Credit: M. XERIDAT

Money talks, so they say. But is it whispering covertly through the pages of leading science journals? Too often, the answer is yes, claims a letter sent last month to some 200 journals by the Center for Science in the Public Interest (CSPI), a non-profit organization based in Washington.

The letter urged the journals to strengthen their policies on the disclosure of conflicts of interest — and given that its signatories include former editors of The New England Journal of Medicine (NEJM) and The Journal of the American Medical Association (JAMA), it cannot simply be dismissed as the blusterings of a fringe element. “Whether the issue is clinical research, cancer clusters or global warming, corporate interests can hide behind the credibility of peer-reviewed journals,” argues Virginia Sharpe, who heads the CSPI's Integrity in Science project.

As the links between commerce and academia deepen, how to deal with the conflicts of interest that inevitably arise has become an increasingly important issue. Next week sees two meetings devoted to the subject: at Emory University in Atlanta, Georgia, researchers, entrepreneurs and others will meet to debate the 'Commercialization of the Academy'; meanwhile, Warsaw in Poland will host an international conference on conflicts of interest in science and medicine.

The Warsaw meeting will be addressed by the current editors-in-chief of JAMA and NEJM. As research agencies, academic institutions and scientific societies all debate the issue of conflicting interests, journal editors are finding themselves on the front line — the scientific literature is, after all, the main forum for the communication of research results. Some of these editors are thinking hard about strategies to minimize the potential for publications to be biased by commercial pressures, and to draw their readers' attention to any conflicts of interest that may exist.

Lisa Bero (top) and Sheldon Krimsky worry that vested interests are distorting published science. Credit: J. WATSON; M. MORELLI

Lisa Bero, a pharmacologist at the University of California, San Francisco, who studies how science influences clinical practice, has no doubt that commercial interests are biasing the scientific literature. “Studies that are sponsored by a single company are biased compared with studies with multiple or other sponsors,” argues Bero, who signed the CSPI letter. “When research is funded by one company that has an interest in the outcome it is much more likely to have a favourable outcome for the sponsor's product.”

In 1986, Richard Davidson of the University of Florida College of Medicine in Gainesville reviewed 107 published clinical trials and found that those sponsored by drug firms were more likely to report favourably on the treatment being tested1. Since then, several studies have provided support for Davidson's conclusion2. One famous example reviewed 70 articles commenting on the safety of calcium-channel antagonists, a class of drugs used to treat cardiovascular disease. Among the authors of original research papers, reviews and letters to the editor that were supportive of the drugs' use, 96% had financial relationships with the drugs' manufacturers; for publications deemed neutral or critical the figure was only 60% and 37%, respectively3.

Clinical precision

The factors that underlie such biases remain unclear. “It could be because negative or unfavourable studies aren't published; it could be that companies are not dumb enough to fund a study that is not going to work out; it could be that they are conducted or designed in a poor way,” Bero speculates.

Concerns about commercial conflicts have been most acute in clinical medicine — where human lives may be at stake. Medical journals have led the way in introducing editorial policies to deal with conflicting interests, and they continue to blaze a trail (see 'Box 1 Who controls the data?'). At the heart of each policy lies the concept of disclosure: if everyone is made aware of authors' financial interests, goes the argument, the potential for bias can be borne in mind by the reader. But how full that disclosure should be, and how to encourage authors to comply with a stated policy, remain matters for debate.

Leading multidisciplinary journals such as Nature and Science have also adopted conflict-of-interest policies — as have journals in fields such as nutrition, where commercial conflicts have become a serious concern. But in many scientific disciplines the issue is still not on editors' radar screens, despite the best efforts of the CSPI and its supporters. “This is a potential problem in any science in which there is a commercial interest,” says Sheldon Krimsky of Tufts University in Medford, Massachusetts, who has studied the conflict-of-interest policies of leading journals and will speak at next week's meeting in Atlanta.

The CSPI's letter was sent to journals in fields including climate research, environmental science and chemistry. Bruce Coull of the University of South Carolina in Columbia was moved to sign the letter by his concerns about commercial conflicts in his own discipline of marine ecology. “Many scientists in our field are sponsored by companies or work as consultants,” he says. “I would like to know that when I read their publications in ecotoxicological journals, but most of these journals don't have disclosure policies.”

Orrin Pilkey, a coastal geologist at Duke University in Durham, North Carolina, and another signatory of the CSPI letter, feels the same way. “I see more and more examples of studies with a huge amount of optimism, written by scientists who also work as consultants for companies,” he says.

In agreement: Donald Kennedy (left) and Philip Campbell back disclosure to readers but not referees. Credit: SCIENCE

But it seems that the CSPI has some way to go in convincing journal editors to address the issue. “Life is pretty complicated already and this would be another layer of paperwork,” says Charles Finkl of Florida Atlantic University in Boca Raton, and editor-in-chief of the Journal of Coastal Research. Other editors fear that rigid policies could make researchers send their papers elsewhere. “Every time you put another barrier in the way, people can go to another journal,” observes Martin Blume, editor-in-chief of the American Physical Society, which publishes Physical Reviews, Physical Review Letters and Reviews of Modern Physics.

Blurred vision

Even among editors who have embraced the general principle of disclosure, there is considerable divergence over what this should mean (see table). One controversial issue, for instance, is whether authors' conflict-of-interest statements should be given to the referees asked to review a manuscript. Donald Kennedy, editor-in-chief of Science, explains why his journal does not do this: “That's not part of the referees' job. They are supposed to review the quality of the science; they are not ethicists.” Nature has the same policy of not disclosing conflict-of-interest statements to reviewers.

Table 1 Sample policies from leading journals
Full size table

But some journals do send the disclosure statements to referees — for good reason, argues Krimsky. “Personal financial interest is a relevant factor in raising the scepticism of peer reviewers,” he claims. “There are too many articles that are not sufficiently rigorously reviewed, so the flaws in them are only disclosed after publication.”

Another issue that divides the journals is how to deal with the financial interests of the referees themselves. Jeffrey Drazen, editor-in-chief of NEJM, says that his journal pays close attention to this potential source of bias. “We ask them to provide information that throws a light on possible commercial and intellectual prejudice. When a referee gives negative advice about a study we liked very much ourselves, we make some calls,” he says. “We have seven editors spending all their time reading reviews and judging the relevance of negative and positive remarks.”

Nature asks referees to disqualify themselves if they think judging a certain manuscript would create a conflict of interest. “But we do realize that it can happen that a referee judges a manuscript he or she should have pushed aside,” says Nature's editor, Philip Campbell. “We often use three referees, which helps to avoid such problems.”

Opinion toll

Jeffrey Drazen: referees may introduce bias. Credit: H. STIMPSON

Other journals that demand disclosure of conflicts by authors have no formal policy for the interests of referees — the American Heart Journal is one example, although some of its referees have, on occasion, declined to review a manuscript, citing a conflict of interest.

The authorship of opinion and review articles has also emerged as a contentious issue. Although most journals adopt the same rules as for original research papers, NEJM has, since 1990, required authors of such pieces to have had no commercial ties for at least two years with companies whose products they are writing about. But the policy is becoming increasingly difficult to enforce, admits Drazen, and is now under review. “Since the introduction of that rule, the entanglement of scientists and companies has grown,” he says. “Therefore we have to exclude a lot of scientists from writing comments. We have collected data to see if that's still the right policy for the journal.”

Editors, too, can have financial interests, and the major medical journals have strict rules on the commercial ties of their staff. Catherine DeAngelis, editor of JAMA, had to sign a statement that she had no conflicts of interest when she took up her post in 2000. “I am very careful,” she says. “When I received a call that my uncle had left me stocks in the pharmaceutical company Johnson & Johnson, I said immediately that they should be given to my sisters. I don't want stocks.”

Catherine DeAngelis: no stocks, no conflict. Credit: J. C. HARRIS

Drazen, an asthma researcher at Harvard University, had to sever his financial ties to 20 drug companies before he took over the helm at NEJM in 2000. “The proceeds from stock investments I gave to charity,” he says. Drazen also agreed not to deal for two years with manuscripts involving products from companies in which he previously had a financial interest.

Nature requires editorial staff to declare to their managers any interests that might be perceived to influence their editorial judgement. Managers then decide how any potential conflicts should be dealt with.

But move beyond front-line journals, and things become rather murky. “There are journals where the financial ties of the editor will determine what gets published,” claims Mildred Cho, a bioethicist at Stanford University in California. She is particularly concerned about journals publishing papers about medical devices. “After publication, those papers are used as marketing tools by the companies that produce the devices,” says Cho. “But we don't know the prevalence of editorial staff having commercial ties.”

Hidden agendas

There are journals where the financial ties of the editor will determine what gets published. Mildred Cho Credit: M. VAN DE RIJN

Even if a journal has a clear conflict-of-interest policy, it is of limited use if it is widely ignored. Unfortunately, it seems that this is often the case. Last year, Krimsky published a study of articles appearing in 1997 in 1,396 high-impact journals4. Only 15.8% of them had an explicit conflict-of-interest policy, of which almost 90% were medical journals. Among the journals with a stated policy, only 0.5% of papers included a disclosure of conflicting interests, and 65.7% of these journals published zero disclosures. In an earlier study, Krimsky unearthed evidence of lead authors with relevant commercial interests in 34% of a sample of 789 papers5, so he does not believe that so few authors had conflicts to declare. “Poor compliance is the more likely explanation,” says Krimsky.

Nature introduced its conflict-of-interest policy in October 2001. Again, disclosure rates are relatively low: of the first 110 papers accepted under the policy in 2002, only five included a declaration of a financial interest.

Staff at NEJM know from bitter experience all about the subtle difficulties of policing a conflict-of-interest policy. Following revelations in the Los Angeles Times that some authors of review articles in NEJM had commercial interests in the treatments they were writing about, the journal conducted an internal review. In February 2000, NEJM revealed that, since January 1997, 19 of the 40 drug-therapy review articles it had published were written by scientists with industrial links that should have disqualified them under the spirit of the journal's strict policy6. The authors had slipped through a loophole that exempted financial support given to their institution, rather than to them as individuals.

Richard Smith: sees a rise in declarations. Credit: U. PREUSS/BMJ

Richard Smith, editor of the British Medical Journal (BMJ), agrees that it is hard for journals to enforce their policies. “A lot of researchers still think they are immune to the influences of their sponsor and don't realize that bias can slip into their research very subtly,” he says. “Some see it as an infringement on their freedom. But I must say the numbers of disclosure statements we get are increasing. Maybe the culture has started to change.”

Cultural revolution

A recent study by Smith reinforces that view. Counting the disclosures in editorials, original research papers and letters to the editor in five major medical journals — the BMJ, JAMA, NEJM, The Lancet and Annals of Internal Medicine — in 1989, 1994, 1996 and 1999, Smith found an increase, from two declarations in 1989, eight in 1994 and four in 1996, to 38 in 1999 (ref. 7). But still, the vast majority of the 791 articles published by the journals in 1999 contained no disclosures.

Some researchers who have become embroiled in rows over conflicts of interest agree that disclosure is to everyone's benefit. In October 1997, toxicologist Stephen Safe of Texas A&M University in College Station wrote an editorial for NEJM in which he argued that environmental oestrogens such as polychlorinated biphenyls do not cause breast cancer, and attacked public “chemophobia” fed by “paparrazi science”8. When it emerged that Safe had previously received funding of $150,000 from the Chemical Manufacturers Association, the editorial became mired in controversy.

Safe defends his integrity: “My views on endocrine disrupters have been fairly consistent over the years and based on the scientific data; needless to say, I was upset over the commotion.” But in retrospect, he now agrees with the idea of full disclosure of current and prior funding sources.

As to whether well-enforced disclosure policies would reduce commercial biases in the scientific literature, no one can say for sure. “Transparency about conflicts of interests is a bare minimum,” argues Bero. “People shouldn't have the idea that everything is OK just because financial conflicts of interest are disclosed.”

Krimsky notes that disclosure means different things for different journals — some merely require authors to tick a box indicating whether they have financial interests, whereas others demand a full breakdown of what those interests are. He argues that investigating the effectiveness of different types of conflict-of-interest policy is an important avenue for future research — but says it is difficult to get funding for studies in this area.

Many important questions remain unanswered, agrees Bero. “Do disclosure policies discourage investigators from submitting to journals? How do readers use conflict-of-interest information — do they take it into account when reading an article? Are disclosures that are published in journals accurate?”

Given the increasing number of papers getting bogged down in accusations of commercial bias, perhaps it is time to start searching for some answers.

Integrity in Science project → http://www.cspinet.org/integrity

Commercialization of the Academy conference → http://www.emory.edu/PROVOST/SamNunnForum

International Conference on Conflict of Interest and its Significance in Science and Medicine → http://surfer.iitd.pan.wroc.pl/events/ConferenceApril2002.html

Nature's policy → http://www.nature.com/nature/submit/competing/index.html