CONFLICTING INTERESTS, BIAS, AND SCIENTIFIC PUBLISHING
Unhappiness about the role of vested interests and the extent of biases in research publishing is increasingly evident in the reflections of scholarly publishing. “Most claimed research findings are false,” says medical professor John Ioannidis in one of the most cited papers, “[c]onflicts of interest are very common in biomedical research, and typically they are inadequately and sparsely reported” (1, p. 124). DeAngelis reports researchers’ complaints that corporations commonly “encourage researchers to spin the findings they report in academic journals” (2). Chan et al. are among many scholars raising the alarm about selective reporting of trials (3, p. 28). Krumholz and Ross advocate tighter regulation of commercially funded research (4).
Through two hundred years of organized research, the peer review system has remained the best way to identify biases that might be present in the design, conduct, and conclusions of research. Peer reviewed publishing aspires to objectivity in assessing a research paper’s merits. Peers, it is important to note, are not just people researching in the same field but publishing in it, and therefore subject to the same discipline, while being sufficiently independent of the work to be published.
Authors and reviewers are asked to declare relevant interests. This is noted by editors, by reviewers in those cases where it is made known to them (this varies from journal to journal), and, on publication of the paper, by readers. Such declarations also help scientific publishing to track biases that might emerge across a field and many papers, rather than show up as flaws in individual studies.
Defining vested interest
Many have argued that discussion is too centered on commercial interests. Vested interest biases in research and reporting arise not simply as a result of commercial conflicts of interest, but also because of attachment to current professional practices, political and campaign positions, or funding from non-commercial sources with set objectives. Hames, in a comprehensive guide to the responsibilities of editors, identifies five types of possible conflicts of interest:
· Financial (research funding, personal financial interests, and employment related)
· Political, religious, racial, or gender related (prejudice) (5)
Interests need to be differentiated within these too. “Industry-funded study” could refer to a wide range of relationships. Researchers approaching an organization to fund a piece of research that has been independently determined is different from the same organization designing and commissioning the research. It would seem straightforward that there are two different agendas here. In practice, however, it is not straightforward. Angell has noted that the priorities of prospective funders may be internalized by researchers (6).
Restricting commercial interests?
Few would suggest that papers should automatically be turned down if there might be a commercial incentive to publish them. It is widely accepted that a paper should be published on its merits as a contribution to the field. Indeed, some commercial organizations are obliged to conduct research into the risks and benefits of their products, not least for licensing purposes, and these studies are expected to be open and scrutinized.
However, there is growing interest among journals and commentators in the intervening governance of the research that might be encouraged.
Some kinds of interests, though, are seen to undermine research fundamentally. In 2009, the Royal College of Physicians in London published a report of a working group on the relationship between physicians, the pharmaceutical industry, and the National Health Service that urged the creation of independent sources of evidence about the effectiveness of different prescription drugs for consumers (7).
There are some conflicting interests that are so likely to have an impact on the outcome of research as to cast doubt over the reliability of the results reported, or to make it impossible for the reviewer and subsequent readers to ascertain. However, even this is not a question that is resolved entirely without reference to the content of a paper. A double blind trial in which there is clearly follow up of all participants might be considered worthy of publication even where the author has a substantial conflict, whereas clinical observations, long established as vulnerable to wishful thinking, might not be. Fatal conflicts are not therefore the same as adopting an outright disqualification approach.
Increasing confidence in published research
While there is scope for developing practices to address evident abuses, conflict of interest statements in themselves can offer no guarantees about the confidence we should place in a piece of research. There are many other sources of bias in research publishing and poor or misleading research is not necessarily directly a product of vested interests.
The role of declarations of interest in safeguarding the research base is predominantly as a supplement to the broader activity of assessment, peer review, and post-publication scrutiny. However, while a potential conflict of interest is treated as cautionary information by journal editors, in wider public discussion it has often been seen as cause for dismissal of the findings or for cynicism about whether they present a full picture. The characteristics of source material are important determinants of public attitudes, for example, those pertaining to genetic engineering (8). Unfortunately, the author's declared interests are liable to be more readily understood by non-specialists than the validity (or scientific persuasiveness) of the research paper, and might prove conclusive in determining public, media, and policy attitudes or confidence in the material.
All of this suggests the need for more consideration of governance, alongside greater public emphasis on the journal’s use of independent review, and the importance of replication. Web 2.0 technologies are facilitating post-publication scrutiny, which in turn provides the opportunity to move away from simplistic shortcuts towards establishing the validity of research results.
Publishers and editors, while developing a stronger and more consistent interest in research governance, must at the same time press for a more nuanced understanding of the different kinds of interest and the effects they might have on research reporting, and provide statements on independent peer evaluation rather than assume that this is understood. They might state more widely, for example, that it is not always possible to detect bias in research but on the basis of the methodology, data and conclusions supplied, and at least two independent reviewers who are publishing work in the same field of enquiry have judged a paper to be worthy of publication.
1. Ioannidis, J. P. A. (2005). Why most published research findings are false. PLoS Medicine,2, 124.
2. DeAngelis, T. (2003). Does industry funding deserve a bad rap? Monitor on Pyschology,34, 28.
3. Chan, A. W., Hrobjartsson, A., Haahr, M. T., Gotzsche, P. C., & Altman, D. G. (2004). Empirical evidence for selective reporting of outcomes in randomized trials: comparison of protocols to published articles. JAMA,291, 2457-2465.
4. Krumholz, H. M., & Ross, J. S. (2009). Relationships with the drug industry: More regulation, greater transparency. BMJ,338, b211.
5. Hames, I. (2007). Peer review and manuscript management in scientific journals. Oxford: Blackwell Publishing.
6. Angell, M. (2000). Is academic medicine for sale? N Engl J Med,342, 1516-1518.
7. (2009). Innovating for health: Pateinets, physicians, the pharmaceutical industry and NHS. London.
8. Frewer, L. J. (1999). Reactions to information about genetic engineering: Impact of source characteristics, pereceived personal relevance, and persuasiveness. Public Understanding of Science,8, 35-50.