In a recent issue of the IJC, we published an article reporting the analysis of a case control study of lung cancer,1 in which smoking was assessed as a potential confounder of fruit and vegetable consumption. Unfortunately, although the authors had revealed (as we require) the funding sources for this study, we failed to recognize that one of them was an institution funded by the tobacco industry, and the authors did not reveal that to us. Worse than that, one of the authors who served for many years as a consultant to the tobacco industry2, 3 failed to disclose that connection.
This incident, originally brought to our attention by the UICC from a staff member's review of the online version of the article, has led to a reconsideration of our editorial policy, with regard to potential conflicts of interest.
Conflicts of interest have been clearly defined by the International Committee of Medical Journal Editors in their Uniform Requirements for Manuscripts Submitted to Biomedical Journals.4 Our instructions to authors have always requested that potential conflicts of interest should be revealed to the editors at the time a manuscript is submitted for consideration. However, unlike other journals, we have not required authors to sign a conflict of interest form. This lacuna will now be filled. As of now, all authors will be required to sign a “conflict of interest” form upon submission, and we announce that if, after a manuscript has been accepted, we learn of any undisclosed conflict of interest, we shall withdraw the manuscript, whatever the stage of the production process it has reached. Our author instructions have been revised accordingly.
However, we acknowledge that authors are entitled to some guidance from us as to what we regard as a potential conflict of interest. Clearly, in the incident cited earlier, a potential conflict of interest existed. A similar conflict would relate to a study of cancer in workers exposed to a putative carcinogen if the study was financed by the relevant industry, if one of the authors received financial support from the same industry or if one of the authors was an employee of the relevant industrial organization. A study of diet and cancer financed by the food industry would fall under exactly the same heading, as would a study of the risk from a potential oncogenic virus, if the firm that had developed the test used had provided support in any form to the study (including providing the test free of charge to the investigators). Similarly, a clinical trial of a drug used to treat cancer, financed in part or in whole by a pharmaceutical industry (or with the drugs provided free of charge), would fall under the same heading. We are certain that other examples may also occur to many of our readers; however, here, our purpose is not to be all inclusive, but to provide relevant examples.
Whether or not a potential conflict of interest is of sufficient importance to result in risk of bias in a manuscript is clearly a matter of judgment. If the editors regard the potential conflict as sufficiently serious that we have reservations on the validity of the manuscript, we reserve the right to reject the manuscript on these grounds. But we recognize that opinions may differ as to what constitutes a conflict of interest, and we also reserve the right, if we regard the manuscript of sufficient importance, to publish such an article, with the potential conflict clearly revealed, leaving it to our readers to make their own interpretation on the validity of the findings reported.
We therefore urge all potential authors, mindful as we are of the necessity to preserve the integrity of the scientific process, to regard our request that they reveal potential conflicts of interest to us extremely seriously, and we guarantee that we shall consider each case strictly on its own merit.