A recent article by Michael Mullins in The Scientist and an accompanying editorial by the publication’s editor-in-chief point to the dangers of disseminating un-peer-reviewed biomedical research in the form of preprints. Both contain factual inaccuracies and misunderstandings.
The Mullins article cites a medRxiv preprint by Didier Raoult and colleagues on hydroxychloroquine as an example of the danger. Hydroxychloroquine was widely used to treat COVID-19 early in the pandemic, in part out of desperation and in part because it was championed by influential figures such as then–US President Donald Trump. The article implies that this was due to the medRxiv preprint. In fact, Raoult made his findings public on YouTube and his institution’s website prior to posting them on the preprint server. Moreover, the manuscript was rapidly published in a peer-reviewed journal, the International Journal of Antimicrobial Agents, that very same day, March 20, 2020, as Mullins points out. It was this journal article, and not the preprint, that was specifically highlighted by President Trump in a tweet dated March 21, 2020.
There are two important lessons here. First, the universal availability of the internet and social networks mean that this type of information can be easily disseminated independently of preprints. Second, peer-reviewed journals may not effectively function as gatekeepers: Raoult’s paper was published after alleged peer review despite its flaws and, as of today, still has not been retracted. Preprints provide an opportunity for the scientific community to discuss new work, and indeed many researchers pointed out the flaws in the Raoult manuscript in medRxiv’s comment section and elsewhere. Additionally, the “more-sober analysis” Mullins refers to showing “HCQ has no proven role” was itself a preprint posted to medRxiv in July 2020.
Elsewhere, Mullins notes there are “questionable studies [of ivermectin] lacking peer review.” This is true, but it is also true that there are questionable positive ivermectin treatment claims in the peer-reviewed literature. The drug’s efficacy in treating COVID-19 remains controversial, and at this point, the major problem appears to us to be investigator fraud/error coupled with lobbying by interest groups and social media amplification rather than preprints.
Mullins proposes that problems associated with preprints could be addressed if “preprints no longer receive[d] a permanent DOI.” It is unclear what exactly he means. Is he suggesting the existence of a DOI in some way legitimizes the information as accepted science? This would be an odd suggestion given that even pornographic videos can receive DOIs. If the suggestion is instead that the DOIs should not be permanent, this is at odds with the essential aspect of DOIs that they are persistent identifiers. Mullins adds that preprints “should have a limited shelf life with a link that expires within 12 months.” This is inconsistent with the way the internet works (search engine traces persist, files can be downloaded, and material is archived elsewhere) and with the importance of preserving a transparent scientific record. Removing links to preprints would also be far more likely to prompt conspiracy theories and accusations of “cover-ups” if the content were controversial.
We and the other cofounders of medRxiv are experienced biomedical editors and thus well aware of the challenges presented by biomedical preprints. We recognize the need to balance their undoubted advantages (which have been particularly evident during the pandemic, when they have allowed researchers to quickly share information about promising research avenues and treatments) with the potential drawbacks. medRxiv papers go through extensive screening for dangerous material, and we have previously detailed the reasons for declining certain manuscripts out of an abundance of caution. Meanwhile, as the growth of preprints on bioRxiv and medRxiv demonstrates, the scientific community is becoming acclimatized to a new norm in which research is available for discussion and comment prior to formal review.
The Scientist editorial notes that, “[T]he proper framing of science is essential.” We agree. In this instance, Mullins’s opinion piece was not properly framed and contained inaccuracies and misunderstandings that perpetuate flawed narratives about preprints.