Open publishing can improve access to knowledge

There have been increasing demands for more transparency in the reporting of research findings, which has built up pressure on all those involved in making and testing potential new drugs, biologics and medical devices. The lack of systematic reporting of both positive and negative results and the resulting ‘publication bias’ is not only pertinent to research carried out by industry: it is also being hotly debated throughout academia, because our success in discovering new medicines is reliant on a more complete understanding of the underpinning science. 

Publication bias has many causes. For example, most researchers and particularly commercial organisations are not keen to share negative and null findings because of the perceived impact on their reputation or the commercial value of their product. Many journals are also not keen to publish such findings because they are thought to be less newsworthy and may negatively affect their Impact Factor. In addition, there are inherent biases in the traditional anonymous prepublication peer review system. Here I argue for the importance and benefits of moving to an open science publishing approach, and detail what this means in practice.

Problems with the traditional publishing process
One of the biggest problems with traditional scientific publishing is the time it often takes from when new research is ready to be shared to when others actually get to read and benefit from it – usually several months, and many researchers have examples of papers that took years to publish. In a digital age where most ideas and information can be shared immediately at the touch of a button, such delays are unnecessary, and worse, bad for scientific discovery and ultimately for the development of new medicines.

The delays are mostly caused by the peer review process: firstly, because it is a closed process that takes place before publication; and secondly, because the majority of journals still ask referees and editors to assess the value of a study based on an article’s perceived importance. 

The current prevailing process wastes the time of busy referees as each paper is assessed repeatedly by many experts while it is submitted from one journal to the next until it is published (with valid criticism often getting lost if it’s not addressed before submission to the next journal). This system also wastes the authors’ time in restructuring and resubmitting the article to multiple journals, each with their own specific article guidelines and policies.

Referees, by their very nature, are experts in the field in question and therefore likely to be working in the same area as the authors. The highly competitive academic and industry environment creates a strong incentive to be overly negative about a competitor’s study or to hold up publication of their work. The anonymous nature of peer review in most biology and medical journals means that there are no obvious consequences for referees who block a publication. Editors also have their own biases^1,2, and have an interest in favouring apparently impactful research that will boost their journals’ profile, influencing their role as final decision maker of what gets accepted for publication.  

It is clear that peer review is important for improving the paper and in providing a check on the science of newly published research; however, it often fails to achieve these crucial goals. There have long been calls to fix the traditional anonymous pre-publication peer review process but only recently have there been major initiatives experimenting with new forms of peer review.  

Recent developments in publishing
The BioMed Central (BMC) series was launched in 2000 with the specific aim of publishing all sound science, uncoupling the question of importance and impact from the basic validity of the findings. PLOS ONE then became the first megajournal across life sciences with this ‘all sound science’ ethos. There has also been a move towards open peer review, especially in medical publishing. For example, BMJ Open, the medical BMC-series journals, and the Frontiers series publish the complete refereeing history of articles they accept, including the names of the peer reviewers. 

All these models represent a step in the right direction but still lack transparency with regards to papers that were rejected because of the potential bias of the referees and editors making decisions on those manuscripts. Some critics have called for double-blind peer review, which brings some advantages but even if referees cannot guess who the authors are, the reasons for bias are still present and the process is still hidden from view.

F1000 Research was launched in January 2013 to address these fundamental outstanding problems with the traditional scientific publishing process. The publication delay is removed by publishing articles right away once they pass a rapid in-house pre-publication step to check that the work represents proper science, is readable, and meets all necessary ethical requirements: it takes on average under a week from submitted manuscript to full citable publication.

Articles then undergo a completely transparent refereeing process, using formally invited experts whose names, affiliations and full reports are published alongside the article, together with all author responses. Authors are invited to revise their articles as many times as they wish by publishing a new version, but they are under no obligation to do so. Once an article passes this peer review process, it is sent to the major bibliographic databases such as PubMed for indexing.

Lack of transparency and publication bias
In recent years, there have been increasingly strong calls for data transparency and the publication of all results regardless of outcome. Initiatives such as AllTrials, which calls for the publication of the results of all trials, have certainly raised awareness, and been signed by many major publishers such as BMJ, PLOS, BMC, Royal Society of Medicine, SAGE, Wiley and F1000 Research. PLOS Medicine, BMJ and BMC Medicine have also shown their support for the RIAT (Restoring Invisible and Abandoned Trials) initiative, which in June 2013 called upon:

“[I]nstitutions that funded and investigators who conducted abandoned trials to publish (in the case of unpublished trials) or formally correct or republish (in the case of misreported trials) their studies within the next year.”  

Beyond the ethical obligations to publish this information, there are several mandatory ones such as the FDAAA Section 801, shortly a European equivalent from the EMA, and the 2010 Joint Position on the Disclosure of Clinical Trial Information via Clinical Trial Registries and Databases. In line with these developments, several pharma companies have stated that they will release all the data from their clinical trials going forward eg GSK, Pfizer and Johnson & Johnson.

Several initiatives in publishing are working to improve data transparency and avoid publication bias. Many journals are collaborating to develop data policies to ensure that data are hosted in a suitable and stable location, with appropriate access controls where required.

The major medical journals are typically willing to publish ‘negative’ findings from the larger trials, but often are less interested in the many small studies that on their own don’t seem to show much.

Collectively these smaller results could become significant, however – if they are published routinely. They often concern compounds that won’t be taken further but the findings may have already been shared at conferences and therefore already have passed the often lengthy legal approval process. The sharing of these results can contribute significantly to the company’s public profile, especially if they are published Open Access for anyone to read. 

In a recent audience survey during an ISMPP University webinar of 107 publication planners and medical writers, few couldn’t see a benefit in writing up negative results (9 per cent) or were worried about the impact on the company profile (6 per cent). Most said they were not currently publishing small negative studies because they believed journals wouldn’t publish them (25 per cent), it would take too much time (30 per cent), or it would cost too much (30 per cent). 

In fact, several journals explicitly state that they will publish these smaller negative/null findings (see table below for a non-exhaustive list). Interestingly, the Journal of Negative Results in Biomedicine recently moved to an open peer review model “to remove concerns that these papers would be filtered by the peer-review firewall”. And a process that allows authors to publish almost immediately and communicate more directly with referees should remove the need to resubmit articles to a cascade of different journals, impacting both time and cost.

A strong push
The push towards data transparency and removing publication bias is set to only get stronger. While pharma grapples with the challenges around releasing negative data from the larger clinical trials, the systematic publication of smaller negative results could have an equally significant benefit on science, healthcare and the public view of the pharma industry. An increasing number of publishers are experimenting with new models of publication and peer review, but many of these are still comparatively small tweaks to a process that has systemic problems. These major issues need to be addressed head-on, using an open invited peer review process, after publication. This can reduce bias in what is published and make the publication of all findings, including small negative/null studies, a rapid and cheap process.

References
1    Peters D, Ceci S. Peer-review practices of psychological journals: the fate of submitted articles, submitted again. Behav Brain Sci 1982; 5: 187-255. 
2    Smith R. Peer review: a flawed process at the heart of science and journals J R Soc Med. Apr 2006; 99(4): 178–182. doi:  10.1258/jrsm.99.4.178