Recent incidents of scientific misconduct have made researchers and their institutions more aware that credit given on papers is not always credit due. But major research universities and journals in the life sciences have taken few steps to develop policies or guidelines on responsible co-authorship, according to an informal study by The Scientist.
The School of Basic Health Sciences at Virginia Commonwealth University adopted such a policy in August in response to national concern and because administrators felt faculty had not been given sufficient guidance, according to the dean, S. Gaylen Bradley. It includes criteria on how to determine if co-authorship is warranted (see sidebar)…
The editor of Annals of Internal Medicine, Edward Huth, last year adapted guidelines approved in 1985 by the International Committee of Medical Journal Editors. Its criteria define authors as scientists who have "conceived and planned the work that led to the paper, or interpreted the evidence it presents, or both; wrote the paper or reviewed successive versions and took part in revising them; and approved the final version."
Other universities have taken small steps. At Stanford University, a speech on academic authorship by President Donald Kennedy in September 1985 prompted a survey of nearly 1,000 graduate students, research associates, postdocs and faculty. A preliminary analysis of the results by the school's Committee on Research, according to Pat Devaney, associate dean of research and academic administration, has found "there is not widespread concern" about the issue.
She speculated that university administrators will suggest that each department develop its own guidelines and that graduate students be urged to discuss the subject with their advisers.
Harvard Medical School, stung by the discovery of misconduct by John Darsee in 1981 and the retraction this past November of a paper by researchers at Dana-Farber Cancer Institute, is forming a committee on the prevention of scientific misconduct. Darsee was a cardiology researcher at Emory and Harvard universities who admitted to one incident of fraud. Subsequent investigations by both universities and NIH found he had falsified or fabricated data in dozens of other cases, and questions have been raised about the responsibility of his many co-authors in uncovering such misconduct. The incident at Dana-Farber involved a March 1986 paper on a new immunological response that a co-author now admits was based on fabricated data. It too has raised the question of the proper role of coauthors in team efforts.
Taking the Heat
But action is the exception rather than the rule. Editors and administrators point to the difficulty of setting standards and disagreement on the meaning of co-authorship. Some argue that guidelines would do more harm than good by stifling the creative process.
The debate on co-authorship touches on some of the most important and sensitive subjects in science: trust among colleagues, the mentor relationship in a research laboratory, the growing importance of collaboration, and the role of productivity and publication in career advancement.
"The problem is clear-cut in that, if you take credit, you also take responsibility," said Samuel Thier, president of the Institute of Medicine and a faculty member at Yale Medical School. "Beyond that, there needs to be a level of sophistication in dealing with it." The Institute, in fact, may undertake a study on scientific misconduct that could encompass the duties of co-authors. Executive officer Charles Miller said he expected a decision on the project, including its scope and sponsors, to be made by the end of this month.
A recent incident at Columbia University appears to touch on Thier's point. Last month Ronald Breslow, a professor of organic chemistry, notified the Journal of the American Chemical Society that he was withdrawing three recent communications that described novel catalytic steroid reactions because the results could not be duplicated. Breslow was co-author of the papers with a graduate student, Monica Mehta, who has since resigned.
Multiple authorship is increasingly common. A comparison between 1969 and 1979 of the number of authors on original articles in the Annals of Internal Medicine and the New England Journal of Medicine, done in 1982 by Lt. Gen. Kenneth Burman of Walter Reed Medical Center, showed an in-crease of 50 percent (from 3.2 to 4.7 authors) in the Annals and a rise of 37 percent (from 3.8 to 5.2) in NEJM.
But scientists and journal editors disagree on the meaning of being listed as a co-author. Few support the use of purely honorific co-authorship. Listing scientists in return for a contribution of money or space, but not data or analysis, is frowned on, although some researchers say it is common. Most investigators believe the contribution of data to research is necessary, but not sufficient, for someone to qualify as a co-author.
"What a co-author should insist on is not only data gathering, but also analysis and interpretation," said Robert Petersdorf, new president of the Association of American Medical Colleges and former dean of the medical school at the University of California-San Diego. In October a university committee there issued recommendations on co-authorship as a result of its investigation into the misconduct of cardiology researcher Robert Slutsky. Slutsky was found to have fabricated data in as many as 68 papers based on his research, again raising the issue of the role of his co-authors in uncovering such misconduct.
Added Thier, "co-authors should be contributors of ideas, data and critique. It is conceivable that one could contribute ideas and critique but not data, if they were very important."
A co-author's familiarity with all aspects of the work in the paper is a central issue. Scientists disagree on the appropriate level of understanding.
"I would not want anyone to be a co-author without full intellectual understanding of the whole work. To me, that's what authorship means," said Neil Moran, professor and chair of the Department of Pharmacology at Emory University School of Medicine. Moran chaired the committee that investigated Darsee's fraudulent conduct at Emory.
Other scientists contend, however, that investigators should be able to rely on a co-author's expertise in a particular specialized field. A requirement that each collaborator understand every other person's work, they argue, would discourage cooperation and retard progress. The size and "culture" of each laboratory should not be ignored, they add, pointing to the situation where the chief of a small lab may make a legitimate contribution to each publication by lab members.
Journals have taken some steps to deter scientists from inappropriately placing their names—or those of colleagues—on papers. The NEJM and Annals, for example, require all co-authors to sign a copyright release form. The form serves as an acknowledgement of their names being used on the paper.
A Plea for Freedom
Some editors do not believe they should define the responsibilities of co-authors. "It would be inappropriate for a journal to set up [coauthorship] standards," said Benjamin Lewin, editor and publisher of Cell. His journal and the Proceedings of the Federation of American Societies for Experimental Biology, for example, permit the corresponding author to sign for all co-authors. Cell's policy stems from the difficulty of tracking down each author, Lewin said, as well as from the assumption "that people are acting in good faith."
Despite recurring instances of misconduct, many scientists are loathe to question the trust upon which most research is based. John Ultmann, dean of research and development at the University of Chicago medical school, thinks that any policy on co-authorship is inappropriate. "Research is a creative act," he said. "You don't tell people how to do it in advance. The more freedom they have, the better [the results]."
But even those who support guidelines to prevent misconduct may find it hard to create a workable policy. At Emory, as at many schools during the first part of this decade, medical school officials tried to go beyond a policy to investigate alleged fraud to one that could prevent misconduct. But they wound up recommending that departments establish individual guidelines on co-authorship, ownership of data and related issues. Within the department of medicine, for example, each co-author and the deputy department chief must sign off on any article written by a member of the department.
One idea gaining support is a limit on the number of papers that tenure and promotion committees review. But even that straightforward change is not a simple matter. Thier cautioned that such a system could penalize investigators whose findings are reported in a series of papers, none of which seems especially significant on its own. And .Petersdorf, who supports the concept, noted that many panels rely on quantity because they do not have the expertise to judge specialized research.
An editorial in the January issue of Annals by Thomas Stossel, editor of the Journal of Clinical Investigations, indicates that such a policy would be a drastic change for most medical schools. His informal poll of the practices of 17 leading departments of internal medicine found that three require a minimum number of papers for appointment and in only five cases would a small number of high-quality papers "always" ensure promotion. "Despite pious assertions to the contrary," Stossel observed, "promotion practices may signal more value for quantity than quality."
One Approach to the Problem
The overriding concern of the policy adopted recently at Virginia Commonwealth University, developed largely by the dean of the School of Basic Health Sciences, S. Gaylen Bradley, is the ownership of intellectual property. The proper use of co-authorship is part of a larger effort to assign appropriate credit to intellectual achievements, a process that may include professional, scientific and/or economic rewards.
The determination of intellectual ownership "has become increasingly complex," Bradley writes in the policy, "as new constituencies assert partial ownership based upon contributions to a research project."
According to the policy, to merit credit on publications, "the individual must
- Contribute significant ideas and experimental design for the project;
- Take part in the actual experimentation and data analysis; and
- Be able to present and defend the work in context at a scientific meeting. Exceptions will need to be made in publications in which one author has carried out a unique, sophisticated study or analysis."
The policy also discusses protocol in listing authors. It says, "In general, the principal author is the first author, and a laboratory director is listed last. Other authors are more or less listed in descending order of their contributions to the research being reported. The submitting author is usually the first or last author listed."
The policy emphasizes the need to discuss the question of proper credit at every point in the research process and with every person involved in the effort. Taking note of the hierarchical structure of a research laboratory, and the lab's relationship to the department, the institution, and the various contributors to that institution, the policy cautions that "the rights and responsibilities of the various interest groups are not well understood and are rarely discussed openly…. The strategy for assigning credit and responsibility should be established early in a research project, reviewed periodically, and revised if appropriate?"