The Corona pandemic brings a lot of challenges with it and will change our lives forever. One very specific thing changing is how we view science and science communication. The way science is published has changed. It is faster and more open – which is great, but may lead to miscommunication and at its worst: exploration by bad journalists.
Germany is lucky in having Prof. Drosten as well-established virologist not only working on Coronavirus, but also communicating about the virus and disease (in an understandable way) in a podcast. While many are grateful to him, it came with a downside: a hype around Prof Drosten, which led to a nasty dispute with the Bild newspaper and death threats.
At the core of the problem is a study, in which Prof. Drostens team compares the viral load (number of virus particles in the throat of a person) of children and adults. A loaded topic. Not much data is available on children yet – so any results are deemed of large consequence for the decision to re-open schools and daycare. In the study, Drosten and his team claim “that viral loads in the very young do not differ significantly from those of adults.” and that “Children may be as infectious as adults.”. They further issue a (mild) warning against reopening child-care. Soon after publication, several statisticians (1,2,3) reviewed the paper, critiqued the statistical methods used and in part reanalysed the data. Given their results they suggested different analysis approaches to improve the study. Some media outlets have taking this public scientific exchange to suggest Drosten is a liar and his study as wrong. *SPOILER: not true, never believe what Bild says*
The thing is, this is common practise in science. But normally it happens behind closed doors and long before the study results are shared and published.
Peer-review and preprints
Standard procedure, although currently changing, is that once a scientist has made a discovery and finished experiments to prove a theory or hypothesis, he writes a paper. Just like Drosten did. They send their data and interpretation of it to a journal, which sends the manuscript to other experts on the given topic to review: to judge if the manuscript proofs its hypothesis, to make suggestions for improvement and to advice on whether to publish the manuscript as is or modified or not at all. This process is called peer-review. While Drosten surely sent his manuscript out and is awaiting peer-review, he already deposited a so called preprint of his results online and shared the results on social media.
Publishing preprint manuscripts online is becoming more and more common and has the advantage of making data widely accessible quickly. Due to the peer-review process and publication times, journal publishing is slower than posting a preprint online. (Side-note: normally preprints are not simply posted on the institutes website, but deposited on so called preprint servers, which give the science community a chance to comment on the manuscripts directly on the website). Reading preprint manuscripts is free, making them widely accessible.
During the pandemic, when research into the Coronavirus is quickly evolving, this is great, as new results, which help understand the biology of the virus or identify therapy or vaccination targets, can be quickly shared and speed up finding a cure. But while scientists know to read preprint data carefully and critically (basically a hive mind of scientist reviewing each paper), the public and journalists, digesting the info for the public, still needs training. It appears that some people sometimes forget what a preprint entitles. They assume the data published is set in stone. Like in case of the Drosten study, where the preprint has been taken as final statement and when it was reviewed and critiqued, Bild suggested this criticism is a criticism of Drosten himself and not a way to improve his work. They even implied he lied and misled on purpose. But this is by no means the case. Drosten has accepted the suggestion, as is normal in peer-review, and stated that a new improved analysis will be published soon. *Update: the new manuscript is available now.
It is important to keep in mind: this is how science works.
It is a collaboratory affair. Sometimes there are inconsistency in a study design, additional experiments or better methods that can be applied to analyse the data. That is why other expert evaluate their colleagues work. Traditionally, this happens before the results are widely shared. But in these times, when research on Corona is highly valuable for the public, the publishing process is not only speed up, but also it is reported on more and criticism is more public as well. This does not mean the underlining study is flawed or intrinsically wrong. It means it can be improved by the hive mind of experts. A good thing – not a bad!
The role of preprints during this pandemic
Preprints play a big role in science reporting during this pandemic and maybe for all science in the future. A staggering amount of papers on the coronavirus is first deposited on preprint servers: 4464 preprints could be found on medRxiv and bioRxiv when this blog went online. In comparison, searching “immunotherapy” – a much older field investigating the use of the immune system in curing cancer and other diseases- gives 2,589 Results on bioRxiv and 191 on medRxiv. In addition, preprints concerning the Coronavirus are accessed and distributed 15 times more than preprints on other topics (data from a preprint by Fraser et al.). While it is not clear if the trend of more results first available as preprint and the higher useage will continue or carry over into other fields, the article indicates that journalist have discovered preprints as source. The articles are also shared more often on social media (Twitter). Increased usage or distribution of preprints in the media, if not reported carefully, can be misleading if the preprint does not make it through peer-review. Rules on how to report on preprints should therefore be made and people sensitized to preprint and their implications. It is key to understand that these data are not reviewed by the scientific community and that their conclusions may change with time. Alongside the preprint, it may be of value to look at and also report on responses to the preprint.
Preprint servers bioRvix and medRvix have now added a disclaimer on their sites, we need to see more often:
Caution: Preprints are preliminary reports of work that have not been certified by peer review. They should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information.
*Update: the new manuscript published by Prof Drostens team also includes a disclaimer – making it Kristall clear what a pre-print is:
Preprint: This document is a snapshot of research work in progress. It is released to provide an impression of viral loads based on diagnostic testing. It reports new medical research that has yet to be evaluated. As with other preprints, it should not be used to guide clinical practice.*
Picture: MedicalGraphics.de – Lizenz: CC BY-ND 4.0 DE