Some articles about the coronavirus have been extremely widely disseminated. One preprint in Medrxiv about the stability of the virus on different surfaces has been downloaded more than 660 000 times, tweeted about by more than 13 000 persons, and mentioned by more than 320 news sources.

Chasing quick results

Svenska 2020-04-28

The need to quickly increase knowledge about the corona pandemic has led to a tidal wave of new research findings, which are published without first having been reviewed by other researchers. They are known as ‘preprints’, and make it possible to quickly share and partake of results, but they also increase the risk of errors being spread.

Traditionally, scientific journals play an important part in the research world by conducting peer reviews and disseminating research results. But this process takes time. Months can pass before a scientific manuscript submitted to a journal might possibly be published.

The corona pandemic underlines the need for fast scientific communication to increase knowledge of the virus, and how it can be fought. This is also the purpose of the ‘preprints’. These are reports from scientific studies that are made public before they have gone through peer review.

“The point of preprints is their speed. A preprint may be published within 24–48 hours, other researchers can read it straight away, and continue building on it. This is becoming even more important now, with the corona pandemic – when we really want things to move forward as quickly as possible,” says Richard Sever, Assistant Manager at Cold Spring Harbor Laboratory Press and co-founder of the preprint archives Biorxiv and Medrxiv.

More than 50 preprints per day

The phenomenon of preprints has been around for a long time, for example in physics. In biology and medicine, their use is not as wide-spread, but it is growing.

The phenomenon received a major boost in 2013, when the Biorxiv archive for preprints in biology and life sciences was started by the research institution Cold Spring Harbor Laboratory, CSHL. In 2019, the preprint archive Medrxiv was launched for results from clinical research by CSHL, Yale University and the publisher BMJ.

In January 2020, Biorxiv published its first preprint on the new coronavirus. Since then, numbers have risen rapidly, and as we write there are just over 1 900 articles about COVID-19 and the new coronavirus in Medrxiv and Biorxiv together.

“Right now, we are publishing more than 50 preprints daily on COVID-19 and SARS-CoV-2, and the number is growing,” says Richard Sever.

Important function, but risk of errors

Anna-Lena Spetz

“In principle, preprints are good and make sure that not everyone is working on the same thing,” says Anna-Lena Spetz, Professor of Immunology at Stockholm University, who is leading an EU-funded project focusing on treatment of the new coronavirus. But she can also see certain risks.

“Big, strong laboratories can pick up ideas from preprints, conduct their own studies and then publish the results in a prestigious, peer-reviewed journal before the team that uploaded the preprint can do so. For science this is good, but for a minor researcher there might be a risk in publishing.”

She also emphasises the risk that preprints can be fast but faulty, which could lead other researchers astray.

Matti Sällberg

Matti Sällberg also thinks that preprints fulfil an important function. He is Professor of Biomedical Analysis at Karolinska Institutet, and leads an EU-funded project focusing on vaccine against the coronavirus. His team reads many preprints and circulate them among themselves for discussion.

“They are extremely important; we find things out quickly through them. They are not peer reviewed, but those who know the subject can judge what is sensible and what is rubbish. Of course, there is some risk that they are not solid data, but then neither is everything that is published in journals.”

Extra check

Before articles are published in Biorxiv or Medrxiv, a quick check is carried out to ensure the contents are scientific, and that the material does not constitute a health risk. Ahead of publication in Medrxiv, several checks are carried out. Clinical studies must be registered, funding sources and possible conflicts of interest must be stated, and a certificate that ethical guidelines have been complied with and relevant permits obtained must be appended.

For texts relating to the new coronavirus, an extra check has now been introduced, Richard Sever tells us.

“We saw that many more people are now reading the articles, not just researchers. Therefore, we decided that all articles on the coronavirus shall be reviewed by a team of virologists, to check that they don’t contain anything that can be dangerous if it is untrue.”

The new check was introduced in conjunction with a publication in Biorxiv at the beginning of February. The article claimed that parts of the new coronavirus showed an “uncanny similarity” with HIV. This claim received scientific criticism straight away from other researchers, and was quickly retracted. But the result, and the interpretation that the virus was a manufactured, biological weapon, did have time to gain some spread.

After this, Biorxiv also introduced an extra warning text that preprints relating to the coronavirus constitute preliminary reports. They should not form the basis for any clinical activity or health-related behaviour, and must not be reported as established information in the news media. Before then, notices that preprints are not peer reviewed were already included in both Biorxiv and Medrxiv, and also in other preprint archives.

Advantages outweigh disadvantages

While most preprints do not attract much attention beyond their own scientific field, certain corona-related preprints are now disseminated extremely widely. On the website Rxivist, which traces traffic on Biorxiv, articles about the coronavirus have repeatedly headed the list of the week’s most downloaded preprints. Several are among the archive’s most ever downloaded.

One preprint in Medrxiv about the stability of the virus on different surfaces has been downloaded more than 660 000 times, tweeted about by more than 13 000 persons, and mentioned by more than 320 news sources.

When preprints are picked out by the media, the researcher behind the results has great responsibility, Matti Sällberg considers.

“Researchers must make it clear that these are unreviewed data that no independent expert has looked at, even if those who published the data believe in them. But the advantage of researchers being able to quickly partake of others’ results outweigh the risks of preprints, in my view.”

On several occasions during the corona outbreak, Swedish media have reported on research results published as preprints. Several of these articles contain information that the research has not passed through the normal review process, but not all.

Several journals accept preprints

Most scientific journals now accept that article manuscripts are published in preprint form before they are sent to the journal for possible publication. Some journals also collaborate with preprint archives; examples are The Lancet with SSRN/First Look from Elsevier, The BMJ with Medxriv, and Nature with In Review from Research Square.

During the corona pandemic, several medical journals have gone one step further than usual, and are encouraging researchers to publish their results as preprints. Many have also signed the British research funding body Wellcome Trust’s request to researchers, journals and funding bodies to share data and research results of importance to the outbreak quickly and openly. Among the signatories are Science, Nature, The Lancet, and New England Journal of Medicine, NEJM.

“The pandemic threatens public health, and is news that changes from hour to hour. Doctors and public health authorities need to get information as quickly as possible in order to act. Therefore, we encourage them to upload to archives for preprints as a way of sharing information straight away,” writes Jennifer Zeis, Communications Manager at NEJM in an email to Curie.

Rapid evaluation of manuscripts

In its policy for handling article manuscripts about the corona pandemic, NEJM writes that they will evaluate submitted manuscripts quickly, and make articles for publication accessible as fast as possible. According to Jennifer Zeis, this does not impact on the standard of what is published; instead it is made possible by speeding up all the steps.

“A process that can take months has in many cases been condensed to 48 hours, or less. We are grateful to our reviewers, who conduct their reviews quickly,” she writes.

Like Nature, The Lancet, Science and several other journals, NEJM has also made its material on COVID-19 and SARS-CoV-2 freely accessible.

Do not comply with the journal’s quality standard

Curie also contacted the journal Science, which has received more than 320 COVID-19-related articles and research reports since 1 January. Up until 17 April, only 14 of these had been published.

“Many of the COVID-19 manuscripts sent in have been produced quickly, and do not comply with the journal’s quality standard. By publishing the research with the strongest data, we help our readers to focus on the most important advances,” the editors of Science write in a comment to Curie.

Text: Sara Nilsson
Photo: Obi Onyeador