One in five Covid research papers corrected

One in five pieces of COVID-19 research published in the first five months of the year had to be corrected in the top medical journals in which they appeared.

As the Covid pandemic took hold, the need to share data rapidly became imperative but quality was seemingly compromised.

One in five pieces of COVID-19 research published in the first five months of the year had to be corrected in the top medical journals in which they appeared, almost three times the usual number.

The median time between submission and publication dropped from 139 to 23 days and was likely the cause of the high rate of corrections, according to University of Queensland researcher Michael Reade, the lead author of the new study published on Monday in the Medical Journal of Australia.

As the pandemic took hold, the need to share data rapidly became imperative but quality was compromised, Professor Reade said.

“We, like many people, had been reading the clinical and scientific literature over the first few months of this year, thinking it was informative, but perhaps not quite of the same sort of quality that we were used to reading in the New England Journal of Medicine, the Journal of the American Medical Association and so on,” Professor Reade, the Australian Defence Force Chair of Military Medicine and Surgery, said.

“And we had been reading some editorials in those journals and others somewhat bemoaning the quality of what had been published to date about COVID-19, but without really any evidence to back up those criticisms.”

Professor Reade’s team compared a random sample of publications in the NEJM, JAMA, The Lancet, the British Medical Journal and Annals of Internal Medicine between January 1 and May 31 in 2019 against COVID-19 research in the same journals for the same period in 2020 to determine what evidence existed for the growing disquiet about the quality of scientific papers dealing with the pandemic.

They found, as expected, that there were fewer randomised trials reported: 5.2 per cent in the COVID-19 period compared to 35.2 per cent in the same period last year. “That’s not a criticism, there wouldn’t have been the opportunity to set up and run those studies,” Professor Reade said.

However, the findings about the much shorter time between submission and publication and the data on corrections — rising from 7.4 per cent of papers in the 2019 period to 20.9 per cent for the 2020 period — were concerning. “So one in five of the COVID-19 papers published a correction, most typically within a few days of being published,” Professor Reade said. “On top of that, none in the 2019 random sample were retracted, whereas three of COVID-19 papers were retracted. Certainly, some of those papers were very high-profile and influential.”

The best known was an investigation into the safety and effectiveness of the drug hydroxy-chloroquine, published in The Lancet in May and retracted in June. “This resulted in the temporary cessation of the World Health Organisation’s trial. This observational study, in particular, was published, it would seem, on evidence that was not sound. That this halted recruitment in these important trials was really very problematic.”

Professor Reade said publications since May 31 were of quite high quality “but … there’s almost certainly going to be another public health event of significance that the journals are going to have to respond to”. He and his co-authors endorsed recommendations including a two-track review process for pandemic and non-pandemic research and mentored peer reviewing by research trainees.

“In the same way a public health authority has a plan for augmenting its contact-tracing teams and all the things that go into pandemic preparedness, a journal should have a plan to augment its review teams when they have a deluge of papers on a particular subject and … to develop some subject-specific expertise on that particular topic,” he said.