An economic evaluation of acupuncture sets the extra cost of adding a course of acupuncture against the extra health it buys, and expresses the result as a cost per quality-adjusted life year (QALY). A QALY is one year lived in full health; a year at half the quality of life scores 0.5. The ratio of extra cost to extra QALYs is the incremental cost-effectiveness ratio (ICER), and a health system judges it against a threshold, the most it is prepared to pay for one QALY. The trial-based studies below each reported a ratio and a verdict, but the verdict also depends on four choices made before any patient was treated: what acupuncture was compared with, what a session was assumed to cost, how many sessions a course held, and how long the benefit was followed. This page reports the studies with those choices stated, and shows how the National Institute for Health and Care Excellence (NICE) in England reached different conclusions from overlapping evidence.
How is the cost-effectiveness of acupuncture measured?
The trial-based studies below compared acupuncture plus usual care with usual care alone, recording each patient's use of health services and their quality of life on a standard questionnaire. The difference in cost is divided by the difference in QALYs over the trial's follow-up. A second kind, a decision model, pools results from several trials and adds assumptions about costs and how long an effect lasts; NICE built such models for two of its guidelines. Trials against sham acupuncture are rarely used for costing, because, as the NICE chronic pain analysis put it, sham is not used outside research, so it is not a real alternative a health service could buy. Sham acupuncture covers what those trials measure instead.
Trial-based evaluations
The table gives each study's country, condition, comparison, result and the threshold its authors judged it against, in the currency and price year the paper reports. Where a cell says the threshold is unnamed, the abstract states a verdict without giving a figure.
| Study | Country and condition | Comparison and follow-up | Cost per QALY | Threshold judged against | Currency and price year |
|---|---|---|---|---|---|
| Wonderling, 2004 | England and Wales; chronic headache, mostly migraine; 401 patients | Up to 12 treatments from physiotherapists over 3 months plus usual care, against usual care; 1 year | £9,180 | £30,000, which the authors describe as consistent with NICE's decisions; 92% probability below it | £, 2002–3 prices |
| Ratcliffe, 2006 | York, England; persistent non-specific low back pain; 241 adults | 10 individualised treatments from acupuncturists trained in Chinese medicine plus usual care, against usual care; 2 years | £4,241 | £20,000; more than 90% probability below it | £, 2002–3 prices |
| Whitehurst, 2011 | UK; knee osteoarthritis; 352 adults aged 50 or over | Advice and exercise plus acupuncture from NHS physiotherapists, against advice and exercise; 12 months | £3,889 | £20,000; 77% probability below it | £; price year not stated in the abstract |
| Stamuli, 2012 | UK; irritable bowel syndrome; 233 patients | Acupuncture plus usual care, against usual care; 1 year | About £62,500; usual care cheaper and more effective when missing data were imputed; £6,500 in the most severe cases | £20,000 to £30,000 | £, 2010 prices |
| Spackman, 2014 | UK; depression (ACUDep trial) | Acupuncture plus usual care, against usual care and against counselling | £4,560 against usual care | £20,000; probability 0.62 | £; price year not stated in the abstract |
| Essex, 2017 | UK; chronic neck pain (ATLAS trial) | Up to 12 sessions plus usual GP care, against usual GP care | £18,767 (95% confidence interval £4,426 to £74,562) | £20,000 | £, 2012/13 prices |
| Witt, 2006 | Germany; chronic low back pain; 3,093 randomised | Acupuncture plus routine care, against routine care; followed to 6 months | €10,526 | Unnamed; authors call the result relatively cost-effective | €; price year not stated in the abstract |
| Willich, 2006 | Germany; chronic neck pain; 3,451 patients | Acupuncture plus routine care, against routine care; 3 months | €12,469 | Unnamed “international” threshold values | €; price year not stated in the abstract |
| Witt, 2008 | Germany; primary headache; 3,182 patients | Acupuncture plus routine care, against routine care; 3 months | €11,657 | Unnamed “international” threshold values | €; price year not stated in the abstract |
| Reinhold, 2008 | Germany; knee or hip osteoarthritis; 489 patients in 255 practices | Acupuncture plus routine care, against routine care; 3 months | €17,845 | Unnamed | €; price year not stated in the abstract |
| Taylor, 2014 | Australia; chronic non-specific low back pain; model built on pooled trials | Acupuncture added to standard care, against standard care | About A$48,562 per disability-adjusted life year (DALY) avoided | The WHO benchmark of GDP per head, about A$52,000 | A$, 2009 |
| Herman, 2026 | United States, three health systems in California and Washington State; older adults with chronic low back pain | Standard acupuncture (up to 15 sessions over 12 weeks) or enhanced acupuncture (up to 6 more) plus usual medical care, against usual medical care; 1 year | Standard: $52,897. Enhanced: $491 lower cost per person and a gain of 0.037 QALYs, so cost-saving | $100,000 | US$, 2022 |
United Kingdom
The UK studies share a design: a pragmatic trial in NHS primary care, acupuncture added to usual care, and costs taken from the NHS's own perspective. Wonderling, Ratcliffe, Spackman and Essex each attribute the extra cost mainly to the acupuncture itself. Wonderling and colleagues report that the ratio fell substantially once they projected QALY differences into the years after the trial. Whitehurst and colleagues found that a third arm receiving non-penetrating sham needles gave cost-effectiveness estimates similar to real acupuncture, and wrote that the economic benefit could not be attributed to the needle going through the skin. Stamuli and colleagues concluded that acupuncture was not cost-effective for the whole IBS population, while it may be for the most severe cases. Spackman and colleagues found that their ranking of acupuncture against counselling reversed when an acupuncture session was priced the same as a counselling session. NICE's 2021 analysis notes that 40% of the data were missing in the ATLAS acupuncture arm, and that imputing them raised the ratio to £43,838.
Germany
The German evaluations, by Claudia Witt, Stefan Willich, Thomas Reinhold and colleagues, ran alongside very large trials in routine care, with costs taken from the patients' health insurance fund data. Their follow-up was mostly three months, and Willich and colleagues note that over-the-counter medicines did not appear in the insurance data. Reinhold and colleagues found women achieved a better ratio than men.
United States and Australia
Herman and colleagues ran the US evaluation from the health care sector and Medicare perspectives. The quality-of-life gain for standard acupuncture was not statistically significant, and the enhanced course carried the saving. Taylor and colleagues in Australia used disability-adjusted life years and the World Health Organization's benchmark of national income per head rather than a QALY threshold; they found acupuncture as a substitute for standard care not cost-effective unless relief of coexisting depression was counted.
What did NICE conclude about the cost-effectiveness of acupuncture?
Four NICE guidelines weighed the cost-effectiveness of acupuncture, and their committees disagreed about which comparison should decide whether acupuncture is worth paying for.
| Guideline | Condition | Economic evidence used | Cost per QALY and threshold | Recommendation |
|---|---|---|---|---|
| CG88, May 2009 | Persistent non-specific low back pain | No model of its own for acupuncture; relied on Ratcliffe 2006 | Took Ratcliffe's result, judged against £20,000 | Consider a course of up to 10 sessions over up to 12 weeks |
| CG150, September 2012 | Migraine prevention | Its own decision model over 6 months, against no treatment and against drug prophylaxis | £29,882 against no treatment, above the £20,000 threshold; acupuncture likely the least cost-effective of the options modelled | Consider up to 10 sessions over 5 to 8 weeks, for migraine only once propranolol, topiramate and amitriptyline have not worked or are unsuitable (as amended 2025), and for chronic tension-type headache |
| NG59, November 2016 | Low back pain with or without sciatica | Reviewed the Ratcliffe and Thomas analysis but set it aside | Not weighed | Do not offer acupuncture; the 2016 recommendation stands on the page as last updated in July 2026 |
| NG193, April 2021 | Chronic primary pain | Its own model pooling trials of acupuncture plus usual care against usual care, at an extra £350 a course | £5,710 over a lifetime with the benefit extrapolated; £14,552 without extrapolation; both below £20,000 | Consider a single course, delivered in the community by a band 7 or lower professional within 5 hours of professional time, or elsewhere at equivalent or lower cost |
The 2012 headache model differed from Wonderling's trial on two assumptions, which the guideline names: it costed a course of 15 sessions, the average of the efficacy trials, against about 9 in the trial; and it took the outcome of “no treatment” from the sham arms of trials rather than from usual care, which the authors wrote could underestimate acupuncture's cost-effectiveness. Run with the trial's session count, the model found acupuncture cost-effective against no treatment, and a threshold analysis put the break-even point at 10 sessions. The recommendation sets the same limit of 10 sessions.
The 2016 reversal on back pain was not an economic finding. The NG59 committee decided that the sham-controlled trials showed no consistent treatment-specific effect, attributed the gains over usual care to contextual effects, and so treated cost-effectiveness as not relevant. It also questioned how the Thomas trial could show a quality-of-life gain without a benefit in pain or function. The 2021 chronic pain committee took the opposite route on the comparator: it reviewed sham trials to judge whether a specific effect existed, then costed acupuncture only against usual care, as the real alternative. AcuiQ's clinical guidelines page sets these recommendations beside other countries', and Low back pain and Migraine report the clinical evidence the committees weighed.
Systematic reviews of the economic evidence
Kim and colleagues in Korea reviewed 17 economic evaluations run alongside randomised trials, searching 11 databases to March 2011. All nine that measured QALYs found acupuncture, with or without usual care, cost-effective against a waiting list or usual care alone, with ratios from €3,011 (period pain) to €22,298 (allergic rhinitis) in the German studies and from £3,855 (osteoarthritis) to £9,951 (headache) in the UK studies. They rated those nine as well designed with a low risk of bias, but not the eight that measured symptom scores instead of QALYs, and cautioned against generalising across diseases and countries where acupuncture's status differs.
What the economic literature cannot settle
Every trial-based evaluation above compares acupuncture with usual or routine care, so none separates the effect of needling from the visits, attention and expectation that come with it; NICE's back pain and chronic pain committees read that same limitation in opposite directions. Most follow-up is short: three months in the German studies and one or two years in the UK and US ones, so whether a benefit lasts, which drives the ratio, rests on extrapolation. Prices do not transfer: the ATLAS trial costed a session at £35 and a German neck pain study at €35, both of which NICE's 2021 analysis judged likely lower than current NHS staff costs, while the US trial paid its acupuncturists $90 a visit; Spackman's result turned on the price of a session. Thresholds differ by country and some German abstracts name none. The studies cover a handful of conditions, almost all chronic pain, in a few high-income health systems.