Review paper highlights substandard COVID-19 research in both peer-review journals and pre-review websites
Simon Collins, HIV i-Base
A review of medical publications on COVID-19 has highlighted the number of substandard studies that have either been published in peer-reviewed journals or that are getting wider circulation ahead of peer-review on the medRXiv and bioRxiv websites (currently with more than 8000 papers), which have fewer quality checks.
The review, in the Journal of Medical Ethics, also notes the many registered COVID-19 studies – more than 1220 on clinicaltrial.gov by 7 May 2020 – and that these are largely and rightly driven by the urgency of finding effective options for treatment and prevention.
However, the paper is also concerned about the risk of honest errors as well as misconduct. By the end of July 2020, 33 papers (19 published and 14 preprints) have been retracted, withdrawn, or raised serious doubts, including two in the Lancet and New England Journal of Medicine. Most of these papers (19/33) came from Asia, with 11/19 coming from China. Reasons for the change included data falsification, methodological concerns, and concerns about interpretation of data and conclusions, as well as authorship and research participant privacy issues
The paper argues that the rush to publish results is straining the integrity of research and publications and that this has implications for patients, clinicians, and potentially government policy.
The caution for any articles published ahead of peer-review is important. Many of the older papers posted on MedRxiv and BioRxiv might also have outdated results and might never be published.
It is good that both sites update listings when papers are published following peer review.
It is also positive that many subscription journals are publishing COVID-19 research as open-access articles.
Bramstedt K et al. The carnage of substandard research during the COVID-19 pandemic: a call for quality. Journal of Medical Ethics. DOI:10.1136/medethics-2020-106494. (1 October 2020).