Rewiring the brain with acupuncture

– does sophisticated MRI data point us to optimal treatment?

This blog was first published on 25th January 2018 on


For some time, a group on the US east coast have been quietly scanning brains with functional magnetic resonance imaging (fMRI) and related techniques to examine the effects of acupuncture. They hit the headlines in rather a dramatic manner in 2017.[1,2]

At the beginning of the fMRI story we saw some extraordinary claims concerning point specificity in acupuncture from a famous name in the development of fMRI, but these were later retracted.[3] I was relieved to see the retraction, as the claims did not seem mechanistically credible from a neurophysiological point of view. The authors retracted their paper because the results were not in line with the body of developing evidence that acupuncture with typical deep tissue sensation (I prefer this term, but it equates to de qi) seems to cause a general deactivation of limbic structures,[4,5] rather than very specific and targeted functional activations.

So why am I a year behind the headlines? Well I was asked to discuss the research for a television programme, so I read the paper thoroughly in preparation and discovered an interesting observation that had previously escaped my notice. Then there was a mix-up with storyboards and we discussed other research instead. So I thought I had better put all those hours of preparation to good use by describing my thoughts on this rather complex area of research.

The team concerned here first came to my attention when they demonstrated a change in cortical mapping of the second and third fingers (D2/D3) in patients with carpal tunnel syndrome (CTS) treated with acupuncture.[6] This was a departure from the fMRI studies prior, which had tended to simply watch what happened in the brain after acupuncture or control procedures in healthy subjects. Whilst it was an interesting finding, the study was observational rather than a strict RCT, and I felt that the change in cortical mapping was likely to be downstream of the main effects of acupuncture rather than a direct effect, ie a consequence of the acupuncture mechanism rather than part of the mechanism. This is always the problem with observational data – is the observed association causal or consequential?

The team continued to study CTS, and the research published in 2017 was a three-armed study of 80 patients with CTS [2] – a relatively large study in fMRI terms, but small and underpowered in terms of standard clinical trials of acupuncture in pain. The 3 different interventions principally involved electroacupuncture (EA): local EA (PC7–TE5), distant contralateral EA (SP6–LR4) and ipsilateral regional sham EA (non-points on the flexor aspect of the mid forearm). Manual points were included in the same regions as the EA in each group.

The symptom scores in all groups declined over the course of treatment with no significant differences, although noticeably bigger change scores in the local and sham groups, where the focus of treatment was in the correct limb. Despite this, the median nerve conduction latency improved in both EA groups and deteriorated in the sham EA group. D2/D3 cortical separation distance improved marginally more (not significant) with local EA than distant EA, and not at all in the sham.

…median nerve conduction latency and D2/D3 cortical separation distance improved in both EA groups

So despite there being no difference between groups in terms of symptoms, there was a clear difference in objective measures of nerve function and brain function. And there is more! The degree of improvement in D2/D3 cortical separation distance immediately after the 8-week 16-session treatment course predicted (correlated with) the symptom score at 3 months follow-up. That is very interesting, and somewhat counters my assumption that the cortical remapping is downstream (ie a consequence rather than a cause) of the effect of EA.

The degree of improvement in D2/D3 cortical separation distance correlated with the symptom score at 3 months follow-up

Another interesting aspect is the rate of deterioration in symptom score of the sham group after they were unmasked, and the continued improvement of the distal group after they learned that they had a genuine treatment. This makes me ponder over the influence of other brain centres – those related to cognition analysis and expectation – and how these can add unwanted noise in group means for subjective outcomes.

And there is still more! The bit I originally missed because it was just too much effort to read and understand. The team studied the microstructure of the white matter adjacent to the relevant areas of the primary somatosensory cortex (S1). I didn’t even know this was possible, but it has been around for about 15 years. Fractional anisotropy is a measure of order in the structure of white matter based on diffusion of water. A perfect isotropic material would have an even pattern of diffusion in all directions, but uniform tracts of myelinated neurones will disturb this, and cause a degree of anisotropy. Got it? Anyway, the team discovered that the changes after real EA (local & distant) in fractional anisotropy near the S1 cortex related to the contralesional hand correlated with latency changes in the median nerve. This was not true of sham. Even more interesting is that this correlation between changes in fractional anisotropy and median nerve latency occurred in different areas of the ipsilesional SI cortex depending on whether the EA was local or distant.

This has to be considered speculative, since it was a bit of a fishing trip, but it is very exciting to speculate that in the future we may be able to develop ways of tracking the course of plastic changes in the central nervous system and design optimal treatment approaches as a result; moving us from ancient philosophy, through neuroscience from the last century (segmental neuromodulation) perhaps to real-time neural remodelling.


  1. Ditch the paracetamol and try ACUPUNCTURE. Daily Mail Online 2017. (accessed 24 Jan2018).
  2. Maeda Y, Kim H, Kettner N, et al. Rewiring the primary somatosensory cortex in carpal tunnel syndrome with acupuncture. Brain 2017;140:914–27. doi:10.1093/brain/awx015
  3. Cho ZH, Chung SC, Lee HJ, et al. Retraction. New findings of the correlation between acupoints and corresponding brain cortices using functional MRI. Proc Natl Acad Sci U S A 2006;103:10527. doi:10.1073/pnas.0602520103
  4. Wu MT, Hsieh JC, Xiong J, et al. Central nervous pathway for acupuncture stimulation: localization of processing with functional MR imaging of the brain–preliminary experience. Radiology 1999;212:133–41. (accessed 28 Aug2011).
  5. Hui KK, Liu J, Makris N, et al. Acupuncture modulates the limbic system and subcortical gray structures of the human brain: evidence from fMRI studies in normal subjects. Hum Brain Mapp 2000;9:13–25.
  6. Napadow V, Liu J, Li M, et al. Somatosensory cortical plasticity in carpal tunnel syndrome treated by acupuncture. Hum Brain Mapp 2007;28:159–71. doi:10.1002/hbm.20261

Declaration of interests MC

EA for stress urinary incontinence

– perhaps via direct pudendal nerve stimulation

This blog was first published on 4th December 2017 on

blue clean clear dew
Photo by Pixabay on

Earlier in 2017 this paper popped up in JAMA.[1] It was from the same group that had published a huge multicentre trial of electroacupuncture (EA) for chronic severe functional constipation,[2] which was featured in a previous blog. This was a large sham controlled trial (n=504) although not as big as the previous one on constipation (n=1075).

EA can influence symptoms in overactive bladder

We have known for some time that EA can influence symptoms in overactive bladder, albeit under another name.[3] These are mainly urge symptoms, and a similar mechanism that reduces pain perception might be postulated to explain the results. Urinary stress incontinence (SUI) seems entirely different, and so the results of this trial were rather unexpected (by me at least).

I revisited the paper to examine the intervention in more detail and consider the anatomy involved. The active EA intervention involved just two points on each side (I love the simplicity!) – BL33 and BL35. 75mm needles were employed, and inserted 50 to 60mm. At BL33 the angulation (oblique inferiomedial) aimed to have the needle tip entering the S3 foramina. At BL35 (just lateral to the coccyx) the needle was angled superiolateral. In terms of safety, the structure to be avoided in this area is the rectum, and a paper in Acupuncture in Medicine studied the position of needle tips relative to the rectum when inserted into sacral foramina,[4] but not BL35. Needles inserted into S4 got close to the rectum (4-8mm), but there was a bigger margin in S2 and S3 (19-29.5mm). At BL35, a sufficient lateral angulation would avoid getting close to the rectum, but this angulation would be critical.

Figure 1 from Lui et al JAMA 2017[1]

In terms of physiological effects the needle placement at the S3 foramina would certainly stimulate the dorsal ramus of S3, and possibly the ventral ramus, depending on depth. A superiolateral angulation at BL35 might approach the pudendal nerve as it passes over the sacrospinous ligament, lateral to the sacrum at about the level of S4 (sacral hiatus). Direct stimulation of the pudendal nerve has been described in a small case series of patients with spinal stenosis refractory to less invasive EA,[5] and nerve stimulation was confirmed by perineal sensation. The study in JAMA on SUI does not mention perineal sensation but just mentions mild shivering of the skin around the points, presumably secondary to muscle contraction. EA at 50Hz was applied at a current of 1-5mA. If either needle tip was close to motor fibres, the relevant muscle would have contracted quite strongly for the entire period of stimulation (30 minutes). The authors do not report such effects, but they do discuss the possibility of S3 and pudendal nerve stimulation. If direct electrical nerve stimulation occurred, the effect may have been in some part attributed to pelvic floor muscle contraction induced by motor nerve stimulation. With this in mind it is interesting to note that the effect observed after 18 sessions of acupuncture over 6 weeks was similar to 12 weeks of a pelvic floor exercise programme.

18 sessions of acupuncture over 6 weeks was similar to 12 weeks of a pelvic floor exercise programme

I should note that the sham technique involved no skin penetration and no electrical output, so if there was direct motor nerve stimulation in the active group, there would have been quite a considerable difference in terms of the physiological stimulation applied between the groups.


  1. Liu Z, Liu Y, Xu H, et al. Effect of Electroacupuncture on Urinary Leakage Among Women With Stress Urinary Incontinence: A Randomized Clinical Trial. JAMA 2017;317:2493–501. doi:10.1001/jama.2017.7220
  2. Liu Z, Yan S, Wu J, et al. Acupuncture for Chronic Severe Functional Constipation: A Randomized Trial. Ann Intern Med 2016;165:761–9. doi:10.7326/M15-3118
  3. Peters KM, Carrico DJ, Perez-Marrero R a, et al. Randomized Trial of Percutaneous Tibial Nerve Stimulation Versus Sham Efficacy in the Treatment of Overactive Bladder Syndrome: Results From the SUmiT Trial. J Urol 2010;183:1438–43. doi:10.1016/j.juro.2009.12.036
  4. Katayama Y, Kamibeppu T, Nishii R, et al. CT evaluation of acupuncture needles inserted into sacral foramina. Acupunct Med 2016;34:20–6. doi:10.1136/acupmed-2015-010775
  5. Inoue M, Hojo T, Nakajima M, et al. Pudendal nerve electroacupuncture for lumbar spinal canal stenosis – a case series. Acupunct Med 2008;26:140–

Declaration of interests MC

The Acupuncture Trialists’ Collaboration IPDM update 2017

– more data, new insights…

This blog was first published on 29th December 2017 on


In 2012, the first individual patient data meta-analysis (IPDM) in the field of acupuncture was published.[1] It was also one of the first in the field of pain research. It was a struggle to publish, principally (I guess) because the IPDM method was relatively new to journal editors, and because the general medical journals tend towards conservatism and orthodoxy, so positive research on unorthodox interventions probably have higher hurdles to clear prior to publication.

The 2012 IPDM included data from 29 trials and 17 922 patients. The update has just been published, and it includes data from 39 trials and 20 827 patients.[2]

It should be noted that trials were only included where allocation concealment was unambiguously determined to be adequate, so this is a data set built on what used to be called the highest quality trials.

Trials were grouped into 3 clinical categories for analyses: osteoarthritis, chronic headache, non-specific musculoskeletal pain (back and neck pain), and shoulder pain. Acupuncture was shown to be superior to sham controls with effect sizes of 0.24 (previously 0.26), 0.16 (0.15), 0.30 (0.37) and 0.57 (0.64) respectively. Acupuncture was also shown to be superior to no acupuncture controls with effect sizes of 0.63 (0.57), 0.44 (0.42) and 0.54 (0.55) respectively for osteoarthritis, chronic headache, and non-specific musculoskeletal pain. [Effect sizes here are standardised mean difference – SMD – using fixed effects; those in brackets are from the 2012 IPDM].

Acupuncture is superior to sham with an effect size of about 0.2
and superior to no acupuncture with an effect size of about 0.5

Overall we see an effect over sham of about 0.2 and an effect over no acupuncture of about 0.5 SMD. The question remains over whether decisions to use acupuncture should be made with the former or the latter comparison. The effect over sham excludes the context effects that naturally occur in practice, and is small – NICE guideline development groups tend to use this. The effect over no acupuncture is the pragmatic comparison, and closest to real life practice, but naturally includes different context effects for different interventions, and this is favourable to interventions such as acupuncture that might be seen as more dramatic to the patient than say a pill. It is currently impossible to separate out the different context related factors, and the compassionate touch of an acupuncture practitioner should be included in the therapeutic effect, whereas the expectation of the patient being needled perhaps should not. Sham acupuncture includes both of these and is much more effective than most other shams including placebo pills.[3]

On the whole I am a pragmatist in the complex environment of clinical practice, but I like to be a reductionist when it comes to mechanistic analysis and planning interventions. So does this huge data set allow us any further insights through purely statistical analysis? Well I am afraid that with conventional levels of statistical probability very little is revealed; however, if we are not so rigid and examine the trends with a mechanistic eye a few interesting things start to pop out.

In terms of the characteristics of acupuncture, the only factor that had a clear influence was the number of treatments, and this was only apparent in the comparison against no acupuncture. This perhaps does not come as a surprise, but it does suggest that we should be more focussed on providing enough treatment sessions and not worrying as much about other aspects of the acupuncture approach.

Many readers will know that I have a particular interest in electroacupuncture (EA), so I am keen to highlight the fact that in the more mechanistic comparison of acupuncture over sham the use of EA as a treatment characteristic had the largest effect size (0.32) and the lowest p value (p=0.14) of any characteristic studied. This does not reach the commonly adopted level of statistical probability, but equates to a 6:7 chance of being a real effect.

Image taken from

Going back to the more pragmatic comparison of acupuncture over no acupuncture, there are a couple of interesting trends apparent. The largest was for ‘de qi attempted’, and this reached 0.74 with p=0.063. This might also suggest a dose effect, but strangely the characteristic ‘manual stimulation allowed’ actually had a moderately negative effect.

Another characteristic that approached significance in the pragmatic comparison was ‘male practitioner’ at a p value of p=0.084. The effect size associated with this was very small and negative at -0.07, but it is tempting to view this against a backdrop of mechanistic research on nocebo in pain, and suggest that male practitioners might think about channelling their more feminine sides during consultations.

male practitioners might think about channelling their more feminine sides during consultations

There are a couple of insights that have relevance to future research. First, there was a significant difference in the effect size measured between penetrating and non-penetrating shams. There was a smaller difference between acupuncture and penetrating shams, or needling against needling, as I like to refer to this comparison. I should note that this result was not maintained when outlying trials with large effect sizes were excluded; however, I would still advise researchers to avoid testing needling against needling in clinical trials! Second was in the pragmatic comparison with no acupuncture controls. The effect size of acupuncture was significantly smaller when compared with controls that were classified as high intensity, for example a course of individualised supervised physical therapy.

The team (ATC – acupuncture trialists’ collaboration) also studied the time course of acupuncture effects by analysing the change in effect size at different time points. This seems most relevant for the pragmatic comparison against no acupuncture, but they analysed both. The effects of acupuncture held up well against no acupuncture controls with an estimated drop of only 15% of the effect after one year.

There was an estimated drop of 25% at one year in the effect of acupuncture over sham acupuncture, but heterogeneity was significant. When neck pain trials were excluded the heterogeneity disappeared, and the drop in effect size reduced to about 6%. It is difficult to imagine what this means clinically, so I investigated a little. The overall effect size for acupuncture over sham in neck pain was the most remarkable at 0.83, which sits in stark contrast to the tiny 0.17 for back pain. Neck and back pain are not the same of course, with a higher proportion of soft tissue pain being the likely reason that acupuncture has greater effects in the former. But those huge effects on neck pain seem to disappear quickly in the statistics and generate heterogeneity by contrast with small effects of a longer time course in back pain. When you look more closely you see that there are only 3 trials in neck pain, and one of them is a clear outlier.[4] I checked the original paper. It is the largest sham controlled trial in neck pain, and the control group received mock TENS. The treatment period was for 3 weeks and involved 5 treatment sessions. The mean pain score (0-100mm scale) in the acupuncture group dropped from a baseline of 68.7 to 26.6 after treatment and 27.6 after 6 months. The pain score in the control group was 72.3 at baseline, 58.3 after treatment, and 45.5 after 6 months. So you can see from these figures that the effect of acupuncture was actually maintained fully for 6 months, but the control group improved over 6 months to reduce the difference by about 50%, either by natural history or perhaps patients seeking other treatments. Just looking at the summary statistics of differences between groups gives a very different impression from the within group changes. This is one of the dangers of drawing conclusions from between group change values, without checking the within group data. We are given a false impression that the effect of acupuncture in neck pain is short-lived, when in fact it is not. The false impression comes from an improvement in the comparator, rather than a degradation of the effect in the treatment group. This change is almost certainly responsible for the statistical heterogeneity observed, and the consequent uncertainty is inappropriately directed at the acupuncture effect.


Well I think I’ll leave it there… please go and pour over the numbers in this update, it is a truly fabulous endeavour, which is giving us ever-greater clarity as the data grows, as well as more questions to debate.


  1. Vickers AJ, Cronin AM, Maschino AC, et al. Acupuncture for chronic pain: individual patient data meta-analysis. Arch Intern Med 2012;172:1444–53. doi:10.1001/archinternmed.2012.3654
  2. Vickers AJ, Vertosick EA, Lewith G, et al. Acupuncture for Chronic Pain: Update of an Individual Patient Data Meta-Analysis. J Pain Published Online First: 30 November 2017. doi:10.1016/j.jpain.2017.11.005
  3. Meissner K, Fässler M, Rücker G, et al. Differential effectiveness of placebo treatments: a systematic review of migraine prophylaxis. JAMA Intern Med 2013;173:1941–51. doi:10.1001/jamainternmed.2013.10391
  4. Vas J, Perea-Milla E, Méndez C, et al. Efficacy and safety of acupuncture for chronic uncomplicated neck pain: a randomised controlled study. Pain 2006;126:245–55. doi:10.1016/j.pain.2006.07.002

Declaration of interests MC

EA for chronic severe functional constipation

– it seems to work after an intensive course…

This blog was first published on 28th November 2017 on

adult attic bathroom bathtub
Photo by Pixabay on

About a year ago I was surprised to see a sham controlled RCT of electroacupuncture (EA) published in Annals of Internal Medicine.[1] I was surprised for several reasons: I review for Annals, and I had not seen the paper for review; it was on chronic severe functional constipation – a subject with few previous RCTs of acupuncture; it was positive; it was the biggest 2 arm RCT in the acupuncture literature to date (n=1075).

The treatment was relatively straightforward – EA to the rectus abdominis muscle bilaterally (ST25—SP14) and manual stimulation of a point in tibialis anterior bilaterally (ST37). The sham control involved shallow needling of points close by, but not on the meridians, avoidance of typical needling sensation (de qi) and sham EA (ie no electrical stimulus, but the power indicator and sound as if it were real EA). The frequency used varied from 10 to 50Hz and the intensity was between 0.1 and 1mA. The frequency covers the best range for autonomic modulation (as judged by visceral blood flow), but the intensity was varied from sub-threshold (in terms of nerve stimulation you generally need 0.3 to 0.5mA for the lowest electrical threshold nerves to respond) to a rather gentle stimulus of 1mA. In my experience, many patients do not perceive sensation until around 1.5mA when needle tips are placed into muscle.

Twenty-eight EA sessions were provided over 8 weeks

Twenty-eight EA sessions were provided over 8 weeks – 5 sessions per week for the first 2 weeks, and 3 sessions per week for the next 5 weeks. So it was quite an intense course of treatment. There was no comment about rotation of needling sites – with such an intense treatment regime it is common to alternate similar protocols so that needle sites do not get sore.

The primary outcome was based on a stool diary and was the number of complete spontaneous bowel movements (CSBM) per week. EA performed significantly better than sham at 8 weeks and the effect was maintained at 12 weeks follow-up.

Figure 2 from Lui et al Ann Int Med 2016[1]

Blinding appears to have been maintained at 4 and 8 weeks from a random sample of 140 participants. At 8 weeks about 90% from each group guessed that they had received the real EA.

The size of the effect appears to be clinically relevant, with a change from baseline in CSBM of greater than 1 per week,[2] although the group average does not exceed 3 per week so most participants in the active group would still be considered to have constipation.

Of interest is a follow-up paper published recently that examines factors associated with the response to acupuncture in this trial. It found that age was inversely related to response to acupuncture (ie younger patients did better), and comorbidities reduced the likelihood of response.[3]


  1. Liu Z, Yan S, Wu J, et al. Acupuncture for Chronic Severe Functional Constipation: A Randomized Trial. Ann Intern Med 2016;165:761–9. doi:10.7326/M15-3118
  2. Johanson JF, Wald A, Tougas G, et al. Effect of tegaserod in chronic constipation: a randomized, double-blind, controlled trial. Clin Gastroenterol Hepatol 2004;2:796–805. doi:10.1016/S1542-3565(04)00356-8
  3. Yang X, Liu Y, Liu B, et al. Factors related to acupuncture response in patients with chronic severe functional constipation: Secondary analysis of a randomized controlled trial. PLoS One 2017;12:e0187723. doi:10.1371/journal.pone.0187723

Declaration of interests MC

Segmental EA falls short in ovulation induction

…in women with polycystic ovarian syndrome (PCOS)

This blog was first published on 18th November 2017 on

affection beautiful blur couple
Photo by Pixabay on

Finally I have come to address this topic, some months following publication of an eagerly awaited (at least by me) large clinical trial.[1] I have been a colleague and friend of the second author (ES-V) since proofreading her PhD thesis over 20 years ago. In those days I focussed solely on muscle so any suggestion of acupuncture being used to influence visceral function or blood flow was a bit of a conceptual stretch for me. After all my needles were going directly into the target that I wanted to influence, and I was just about comfortable with the idea that the needle alone actually did something useful without the need for injecting a drug. So the idea that acupuncture or electroacupuncture could have any useful effect through indirect influences only really arose when I read Lisa’s thesis.

Her early work stimulated interest in the use of acupuncture in fertility and augmented reproduction,[2,3] although the subsequent plume of clinical research that occurred in this field seemed to go a little off course from a basic science perspective, with an unwarranted focus on embryo transfer as part of IVF.[4] Lisa observed this, but continued with her research path, which was by then on PCOS. She clearly showed that segmental electroacupuncture (EA) could have positive influences on the condition, both in terms of hormonal and metabolic markers, and apparently in terms of ovulation rates.[5,6]

Lisa regularly runs research updates for the BMAS, and we were all excited to hear of her involvement in this huge clinical trial in China on women with PCOS. With 1000 women to be randomised and treated the trial was a considerable undertaking, and several years passed with no news. Then on the 27th June 2017 the results were out… clomiphene was nearly twice as good as segmental EA, and segmental EA was no better than a very minimal non-segmental sham.

Figure from Wu et al JAMA 2017 [1]

It did not seem to make sense from the basic science perspective! The numbers were big enough to power the comparison with sham (assuming similar size effects to those we see in clinical trials of chronic pain). The intervention appeared sufficient in neurophysiological terms, to generate the effects that had been demonstrated in the basic science experiments that had led up to this trial. Yes it was a penetrating sham, but the physiological stimulus of the sham intervention would not have generated any effect in the laboratory in terms of somatovisceral reflexes. In the clinical realm, with conscious humans, sham always seems to have a substantial context effect, but still I would have expected some physiological effect from the segmental EA.

Well there was a difference between real and sham EA in terms of adverse events. In the segmental EA group the rate of diarrhoea was 3 times that in the sham EA group, perhaps indicating an excess effect in somatovisceral reflexes in a small proportion of women. It should be noted that the absolute rate of diarrhoea was low at 1.6 and 5%, in sham and real segmental EA respectively.

The primary outcome was live birth rate. This is the most valid outcome for trials of this nature, but it is not the same as ovulation induction of course, so it is not a direct measure of the putative physiological effect of segmental EA. This could add noise to the statistics, but even so, there was not even a trend in favour of segmental EA.

The slightly curious thing is that both acupuncture groups seemed to substantially outperform metformin, which, in a large comparative trial with clomiphene resulted in a live birth rate of just 7.2%.[7] The populations are not easily comparable though as there were notable differences in BMI that would favour acupuncture. The Chinese women were normal weight compared with an average BMI of about 35 in the metformin group in the prior comparative trial, and BMI is inversely related to outcome.[8] Could that explain the difference between 7.2 and 15.4%? Well frankly, I’m afraid it probably can!

So where does that leave the acupuncture in the fertility arena? There is little or no high quality clinical data to support its use, just a lot of experimental data that did seem encouraging, although the results of this trial should give pause to the assumptions of advocates that anything demonstrated at the bench should automatically imply useful effects at the bedside.

For now at least we must encourage women with PCOS to consider clomiphene before acupuncture for ovulation induction.


  1. Wu X-K, Stener-Victorin E, Kuang H-Y, et al. Effect of Acupuncture and Clomiphene in Chinese Women With Polycystic Ovary Syndrome. JAMA 2017;317:2502. doi:10.1001/jama.2017.7217
  2. Stener-Victorin E, Waldenström U, Andersson SA, et al. Reduction of blood flow impedance in the uterine arteries of infertile women with electro-acupuncture. Hum Reprod 1996;11:1314–7.
  3. Stener-Victorin E. Reproductive medicine: Research projects in acupuncture. Acupunct Med 1998;16:80–2.
  4. Carr D. Somatosensory stimulation and assisted reproduction. Acupunct Med 2015;33:2–6. doi:10.1136/acupmed-2014-010739
  5. Stener-Victorin E, Maliqueo M, Soligo M, et al. Changes in HbA1c and circulating and adipose tissue androgen levels in overweight-obese women with polycystic ovary syndrome in response to electroacupuncture. Obes Sci Pract 2016;2:426–35. doi:10.1002/osp4.78
  6. Johansson J, Redman L, Veldhuis PP, et al. Acupuncture for ovulation induction in polycystic ovary syndrome: a randomized controlled trial. Am J Physiol Endocrinol Metab 2013;304:E934-43. doi:10.1152/ajpendo.00039.2013
  7. Legro RS, Barnhart HX, Schlaff WD, et al. Clomiphene, metformin, or both for infertility in the polycystic ovary syndrome. N Engl J Med 2007;356:551–66. doi:10.1056/NEJMoa063971
  8. Legro RS, Zhang H, Eunice Kennedy Shriver NICHD Reproductive Medicine Network. Letrozole or clomiphene for infertility in the polycystic ovary syndrome. N Engl J Med 2014;371:1463–4. doi:10.1056/NEJMc1409550

Declaration of interests MC

Burning nerves with needles in back pain – stop the burning, just use the needles!

This blog was first published on 27th July 2017 on

I subsequently attended the Spanish Pain Society conference (#SEDPalma2018) and there was a whole session devoted to this paper and why it did not apply to the way these specialists used the technique :-/.


This piece has also been stimulated by a publication in JAMA, this time evaluating the use of denervation of joints in spinal pain.[1] It is a set of three large (n=251, n=228, n=202), probably definitive, pragmatic trials that evaluate the use of radiofrequency denervation (RFD) as an addition to a 3 month standardised exercise programme. The design is such that the intervention was given the greatest possible opportunity to demonstrate an effect, that is both the specific effect of the intervention plus the context in which it is provided. This is rather similar to the Acupuncture in Routine Care (ARC) trials performed as part of the German Modellvorhaben Akupunktur;[2] all of which were markedly positive for acupuncture.

…at no time point did the addition of RFD reach clinical significance for the primary outcome

The results seem clear – at no time point did the difference been intervention and control reach clinical significance in terms of pain intensity (the primary outcome); and in only one of 18 time points across the three trials did the difference reach statistical significance (the 3 week outcome in the sacroiliac joint trial). The measure used for clinical significance here was 2 points on a 0-10 scale of pain, or 20mm on a 100mm visual analogue scale, but at no point did RFD achieve the lower level of 1 point set by NICE in NG59.[3]

The data from NG59 comparing acupuncture with no acupuncture controls (the closest equivalent comparison to the current trials of RFD) gave a pooled result at less than 4 months of more than 60% greater than the best outcome recorded in these trials of RFD. A result that was both statistically significant and clinically relevant by the standards used in NG59, although it would not have reached the standard set in these trials.

So what is RFD, and why am I drawing attention to this? RFD is a method of burning nerves, and the idea is that by denervating a pain source in the spine you might achieve sustained pain relief in chronic back pain. The typical targets are facet joints, sacroiliac joints and intervertebral joints. The radiofrequency term is unnecessarily confusing since the method uses electrical pulses at about 5000Hz rather than electromagnetic radiation in the radiowave spectrum. The latter stretches roughly from 3×103Hz to 3×109Hz equating respectively to wavelengths from 100km to 1mm, but that’s enough physics for now. Basically the high frequency electrical pulses cause a heating effect at the tip of the RFD probe and it typically reaches 80 degrees C, which coagulates the tissues at the tip.

Isn’t burning nerves a bit of an archaic technique? Yes it is, and in principle modern pain medicine tends to try to avoid neurolytic treatments. Damaging nerves can cause neuropathic pain in some individuals, and the nerves can grow back anyway.

Why was it recommended in NG59? This decision was controversial because the guideline development group (GDG) for NG59 recommended RFD based on quite limited data from very small trials, and the current Cochrane review clearly concluded that the evidence was insufficient to recommend it.[4] The decision to recommend RFD and recommend against acupuncture was also controversial because of potential conflicts of interests of interventionists on the GDG and how these were addressed.[5]

In reviewing the data on RFD used by NG59, I discovered that there was quite disproportionate weighting given to one particular trial despite it only having 20 patients per group (Tekin 2007 [6]). The reason for this is the meta-analytic software favours trials with low standard errors.

K.16.1 from Appendices K-Q of NG59

Anyway, I thought I would take a closer look at this paper, and discovered that the control group dropped from 6.8 to 4.3 on a 0 to 10 pain score from pre- to post-procedure. The slightly funny thing about this was that the control procedure involved exactly the same intervention as was used to determine eligibility for the trial – a diagnostic medial branch block ie a local anaesthetic block to the nerve that would then be coagulated or burnt in the active (CRF) group. In order to get in the trial the patients had to have a reduction in pain score of 50% or more, yet the control group who went on to have the same procedure again only dropped by about 37% (see Table 2 from Tekin 2007 below).

Table 2 from Tekin 2007

Well it all just goes to show that small trials are unreliable, but what should we do now? We should ask the centre that conducted the guideline (the National Guideline Centre hosted by the RCP) to perform an urgent review on the grounds of safety. As it stands NG59 has all but stopped NHS acupuncture for back pain and is likely to result in a vast increase in the use of RFD, which now we see doesn’t actually do anything worthwhile for patients, but may boost the Maserati-purchasing power of certain interventionists.


  1. Juch JNS, Maas ET, Ostelo RWJG, et al. Effect of Radiofrequency Denervation on Pain Intensity Among Patients With Chronic Low Back Pain: The Mint Randomized Clinical Trials. JAMA 2017;318:68–81. doi:10.1001/jama.2017.7918
  2. Cummings M. Modellvorhaben Akupunktur–a summary of the ART, ARC and GERAC trials. Acupunct Med 2009;27:26–30. doi:10.1136/aim.2008.000281
  3. NICE guideline on low back pain and sciatica in over 16s: assessment and management. 2016.
  4. Maas ET, Ostelo RWJG, Niemisto L, et al. Radiofrequency denervation for chronic low back pain. Cochrane database Syst Rev 2015: CD008572. doi:10.1002/14651858.CD008572.pub2
  5. Cummings M. 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. 2017;356.
  6. Tekin I, Mirzai H, Ok G, et al. A comparison of conventional and pulsed radiofrequency denervation in the treatment of chronic facet joint pain. Clin J Pain;23:524–9. doi:10.1097/AJP.0b013e318074c99c

Declaration of interests MC

Why not needles for OA – no steroid, just the needles!

This blog was first published on 30th May 2017 on


This piece has been stimulated by a recent publication in JAMA evaluating the use of regular intra-articular corticosteroid injections for symptom management and cartilage volume in osteoarthritis (OA) of the knee.[1] Previous research had suggested that the inflammatory process in the knee was associated with both pain and progression of cartilage loss, however, this trial clearly demonstrated a greater loss of cartilage after two years of 3 monthly intra-articular triamcinolone injections compared with the same frequency of saline injections.

It seems pretty clear then that we should avoid long-term use of intra-articular steroid within the knee and probably other synovial joints.

Previous research, also published in JAMA, indicates that steroid can also have a negative impact in the long term on lateral epicondylalgia.[2] Furthermore, a systematic review of the effects of local corticosteroid on tendon clearly concluded that the impact was negative both in vitro and in vivo.[3]

In shoulder pain it does not seem to matter whether or not the steroid is injected into the presumed target based on imaging, or whether it is injected into the buttock.[4] Moreover, given the anatomical vulnerability of the human supraspinatus tendon and its propensity for self destruction with age,[5] combined with the known negative effects of steroid on tendon, it looks as though we should avoid steroid in the shoulder too.

So what do we do if we do not inject steroid into our peripheral sources of musculoskeletal pain? We can try injecting other things I guess. Diclofenac, botulinum toxin or maybe normal saline – the latter seems to do very well when used as a control procedure in trials, in terms of the change from baseline. Having been brought up to accept steroid injection as a standard conventional procedure, it was a major surprise to find that needles alone (dry needling or local acupuncture) in tender muscle appeared to be highly effective.[6] Having got over this surprise I was guided through the process of my first systematic review to find that virtually none of the trials of needling and injection therapies in myofascial trigger point pain demonstrated superiority for any individual technique.[7] Indeed, saline injection (the intended control procedure) proved superior in most outcomes of one particularly good quality trial.[8]

Despite saline injection being no less effective in terms of pain relief than an ‘active’ comparator, and being associated with clinically meaningful changes from baseline in trials, we do not use it in practice. Well it has not been tested in a double blind randomised controlled trial… but what would we use in the control group of such a trial? Perhaps the needle without an injection. Then we have the challenge of blinding the practitioner, and we are getting closer to the dilemma of acupuncture research.

Acupuncture needles are less traumatic than hypodermic needles and push in fewer bugs from the skin…

…and there is no reason for them to go inside the joint

Acupuncture needles are less traumatic than hypodermic needles, and carry no risk related to the injected substance. They also carry less from the outside of the organism (skin flora and contaminants) into the internal environment because they lack the hollow bore of a needle for injection. In general we avoid needling into joint spaces with acupuncture needles despite the reduced theoretical risk of carrying in bugs from the outside. The best quality evidence for acupuncture in chronic pain related to osteoarthritis demonstrates and effect size (standardised mean difference) of 0.26 over sham acupuncture (minimal needling in the biggest trials) and an effect size of 0.57 over no acupuncture controls (waiting list, usual care, or guideline-based conventional care).[9] For comparison, topical non-steroidal anti-inflammatory drugs have an effect size of 0.4 over placebo and oral preparations range from 0.29 to 0.44.[10] So 0.57 looks pretty good if you don’t mind buying a bit of the relatively safe context of acupuncture, or if you prefer something more potent you might go for oral opiates which come in at 0.78… but we all know the path from there on, and it does not look so rosey.

Well I would go for the needles, probably with a little umph added from electrical impulses as Jorge Vas did in 2004,[11] with an effect size of 1.21 (this was an outlier in the Vickers IPDM,[9] but the only trial to use electroacupuncture (EA) to muscles around the knee compared with non-penetrating sham EA).


  1. McAlindon TE, LaValley MP, Harvey WF, et al. Effect of Intra-articular Triamcinolone vs Saline on Knee Cartilage Volume and Pain in Patients With Knee Osteoarthritis: A Randomized Clinical Trial. JAMA 2017;317:1967–75. doi:10.1001/jama.2017.5283
  2. Coombes BK, Bisset L, Brooks P, et al. Effect of corticosteroid injection, physiotherapy, or both on clinical outcomes in patients with unilateral lateral epicondylalgia: a randomized controlled trial. JAMA 2013;309:461–9. doi:10.1001/jama.2013.129
  3. Dean BJF, Lostis E, Oakley T, et al. The risks and benefits of glucocorticoid treatment for tendinopathy: a systematic review of the effects of local glucocorticoid on tendon. Semin Arthritis Rheum 2014;43:570–6. doi:10.1016/j.semarthrit.2013.08.006
  4. Ekeberg OM, Bautz-Holter E, Tveitå EK, et al. Subacromial ultrasound guided or systemic steroid injection for rotator cuff disease: randomised double blind study. BMJ 2009;338:a3112. doi:10.1136/bmj.a3112
  5. Vincent K, Leboeuf-Yde C, Gagey O. Are degenerative rotator cuff disorders a cause of shoulder pain? Comparison of prevalence of degenerative rotator cuff disease to prevalence of nontraumatic shoulder pain through three systematic and critical reviews. J shoulder Elb Surg 2017;26:766–73. doi:10.1016/j.jse.2016.09.060
  6. Cummings TM. A computerised audit of acupuncture in two populations: civilian and forces. Acupunct Med 1996;14:37–9. doi:10.1136/aim.14.1.37
  7. Cummings TM, White AR. Needling therapies in the management of myofascial trigger point pain: a systematic review. Arch Phys Med Rehabil 2001;82:986–92. doi:10.1053/apmr.2001.24023
  8. Frost FA, Jessen B, Siggaard-Andersen J. A control, double-blind comparison of mepivacaine injection versus saline injection for myofascial pain. Lancet 1980;1:499–
  9. Vickers AJ, Cronin AM, Maschino AC, et al. Acupuncture for chronic pain: individual patient data meta-analysis. Arch Intern Med 2012;172:1444–53. doi:10.1001/archinternmed.2012.3654
  10. Birch S, Lee MS, Robinson N, et al. The U.K. NICE 2014 Guidelines for Osteoarthritis of the Knee: Lessons Learned in a Narrative Review Addressing Inadvertent Limitations and Bias. J Altern Complement Med 2017;23:242–6. doi:10.1089/acm.2016.0385
  11. Vas J, Mendez C, Perea-Milla E, et al. Acupuncture as a complementary therapy to the pharmacological treatment of osteoarthritis of the knee: randomised controlled trial. BMJ 2004;

Declaration of interests MC

TENS and acupuncture appear cost-effective in knee osteoarthritis


This blog was first published on 9th March 2017 on

Figure 3 from Woods B et al PLoS One 2017[1]

This figure may seem familiar to some who follow big data in the acupuncture field. It comes from another big project at the Centre for Health Economics, University of York.[1] It is effectively a repeat of their first large network meta-analysis (NMA) that included acupuncture and sham acupuncture in knee osteoarthritis (OA),[2] but this time replacing pain outcomes with health-related quality of life in the form of the EQ-5D aka Euroqol.

There are also overlaps with Saramago et al from 2016,[3] which I wrote about on this blog under the title Quality sham. This paper by Woods et al narrows the view from chronic pain to OA knee alone and extends the analysis to a full cost comparison of non-pharmacological interventions.

This is a thorough piece of work from a well-recognised centre. Whilst data for some interventions was limited, the data for acupuncture and muscle strengthening exercise for example appears fairly reliable; that is, the confidence intervals are tight and the point estimate consistent in both analyses illustrated in the figure above. In total the NMA included 88 RCTs (randomised controlled trials) and 7507 patients.

I suppose the major limitation of this analysis is that there was only data available to calculate outcomes at 8 weeks ie after a course of treatment rather than in the long term. Woods et al cover this aspect in their discussion and put forward an argument for positive commissioning decisions rather than waiting for more evidence.

In terms of EQ-5D outcomes, acupuncture appears to do well, but costs of performing a course of treatment must also be taken into account. When this is done, TENS is the most cost effective intervention, coming in at £2690 per QALY (quality adjusted life year) versus usual care. When only trials with a low risk of selection bias were considered the effect size of TENS dropped and it then came in at £6142 per QALY versus usual care. In this analysis acupuncture then became cost effective at £13 502 versus TENS.

I should note that when all non-pharmacological interventions are considered TENS and acupuncture are the two most cost effective. In the latest clinical guideline from NICE on osteoarthritis (CG177)[4] TENS is recommended, but acupuncture is not.

…when all non-pharmacological interventions are considered TENS and acupuncture are the two most cost effective, but NICE only recommends TENS


  1. Woods B, Manca A, Weatherly H, et al. Cost-effectiveness of adjunct non-pharmacological interventions for osteoarthritis of the knee. PLoS One 2017;12:e0172749. doi:10.1371/journal.pone.0172749
  2. Corbett MS, Rice SJC, Madurasinghe V, et al. Acupuncture and other physical treatments for the relief of pain due to osteoarthritis of the knee: network meta-analysis. Osteoarthritis Cartilage 2013;21:1290–8. doi:10.1016/j.joca.2013.05.007
  3. Saramago P, Woods B, Weatherly H, et al. Methods for network meta-analysis of continuous outcomes using individual patient data: a case study in acupuncture for chronic pain. BMC Med Res Methodol 2016;16:131. doi:10.1186/s12874-016-0224-1
  4. NICE guideline update on osteoarthritis: the care and management of osteoarthritis in adults. 2014.

Declaration of interests MC

Precision needling in myofascial pain

This blog was first published on 9th February 2017 on

Inspired by Wang et al Acupunct Med 2017 [1]

Image taken from Cummings M Acupunct Med 2009 [2]

I got interested in swapping my hypodermic needles for filiform ones some 25 years ago, and was encouraged by my early success treating myofascial pain in a military population.[3] I became more and more expert at identifying these targets we call trigger points, touching them briskly with the tip of my fine filiform needle, and seeing them twitch with almost immediate relief of pain and tightness in the muscle. The twitch seemed to go along with immediate results, but it could be elusive, and other colleagues claimed similar success with less 3 dimensional accuracy – superficial needling or simply needling an acupuncture point nearby. As I have followed the clinical research in acupuncture I became less and less convinced that my accuracy, and the accuracy I tried to teach would be validated since there was so little difference between even real and sham needling.

I became less and less convinced that precision of trigger point needling would be validated

My early research was a review of all both wet and dry needling in myofascial pain.[4] There was a strong suggestion that when injecting trigger points (wet needling), the substance in the syringe did not seem to matter, but all groups appeared to improve dramatically. At the time there were not many trials using filiform needles.

Now we have a selection of trials that can be combined in meta-analysis, and the tentative conclusion is that targeting trigger points seems to have some specific effect over sham, but that targeting acupuncture points is not clearly superior to sham.[1]

Figure 3 from Wang et al Acupunct Med 2017 [1]

It is always worth having a careful look at Forest plots – so easy to miss the wood for the trees, so to speak, or even overlook some very strange trees! There are some issues to note here. The results of one trial (1.1.4 Tekin 2013) got included twice, albeit at different time points – they probably should have just decided on using one time point. Then there is a noticeable outlier in the lower plot (1.1.5 Chou 2009) – the effect size of this trial was huge in comparison to all the others. Under these circumstances it is always worth doing a sensitivity analysis excluding outliers. In this case it led me to check the original paper, and whilst the authors of this review classified it under acupuncture point treatment rather than trigger point treatment, the paper seems to suggest it used a trigger point needling technique to obtain multiple local twitch responses (LTRs) from remote trigger points that happened to be also at acupuncture point sites – in this case LI11 & TE5. Tricky to know how to classify this one then, but wait, there is another paper that used remote needling of a trigger point and measured an effect on upper trapezius myofascial pain (1.1.4 Tsai 2010). Maybe we should exclude that one as well in sensitivity analysis? Well I would have done all that for you, but given the small total number of total participants and the risk of bias, any conclusions would be unlikely to rise beyond a tentative suggestion.

So there you have it, perhaps the first meta-analysis of filiform needling in myofascial pain that points towards more accurate targeting of trigger points – but we have a long way to go!


  1. Wang R, Li X, Zhou S, et al. Manual acupuncture for myofascial pain syndrome: a systematic review and meta-analysis. Acupunct Med 2017. doi:10.1136/acupmed-2016-011176
  2. Cummings M. Myofascial trigger points: does recent research gives new insights into the pathophysiology? Acupunct Med 2009;27:148–9. doi:10.1136/aim.2009.001289
  3. Cummings TM. A computerised audit of acupuncture in two populations: civilian and forces. Acupunct Med 1996;14:37–9. doi:10.1136/aim.14.1.37
  4. Cummings TM, White AR. Needling therapies in the management of myofascial trigger point pain: a systematic review. Arch Phys Med Rehabil 2001;82:986–92. doi:10.1053/apmr.2001.24023

Declaration of interests MC

Acupuncture for infantile colic – misdirection in the media or over-reaction from a sceptic blogger?

This blog was first published on 26th January 2017 on At the time I was in Cape Town on holiday, trying to get a rapid response published to the NG59 summary in the BMJ. It was critical of NICE, and I was negotiating over content with a legal expert from BMJ! The response took three weeks to go up, by which time it was too late to be noticed. In the meantime I created a bit of a storm with this blog, and my use of the term ‘old sceptic blogger’ in the title. This is (mostly) the version edited by BMJ.


So there has been a big response to this paper press released by BMJ on behalf of the journal Acupuncture in Medicine. The response has been influenced by the usual characters – retired professors who are professional bloggers and vocal critics of anything in the realm of complementary medicine. They thrive on flexing their EBM muscles for a baying mob of fellow sceptics (see my ‘stereotypical mental image’ here). Their target in this instant is a relatively small trial on acupuncture for infantile colic.[1] Deserving of being press released by virtue of being the largest to date in the field, but by no means because it gave a definitive answer to the question of the efficacy of acupuncture in the condition. We need to wait for an SR where the data from the 4 trials to date can be combined.

On this occasion I had the pleasure of joining a short segment on the Today programme on BBC Radio 4 led by John Humphreys. My protagonist was David Colquhoun, who spent his short air-time complaining that the journal was even allowed to be published in the first place. Why would BBC Radio 4 invite a retired basic scientist and professional sceptic to be interviewed alongside one of the journal editors – a clinician with expertise in acupuncture (WMA)?

At no point was it made manifest that only one of us had ever been in a position to try to help parents with a baby that cries excessively. 

So what about the research itself? I have already said that the trial was not definitive, but it was not a bad trial. It suffered from under-recruiting, which meant that it was underpowered in terms of the statistical analysis. But it was prospectively registered, had ethical approval and the protocol was published. Primary and secondary outcomes were clearly defined, and the only change from the published protocol was to combine the two acupuncture groups in an attempt to improve the statistical power because of under recruitment. The fact that this decision was made after the trial had begun means that the results would have to be considered speculative. For this reason the editors of Acupuncture in Medicine insisted on alteration of the language in which the conclusions were framed to reflect this level of uncertainty.

David Colquhoun has focussed on multiple statistical testing and p values. These are important considerations, and we could have insisted on more clarity in the paper. P values are a guide and the 0.05 level commonly adopted must be interpreted appropriately in the circumstances. In this paper there are no definitive conclusions, so the p values recorded are there to guide future hypothesis generation and trial design. There were over 50 p values reported in this paper, so by chance alone you must expect some to be below 0.05. If one is to claim statistical significance of an outcome at the 0.05 level, ie a 1:20 likelihood of the event happening by chance alone, you can only perform the test once. If you perform the test twice you must reduce the p value to 0.025 if you want to claim statistical significance of one or other of the tests. So now we must come to the predefined outcomes. They were clearly stated, and the results of these are the only ones relevant to the conclusions of the paper. The primary outcome was the relative reduction in total crying time (TC) at 2 weeks. There were two significance tests at this point for relative TC. For a statistically significant result, the p values would need to be less than or equal to 0.025 – neither was this low, hence my comment on the Radio 4 Today programme that this was technically a negative trial (more correctly ‘not a positive trial’ – it failed to disprove the null hypothesis ie that the samples were drawn from the same population and the acupuncture intervention did not change the population treated). Finally to the secondary outcome – this was the number of infants in each group who continued to fulfil the criteria for colic at the end of each intervention week. There were four tests of significance so we need to divide 0.05 by 4 to maintain the 1:20 chance of a random event ie only draw conclusions regarding statistical significance if any of the tests resulted in a p value at or below 0.0125. Two of the 4 tests were below this figure, so we say that the result is unlikely to have been chance alone in this case. With hindsight it might have been good to include this explanation in the paper itself, but as editors we must constantly balance how much we push authors to adjust their papers, and in this case the editor focussed on reducing the conclusions to being speculative rather than definitive. A significant result in a secondary outcome leads to a speculative conclusion that acupuncture ‘may’ be an effective treatment option… but further research will be needed etc…

Now a final word on the 3000 plus acupuncture trials that David Colquhoun mentions. His point is that there is no consistent evidence for acupuncture after over 3000 RCTs, so it clearly doesn’t work. He first quoted this figure in an editorial after discussing the largest, most statistically reliable meta-analysis to date – the Vickers et al IPDM.[2] He admits that there is a small effect of acupuncture over sham, but follows the standard EBM mantra that it is too small to be clinically meaningful without ever considering the possibility that sham (gentle acupuncture plus context of acupuncture) can have clinically relevant effects when compared with conventional treatments. Perhaps now the best example of this is a network meta-analysis (NMA) using individual patient data (IPD), which clearly demonstrates benefits of sham acupuncture over usual care (a variety of best standard or usual care) in terms of health-related quality of life (HRQoL).[3]

Key to abbreviations

  • BMJ – British Medical Journal (company)
  • EBM – evidence-based medicine
  • HRQoL – health-related quality of life
  • IDP – individual patient data
  • IDPM – individual patient data meta-analysis
  • MCID – minimal clinically important difference
  • NMA – network meta-analysis
  • SR – systematic review
  • VAS – visual analogue scale (usually a 100mm line)


  1. Landgren K, Hallström I. Effect of minimal acupuncture for infantile colic: a multicentre, three-armed, single-blind, randomised controlled trial (ACU-COL). Acupunct Med 2017: acupmed-2016-011208. doi:10.1136/acupmed-2016-011208
  2. Vickers AJ, Cronin AM, Maschino AC, et al. Acupuncture for chronic pain: individual patient data meta-analysis. Arch Intern Med 2012;172:1444–53. doi:10.1001/archinternmed.2012.3654
  3. Saramago P, Woods B, Weatherly H, et al. Methods for network meta-analysis of continuous outcomes using individual patient data: a case study in acupuncture for chronic pain. BMC Med Res Methodol 2016;16:131. doi:10.1186/s12874-016-0224-1

Declaration of interests MC