Dr. Adrian White was my coworker from about 1994 until 2005 in the Department of Complementary Medicine at the University of Exeter. He became a trusted colleague and eventually the deputy head of my department. When I discovered that my trust had been misplaced, I did not prolong his contract (for details, see my memoir [Ernst 2015]). Subsequently, White got a post as the senior research fellow at the department of general practice, where he continued his research into acupuncture for another ten years.

White had always been an acupuncturist at heart (in fact, I had never before met anyone so utterly convinced of the value of this therapy). When he joined my team, he was scientifically naive, and we spent many months trying to teach him scientific methodologies and how to think like a scientist. Initially, he found it very difficult to be critical about acupuncture. Later, I assumed the problem was under control. Yet most of his research in my department was guided and supervised by me. I tried to make sure that all our studies were testing rather than promoting so-called alternative medicine (SCAM; Ernst 2018) and that our reviews were critical assessments of the published evidence. After White left my unit, he was able to work mostly without supervision.

Thus, there exist two separate and well-documented periods of a pro-acupuncture researcher: ten years guided by me and members of my team and ten years largely unsupervised. The aim of this analysis is to compare these two periods.

Methods

I conducted a Medline search of White’s publications on acupuncture and selected those articles that reported factual new data. The endpoints for my analysis were the directions of the conclusions of White’s papers. I therefore extracted, dated, and rated the conclusions of the included articles as follows:

• P = positive from the point of view of an acupuncture advocate.
• N = negative from the point of view of an acupuncture advocate.
• P/N = not clearly pointing in either direction.

To render this exercise transparent (occasionally, I was not entirely sure about my ratings), I copied all the conclusions and provided links to the original papers so that anyone interested would be able to check my judgements.

Results

The Medline search identified 120 papers authored by Adrian White on acupuncture. Of those 120 articles, fifty-two fulfilled the inclusion criterion of reporting factual new data. They were mostly clinical trials and systematic reviews. These publications were distributed almost equally within the two ten-year periods. Articles 1–27 were published after White had left my department; articles 28–52 were from the time while he worked with me.

Considering that one would not expect unbiased studies or reviews of acupuncture to generate mostly positive conclusions, the findings are remarkable.

After White had left my department:

• 0 times N
• 5 times N/P
• 22 times P

While White was working in my department:

• 17 times N
• 7 times N/P
•  0 times P

From these figures, it is easy calculate the ratios for both periods of negative : positive conclusions; they are zero versus infinite.

Figure 1 depicts the positive and negative conclusions of all of White’s articles over time. A sample of six of White’s conclusions appears in Table 1. A full list of the fifty-two conclusions can be found on my blog at https://edzardernst.com/2021/01/a-demonstration-of-a-powerful-non-financial-conflict-of-interest-in-scam-research/.

 

Discussion

White’s case might be unique, because it allows us an observation of bias over two decades. It demonstrates that an ardent believer in acupuncture managed under close supervision to learn the scientific methodology. This enabled him for over a decade to publish numerous papers on his subject that, despite his belief, were based on the evidence and thus questioned the value of acupuncture. Once he had left my department, White was allowed to work unsupervised. He seems to have then abandoned critical thinking, and he published articles with predominantly positive conclusions.

This striking contrast requires an explanation. It might mean that while under my supervision White published reliable evidence and when working unchecked during the second period, his pro-acupuncture bias got the better of him. In this first explanation, his “ideological” bias was so strong and durable that it re-emerged as soon as it was left unchecked.

However, there is another possibility. According to a second explanation, White did not dare tell the truth while working for me and, only after being liberated, was he able to publish unbiased conclusions.

To decide which explanation is correct, we need to determine whether acupuncture is generally effective or not. Those who believe it is would probably favor the second explanation. Those who believe it isn’t would prefer the first. Attempting to decide who is correct presents a dilemma: there are plenty of articles on acupuncture that suggest it is effective for a wide range of conditions (Wang et al. 2013); similarly, there are numerous papers that suggest the opposite (Colquhoun and Novella 2013). The former type of publication is usually written by researchers such as White who might be criticized (by skeptics) for their pro-acupuncture bias. The latter type of article is often authored by skeptics who might be criticized (by acupuncturists) for not fully understanding the subject matter.

The solution could be to rely on reviews conducted by the Cochrane Collaboration. These are carefully designed to minimize bias, usually include acupuncturists among the authors, and are generally regarded as the most trustworthy evidence available. In June 2020, there were fifty-two Cochrane reviews of acupuncture. Of those, only two reached a positive conclusion based on more than one high-quality primary study (Ernst 2020).

It is thus likely that the first explanation applies, and White’s conclusions in the second part of his career did indeed lack critical input and were thus false-positive. There is further evidence for this assumption: in several articles (for instance, 5, 20, and 27 from the full list), White seems to spin an essentially negative finding into a positive conclusion. Moreover, he conducted a negative clinical trial of acupuncture for smoking cessation (49) while working with me and later published a positive review on the subject, seemingly distrusting his own study because it had not generated the result he had hoped for. Lastly, the second explanation would surmise that I have a strong anti-acupuncture bias. This is not the case and can easily be demonstrated by the existence of my systematic reviews that supported acupuncture when the totality of the best evidence was positive (Manheimer et al. 2005; Editors 2005; Ernst and Pittler 1998).

Of course, this analysis is merely a case study and therefore its findings are not generalizable. However, in my personal experience from twenty-five years of SCAM research, the described phenomenon is by no means an exception. On my blog (edzardernst.com), I regularly publish a growing list of SCAM researchers who managed the extraordinary feat of never publishing any negative conclusions about their specific SCAM; I’ve christened it “The Alternative Medicine Hall of Fame.” As of January 2021, the list contains just seventeen names but is expected to grow steadily. “Ideological” conflicts of interest seem to have a more powerful influence in SCAM research than financial ones.

As the influence of ideological bias can be so dominant in SCAM, it is important to know about it. Consumers need to be aware when SCAM research originates from unreliable sources, and funders of SCAM research would be wise to stop supporting researchers who are prone to generating findings that are false-positive.

In the final analysis, research by hopelessly biased SCAM researchers is worse than no research at all.

References

Colquhoun, D., and S.P. Novella. 2013. Acupuncture is theatrical placebo. Anesthesia & Analgesia 116(6): 1360–63. DOI: 10.1213/ANE.0b013e31828f2d5e. PMID: 23709076.

Editors. 2005. Correction: Meta-analysis: Acupuncture for low back pain (erratum). Annals of Internal Medicine 142(11): 950–51. DOI: 10.7326/0003-4819-142-11-200506070-00021. PMID: 15838072.

Ernst, Edzard. 2015. A Scientist in Wonderland: A Memoir of Searching for Truth and Finding Trouble. Exeter, UK: Imprint Academic.

———. 2018. SCAM: So-Called Alternative Medicine. Exeter, UK: Societas.

———. 2020. Acupuncture: An update of the most reliable evidence—part 3. Edzardernst.com (June 5). Available online at https://edzardernst.com/2020/06/acupuncture-an-update-of-the-most-reliable-evidence-part-3/.

Ernst, E., and M.H. Pittler. 1998. The effectiveness of acupuncture in treating acute dental pain: A systematic review. British Dental Journal 184(9): 443–47. DOI: 10.1038/sj.bdj.4809654. PMID: 9617000.

Manheimer, E., A. White, B. Berman, et al. 2005. Meta-analysis: Acupuncture for low back pain. Annals of Internal Medicine 142(8): 651–63. DOI: 10.7326/0003-4819-142-8-200504190-00014.

Wang, S.M., R.E. Harris, Y.C. Lin, et al. 2013. Acupuncture in 21st century anesthesia: Is there a needle in the haystack? Anesthesia & Analgesia 116(6): 1356–59. DOI: 10.1213/ANE.0b013e31828f5efa. PMID: 23709075.

 

Table 1.   A Sample of Six Conclusions from Adrian White’s Articles

[P = positive from the point of view of an acupuncture advocate; N = negative from the point of view of an acupuncture advocate]

1. The available evidence suggests that adding acupuncture to symptomatic treatment of attacks reduces the frequency of headaches. Contrary to the previous findings, the updated evidence also suggests that there is an effect over sham, but this effect is small. The available trials also suggest that acupuncture may be at least similarly effective as treatment with prophylactic drugs. Acupuncture can be considered a treatment option for patients willing to undergo this treatment. As for other migraine treatments, long-term studies, more than one year in duration, are lacking. (2016) P

2. 70% of those patients eligible to participate volunteered to do so; all participants had clinically identified MTrPs; a 100% completion rate was achieved for recorded self-assessment data; no serious adverse events were reported as a result of either intervention; and the end of treatment attrition rate was 17%. A phase III study is both feasible and clinically relevant. This study is currently being planned. (2010) P

3. Acupuncture appears to offer symptomatic improvement to some patients with fibromyalgia in a tertiary clinic who have failed to respond to other treatments. In view of its safety, further acupuncture research is justified in this population. (2007) P

4. Acupuncture is not superior to sham treatment for recovery in activities of daily living and health-related quality of life after stroke, although there may be a limited effect on leg function in more severely affected patients.  (2005) N

5. Acupuncture has not been demonstrated to be efficacious as a treatment for tinnitus on the evidence of rigorous randomized controlled trials. (2000) N 6. It is therefore concluded that, according to the data published to date, the evidence that acupuncture is a useful adjunct for stroke rehabilitation is encouraging but not compelling. More and better trials are required to clarify this highly relevant issue. (1996) N