All scientific content of the BMJ should declare authors' conflicts of interestBMJ 1998; 317 doi: http://dx.doi.org/10.1136/bmj.317.7154.351 (Published 01 August 1998) Cite this as: BMJ 1998;317:351
- David F Marshall, Paediatric surgical research fellow*
EDITOR— The BMJ's practice of declaring any conflict of interest that the authors of papers may have encourages critical appraisal. In medical literature, however, as in politics, attempts to make public the interests of policymakers, sometimes voluntarily, seem now to require reinforcement. The BMJ has published a news item about the New England Journal of Medicine's failure to mention a conflict of interest for the reviewer of a book linking environmental pollutants to cancer.1
The BMJ risks similar revelations because most of its scientific content failsto declare authors' conflicting interests. This is particularly incongruous in the Editorials and Letters sections, where opinions, often on controversial subjects, are expressed without challenge(such challenge often occurs in the Education and Debate section). The editorial on the health risk of silicone breast implants is an example of this non-declaration of interests.2 Despite the media and public interest and the legal and commercial implications surrounding this issue, the authors omitted the data that presumably provoked the Food and Drug Administration's limited ban on such devices, and they dismissed other studies suggesting an association between silicone breast implants and connective disease. Yet readers of this editorial, and of a letter on the same subject in that issue,3 are ignorant of any relevant interests that the authors have. Thankfully, because the papers in the BMJ do mention conflicts, we learn that the paper that prompted the above editorial was funded, albeit indirectly, by Dow-Corning Corporation, although we are not informed that this isthe manufacturer of the implants in question.4
Many readers favour the BMJ's editorials above the papers, hoping for an expert's evidence based review of the literature. The author, though, must select which published data to include and may also incorporate unpublished or anecdotal observations. This whole process may be coloured by subjective personal experience and preference.
There can be few reviewers whose final article differs in conclusion from their previously heldopinions. A notable exception is the case of an atheistic researcher who attempted to disprove thevalidity of the Bible: during his studies he became so convinced by the quality and quantity of evidence that he became a Christian.5 Without such divine intervention, or at least a declaration of any conflict of interest, readers of BMJ editorials (and its other scientific content) will never know whether the doctrine printed depends on unbiased, evidence based criteria or, unwittingly, on conflicting interests. The BMJ must not hinder informed critique.
Conflict of interest None