Download Lectures on Biostatistics (1971). Corrected and searchable version of Google books edition
Download review of Lectures on Biostatistics (THES, 1973).
This week’s edition of Nature Neuroscience carried a paper with the title “Adenosine A1 receptors mediate local anti-nociceptive effects of acupuncture“. The paper was not without interest, but it tells one nothing about acupuncture in humans.
The mice had their paws injected with rather unpleasant stuff called Complete Freund’s Adjuvant. This makes them inflamed and sore, so they are unusually sensitive to pain. When the mice were pierced with human-sized needles, and the needles rotated, they responded a bit less to painful stimuli.
The newspaper reports of this paper were even worse than usual.
The worst of the lot was Richard Alleyne, in the Daily Telegraph: “Acupuncture does work as it stimulates a natural pain killer, scientists find“.
Fiona McCrae in the Daily Mail was marginally better, with “Let’s get straight to the point, acupuncture DOES ease pain“. At least she manages to include a brief critical comment from Edzard Ernst.
Even the usually-excellent Ian Sample in the Guardian was unusually gullible with “Acupuncture may ease pain by triggering release of natural painkiller“. The subheading did say the study was done in mice not man, but there was no critical comment and, as in the other reports, no attempt to summarise what’s known already about acupuncture.
Worse still, it wasn’t just the newspapers. There were similar poor reports in Science magazine, in Scientific American, and even the Nature news blog left a lot to be desired.
When this sort of misreporting occurs, it often seems to come from the press release (another case documented here). Many journalists never get any further than the press release, Here is Nature‘s press release.
|
How needles pierce pain DOI: 10.1038/nn.2562 Acupuncture can relieve many kinds of pain, but it remains unclear how it might work, beyond the possibility of a strong placebo effect. A study published online in Nature Neuroscience now shows that acupuncture locally activates pain-suppressing receptors which could be the key to the treatment’s ability to relieve pain. Maiken Nedergaard and colleagues inserted fine needles into the mouse equivalent of a traditional acupuncture point near the knee, and rotated these needles intermittently as is practiced by acupuncturists. This alleviated the pain reactions of mice with an inflamed paw, and it also strongly increased the local tissue concentration of the neurotransmitter adenosine. Pain relief required the presence of a particular adenosine receptor. It is known that this receptor resides on pain-transmitting nerve fibers and can reduce the activity of these fibers. The authors found that no pain relief or adenosine elevation was observed when the needles were simply inserted into the acupuncture point without rotation. They also noted that a drug that prolongs the lifetime of adenosine in live tissue helped to prolong the pain-attenuating effect of mouse acupuncture. It should be noted that while this work suggests a mechanism for local pain relief by acupuncture, it does not in any way endorse the ancient mystical idea that the needles work by correcting some aberrant “qi” energy flow along “meridians”. Instead, the authors propose a model whereby the minor tissue injury caused by rotated needles triggers adenosine release, which, if close enough to pain-transmitting nerves, can lead to the suppression of local pain. Author contact: |
The title, and the first sentence, are tendentious, to put it mildly. The false assertion that "Acupuncture can relieve many kinds of pain" misrepresents entirely the current state of the evidence. A lot has been written about acupuncture, not least on this blog, but the current position can be summed up roughly as follows.
- It has been found time and time again that no difference can be detected between real and sham acupuncture in blind comparisons, and
- a non-blind comparison between acupuncture and no-acupuncture usually shows some advantage for the acupuncture, but the effect is small, transient and very variable. A recent review from the Nordic Cochrane group concluded the effect was barely big enough to be clinically significant.
The odds are that acupuncture is no more than a theatrical placebo.
The press release should have made all this clear, but failed to do so. But then the paper itself also failed to make the context clear.
The abstract of the paper starts "Acupuncture is an invasive procedure commonly used to relieve pain." The second paragraph starts "Although the analgesic effect of acupuncture is well documented, little is understood about its biological basis". It is simply not true that "analgesic effect of acupuncture is well documented". It is not. Try reading Barker Bausell’s excellent book, “Snake Oil Science” if you want to know about the strength of the evidence.
The whole paper is written with the underlying premise that acupuncture works, and all that needs to be done is to explain why. How the paper got past the referees in its present form beggars belief.
Journals like Nature Neuroscience, like it or not, have a certain prestige. Of course, the prestige is exaggerated, and like all journals, it has some good papers and some poor ones. With that prestige comes responsibility to ensure that press releases are accurate. But one often has the impression that press releases are written with the aim of getting wide press coverage, and prestige for the journal, rather that as critical evaluations of the science. This is one such case.
Martin Rees said today, in his first Reith lecture, that "science is organised scepticism". It should be, and in the long run, it is. This paper is an exception and sadly it isn’t alone.
Other commentaries on this paper
There have already been some excellent commentaries on blogs. As so often, the best bloggers are sharper than the scientific journals. If it were not for the fact that my comments on Twitter got some criticism from Nature, I wouldn’t have bothered to add to the following excellent analyses.
Ed Yong, in the Discover site was quick off the mark with A biological basis for acupuncture, or more evidence for a placebo effect?. This provides more detail than I’ve given here. It also has an intriguing postscript.
“PS: The paper notes that the authors have no competing financial interests that might have affected their work. However, it is worth noting that one of the co-authors, Jurgen Schnermann, is married to one Josephine Briggs. Briggs is the director of the National Center for Complementary and Alternative Medicine [NCCAM], an institute that has had its fair share of controversy in the past.”
NCCAM is, of course the branch of the US National Institutes of Health that has spent over a billion dollars of US taxpayers’ money without coming up with a single good treatment. This is rather fascinating, because the report in the Telegraph ends with a laudatory comment from, ahem. Dr Josephine Briggs.
Dr Josephine Briggs, director of the National Centre for Complementary and Alternative Medicine at the National Institutes of Health in the US, said: “It is clear that acupuncture may activate a number of different mechanisms.
“This carefully performed study identifies adenosine as a new player in the process. It is an interesting contribution to our growing understanding of the complex intervention which is acupuncture.”
James Cole at the Stuff and Nonsense blog has another take on the same work, under the ironic title Acupuncture Works, Say Scientists
At Respectful Insolence, Orac gives a superb and thorough account. Read it: When what an acupuncture study shows is much more interesting than what acupuncture believers think it shows.
Follow-up
14 Responses to Why was an study on ‘acupuncture’ reported so badly?
Leave a Reply
This site uses Akismet to reduce spam. Learn how your comment data is processed.
[…] This post was mentioned on Twitter by David Colquhoun, Jo Brodie. Jo Brodie said: Fab RT @david_colquhoun My take on Nature Neurosci acupuncture paper http://www.dcscience.net/?p=3136 See also Orac http://bit.ly/avdyma […]
As I contributed to the comments in the Guardian article people seem to think the placebo effect is magic and can have no biological mechanism. Saying acupuncture works as a placebo is not a mechanistic explanation. For that you need mechanisms like adenosine release locally or endorphin release centrally.
All this paper has done is uncover a mechanism for one aspect of the placebo effect. It is a great pity as you say David that they let their assumptions cloud their interpretation of the experiment.
Apart from the Nature Neuro press release, the lead author also seems to have, er, done little to damp down the more overblown claims
From Ian Sample’s Guardian story:
The last sentence in particular smacks of an uncritical enthusiast, to say the least.
I went to a talk this lunchtime where the speaker mainly analysed other people’s (genetic expression) data. He made the very interesting comment that the quality of the data tended to be poor for papers published in journals like Nature and Science, where the reports were brief and people had clearly rushed to get the paper out, than for high-ranking but less prestigious journals, such as J. Neuroscience, where the articles were longer and had clearly been reviewed by experts. I would add that the bias towards highly exciting and topical papers in some journals may also make editors less critical.
Made me think there’s a little project out there for someone, to do a blind methodological evaluation of papers for journals of different levels of impact factor.
It *ought* to be the case that the top journals are the most methodologically rigorous, but the acupuncture paper would be in line with an alternative hypothesis that there’s a U-shaped function, with the peak methodological rigour in journals that have high, but not excessively high, impact.
PS Please do promote the Orwellian prize, which aims to clean up newspaper reporting of science – not appropriate here, i think, because the journalists were merely picking up on the press release, but I’m sure you will find good candidates in future.
http://deevybee.blogspot.com/2010_06_01_archive.html
[…] Why was a study on ‘acupuncture’ reported so badly? […]
@deevybee
Thanks for that very interesting comment.
This is certainly not the first time such suggestions have been made.
In 2007, Brown & Ramaswamy looked at the quality of published crystal structures. They concluded
Are the results that you described about expression data published anywhere?
Does anyone else know of other similar examples?
It is funny you used the term “impact factor” … I know of a website that grades impact factor for medical journals:
http://impactfactor.weebly.com/index.html
Being a crystallographer and a supporter of open access publication, I along with many others believed the oft mentioned belief that impact factor was negatively correlated with structure quality.
However, a more recent analysis than that of Brown & Ramaswamy has concluded that in fact there is little evidence for this once other confounding factors (e.g. size of molecule, resolution) are taken into account. The study I mention is by Kleywegt & Read
Section 4.1 specifically deals with correlations to journal impact factor. Hope that is of interest to you.
It all seems so simple to me. If you have a pain, stick some needles and/or pins into the tissue around that pain. The new pain that doing this causes, will take your mind of the original pain. Acupuncture works!
Alternatively, you could hit the acupuncture points with a hammer or wrench. Big pain requires big tools.
Here’s an interesting post that I just came across.
“Academic Incentives Gone Awry: Why Junk Persists in the Scientific Literature”, by Kent Anderson
Academic Incentives Gone Awry: Why Junk Persists in the Scientific Literature
@Moochie
Some years back I was browsing through one of the many bookshops in Hay-on-Wye and came across a book printed in 1930 entitled ‘Impact Therapy’.
It involved placing a sandbag over the affected area and hitting it with another sandbag. I thought of introducing it into the NHS, but took the view that the ethics committee may not approve!
As to you theory that one pain hides another, I would suggest it’s a but more complicated than that. But then that doesn’t make acupuncture justified in terms of cost benefit.
[…] therapies” built on shaky science) pointing out the true interesting result of the paper, and DC’s Improbable Science dissecting the media coverage and press release. Posted by – Rob […]
The acupuncture point ST-36 (Zusanli) is supposedly applicable for the following conditions: gastric pain,vomiting, abdominal distention, indigestion, diarrhoea, constipation, mastitis, aching of the knee joint, breast pain, oedema of the limbs, difficult urination, incontinence, paralysis, seizures, gastric ulcers, and other abdominal symptoms. The point is not known for use in controlling general pain. So it is mystifying why the researchers claim they are trying to hit this point on a mouse to test analgesic effects. Indeed it is mystifying why they are trying to hit any acupuncture point at all, given the problems of scale.
On a human, a needle about twice the diameter of the 0.16 gauge needle used in these experiments would used to needle into this location – on an area of skin about 1 cm wide. What they did on these poor mice was the equivalent of shoving something the size of a scaffolding pole into the lower leg of a human. It has nothing to do with acupuncture.
The authors also state that acupuncture has been practised for 4000 years. Their lamentable lack of historical knowledge makes one wonder why they even bother mentioning this in a scientific paper if they can’t check their facts. The claim supports the usual myth that the system and doctrine by which sharp things are stuck into the body represents a single body of knowledge going back to 2000 BC. Nothing could be further from the truth. In fact, the earliest known Chinese manuscripts relating to medical practices are the Mawangdui silk texts (168 BC) which describe rituals of blood letting to let the demons out. This has absolutely nothing to do with our general conception of acupuncture in the 21st century. The Huangdi neijing was written later and took a more cosmological view of the cause of illness, but even its description of using needles hardly constitutes the sort of thing that is found today in China.
The primitive kind of shamanistic exorcism that seems to have been practised in China 2200 years ago is as far away from modern acupuncture as is using needles to vaccinate kids against polio. So why do the researchers even mention this fake pedigree in their article? It seems to be mentioned as if it provides some form of validation for the practice of sticking needles into afferent nerves. Clearly it is mentioned to act as a hook for all the lazy journos who write soft articles for women’s magazines – who then unquestioningly repeat the 4000 year old myth, and the unquestioning acceptance that in some way “acupuncture works”. This article certainly did nothing to prove that, because they weren’t performing “acupuncture” by any stretch of the imagination.
[…] for acupuncture. But the advantage is usually too small to be of any clinical significance. In all probability it is a placebo effect -it’s hard to imagine a more theatrical event than having someone in a white coat stick long […]