San Francisco

Authors are ignoring journal policies that require them to state any conflict of interest over their papers, a new study suggests. But editors of leading journals disagree about whether this poses a real problem and, if it does, what to do about it.

Limited disclosure: conflicts are rarely declared.

The study, from Tufts University in Massachusetts, examined 61,134 articles that appeared in 1997 in 181 leading journals that had disclosure policies. The journals examined included Science, The Journal of the American Medical Association and Proceedings of the National Academy of Sciences. But Nature and Cell, for example, did not have disclosure policies, and so were excluded.

The researchers counted every occurrence of a positive disclosure, including honoraria, patents pending, stock holdings or other forms of personal or financial interest. The research appears in Science and Engineering Ethics (7, 205–218; 2001). Only a third of the journals surveyed contained papers carrying any disclosure. Of those, fewer than 1% of papers contained a disclosure.

A possible explanation is that few researchers had anything to disclose. But the study's lead author, Sheldon Krimsky of Tufts' Department of Urban and Environmental Policy, says this is unlikely. “What is the chance that in two-thirds of the journals there was no one with a patent, equity interest or honorarium?” he asks.

But some journal editors say that such information is of little value to readers. Kevin Davies, editor-in-chief of Cell Press, says editors usually consider that good science stands on its own merits. “It's the quality of the research that counts,” he says.

Science's editor-in-chief Donald Kennedy agrees. But the journal's authors must provide information on potential conflicts of interest. These disclosures are reviewed by editors, but not given to external reviewers. “The peer reviewers' job is to evaluate the science and not to be our ethicists,” he says.

But Krimsky argues that complete disclosure would help to ensure fair review. “If the article is on a safe cigarette, it would help the reviewer to know if that person had been in the public sector or employed by a cigarette company for 20 years,” he says.

The Proceedings of the National Academy of Sciences requires contributors to disclose relevant commercial ties. But not all authors respond to the requirement without prompting, according to editor-in-chief Nicholas Cozzarelli. The journal has on occasion published corrections noting an undisclosed conflict of interest after vigilant readers complained.

Cozzarelli says the problem is one of differing interpretations of what constitutes a conflict of interest rather than any intent by authors to hide the truth.

Nature's lack of disclosure requirement is set to change, according to its editor Philip Campbell. Later this year, all Nature journals will ask contributors to declare any financial interest in the research they submit for publication. Contributors will have the option not to declare, but this will also be noted in the paper, Campbell says.

The shift is partly in response to evidence that financial ties do matter. For example, a study in The Journal of the American Medical Association (282, 1453–1457; 1999) found that research funded by pharmaceutical companies was eight times more likely than publicly funded research to reach a favourable conclusion about the usefulness of a drug.

Although the evidence is not conclusive, recent debates on genetically modified crops, for example, have highlighted public concerns about a possible undermining of scientific independence, says Campbell. “Transparency about financial interests should help to minimize such concerns,” he says.