Editorials

Open access to research

BMJ 2008; 337 doi: http://dx.doi.org/10.1136/bmj.a1051 (Published 31 July 2008) Cite this as: BMJ 2008;337:a1051
  1. Fiona Godlee, editor
  1. 1BMJ, London WC1H 9JR
  1. fgodlee{at}bmj.com

    Increases readership but not citations

    This week the BMJ publishes a paper (doi: 10.1136/bmj.a568) that has nothing directly to do with medicine or health care.1 It does, however, have everything to do with access to research results, a topic that should interest authors and readers in any field. The paper asks whether open access (free full text online publication) increases the chances of an article being read and cited compared with subscription access publication (where articles are accessible only to individuals or institutions who pay to subscribe).

    It is a question that many have asked and tried to answer since academics first challenged the subscription based publishing model over 10 years ago. Open access offered an end to what they saw as profiteering by publishers at the expense of the academic community. It restored a public good. If it could also offer higher usage and citation rates, this was icing on the cake. Authors who submitted their work to open access journals might be rewarded with greater visibility, and publishers who launched open access titles or converted existing ones to open access might see their usage figures and impact factors rise.

    Studies in various disciplines have explored this possibility.2 Most have found a correlation between usage and citation rates, as well as a citation advantage from open access. However, all of these have been retrospective observational studies. In what is, to the best of our knowledge, the first randomised trial of open access, Davis and colleagues sampled papers due for publication in a group of physiology journals and randomly allocated them to either open access or subscription access publication. They found significantly higher online usage of open access articles, but no significant difference in citation rates between the two groups in the first year after publication.1

    The study suggests that previous findings of a citation advantage from open access may have been the result of self selection, with more highly citable articles being more likely to be published in open access journals. Indeed, a previous study of medical journals found that the higher the impact factor of the journal in which an article was published, the more likely it was that the article would be available on a non-publisher website.3 Davis and colleagues’ finding that open access provided no citation advantage, despite increased readership, may be explained by the fact that journal readers who generate citations already have subscription access to journals.

    The study confirms that open access articles reach audiences that subscription based access does not reach. Interestingly, it also found that abstracts were downloaded less often when the full text was freely available. Of course, we don’t know whether this means that people are more likely to read the full text rather than just the abstract, but it raises the potential for an additional scientific advantage of open access—that readers and authors are able, should they wish, to build their ideas on a whole article rather than just the summary.

    Can these findings be generalised to medicine? It is hard to see why not. However, one difference between medicine and physiology is the wider press coverage given to medical research. We know that press coverage increases citations.4 If open access increases press coverage, as some have suggested because of the convenience to journalists of being able to access the full text, it is possible that open access could lead to an increase in citations to medical research.

    The BMJ’s own experiment with free or open access publishing of research is now 10 years old and going strong, although as with all experiments it remains under evaluation. In 1998, the BMJ was the first major general medical journal to provide free full text online access to its articles from the moment of publication, to deposit the full text in PubMed Central, and to allow authors of research articles to retain copyright. Access controls were introduced for non-research content in 2006 to protect subscription revenue and to allow us to continue providing the research content free. Since then, the BMJ Group has extended its open access experiment by introducing BMJ Unlocked (http://adc.bmj.com/info/unlocked.dtl), which allows authors submitting research to any of the group’s 19 specialist research journals to pay an author fee in order to make their work open access. (The BMJ itself does not charge author fees.) We have also now made changes to our copyright licence and the information we include in the research articles so that they can be formally listed as open access articles in PubMed Central and other repositories.

    Academic publishing is going through interesting times. We don’t know which model will prevail, or indeed whether there will ultimately be one or several coexisting models. Three things may precipitate a move towards greater openness of access—the demand from funding agencies, such as the US National Institutes of Health and the Wellcome Trust, that grant recipients must self archive or publish in open access journals5; the new Food and Drug Administration (FDA) Amendment Act, which requires trialists seeking FDA approval to deposit their main results in clinicaltrials.gov within 12 months of recruiting the last participant6; and the rapidly expanding and unpredictable influence of web 2.0 on the way we all communicate.

    If greater openness is the future, as it almost certainly is, the academic and publishing communities have some decisions to make. What sort of quality control do we want in the 21st century and how do we fund it? Do we need the current level of peer review and technical editing before dissemination or could medicine move to the physics archive model—dissemination followed by selection for peer reviewed publication (http://arxiv.org/)? How would such a model play in a world in which the media are hungry for new medical breakthroughs, and where manipulation of data and overinterpretation of results is rife? Should quality control continue to be funded largely through subscriptions or can author charges take the strain, either alone as at BioMed Central or with the help of charitable support, as at the Public Library of Science? If funders and institutions support greater access to research results, should they also fund the quality control? Where then would people look for independent verification? Should peer review be centralised and professionalised rather than run by individual journals?

    These are questions that reach to the very heart of the way in which scientists and clinicians communicate. Much will depend in the short to medium term on whether publishers are able to make online usage profitable through advertising rather than subscriptions, whether author charges prove to be a sustainable model, how much pressure is brought to bear by funders and institutions for greater access, and whether the FDA Amendment Act leads to legislation in other countries and beyond randomised trials. These questions also reach to the heart of how academics are acknowledged and rewarded. Hit rates and online profile may soon become as or more important than citations. Whatever the mix of communication models that emerges over the next few years, we need to make sure that science and health care are well served.

    Notes

    Cite this as: BMJ 2008;337:a1051

    Footnotes

    • Research, doi: 10.1136/bmj.a568
    • Competing interests: FG is editor of the BMJ and editor in chief of the BMJ Journals. Her salary is paid by the BMJ Group, which receives revenues from subscriptions, advertising, sponsorship, and reprint sales. The BMJ’s current publication model is open access for peer reviewed research with no author charges and subscription charges for non-research content.

    • Provenance and peer review: Commissioned; not externally peer reviewed.

    References