Annual Meeting Reports

Conflict of Interest: Issues, Policy, and Practice

Companies and journals have different perspectives on conflict of interest (COI) in research. In this session, speakers with experience at companies and scientific journals told the audience how they address the issue.

Laurence Hirsch, of the International Society for Medical Publication Professionals, gave historical background on COI in medical research. He listed articles discussing COI in medicine. The articles resulted in passage of Section VIII of the FDA [Food and Drug Administration] Amendments Act of 2007 and established federal law for clinical-trial registration and results disclosure on Hirsch said that in recent years debates about COI have continued in professional journals—such as the Journal of the American Medical Association, the BMJ, and the Public Library of Science journals—and mass media, such as The New York Times. In addition, disclosures of authors’ COIs by many journals seem to depend on the authors’ relations to industry.

Hirsch asked whether financial association with companies indicates insufficient quality of research. Having worked in industry for nearly 20 years, Hirsch said that it does not.

Furthermore, Hirsch said that to pass FDA examinations, pharmaceutical companies must design reliable clinical trials. Working with pharmaceutical companies does not mean that researchers will distort their research results. Hirsch said that reviewers and editors should judge manuscripts according to their content and quality, not their authors’ affiliations. Companies should design credible studies, provide investigators full access to study data, and respect authorship criteria (including for their own scientists).

Sarah Tegen, of the American Chemical Society (ACS), gave a “chemical perspective” on the subject. In 2005, editors and staff of ACS recognized the importance of COI and revised their policies, Tegen said. They examined the issue and policies of other publications. ACS has now designed specific policies for editors, reviewers, and authors.

Tegen said that all editors at ACS consider manuscripts without regard to authors’ affiliations and that they treat confidential data ethically. She said that reviewers should be sensitive to potential COI, such as research interests and personal relationships. Reviewers should also maintain confidentiality. Authors must reveal COIs to editors and readers. For example, ACS requires authors to read and acknowledge its ethics guidelines before submitting articles and to state their funding sources. Tegen also discussed other ACS policies for editors, reviewers, and authors.

Finally, Katrina Kelner, editor of Science Translational Medicine and former deputy editor of Science for life sciences, shared Science’s experience with implementing policies for COI. Before 1999, Science asked authors to provide the information needed to ensure fair review and prevent COI, but very few provided such information.

In 2002, because Science implemented an electronic manuscript-submission system, it at least could require authors to fill out a form online stating that they had read the statement about COI. Science then received more declarations of COI, but many authors still were not knowledgeable about this topic.

In 2003, Science revised its policy for COI and made the statement more specific. It now requires authors to disclose affiliations (including industry affiliations), all funding sources, and financial relationships, such as patents. After Science revised the policy to request specific categories, it received more disclosures of COI.

Kelner said that researchers working in drug discovery, immunology, cell biology, and similar fields tend to disclose COI more than other researchers do. Not all information disclosed, however, is appropriate to publish. For example, information on patent applications is not appropriate to declare to all readers.

Kelner also mentioned the policy for COI in Science Translational Medicine. The Advisory Board members are required to reveal all COIs on the Web site, and they are not allowed to give input on any paper with which they have a COI.

Kelner said that peer review and replication of experiments are more powerful than disclosing COI in ensuring that research is reliable. When COI is revealed, however, readers may need to judge an article more carefully.