Edzard Ernst et al.’s article “Acupuncture: Does it alleviate pain and are there serious risks? A review of reviews“ was published in the journal Pain in April 2011. It had two parts: (1) it was a systematic review of 57 systematic reviews showing that there was “little truly convincing evidence that acupuncture is effective in reducing pain,” and (2) it tabulated published reports of 5 deaths and 90 other serious complications of acupuncture treatments. I wrote an accompanying commentary, “Acupuncture’s claims punctured: Not proven effective for pain, not harmless.”
William Morris chastised me for not declaring a conflict of interest (!?) in my commentary. Now, in Acupuncture Today, he has criticized the Ernst et al. study itself.
Morris’ first criticism is that the study
did not examine iatrogenic deaths due to conventional drugs, chemotherapy, radiotherapy and surgery. It was more focused on the safety and efficacy of acupuncture.
I laughed out loud when I read this. It was focused on the safety and efficacy of acupuncture because it was a study about the safety and efficacy of acupuncture. Duh! A study of acupuncture for pain is not the appropriate place to examine iatrogenic deaths from four conventional treatment methods that are used for every disease from pneumonia to cancer. We need to know about acupuncture’s efficacy and safety (for pain) before we can even consider making any meaningful comparisons with other treatments (for pain). He says “it would be interesting to see results an [sic] in-depth ‘review of reviews’ on the safety and efficacy of conventional medicine and surgery.” Sure it would, but this is hardly the place. He tries to forestall protests by insisting this is not a tu quoque argument; but it sure sounds like it, or like something worse.
Comparing iatrogenic deaths from different treatments would be a valid subject for a different article. But why does he single out conventional medicine instead of including deaths from all forms of alternative medicine? What about the risk of stroke from neck manipulation? Neck manipulation would be a particularly pertinent comparison to acupuncture because it is used by many chiropractors to treat pain as an alternative to both conventional medicine and acupuncture. And why single out deaths instead of looking at all serious adverse events? Most importantly, why not look at risks in the context of proven benefits?
Curiously, Morris does not address the results of the systematic review of acupuncture’s efficacy or attempt to counter its conclusions in any way. He only briefly criticizes its methodology, and then he focuses on the reports of adverse effects.
The Systematic Review of Systematic Reviews
He criticizes Ernst et al. for lack of detail in describing how they quantified the quality of the studies they included in their analysis. But they supplied a footnote to an article describing their method, indicating that they followed Oxman and Guyatt’s validated checklist. Does Morris think they should have reproduced the entire checklist and evidence for its validation in the text of their article? I can’t fault them for using a footnote to save space.
“For data on efficacy, it doesn’t provide information about how the studies were controlled.” He wants to know whether it was “acupuncture alone, skin penetrating, non-penetrating or adjacent to the point location.” This only highlights the imprecision of acupuncturists’ own definitions of what qualifies as acupuncture and what constitutes appropriate controls. In my commentary I pointed out:
there are various schools of acupuncture with different acupoints, and studies of acupuncture have included “electroacupuncture” (with or without needles), ear acupuncture, cupping, moxibustion, and other loosely related procedures.
If he had any evidence that his preferred version of acupuncture was more effective or safer than other versions, or that one type of control is more appropriate than another, he could have presented it. He expresses doubts rather than offering data.
Morris criticizes Ernst et al. for including case reports where causality was uncertain, but each case report was listed individually and its causality categorized as certain or probable. The deaths were categorized as “certain” causality. Even if you eliminate the uncertain ones, the data show that acupuncture cannot be characterized as risk-free.
Bizarrely, he criticizes Ernst et al. for not including other complications such as fainting, vomiting, or bruising that are more common adverse effects. These are not “serious” risks, which is why they were not included in a study of “serious risks.” In essence, Morris seems to be arguing that acupuncture is even less safe than Ernst et al. depicted.
He says “Ernst and his colleagues do not reference previous studies showing acupuncture as safe,” citing a 2003 article by Lao and Berman published in an alternative medicine journal. That article’s conclusion included the authors’ opinion (a value judgment) that acupuncture is “a generally safe procedure”; but its text identified 202 adverse events, over twice as many as the Ernst et al. article. It included minor events like fainting and nausea, while the Ernst study was limited to serious events. I don’t have access to the full study. I wonder if it is exempt from the same criticisms: did its authors describe how they quantified the quality of the reports they included, did they assess the certainty of causation, and did they follow a validated checklist like Ernst et al. did?
Morris quotes a researcher who says “acupuncture is seen as an extremely safe therapeutic system whose complications are very rare and are easily avoided or rectified.” This is not at all in conflict with the findings of the Pain study. Ernst et al. found a number of complications that was quite small in relation to the large number of acupuncture treatments, and they called for better training of acupuncturists to minimize the chance of complications. The real point is that there is no reason to accept any degree of risk, no matter how little, if there is no benefit to the treatment. Safety by itself is no recommendation: homeopathy is probably the safest treatment going.
Conflict of Interest
Morris criticized me for not declaring a conflict of interest in my commentary. I have no conflict of interest: it makes no difference to me whether acupuncture works or not; I would follow the evidence wherever it led and write about it. He, on the other hand, has a clear interest in defending his occupation, and I don’t see any declaration of his conflict of interest in his Acupuncture Today article. The Ernst et al. study is solid: it shows that there is little evidence that acupuncture is truly effective in reducing pain. Taken as a whole, a rigorous evaluation of the published evidence leads to the conclusion that acupuncture is no more effective than placebo, and it is not risk free. Morris doesn’t want to accept this. He is biased in favor of acupuncture and is doing whatever he can to limit the damage from the Pain study. He can’t do much: his efforts are unconvincing and rather sad.
This article was originally published in the Science-Based Medicine Blog