'We Didn't Want to Hurt Them. We Are Polite': When a Retraction Notice Pulls Punches

Retraction Watch Staff

May 14, 2021

A group of anesthesiology researchers in China have lost their 2020 paper on nerve blocks during lung surgery after finding that the work contained "too many" errors to stand. But after hearing from the top editor of the journal, it's pretty clear "too many errors" was a euphemism for even worse problems.

The article, "Opioid-sparing effect of modified intercostal nerve block during single-port thoracoscopic lobectomy: A randomised controlled trial," came from a team at Anhui Medical University. The senior author was Guang-hong Xu. The paper appeared online in early December in the European Journal of Anaesthesiology .

At which point it caught the attention of a reader in Australia, who emailed the journal to point out fishiness in the data.

Charles Marc Samama, the editor-in-chief of the EJA, told us:

I wasn't aware of this paper because it had been handled and accepted by the former EJA Editorial team last summer (my team took over the Ed Board mid-September). The article became available ahead of print on December 6th. The day after, we received an email from an Australian colleague with several concerns focussed on the tables, figures, inconsistencies, and doubts about data. For instance: demographics looked « too nice » to be true; mean age differing by one year in this small study; number of males differing by just one as well; conversion to open surgery same in both groups; etc, etc.
I checked the whole reviewing process (handling editor, reviewers and language editor) and I didn't find any signal of data fabrication or fraud. Neither the reviewers, nor the handling editor and the language editor had had any concern. However, in order to be completely reassured, we requested the raw data.
Upon analysis of the raw data, we found that some raw data were difficult to explain based on the described methodology. When we requested additional information to clarify our queries, the authors made subtle changes in methodology. For instance we identified that some investigators (such as the anesthetists doing the anesthesia during surgery) were not blinded to the intervention. Also, based on the adapted protocol, it was impossible to have no remifentanil consumption in one study group. The authors also modified crucial aspects of the protocol such as the infusion speed of the drug of which they were studying the consumption as primary outcome parameter.
Additionally, they had to admit three errors in their analysis of the consumed anesthetics including remifentanil, the drug which was their primary outcome parameter.
So to summarize: errors in reported outcome data were identified once we had analyzed the raw data. When we asked for clarification and explanation, the authors started to modify the alleged methodology. We therefore concluded the data could not be trusted and we retracted the article.

Given all that, the retraction notice — which, we note, the original paper does not link to — is rather generous:

Following a post-publication review and careful assessment, the Publish Ahead of Print version of the article, 'Opioid-sparing effect of modified intercostal nerve block during single-port thoracoscopic lobectomy: A randomised controlled trial', was found to contain too many errors to be corrected, therefore, we have decided to retract this manuscript.

Xu did not return a request for comment. We asked Samama why the soft-pedaling, since the issues certainly look deeper than merely a bunch of errors. He said:

The Chinese authors made a lot of efforts and were not dishonest. We didn't want to hurt them. We are polite.

We're sympathetic to that impulse, but we wonder how well it would have served the editors of other anesthesia journals — and indeed all of biomedicine — had they been so polite to people like Joachim Boldt and Yoshitaka Fujii when dealing with those cases.

Indeed, one of the earliest criticisms of Fujii, lobbed years before the extent of his misdeeds became clear, was that his data in one study were "incredibly nice." The editor who published those data, and the letter pointing out the improbability of the results, chose to look the other way … and Fujii went on publishing for the next decade.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....