Prêmio CBMM
Imprimir Republish

Good practices

Fast error correction

Scientific papers on the novel coronavirus retracted due to misunderstandings and methodological flaws, most committed in good faith

Edited version of photo by Angela Deane-Drummond / Evening Standard / Hulton Archive / Getty Images

The Retraction Watch website, which runs a database of more than 20,000 scientific articles removed due to errors or misconduct, started tracking papers on COVID-19 that have been retracted after their results were challenged. The number of invalidated studies is small and almost all involve mistakes made in good faith, a result of over-eagerness to share results that could help in the fight against the epidemic. Of the more than 10,000 papers published on the disease by mid-May, only seven (0.07% of the total) contained errors or methodological issues that compromised their conclusions, leading to their retraction.

The list of retracted articles includes a manuscript published on the preprints repository medRxiv by Chinese and American researchers who examined the epidemiological and clinical characteristics of the novel coronavirus outbreak in China. The research was based on cases recorded in the country prior to January 26. Since the epidemic exploded in February, the authors requested withdrawal of the paper so that they could conduct a new analysis. This had repercussions on other studies that used their findings, one of which was a report published in the journal Lancet Infectious Diseases (Lancet ID) by Imperial College London, which projected the number of cases and deaths in a range of scenarios and helped many countries to devise strategies to combat the pandemic. The Imperial College group asked that the Lancet ID article be corrected, with the removal of data from the Chinese study, but noted that this did not change the forecasts.

One of the most talked-about cases involved a paper shared on bioRxiv in late January by researchers from the Kusuma School of Biological Sciences in New Delhi, India. The title itself stated that there appears to be an “uncanny similarity” between the novel coronavirus and HIV—the virus that causes AIDS. The manuscript only remained available for two days, but that was long enough to spark a number of conspiracy theories on social media. There was so much criticism of its methodology that the authors quickly removed the paper from the repository.

Reprint repositories were rebuked for instantly sharing scientific results, leaving the task of evaluating data consistency to readers. Retraction Watch editors Ivan Oransky and Adam Marcus believe the criticism of preprints is overblown, noting that even scientific journals with strict and selective peer-review systems sometimes have to retract problematic articles. They point out that one of the benefits of preprint repositories is that they quickly identify problems in manuscripts and correct them. “When scientific journals publish bad or erroneous research, it can take months or years for the papers to be corrected or retracted—if they ever are,” wrote the pair in an opinion piece on the Stat News website.

A group of French researchers requested the retraction of an article published in the Bulletin de la Dialyse à Domicile on April 13 that warned of the danger of contamination through contact with the peritoneal dialysis fluid of patients with COVID-19. The paper described a case study of a dialysis patient with symptoms of the novel coronavirus whose fluid tested positive for the virus. The findings were found to be entirely false because the patient did not have COVID-19. The eight authors attributed the blunder to an innocent mistake rather than negligence.

In one retracted article, however, there were signs of misconduct. On February 26, the Lancet Global Health announced the retraction of a letter describing the experiences of nurses fighting the novel coronavirus in Wuhan, China. “In addition to the physical exhaustion, we are also suffering psychologically,” wrote the two authors: Yingchun Zeng, a nurse, and Yan Zhen, a specialist in traditional Chinese medicine. After publishing the letter, the journal was notified that the authors were not part of the task force mobilized to care for the sick in Wuhan. The pair later admitted that it was not a first-hand account.

The most unusual example is an article published in the Chinese Journal of Epidemiology on March 5 and retracted two days later. The paper addressed the existence of potential false positives when testing for COVID-19 in individuals who had close contact with the disease in China. The journal’s editors did not explain the reasons for the retraction and it was not possible to analyze the original article, which disappeared from the internet. According to a report published in the journal Nature on April 15, retractions of articles about COVID-19 in China may be linked to government attempts to control the dissemination of research results on the disease. University representatives have reported that studies on the origin of the SARS-CoV-2 virus can only be published with permission from the ministries of education and science and technology. The Chinese journal Practical Preventive Medicine retracted a paper published in early March that suggested the SARS-CoV-2 virus could spread as far as 5 meters in aerosol particles—twice the distance generally considered safe by health officials. Again, there was no explanation of what was wrong.

The most recent removal was a preprint uploaded on May 11 by researchers at Raymond Poincaré University Hospital in Garches, France, which discussed the potential of hydroxychloroquine and azithromycin therapy in reducing the number of people admitted to hospital with pneumonia caused by COVID-19. The paper was not conclusive—its title ended with a question mark—but the authors opted to remove it on May 20, citing the controversies about these drugs, which were found to be inefficient in other studies, and the need to review the data. The preprint had been praised on Twitter by French physician Didier Raoult, from the IHU Méditerranée Infection hospital and Aix-Marseille University, author of the first article to highlight the efficiency of the treatment, which sparked global interest. Raoult’s work is now being investigated by Elsevier on suspicion of misconduct. There are indications that the study, published in the International Journal of Antimicrobial Agents on March 20, began before being assessed by an ethics committee, and that its peer review was rushed. The paper was approved for publication just one day after being presented to the journal.