I’m a preprint fan. All the research papers from my final period in academia were posted as preprints before being submitted to journals, and I’ve also utilised preprint peer review services at Research Commons and eLife.

But in my present work as a medial writer, where I’m involved in the publication and dissemination of industry-funded biomedical research, I am slowly realising that many of the things that make preprints so fantastic in the biological sciences simply don’t apply in biomedical science, or not to the same extent.

The PRC model of scientific publication

In recent years, the biological sciences have been progressively moving towards a publish-review-curate (PRC) publication model (see here and here), driven in large part by dissatisfaction with the traditional model of scientific publication. PRC reverses the traditional publication model as follows:

Publish – When an author group feels that a research project is ready for publication, they write it up and post it online as a preprint, ensuring immediate release to the scientific community and the public domain. The decision to publish the research is therefore transferred from journal editors to authors, and the data get released without prior peer review. Preprints carry doi identifiers and can be cited in the same way as papers published in a journal.

Review – With the preprint in the public domain, all of the scientific community (not just the small number of reviewers selected by a journal editor) is able to review the content and provide feedback. To facilitate evaluation of preprints, there are also increasing options for preprint review services such as Review Commons, eLife, PREreview who can commission peer reviewers in a similar way to traditional journal editors. After receipt of feedback or peer reviews, the authors can update their preprint, revising it in a similar way to a paper undergoing review at a journal. The revised, peer-reviewed preprint will again be posted online and linked to the original preprint.

Curate – While the truly avant garde may choose to leave the paper as a living document online, most authors still seek publication in a journal as a final seal of professional approval. In a PRC context, the authors ideally submit their reviewed, revised preprint and the journal is left with a simple yes/no decision (this is exactly the system promoted by Review Commons and its partner journals). Crucially, this also means that the unbelievably tedious job of imposing journal-specific formatting can be left until after acceptance of a manuscript, rather than before submission (a massive, massive efficiency gain).

There’s little question that the growing popularity of the PRC model has had a seismic impact on the biological sciences and publicly-funded research more generally. Preprints are now mainstream, editors are now routinely scouring preprint servers to find papers they want to acquire for their journals, and authors are firmly in the driving seat.

So why doesn’t this seem as good a fit for industry-funded biomedical research, and why is it that I’m not (yet) recommending it to my industry clients? Let’s take a look at the main things preprints offer…

Preprints accelerate research speed

Biological sciences: a good thing. Giving authors control (and responsibility) for when their data is published and thereby released to the wider world makes sense when you’re familiar with the litany of horror stories about papers being gummed up in glacially slow (and occasionally malicious) journal review cycles, with the risk of losing novelty or getting scooped continuously rising.

It also ensures that data reach the public domain when they’re still relatively fresh; all too often, by the time a project is published in a peer-reviewed journal the data are so “mature” that the authors have long ago moved on (cognitively, emotionally, and practically) to the next thing.

There’s been some outstanding meta-analyses demonstrating that preprints usually differ little from their final published versions, so the inherent risk of releasing work that hasn’t been formally peer-reviewed is more than offset by the gain in momentum (see here).

Biomedical sciences: not necessarily a good thing. If there is one word that defines biomedical science, and especially industry-funded biomedical science, it is “caution”. That same caution can be pejoratively misinterpreted as conservatism, but at heart it reflects a fundamental concern for a demographic that biological research generally doesn’t need to worry about: patients.

The risk associated with publishing irreproducible results is far, far higher for industry-funded biomedical research than it is for publicly-funded biological research. It’s worth highlighting too that it was industry, not academia, who first drew attention to the reproducibility crisis in biomedical research (see here), something that was at least partly driven by academic scientists’ desire to publish sensational results before their peers. In my experience, biomedical science has not been infected by hype in the same way as the biological sciences, and it’s all the better for it.

The position of the finishing line is also different for industry. In basic research, publishing first is the goal, but in industry it’s regulatory approval and subsequent market access that’s the Holy Grail. When pharmaceutical companies talk about speeding things up, they’re referring to the 10+ years it requires to get a candidate drug to the patients that can benefit from it. Against that timescale, the pace of scholarly publication suddenly doesn’t seem quite so bad. In a similar way, patent protection looks after intellectual property in an industry setting so the timing of a publication doesn’t impact the value of the work in the same way.

One exception: the COVID19 pandemic. Preprints came of age during the pandemic, and this was one instance where there the need to get data into wider circulation as quickly as possibly outweighed the risks outlined above (don’t forget that this was also how hydroxychloroquine got promoted though).

Preprints redress gatekeeping and establish priority

Biological sciences: a good thing. Much of the case for preprints rests on the historical, much maligned, and frequently misunderstood role of journals and journal editors as gatekeepers. Because journals commissioned peer review and organised the process leading up to acceptance and publication, they controlled when a paper finally entered the public domain. Couple this to the outsized role that publications play on an individual scientist’s career progression, and you end up with a system where journals and journal editors (rather than evaluation panels) are making and breaking scientists’ reputations. Not surprisingly, there is a litany of deeply depressing, sometimes unsavoury, and occasionally unethical anecdotes resulting from this setup.
Preprints, by comparison, are a breath of fresh air – they allow author groups to immediately establish priority for their findings and give the entire scientific community the opportunity to collectively evaluate the work.

Biomedical science: not a good thing. One aspect of peer review that gets overlooked is that it’s not just about QC and scientific evaluation, it’s also a means of peer certification, and therefore also a way of spreading responsibility. We all know peer review is inefficient. We all know of instances where there have been awful and really flawed papers published in journals after peer review, but what’s interesting is that in such cases we invariably ask “How could this have gotten through peer review?” and not “How could the authors have put their names to this?”. It’s implicitly understood that authors are capable of being blinded by their hypotheses/biases/expectations but we hold peer reviewers to a higher and more dispassionate standard.

When scientists talk about releasing their work into the public domain, they generally don’t mean that they want or expect the lay public to read their work – they mean that it’s been released to the scientific commons, the wider scientific community. And when they talk about it being reviewed by the community they’re not even looking for reviews from just anybody – they want feedback from the experts in their field.

In theory, anybody can review preprints but in practice, few researchers have the time to voluntarily review others’ work for fun. I have never, ever received voluntary peer review or even comments on any of the preprints I’ve posted (though this could reflect the extremely esoteric nature of my former research in academia!). It illustrates how a key role of journals is commissioning peer review, although Review Commons and similar organisations can now perform the same function for preprints.

None of these points applies in the same way for preprinted biomedical research. There’s a much higher chance that the lay public will read the work, due to the presence of many active and well-informed patient communities who are following developments because they are interested in future treatments. Peer review and its gatekeeping function is a critical way of ensuring that material reaching the public domain becomes the responsibility not just of the authors but also the reviewers and the journals. Open Access publishing means that work will be publicly and freely available after peer review and perhaps it’s best, and safest, to wait until then.

Preprints help the careers of young scientists

Biological science: 100% a good thing. As already noted, when research results only reach the wider community after publication in a peer-reviewed journal, this strongly affects young scientists’ careers. Publications are what bring postdoctoral fellowships, grants, junior faculty positions, and tenure. Preprints and the PRC model allow faster recognition for young scientists’ work by getting those publications into the public domain at an accelerated tempo and letting them start to accumulate citations. Young scientists can go into job interviews with their research online, visible, citable, and accessible to potential employers.

Biomedical science: less of an issue. Clinicians are not as dependent as biologists on research output for career progression; this applies even more so for pharmaceutical company employees. In medical science, the most important people in a paper are not the authors, but the patients who might be affected by the results being described.

Preprints improve accessibility

Biological sciences: a good thing. While most people are in favour of Open Access publishing, article processing charges can be punishingly high. This makes the cheaper option of paywalled publication more attractive for those without deep pockets. Preprints close this loop by being Open Access, free to view, and not paywalled.
(Industry-funded)

Biomedical research: less of an issue. Biotech/pharma companies have deep pockets and invariably publish Open Access to demonstrate transparency.

In conclusion

In summary then, while I’m very happy to have preprinted my academic research, I’m feeling more and more circumspect about promoting the preprint option in MedComms. Have I convinced myself by writing this? Partly, although I’m still uncomfortable sticking with the “traditional” system of publishing where the slower pace is not just a reflection of caution but also incurs multiple efficiency losses (journal-specific formatting, multiple rounds of peer review).

Clearly the current PRC model isn’t the best fit for publication of biomedical research data…but that doesn’t mean that biomedical research should stick with the more conservative, old-fashioned approach to publishing, and there has to be space for a third way.

Maybe “dark preprints” (i.e. preprints that are online and accessible to the research community but not publicly available until after they’ve undergone peer review) would be an option? This could also be achieved at existing preprint servers if authors could select an option to keep preprints from being published until they can appear alongside peer reviews, or even (by no means mutually exclusive) delaying publication until the preprint has been revised after peer review. Implementation might be challenging (how do you restrict access to researchers only?), but I’d go for that, and I imagine I wouldn’t be alone.

Image generated by MidJourney.