Placebo, Sham Controls and Blinding in TCM Trials

The problem in plain words

A good trial of a drug uses a look-alike pill with no active ingredient, so nobody knows who got what. Many traditional Chinese medicine (TCM) treatments are hard to disguise: a needle is felt, a herbal decoction has a strong smell and taste, a massage is obviously a massage. Researchers therefore use sham (pretend) procedures and placebo (inactive look-alike) products, and each has weaknesses. NCCIH, the US government's complementary health research center, notes that if patients or investigators know who received which treatment, it may affect their impression of improvement however hard they try. [1]

Evidence labels: established, promising but limited, mixed, insufficient evidence.

What counts as a sham in acupuncture

Sham acupuncture comes in several forms, from devices that poke the skin without penetrating, to shallow needling, to needles placed at non-acupuncture points. NCCIH gives the example of a device that touches the skin but does not penetrate it. [2] These are not equally inert:

• A 2014 analysis of individual patient data in chronic pain found acupuncture was superior to every kind of control, but its advantage was smaller when the sham used penetrating needles than when the sham was non-penetrating or needle-free. The authors concluded that penetrating needles appear to have important physiologic activity and recommended avoiding that type of sham. [3]
• So a sham is a comparison with its own effects, not a neutral zero. Which sham a trial used changes the size of the result, which is why trials with different shams are hard to compare. [3]

Reading the results fairly

The largest pooled analysis, of 39 trials and 20,827 people with nonspecific musculoskeletal pain, osteoarthritis, chronic headache or shoulder pain, found acupuncture beat both sham and no-acupuncture control. Differences were close to 0.5 standard deviations against no acupuncture and close to 0.2 against sham, with only about 15% of the effect lost by one year. [4] The authors concluded decreases in pain could not be explained solely by placebo effects. [4]

Another reading of the same data is that, against usual care, about 60% of the effect is attributed to non-specific context effects plus sham and about 40% to the specific benefit of true acupuncture. [5] Both readings agree on one point: the larger effect seen against usual care is mostly attributed to context effects plus sham, not to needle position alone, and a sham arm often has a real therapeutic effect. [5]

Practical consequence: "better than sham" and "better than nothing" are different questions. NCCIH reports, for back or neck pain, 12 studies (8,003 participants) versus no treatment and 10 studies (1,963) versus sham, with a larger difference against no treatment. [2] For a patient, the "no treatment" comparison asks whether a course of care helps; the sham comparison asks whether the needling itself matters. Guidelines also differ: the American College of Physicians lists acupuncture among initial non-drug options for chronic low back pain [6], NICE recommends against acupuncture for low back pain [7], and NICE allows a single limited course for chronic primary pain. [8]

For condition-specific results see Acupuncture for back and neck pain and What is acupuncture?.

Placebo for herbal formulas

A 2019 review of 889 registered Chinese herbal medicine trials in the WHO trial registry platform found 363 (40.8%) used a placebo. Only 46 reported the placebo's composition, 14 reported quality control of the placebo, and 2 (0.6%) assessed whether blinding worked. The authors say a Chinese herbal placebo is difficult to make because of the medicine's colour, taste and smell, and call for design standards. [9]

Also, TCM formulas are often individualized to a "pattern", which adds a further layer. The reporting guideline CONSORT-CHM Formulas 2017 extends the standard trial checklist by introducing the idea of the TCM pattern and the features of herbal formulas. [10]

Other limits specific to TCM trials

• Country and publication bias. A 1998 review found that all acupuncture trials from China, Japan, Hong Kong and Taiwan in the sample were positive, and suggested publication bias as a possible explanation. [11]
• Quality varies. A 2023 appraisal of 64 trials of TCM for COVID-19 found 27 with high risk of bias, and trials published in Chinese were of lower quality than those published in English. [12]
• WHO view. WHO says that for non-medicinal therapies such as acupuncture, randomized trials are extremely difficult, if not impossible, to run well, and that new methods are needed. [13]

What this means for you

Read a study's control group before its headline, and ask what the answer would mean if the comparison were different. How to read a TCM study gives a checklist, and our general guides on randomization, blinding and placebo and understanding evidence explain the basics. Do not stop prescribed treatment based on any single trial.