Can we finally just say that acupuncture is nothing more than an elaborate placebo? Can we?

ResearchBlogging.orgI think my title says it all: Can we finally just say that acupuncture is nothing more than an elaborate placebo? Can we?

The reason I ask this question is because yet another large meta-analysis has been released that is entirely consistent with the hypothesis that acupuncture is a placebo. Because I’ve written about so many of these sorts of studies over the last year or two that I really had a hard time mustering up the will to write about one more. But I got in pretty late last night and therefore knew I could handle this in a reasonably expeditious fashion. Besides, it is a fairly interesting study as far as methodology. Also, I’ve been on a roll writing about antivaccine nonsense, but it’s taken a toll. My foray into the discussion forums resulted in a real assault on my neurons, so much so that, while I had considered taking on the not-so-dynamic duo, two crappy tastes that taste crappy together, otherwise known as David Kirby and Robert F. Kennedy, Jr. in the pages of The Huffington Post. Truly, just when I think HuffPo can’t go any lower, damn if it doesn’t prove me wrong. In any case, I thought it ill advisable to subject myself to more of Kirby and RFK, Jr. so soon on the heels of waves of burning mommy stupid. Better to take on this study, because it’s yet another piece of evidence that backs up what I’ve come to accept after having actually read that acupuncture literature: That acupuncture is nothing more than a placebo.

So let’s take a bit more of a look at the study. Entitled Acupuncture treatment for pain: systematic review of randomised clinical trials with acupuncture, placebo acupuncture, and no acupuncture groups appeared in the BMJ two days ago and was performed by Madsen et al at the Nordic Cochrane Center, meaning it’s a Cochrane review.

A meta-analysis, as I hope you recall, is a systematic methodology for reviewing the peer-reviewed medical literature in which studies are examined and pooled together to try to produce a consensus. Of course, this is the sort of thing that’s fraught with methodological problems. It’s easy to wonder what pooling data means when the data come from multiple studies that may have variable inclusion criteria, different endpoints, different design. However, it can be done, although one should look at the results with a skeptical eye. After all, garbage in = garbage out, which means that study selection is almost always the most important aspect of any meta-analysis. Consequently, that’s the part I always look at first.

This particular meta-analysis examined studies of acupuncture for pain. What’s good about the search strategy used is that it not only discarded various types of acupuncture that aren’t really acupuncture, such as electroacupuncture, but it also only included only trials that used some sort of placebo acupuncture, be it sticking needles into non-meridian points or using the special “placebo needles” that retract and give the illusion of penetrating the skin but do not actually do so. In other words, it used only trials that included an acupuncture group, a placebo acupuncture group, and a no acupuncture group. Another strength of the analysis is that it also included only studies in which both the acupuncture groups and the no-acupuncture groups were given the same medical interventions. If the no treatment group received a treatment substantially different from the acupuncture groups, the study was not included. Finally, trials in which the patients self-reported pain on a visual analog scale or similar pain scale were included.

The results were thirteen trials, with a total of 3,025 patients between them. Unfortunately, this leads me to mention one weakness of this analysis, namely that the studies included covered a rather wide range of conditions, including knee osteoarthritis,w7-w9 tension type headache, migraine, low back pain, fibromyalgia, abdominal scar pain, postoperative pain, and procedural pain during colonoscopy. Moreover, the duration of treatment varied from one day to 12 weeks. I could also complain about the way the term qi and meridians are described in the manuscript as though they had any validity other than as prescientific magical thinking, but what would be the point? Be that as it may, what were the findings?

In essence, zilch, nada, zip.

Well, not quite, but the results were meager. Let’s take a look at these relative risks. Here is the plot of acupuncture versus placebo acupuncture:


The diamond represents the pooled results of all the studies. Although it is statistically significantly different than zero, it is barely so. Indeed, it’s so small that that it falls below the minimally accepted threshold for a clinically noticeable reduction in pain. The authors even speculated that it was so low that it could not be distinguished from bias that might result from the difficulty in truly double-blinding acupuncture studies (generally, it’s very difficult to blind the practitioners doing the acupuncture to who is getting placebo and “real” acupuncture). Incomplete blinding could account for this small difference.

The authors also looked at placebo acupuncture versus no acupuncture. There, as expected given the lack of blinding of patient or practitioner, there was a more significant apparent reduction in pain due to the placebo effect, but even then it only just barely approached the threshold accepted as being a clinically noticeable and relevant analgesic effect. Interestingly, the magnitude of this effect did not depend upon the type of placebo used (retractable needles versus needles stuck in the “wrong” places). Also, the effect was quite variable, with some trials showing large effects that would clearly be clinically relevant and others showing much smaller, clinically irrelevant effects.

Not surprisingly, various press reports, perhaps the most egregious of which, have represented this study as being less negative in terms of whether acupuncture is more than a placebo than it is. For example, the BBC reported Confusion on acupuncture benefit when there really isn’t much confusion when one looks at the science dispassionately. It even quotes acupuncture “experts” thusly:

Dr Adrian White, a researcher into acupuncture at the Peninsula Medical School in south west England, said that “sham” acupuncture might well be having an effect rather than acting as a simple placebo version of the treatment.

“Sham acupuncture often consists of superficial, off point needling, but this may still have a physiological effect.

“Sham controlled studies are of little value in estimating benefits to patients.”
The British Acupuncture Council described the treatment as “a beneficial and effective therapy for a variety of conditions”.

Mike O’Farrell, the chief executive, said the latest study contradicted the majority of previous research in the area.

He said: “Acupuncture does work and research results consistently demonstrate the positive outcomes of treatment on patients.”

Ah, yes. Argument by assertion, no research results actually cited. Also, note the dismissal of sham-controlled studies. There’s a reason why acupuncturists don’t like sham-controlled studies. If you review them, you’ll see that the preponderance of evidence from such studies shows that (1) it doesn’t matter where acupuncturists stick the needles, and the effect is the same no matter where they are stuck; and (2) acupuncture produces no effects greater than placebo.

That’s because it is a placebo. It’s an elaborate placebo, and the relaxation and attention given patients is the only real “healing” going on. That, and the break that looking at this paper allowed me to take from subjecting myself to the burning stupidity I’ve been dealing with lately.


M. V. Madsen, P. C Gotzsche, A. Hrobjartsson (2009). Acupuncture treatment for pain: systematic review of randomised clinical trials with acupuncture, placebo acupuncture, and no acupuncture groups BMJ, 338 (jan27 2) DOI: 10.1136/bmj.a3115