Is acupuncture just NICE enough?

Why is acupuncture now recommended for chronic pain but not for back pain?

Mike Cummings, Medical Director, British Medical Acupuncture Society

Published in JHH 19.1 – Integrative Medicine


It was a couple of chance events early in my medical career that steered me to a career in acupuncture. The first was seeing that acupuncture was being used in the British military in the late 1980s. The second was taking over a small private acupuncture practice from Dr Adrian White in the mid 1990s, so that he could move into fulltime research.

Adrian, who had originally taught me acupuncture, then encouraged me not to necessarily believe all that I had been taught but to look for evidence in support of my conviction that acupuncture was a really good treatment for muscle pain. This naturally led me into systematic reviewing, and the results of my first review led me to start questioning a lot of the orthodox medicine I had naturally assumed to be true. Specifically, that injections of local anaesthetics and corticosteroid for muscle pain were no different to injections of saline or water (Cummings, 2001).

By the mid 2000s I had been a journal editor, a peer reviewer, and a systematic review author for several years, so I was in an ideal position to take an interest in the first guideline from the then National Institute for Clinical Excellence (NICE) that considered acupuncture (NICE, 2008) and every one of them since (NICE, 2008, 2009, 2012, 2014, 2021).

I also took an interest in Cochrane reviews and was stimulated to read a whole review for the first time in 1999 (van Tulder et al, 1999) when its conclusion appeared to be completely at odds with the first meta-analysis of acupuncture for low back pain (Ernst & White, 1998). I discovered that the ‘no evidence’ conclusion of Van Tulder et al was because reviewers disagreed on the interpretation of certain trials, one of which did not actually include any acupuncture. I came away with the strong impression that somebody needed to keep an eye on them, and I took on the job.

Later I had the experience of helping perform a Cochrane review and contributing to a NICE guideline development group (GDG) as an advisor so I came to better appreciate the challenges of both, and I subsequently came to be somewhat less critical. That is not to say that I am uncritical, as the following narrative will demonstrate. We cannot really continue to allow data errors, orthodox biases, or conflicts of interest to influence national provision without at least raising a voice from time to time.

Acupuncture for osteoarthritis

In 2004 we (the acupuncture research community) had a couple of relatively large positive sham-controlled trials of electroacupuncture (EA) in osteoarthritis (OA) of the knee. I realised there was a possibility we might get a positive recommendation for acupuncture in OA before we got one for my favourite condition – muscle pain or myofascial pain. I suggested to a colleague Adrian White that we should perform a systematic review with meta analysis in OA knee (White et al, 2007).

Our review demonstrated efficacy over sham, yet the subsequent NICE guideline (CG59) recommended against the use of EA (NICE, 2008). This decision was based on health economic modelling using efficacy data from a single paper with the term ‘electroacupuncture’ in the title, even though other trials had also used EA. To make matters worse, I had just convinced colleagues at the Royal London Homeopathic Hospital (RLHH) (now the Royal London Hospital for Integrated Medicine – RLHIM), to set up group clinics to treat OA knee patients with EA.

‘We cannot really continue to allow data errors, orthodox biases, or conflicts of interest to influence national provision’

We (Adrian White and I) argued that modelling cost effectiveness based on the difference between real and sham (ie proper needling versus gentle needling) did not make sense (White, 2009) but the health economist responsible argued that it did (Latimer, 2009). Later he came to appreciate that this approach was problematic (Latimer et al, 2012). Fortunately, our group clinics survived the negative recommendation for EA (Berkovitz et al, 2008) but they eventually succumbed after the draft of the second low back pain guideline (NICE, 2016) which triggered widespread decommissioning of acupuncture services across the NHS.

With a more pragmatic cost-effectiveness perspective accepted I thought we had solved the acupuncture for OA problem, but the next time around NICE introduced another hurdle – the dreaded minimal clinically important difference (MCID). The MCID chosen was a standard mean difference (SMD) of 0.5. This is a measure of effect size (ES) where the mean difference (MD) between an intervention and control is divided by the standard deviation (SD) of the difference. For acupuncture in OA (Vickers et al, 2012) the SMD over no acupuncture controls was greater than 0.5, but the SMD over sham was less than 0.5. So, in CG177, published in February 2014, acupuncture sadly fell at the first explanatory hurdle (NICE, 2014).

Acupuncture for back pain

Around the same time as I was reeling from the negative recommendation for electroacupuncture (my favourite treatment approach) in OA (CG59), I was invited to join a NICE GDG to advise on acupuncture for low back pain (LBP). The scope was LBP from six weeks to one year, so mainly aimed at primary care, and the chair was an academic GP. The approach to the evidence was much more pragmatic, and acupuncture was recommended along with exercise and manual therapies based on patient choice in CG88 (NICE, 2009. What I did not see at the time was the reaction from a group of pain interventionists in secondary care who took exception to a recommendation against spinal injections in patients with pain for less than one year. They called an extra ordinary meeting of the British Pain Society (BPS) and put forward a vote of no confidence in the president, who had been a key member of the GDG. He subsequently resigned (Kmietowicz, 2009).

I understand that meetings then took place out of the public eye between BPS and NICE before the announcement of a new updated guideline. None of the key protagonists from CG88 were invited to the new GDG despite applying, and the scope of the guideline changed substantially. Sciatica was included, and any time limitations were removed.

While we could see the writing on the wall based on the interests and opinions of several members of the GDG, it was still shocking to see a flagrant exhibition of double standards in the assessment of research evidence for conventional interventions compared with those applied to acupuncture in NG59 (NICE, 2016; Cummings, 2017).

There is a conference presentation from the BMAS Spring Scientific Meeting in May 2017 that narrates the story of acupuncture recommendations in CG88 to NG59 (BMAS, 2017).

‘…it was shocking to see a flagrant exhibition of double standards in the assessment of research evidence for conventional interventions compared with those applied to acupuncture’

Acupuncture for headache

Having followed the big clinical acupuncture trials from Germany in the 2000s (Cummings, 2009) which went on to dominate the subsequent meta-analyses in terms of statistical power (larger trials get greater weighting in meta-analysis), and having seen the approach of NICE in the OA guidelines, I was not expecting a positive recommendation for acupuncture in the headache guideline (NICE, 2012). Acupuncture works well in headache, but so does sham acupuncture, and the difference would be unlikely to meet an MCID if that were applied.

My expectations were elevated after a neurologist from the GDG attended one of the BMAS foundation courses several months before the guidance was released. He said he had looked at the evidence and while it was not particularly convincing, the evidence for other interventions was equally unimpressive. I was pleased that he attended the entire course, and subsequently used acupuncture for a time in his neurology practice.

When the guideline was finally released, the thing that shocked me was the confident statement in the guideline that the drug topiramate (an anti-epileptic that was the current favourite in migraine prophylaxis) was twice as good as acupuncture. I had already seen one head to head comparison demonstrating superiority of acupuncture (Yang et al, 2011) and the Cochrane review subsequently confirmed a more general advantage of acupuncture over drug prophylaxis (Linde et al, 2016).

This difference is explained by NICE taking the explanatory approach and only considering the effect of acupuncture over a sham form and comparing this with the effect of topiramate over placebo. The data they used gave a responder rate for sham acupuncture higher than that for real topiramate, but this comparison was not used.

Acupuncture for chronic primary pain

Finally, we come to the most contemporary guideline – a new guideline on chronic pain (NICE, 2021; Kmietowicz 2021). NG193 has recommended acupuncture albeit with limitations to the provision. Many groups representing NHS providers questioned why acupuncture was being recommended for chronic pain when it had not been recommended for back pain, which is probably the biggest cause of chronic pain in the community.

Ironically, the reason why acupuncture was recommended for chronic pain was because the trial data on back pain was deliberately excluded. This data includes two very large sham controlled trials in which the response rates were unexpectedly high in the sham acupuncture groups (Haake et al, 2007; Cherkin et al, 2009) thus reducing the group mean difference between real acupuncture and the sham version when included in meta-analysis. By excluding these trials and leaving neck pain, fibromyalgia, myofascial pain and chronic pelvic pain the group mean difference exceeded the pre-defined MCID.

Unlike NG59, this GDG was not prepared to set a double standard and insisted on recommending acupuncture despite pressure to do otherwise from the hierarchy [personal communications with unnamed sources]. They also recommended against the use of a whole swathe of drugs that are still in very common usage in chronic pain.

Interestingly, they left the door open to treating chronic low back pain as well by including the proviso that you first follow the specific guideline (NG59, which recommends against acupuncture), but if the pain or its impact is out of proportion to the underlying condition and would be better managed as chronic primary pain, then acupuncture becomes one of the options you can consider.


After many years of scrutinising meta-analyses and guidelines, as well as manuals of methodology, I have finally come to terms with the reasons behind the sometimes bizarre and often opposing recommendations, conclusions, and rhetoric concerning acupuncture. The difference usually stems from whether the commentator takes an explanatory point of view (acupuncture versus sham/robust needling versus gentle needling) or a pragmatic point of view (acupuncture versus no acupuncture/acupuncture versus other treatments).

If you want the bottom line in big acupuncture data, the place to go is the individual patient data meta-analysis (IPDM) led by Andrew Vickers and including a large group of researchers collectively known as the Acupuncture Trialists Collaboration (ATC). They published the first version of their IPDM after considerable difficulty (Vickers et al, 2012; Vickers et al, 2013) and an update was published in 2018 (Vickers et al, 2018). Interestingly, after comments on the draft of NG59, data from the IPDM was included in the final version, but the data analysts, in error, used SMD data from the IPDM paper in their own meta-analysis of mean differences. This error made the summary value of acupuncture versus sham look even worse than in the draft of NG59, and the GDG over interpreted this error by suggesting that the better the quality of data the smaller the effect of acupuncture. Once a final guideline is published there is no chance for further comment, even if the errors are as gross as using SMD instead of MD.

‘I continue to try to push for a fair and sensible assessment of the evidence for acupuncture’

I continue to try to push for a fair and sensible assessment of the evidence for acupuncture by both NICE and Cochrane; however, Cochrane appears to be moving away from the more pragmatic view of the data in Linde et al, to a more rigid explanatory one, as illustrated by the long-awaited update to the low back pain review (Mu et al, 2020).


Berkovitz S, Cummings M, Perrin C & Rieko I (2008) High Volume Acupuncture Clinic (Hvac) for Chronic Knee Pain – Audit of a possible model for delivery of acupuncture in the National Health Service. Acupunct Med, 26, 46–50. doi:10.1136/aim.26.1.46.

BMAS (2017) Acupuncture in low back pain. Available at: (accessed 9 Feb 2022).

Cherkin DC, Sherman KJ, Avins AL et al (2009) A randomized trial comparing acupuncture, simulated acupuncture, and usual care for chronic low back pain. Arch Intern Med, 169, 858. doi:10.1001/archinternmed. 2009.65.

Cummings M (2017) NG59 used different levels of evidence for conventional interventions compared with those for acupuncture and may not have adequately addressed personal financial COIs of the GDG chair. BMJ, 356, i6748. doi:

Cummings M (2009) Modellvorhaben Akupunktur – a summary of the ART, ARC and GERAC trials. Acupunct Med, 27, 26–30. doi:10.1136/aim.2008.000281.

Cummings TM & White AR (2001) Needling therapies in the management of myofascial trigger point pain: A systematic review. Arch Phys Med Rehabil, 82, 986–92. doi:10.1053/apmr.2001.24023.

Ernst E & White AR (1998) Acupuncture for back pain: a meta-analysis of randomized controlled trials. Arch Intern Med, 158, 2235–41. doi:10.1001/archinte.158.20.2235.

Haake M, Müller H-H, Schade-Brittinger C et al (2007) German acupuncture trials (GERAC) for chronic low back pain: randomized, multicenter, blinded, parallel-group trial with 3 groups. Arch Intern Med, 167, 1892–8. doi:10.1001/archinte.167.17.1892.

Kmietowicz Z (2021) Offer exercise, therapy, acupuncture, or antidepressants for chronic primary pain, says NICE. BMJ, 373:n907. doi:10.1136/bmj.n907.

Kmietowicz Z (2009) President of British Pain Society is forced from office after refusing to denounce NICE guidance on low back pain. BMJ, 339, b3049–b3049. doi:10.1136/bmj.b3049.

Latimer N (2009) NICE guideline on osteoarthritis: is it fair to acupuncture? Yes. Acupunct Med, 27,72–5. doi:10.1136/aim.2009.000802.

Latimer NR, Bhanu AC & Whitehurst DGT (2010) Inconsistencies in NICE guidance for acupuncture: reanalysis and discussion. Acupunct Med, 30, 182–6. doi:10.1136/acupmed-2012-010152.

Linde K, Allais G, Brinkhaus B et al (2016) Acupuncture for the prevention of episodic migraine. Cochrane Database Syst Rev, CD001218. doi:10.1002/14651858.CD001218.pub3.

Mu J, Furlan AD, Lam WY et al (2020) Acupuncture for chronic nonspecific low back pain. Cochrane Database Syst Rev, 12, No CD013814. doi:10.1002/14651858.CD013814.

NICE (2021) Chronic pain (primary and secondary) in over 16s: assessment of all chronic pain and management of chronic primary pain. Available at: (accessed 9 February 2022).

NICE (2016) Low back pain and sciatica in over 16s: assessment and management. Available at: (accessed 9 February 2022).

NICE (2014) Osteoarthritis: care and management of osteoarthritis. Available at: (accessed 9 February 2022).

NICE (2012) Headaches in over-12s: diagnosis and management. Available at: (accessed 9 February 2022).

NICE (2009) Low back pain in adults: early management. Available at: (accessed 9 February 2022).

NICE (2008) Osteoarthritis: the care and management of osteoarthritis in adults. Available at: (accessed 9 February 2022).

van Tulder M, Cherkin D, Berman B, Lao L & Koes B (1999) Acupuncture for low-back pain. Cochrane Database of Systematic Reviews, 2, No. CD001351. doi:10.1002/14651858.CD001351.

Vickers AJ, Cronin AM, Maschino AC et al (2012) Acupuncture for chronic pain: individual patient data meta-analysis. Arch Intern Med, 172, 1444–53. doi:10.1001/archinternmed.2012.3654.

Vickers AJ, Maschino AC, Lewith G, et al (2013) Responses to the Acupuncture Trialists – Collaboration Individual Patient Data Meta- Analysis. Acupunct Med, 31, 98–100. doi:10.1136/acupmed-2013-010312.

Vickers AJ, Vertosick EA, Lewith G et al (2018) Acupuncture for chronic pain: update of an individual patient data meta-analysis. J Pain, 19, 455–74. doi:10.1016/j.jpain.2017.11.005.

White A (2009) NICE guideline on osteoarthritis: is it fair to acupuncture? No. Acupunct Med, 27, 70–2. doi:10.1136/aim.2009.000810.

White A, Foster NE, Cummings M & Barlas P (2007) Acupuncture treatment for chronic knee pain: a systematic review. Rheumatology, 46, 384–90. doi:10.1093/rheumatology/kel413.

Yang C-P, Chang M-H, Liu P-E et al (2011) Acupuncture versus topiramate in chronic migraine prophylaxis: A randomized clinical trial. Cephalalgia, 31, 1510–21. doi:10.1177/0333102411420585.