Acupuncture

What it is

The insertion of fine needles into specific points on the body, traditionally to regulate the flow of qi through the channels.

It is the most extensively researched practice covered by this site — thousands of trials, several large individual-patient-data meta-analyses, and multiple national guideline reviews. That means this article can be unusually specific, and it can afford to disappoint both camps.

Where it came from

Needling with fine metal needles is attested in China from around the Han period, though the earlier Mǎwángduī manuscripts describe channel therapy without points or needles, using cauterisation instead. Stone and bone implements found at earlier sites are sometimes claimed as proto-acupuncture; the identification is disputed.

The practice spread to Korea, Japan and Vietnam, developing distinct national styles — Japanese practice uses notably thinner needles and shallower insertion.

It reached the West in stages, but its modern popularity dates precisely: 1971, when the New York Times journalist James Reston published an account of receiving acupuncture for post-operative pain in Beijing while covering Nixon's diplomacy. The article was a sensation. It is worth noting what it actually described — needling for pain after an appendectomy performed under conventional anaesthesia, not surgery performed under acupuncture, which is how it is often remembered.

How it is done

Sterile single-use needles are inserted at selected points to varying depths, sometimes manipulated by rotation or lifting to produce déqì — a heavy, aching, spreading sensation regarded as a sign of correct placement — and left in place for perhaps twenty minutes. Electroacupuncture applies a small current between needles.

What we can and cannot say

We can be precise, because the research exists.

There is a real effect on chronic pain, and it is small. The strongest evidence is the individual-patient-data meta-analysis led by Andrew Vickers, first published in 2012 and updated in 2018, pooling around twenty thousand patients from high-quality randomised trials of chronic musculoskeletal pain, headache and osteoarthritis. Its findings:

• Acupuncture clearly outperforms no acupuncture — a substantial difference
• Acupuncture also outperforms sham acupuncture, by a small but statistically significant margin, roughly 0.15 to 0.23 standard deviations

The second finding is the interesting one, and both sides misreport it. Advocates cite it as proof acupuncture works. Critics cite the smallness as proof it does not. What it actually shows is that most of the benefit patients experience comes from something the sham procedure also delivers, and a small remainder does not.

The point specificity does not survive. Sham acupuncture — retractable needles that do not penetrate, or real needles inserted at non-points, or shallow needling away from the indicated locations — performs nearly as well as the real thing across a very large number of trials. The large German trials of the 2000s, among the best-conducted in the field, consistently found real acupuncture better than usual care for low back pain and headache, and no better than sham.

If needling anywhere works about as well as needling at the correct point, then the map is not doing the work. That is a direct empirical verdict on the meridian theory, arrived at by testing.

The mechanism does not require any of the theory. Needling produces documented physiological effects: local release of adenosine, activation of endogenous opioid systems, and diffuse noxious inhibitory control, in which a painful stimulus in one location dampens pain perception elsewhere. Add strong expectation, a lengthy attentive consultation, touch, and rest — all of which are known to move pain reports substantially — and the observed results are accounted for.

By condition. The evidence is least bad for chronic low back pain, neck pain, osteoarthritis of the knee, tension headache and migraine prophylaxis, and for chemotherapy-induced nausea. For most other indications — fertility, allergies, smoking cessation, depression, and a long list of others — the evidence is weak, absent or negative. Guideline bodies have moved in both directions over time; the UK's NICE, for instance, has both added and withdrawn acupuncture recommendations for low back pain as evidence accumulated.

Safety. With sterile single-use needles and a trained practitioner it is comparatively safe. Serious adverse events are rare but documented and real: pneumothorax from deep thoracic needling, infection, and retained needle fragments. The greater risk is the one that runs through this whole site — treatment sought for a condition requiring something else.

We cannot say acupuncture does nothing. We also cannot say it does what it says. The honest verdict is that a practice built on a theory that testing does not support nonetheless produces a small genuine benefit for some kinds of chronic pain, largely by mechanisms its own tradition never described.

Further reading

• Vickers et al., "Acupuncture for Chronic Pain: Update of an Individual Patient Data Meta-Analysis" (The Journal of Pain, 2018).
• The German GERAC and ART trials (2002–2007), for the real-versus-sham comparisons.
• Cochrane reviews by condition — they are freely available in summary and differ sharply from one another, which is itself informative.