or standard care.
Reading a result reported against sham
A trial that finds acupuncture no better than sham has shown that, in that trial, the chosen points and depths added nothing measurable to the sham procedure it used. It has not shown that patients did not improve: in the GERAC back pain and knee trials, both arms improved more than standard care. A trial that finds acupuncture better than sham has shown a difference over that particular sham, whose size depends on how active the sham was. The STRICTA reporting standard asks every trial to say why it chose its control and to describe a sham in the same detail as the real treatment, so the report should state what the sham group received.
AcuiQ’s protocols come from trials with every kind of control, and an entry records what a study prescribed, not what it was compared against. Reading a protocol explains what each field holds, and AcuiQ’s low back pain page lists the protocols individual studies prescribed, each with its citation, so the comparison can be read in the paper.
What the sources cannot settle
None of these studies establishes whether the location of a needle matters, because no sham has been shown to be inactive: if real and sham match, either location does not matter or the sham is active enough to hide the difference. The pooled analyses are mostly of chronic pain, so the size of sham effects in other conditions rests on fewer trials. Practitioner blinding remains rare: Takakura’s needle tested masking, not outcomes, and the reviews above drew almost entirely on trials in which the person holding the needle knew which arm the patient was in.