Acupuncture is a complex “alternative” modality because something physical is actually happening – thin needles are being stuck through the skin and manipulated. So it is therefore not impossible that a physiological response is happening. It is much easier to comment on things like homeopathy and therapeutic touch where literally nothing physical is happening and the plausibility for any benefit is therefore zero. So if I try to answer the question in my title, much explanation and qualifications are required. To answer this question – does acupuncture work? – my current best answer based upon available evidence is a qualified no. This answer is not changed by the most recent study of acupuncture that is being touted by the press as evidence that acupuncture works. (Here is the original study, but a subscription is required.)

Let’s first look at this study, which was a German study of acupuncture for back pain. Dr. Heinz Endres studied 1,100 randomized patients with three treatment arms. The first received standard therapy – massage, anti-inflammatories, and heating pads. The second received acupuncture, and the third received sham acupuncture where the needles were inserted but not deeply, and not manipulated, and not in traditional acupuncture points. The study found 47% improvement in the acupuncture group, 44% in the sham acupuncture, and 27% in the standard therapy group after 6 months.

This single study, even taken just by itself, falls far short of demonstrating that acupuncture works. And of course we have to place it in the context of plausibility and the entire acupuncture literature. We also have to identify appropriate sub-questions.

First let us consider the difference between “real” acupuncture and “sham” acupuncture. Acupuncture is based upon the ancient and superstitious pre-scientific notion that there are lines of mysterious life energy (chi) flowing through our bodies, and that the flow of this energy is responsible for health and illness. Acupuncture is supposed to free up blockages in the flow of chi energy. I grant this idea a scientific plausibility of zero – meaning we can safely discard it.

What does the evidence show for the chi theory of acupuncture? The evidence is overwhelmingly negative, and this study supports this negative consensus. Most well-designed studies that compare traditional and sham acupuncture show no difference between the groups. In this study the two groups were 47% and 44% respectively. This means that it does not matter where you put the needles or if you manipulate them in any way – that’s because there are no lines of flowing chi.

It’s interesting that no headlines declare from this data that the study shows chi does not exist, or that traditional acupuncture is a “sham.” (Hey, I like that – someone should use it.)

This means that all claims for acupuncture that rely upon the notion of chi – including all claims for so-called “medical” acupuncture, are without a theoretical basis and are refuted by empirical evidence. So acupuncture won’t cure cancer.

But this still leaves the semi-plausible claim that sticking needles in random locations in the body may provoke a physiological response that is helpful in the symptomatic management of certain chronic syndromes that involve pain, nausea, headaches, or similar symptoms. For this much more narrow claim my current answer is still a tentative no, acupuncture does not work (at least the evidence is not sufficient for me to prescribe it) but I am willing to be convinced by better evidence.

I do not find this study or studies like it convincing for a number of reasons. First, there is the obvious problem that the standard therapy group knew they were not getting acupuncture, so they were not blinded.

The second big problem with this study is that back pain is a complex and hard to treat entity. It may be caused by arthritic pain, inflammation, muscle tightness, soft tissue pain, and nerve pain, or any combination of these factors. So it is very heterogenous, and that is a bad thing for clinical trials. But more importantly, I am not convinced that the standard therapy group was adequate, primarily because 27% improvement seems low. Other studies have shown that as many as 65% of patients with chronic low back pain will improve within 12 weeks. This low response rate to the standard therapy opens the door to a placebo effect in the sham and traditional acupuncture groups, and that may be enough to explain the effect size we are seeing in this study.

Further, more is happening on the acupuncture table than just the needles. Subjects are encourage to relax, often with music and incense to enhance the environment. Often the acupuncture points are palpated prior to needle insertion, and this can serve as a form of gentle massage. In fact an acupuncturist once confided in me that he thought everything leading up to the needle insertion was more important to the symptomatic benefit than the needles. All of this introduces other variables that interfere with our ability to conclude that the needles introduce any physiological benefit.

Conclusion and Recommendations

At this time I think we can say that the theoretical basis of traditional acupuncture, namely chi, can be discarded. We can also confidently conclude on the weight of the literature that needle placement based upon traditional Chinese diagnostic methods and the notion of chi has been proven false. It doesn’t matter where you stick the needles. I think we can also dispose of any claims for medical acupuncture – using acupuncture to cure disease rather than treat symptoms. The evidence for acupuncture and smoking cessation is also sufficient to abandon this as a viable treatment.

What we are left with is using needle insertion for symptomatic treatment of various pain syndromes and perhaps nausea. I think it is important to dispense with the superstitious nonsense that still surrounds acupuncture so we can focus on these remaining questions. Is there any benefit to acupuncture above the placebo effect? Is the needle insertion a necessary component of the acupuncture treatment? (So far I find the evidence unconvincing on these questions.) If it does work, what is the underlying physiological mechanism and are their easier and less invasive ways to exploit it?

Acupuncture must free itself from its superstitious roots to more effectively address these questions, to be taken seriously, and to avoid the chronic problem of CAM where positive findings in a very narrow area are used to justify a large system of pseudoscience. In other words, evidence that perhaps acupuncture may help back pain is used to justify using it for everything, including cancer. The link must be broken.

Meanwhile, what is the next step in addressing the questions above. Study design in acupuncture has been improving, but is not yet sufficient to eliminate the placebo effect and other variables as being responsible for the inconsistent positive effects that have been seen. What we need are truly double blind tests – and not between traditional acupuncture and sham acupuncture (as I said, that ship has sailed) but between any acupuncture and no acupuncture.

Here is a suggestion. We need to develop an experimental acupuncture needle that is housed in an opaque rigid sheath. This has been done with glass sheaths in some studies because when pressing the sheath against the skin the subject cannot tell if a needle is inserted or not (because of limitations in what we call two-point discrimination – the nerves cannot separate the stimuli). This is a good idea to blind the subject, but now we must take it a step further to blind the acupuncturist. Modify this setup so that a plunger is depressed that either will or will not insert a needle into the subject, in a way so that the subject and the acupuncturist cannot know if a needle was inserted. What this will accomplish is to truly isolate the variable of needle insertion.

So far what we have seen in the acupuncture literature amounts to noise (see my earlier post on how to interpret the literature). To rise above these we need better studies. And more importantly the acupuncture community (and the media) needs to stop using narrow scientific evidence to support broad pseudoscientific claims.

________________________________________________________________________________________________

Addendum:

Orac has also commented quite eloquently on this study. He makes the additional point that, according to study protocol, subjects were moved from the responder group to non-responder if they used any proscribed treatments during the study, and this was done for 50% of the conventional therapy group and only 35% and 33% for the acupuncture groups. This difference would tend to artificially increase the apparent superiority of acupuncture over conventional therapy.

Ben Goldacre at Bad Science also discusses this article. He discusses the role of the placebo effect in this study, and also points out that the study included patients who had 8 years of chronic back pain, not previously treated with acupuncture. This introduces a potential bias into the study. The patient were chosen for being refractory to conventional therapy, but were naive to acupuncture. So this might explain why the response rate to conventional therapy was so low, and also probably contributed to the difference in response between conventional therapy and acupuncture.

In other words, imagine you want to compare treatment A to treatment B for condition X. You choose 1000 subjects with X who have all tried treatment A and it failed to work, but who have not tried treatment B. Do you think this is a fair comparison of the treatments?

Finally (at the risk of patting myself on the back) I could not help noticing that the response of the mainstream media to this study was largely credulous – they completely missed the real points of this study and did not place it into a reasonable scientific context. Meanwhile, science bloggers have collectively provided insightful critical analysis with appropriate background for context. It is increasingly apparent to me that on issues such as this science bloggers are consistently serving the public far better than traditional media. Exploring the reasons for this is a blog entry of its own, but my quick analysis is that mainstream science journalism is too diluted by non-science journalists and the blurring of news and entertainment in the media. Meanwhile science bloggers are a self-selected group of scientists and educators with the inclination for promoting the public understanding of science.