Printer Friendly

Publishing biomedical research: roles and responsibilities.

Authors, reviewers, and editors have critical responsibilities to ensure the validity and utility of published biomedical research. Publication of the results of biomedical research in scientific journals is not a mere formality or epiphenomenon in science. It is the culmination of the long process of scientific research, in a sense the fruits of the research. It is the means by which a laboratory scientist or clinical investigator gives a definitive accounting of his work to his peers, thus enabling them to judge its significance, confirm it, test its applicability to their own studies and, if it is found acceptable, to incorporate the work into the growing body of knowledge that constitutes the current understanding of a given field. Collectively, published research reports define the content of a scientific discipline and record its evolution. Without a published record, science simply could not function.

A scientist's publications stand as a permanent record of his personal contribution. They largely determine his reputation and the advancement of his career. Appointments, promotions, awards, and research grants all hinge on the quality and quantity of his publications. The resulting "publish or perish" imperative is easily understandable, but it has unfortunate consequences that will be discussed in more detail below.

The public also has a major stake in scientific publications. Journal articles are an important-probably the most important-source of stories in the popular media about new developments in biomedical science. With the help of the authors and other experts, reporters interpret journal articles for the general public. The public appetite for medical news seems insatiable, and new developments in medicine are often treated in the media as major news events. Assuming that the work is accurately reported by the media, its publication in a reputable scientific journal lends substance to the news story and ensures that the story is based on evidence judged by experts to be at least worthy of serious attention. That is no small benefit, given the frequency with which the public has been misled by premature or inaccurate scientific claims when the news has been released only through media channels without the careful scrutiny associated with journal publication. Editorial Peer Review

Scientific reports submitted to the best scientific journals go through a process of evaluation, criticism, selection, and revision known as editorial peer review. Although the details vary among journals, the basic outlines of the process are generally similar. (1) Authors submit unsolicited reports of their research, prepared in a standard format that has been adopted by the great majority Of biomedical journals.

The format ensures that all essential information will be reported, and it facilitates review, but it also inhibits freedom of stylistic expression and makes scientific papers more boring than they need be. The reports are initially screened by editors and those appropriate for further consideration are referred to outside experts for critical review. Aided by the advice of their referees, editors select those papers most likely to meet the journal's scientific standards, and ask the authors to revise their manuscripts to satisfy the reviewers' criticisms. After further review of the revised manuscript, a final decision is made by the editor. It usually takes about three to four months from submission to decide that a paper is finally acceptable, but less than half that time to reach a negative decision. In the course of this process, many submissions are eliminated. The most competitive peer-reviewed journals may accept fewer than a third or a quarter of the manuscripts they receive, but many other journals publish a much. larger fraction. Authors who persist in resubmitting their rejected manuscripts to other journals will usually succeed in getting their paper accepted somewhere, unless the manuscript is grossly unfit for publication.

The publication of scientific reports therefore involves at least three categories of participants: authors, reviewers, and editors. For the process to work properly, all participants must act responsibly, with due regard for their obligations to their colleagues, science, and the public welfare. What follows is a brief account of what I believe those obligations to be. It is a personal view, derived from long experience in all three of those roles. For many years I did research, wrote scientific papers, and acted as a reviewer for several journals. I also had the privilege of serving a five-year term as editor of the Journal of Clinical Investigation. Currently, for nearly thirteen years, I have been editor of the New England Journal of Medicine-a post that affords a unique perspective not only on the communication of research results by biomedical scientists to their peers and to practicing physicians, but on the reporting of this news to the public. The Responsibilities of Authors

What, then, are the obligations of authors? Most important of all, they should faithfully describe what they did and observed, and they should make every possible effort to avoid self-serving bias. Authors have no obligation to be correct, but they must be honest. They can be wrong in their choice of experimental design or methods and wrong in the analysis and interpretation of their data. Such honest mistakes are permissible because errors are inevitable and, given a candid and intelligible account of what was done and what was observed, they can be identified and corrected during the review of the manuscript or later. Deliberate deception, on the other hand, is not only morally reprehensible, but more difficult to discern. Particularly when it is adroit, fraud usually cannot be detected during manuscript review because there is no practical way for editors or reviewers to know with certainty what actually occurred in the laboratory or in the clinic. When an author reports plausible but fraudulent results there is no choice but to take his data on face value. To do otherwise would undermine the basic collegial trust upon which scientific discourse rests. (2) Scientists habitually question colleagues' methods, conclusions, and interpretations, but without direct evidence they do not challenge their honesty.

Questions about a scientist's honesty and challenges to the integrity of his research claims are most appropriately raised by those who work directly with him. Coworkers and supervisors are most likely to be aware of any suspicious irregularities in a scientist's behavior and it is they who are in the best position to suspect fraud when it is being committed, not those who later read the fraudulent results in a research report. Fraudulent science is almost always the work of one person, not a team. Too often, coauthors and coworkers have not paid enough attention to the culprit's work in the laboratory, or have been too distant from the laboratory to be aware of any problem. To remedy that, Konrad Bloch and I recently proposed that all coauthors should routinely be asked to sign a statement that they have not only read and approved the manuscript being submitted but are willing to take some responsibility for its integrity. (3) Deliberate dishonesty among scientific authors fortunately is uncommon, but when it occurs it is so damaging that strenuous efforts must be made by the scientific community to detect it early and to prevent it whenever possible. Asking each coauthor to vouch for the honesty of the article mill in my view do much more to achieve that goal than the auditing and policing schemes now being proposed by some zealous government officials and others concerned about fraud in science. Institutions where the work is done must also take more responsibility in establishing a climate that encourages honesty, collegiality, and accountability among research workers.

Authors have an obligation to be as clear and concise in communicating their research results as is compatible with full reporting of their work. Unnecessarily lengthy reports waste readers' time and journal space, both of which are scarce commodities. Clarity is no less important than brevity. Obscure, badly written reports full of jargon and highly technical language discourage comprehension by all but those closest to the field. Such reports frustrate one of the primary objectives of scientific publishing, that is, the communication of new ideas and information to a wide circle of colleagues and the establishment of new leads and connections for further work. Of course not all authors are skillful writers and many will need help from copy editors, but all who wish to publish their work in journals have an obligation to write as clearly and concisely as they can, and to be responsive to the efforts of editors seeking to make their work comprehensible to the greatest number of readers.

The obligation to be economical in expression applies as well to the number of reports a scientist chooses to submit. Researchers should be judged primarily by the quality and number of their real scientific contributions, not by the quantity of papers they publish. Unfortunately, however, scientists can gain recognition simply by publishing frequently, regardless of the quality or originality of their work, or the importance of each publication. As a consequence, the desire to add to one's bibliography often tempts researchers to "redundant publication," that is, to the reporting of what is essentially one study in two or more fragmented or overlapping publications. (4) Sometimes, multiple reports on the same study can be justified and the definition of "redundant" publication in a given instance may be a matter of judgment. But authors have an obligation to inform editors when they are submitting a report that overlaps with another paper, so that editors can make their own judgments about the manuscript they are being asked to review. Redundant multiple submissions may serve the ambitions of authors, but not the needs of science. They clutter and confuse the literature and they debase the value of a scientist's bibliographic record as a measure of his productivity.

Authors should be generous in acknowledging the work of others and they should be scrupulous in their historical accounting of the events leading to their own workModern biomedical science is a structure built by the work of innumerable scientists. Virtually all advances-even the so-called "breakthroughs"-are made possible by earlier contributions from others. To understand how a particular study contributes to the flow of new ideas in a field, readers need to know the relevant background, and it is the author's responsibility to provide that information accurately and fairly. Authors should not be reluctant to recognize their indebtedness to others. Research in exciting, fast-moving fields is understandably competitive, but there are ample rewards for all good, original work even when priority is shared. Authors serve no useful purpose-not even their own reputation and professional advancement-by ignoring the work of others, or by claiming more than their fair share of the credit for a new discovery.

Finally, authors ought to be particularly careful to avoid premature and unseemly publicity about their work. In biomedical scientific research, everything depends on the evidence. Unless convincing supporting data are made available, claims and conclusions cannot be evaluated and public disclosure is usually unjustified. The supporting data are best presented in scientific journals where, after peer review, they are fully available for study and criticism. journal publication also affords authors adequate opportunity to discuss the significance and limitations of their research and relate it to other work in the field. Other scientists need the published article if they want to confirm and extend the work, and journalists rely on it for the stories they will release to the public. Before publication of the article, presentation of the results by the authors at a scientific meeting, or review and public release of the main conclusions by an official government or scientific body may serve the same purpose, but usually not as well. In any case, authors should not publicly disseminate news of their work until the data are generally available for their colleagues. This is particularly important in clinical research, where premature claims of new treatments or other medical advances can heighten public interest and anxiety even before physicians have seen the evidence and are in a position to respond to questions from their patients. The New England Journal of Medicine discourages authors from publicizing their work before it has been published, except in rare instances in which immediate publicity is required to protect the public health. There have been many examples of damage done by false and misleading claims when this principle has been ignored. Even when the work is sound, premature release of publicity before full publication prevents any response by the medical profession and invites public misunderstanding. Good scientists usually shun premature publicity, and responsible journalists usually do not encourage it. Unfortunately, not all journalists are fastidious about the sources of their science stories and some actively encourage scientists to circumvent the usual professional channels of communication. At the same time, some research institutions and some scientists, pressured by the increasing competition for funds and public support, seem to be more willing to use the techniques of public relations to further their own interests. In my view, there are good reasons why authors should in most cases delay public announcement of their work until the publication of their data in the scientific literature, or at least until they have presented their results at a scientific meeting. In exceptional cases, the urgency of the findings may justify immediate announcement, but this is a judgment best made by public health officials rather than authors. Reviewers' Responsibilities

Reviewers also have a key role to play in the publication of scientific communications: They are consultants and advisers to the editor in judging the quality and publishability of a manuscript and are chosen for this role because they are experts in the field and are in the best position to give a critical technical assessment of the paper. Editors (and authors) expect reviewers to be not only scientifically competent and rigorous in their refereeing of the work, but also prompt, fair, and temperate.

If they agree to accept the assignment, reviewers are usually given two or three weeks to return the manuscript to the editor with their written commentary. They are asked to offer detailed criticisms and suggestions on a form that mill be mailed to the author. On a separate sheet, for the editor's eyes only, they give their overall assessment of the paper and their opinion as to its publishability. Comments to the authors are supposed to be technical only and usually do not include the general opinions transmitted to the editor. The purpose is to leave the overall assessment to the editor, who may have to reconcile conflicting advice from the reviewers, or who may wish to make a different editorial decision from that favored by the reviewers. The reviewers are the technical experts, but the ultimate decision to publish or reject may hinge on broad nontechnical considerations, such as interest to readers, that are properly the editor's domain.

The conscientious reviewer reads the entire manuscript carefully, including not only the design, methods, results, and the analysis and interpretation of data, but also the abstract, introduction, discussion, figures, tables, and references. No detail is too small to escape his notice, and his appended comments may be voluminous. His critique should be rigorous, but its tone moderate and constructive. Intemperate or insulting language has no place in his commentary and the helpful reviewer avoids extravagant praise at one extreme and harsh condemnation at the other.

Depending on the length and complexity of the manuscript, a careful review of this kind may take several hours to complete. For persons with many commitments this is a significant effort, particularly when they may be asked by various journal editors to review manuscripts frequently. Reviewers are paid nothing or merely a token for their efforts, but if they have the time most accept the invitation out of a sense of duty. As members of the research community, they know they will be submitting their own papers for review and will have the benefit of critiques by other busy scientists who will be donating their own time and effort. Reviewers may also be members of the editorial board of the journal, and may therefore have a special obligation to be helpful to the editor. Another reason scientists are willing to give their time as reviewers raises interesting questions about conflicts of interest. Scientists who are actively working in the field are often asked to act as reviewers because they are most likely to be familiar with the technical aspects of the work. But they may also be competitors of the author or may have some intellectual stake in the problem studied. The manuscript under review therefore may contain information potentially useful to the reviewer, or it may support, or conflict with, his published opinions on the same subject. In any case, an expert reviewer may want to review the author's manuscript because he will gain access to private information not yet generally available, which may be of interest to him in his own scientific work. Such conflicts are hard to avoid, but any bias they may engender can be mitigated by careful editorial oversight and by the fact that several reviewers are consulted.

Manuscripts are of course sent to reviewers as confidential documents. The New England Journal of Medicine asks that manuscripts under review be treated as privileged communications-not to be discussed with or shown to anyone else except by special permission. The journal also asks that reviewers not make copies of the manuscript for their files, thus emphasizing that the manuscript and its contents are the private property of the author. But how can the author's intellectual property be protected from possible exploitation by the reviewer? How can even the most conscientious reviewer expunge from his mind all knowledge of the author's work after he has completed his review and returned the manuscript; It is obvious that he cannot and that the author's ideas can only enjoy incomplete and temporary protection. Reviewers can be enjoined from using the author's ideas while the manuscript is under review, but after that, what is to prevent reviewers from taking advantage of their privileged advance access to the author's data and ideas? Concerned about these risks, some authors stake their priority claim in advance by presenting their work at a scientific meeting, or submitting an abstract for consideration at such a meeting before submitting their manuscript to a journal. Others ask editors not to send their manuscript to particular reviewers who are known to be rivals of the authors and actively working on the same problem. When they think their work is particularly "hot," authors may submit it to a journal known for its rapid publication schedule, or they may ask the editor to expedite the review process.

In my view, these concerns, while real, probably exaggerate the seriousness of the problem. The introduction of commercial interests in some areas of molecular biology seems to be changing the climate, but most scientists in most other areas of medical research still feel comfortable about discussing their current unpublished research with their colleagues, not only in meetings and seminars but in many other informal situations. Most are willing to let the peer review system work, on the assumption that the benefits of the process outweigh the risks. They believe (with justification I think) that few reviewers are likely to use the authors' ideas in a way that would be damaging. Editors can help protect authors against such exploitation not only by insisting that manuscripts be treated as privileged communications, but also by ensuring reasonably prompt editorial decisions so that manuscripts are not held too long in the limbo of the review process.

Some critics have suggested that authors would have more protection if reviewers were uniformly required to sign their comments to the author. Most journals currently do not have such a requirement (the New England Journal of Medicine leaves the decision up to the reviewers; about 85% do not sign), but there is a lively debate on this issue. Those favoring compulsory signing of reviews believe that reviewers would be more responsible and helpful, and less biased; it could also be argued that they would be less likely to pirate the author's ideas. Those favoring anonymous reviews (I among them) think that anonymity favors more outspoken and rigorous criticism. When authors know the reviewer's identity, reviews are likely to be softer, less critical, and therefore less useful. This is particularly true when the author is a senior figure and the reviewer is more junior, or when author and reviewer have personal or professional des that make it difficult for the reviewer to be totally candid. Reviewer bias and the pirating of authors' ideas are, in my opinion, much less of a concern than the timid, perfunctory review likely to result when a reviewer tries to avoid offending an author he knows. I believe that an alert and fair-minded editor is the author's best protection against unfair treatment by the reviewer. This brings my discussion to the role of the editor, the third member of the peer review trio. The Obligations of Editorship

Editors of peer-reviewed scientific journals may view their primary function in at least two different ways. Some consider themselves to be mainly a mediator between authors and reviewers. This view, found chiefly among the editors of specialty journals, sees the reviewer as the primary decisionmaker and the editor simply as the person who counts the reviewers' votes and ensures that the author carries out the changes they recommend. This is appropriate when all the papers are in one particular specialty, the decisions made primarily on the basis of technical considerations, and the editor serves on a part-time basis and therefore does not have the time to become too deeply involved with each submission. The other view of the editor's role is the one projected chiefly by the editors of general medical journals, that is, journals that publish research reports in all fields, as well as a variety of other kinds of articles, and have a broad readership. The editors of general journals usually work full-time at their job and they take a more active role in the selection of material than do most specialty journal editors. They use reviewers as consultants and advisers on the research reports, depending on them mainly for scientific advice, but relying on their own judgment and that of their in-house editorial staff to choose among the technically acceptable scientific reports. As the editor of a general medical journal, my views on the obligations of editorship naturally reflect experience in this second type of editorial role.

With respect to the publication of research articles, the editor's responsibility is to see that the peer review process works properly. In doing so, the editor serves the interests of his readers and of scientists in general. Ultimately, it is society that benefits from the editor's labors. His task is to ensure, to the best of his ability, that authors and reviewers meet their responsibilities and that the selection and editing of articles proceeds in accordance with established policies and standards. The latter, presumably approved by the scientific organization or board responsible for the journal, should be clearly stated so that all concerned understand the rules of the game. The editor and his staff must have the independence and authority to make individual editorial decisions, but the general policies under which they operate must be known and should have broad support-not only from those ultimately responsible for the journal, but from contributors, reviewers, and readers.

Editors of general journals cannot escape the obligation to make decisions. Sometimes editorial decisions virtually make themselves because the material is clearly inappropriate for the journal, or because the reviewers are unanimous in pointing out clear-cut, fatal technical flaws. More often, even with purely scientific articles, the issues are not as stark and the opinions of the experts are not unanimous. Frequently, the editor must make qualitative judgments that do not depend on scientific and technical questions but on such issues as general interest, readability, and appropriateness. Faced with these questions, the editor could poll his reviewers and staff and let the majority prevail, but the result would be an erratic and unpredictable editorial policy and a journal with no character. In my view, the journal is better and the contributors more satisfied if the editor is responsible for the editorial decisions.

Firm editorial authority works well, provided certain conditions are met. The editor must have readily available to him expert help from the reviewers, as well as sound counsel from his editorial staff He should make his own decisions, but only after he has carefully weighed what his advisers have to say. in resolving disputes, the editor should be sure that he has considered all sides and given everyone concerned a fair chance to be heard. He should also be careful to avoid showing any favoritism; he must not give special treatment to friends or associates, or to anyone else. Every manuscript must be judged on its merits, regardless of its source. To avoid any appearance of bias, an editor should have no personal economic ties that might influence his handling of a manuscript and it would be advisable for him to hold his key editorial staff members to the same standard.

Editorial decisions should be clear and consistent. When a paper is rejected purely for reasons of editorial choice (usually without outside review) the editor has no obligation to give the author any explanation beyond the statement that the manuscript was not considered appropriate. However, when a paper is rejected for technical reasons, after scientific review, the author is entitled to an adequate explanation of those reasons. The editor should be willing to consider appeals from rejected authors who believe they can answer the reviewer's criticisms or think that the criticisms are unfair. Such appeals may warrant a fresh review. When a paper is rejected only provisionally and the author is invited to submit a revised version for re-review, the editor must see that the critique is unambiguous and definitive-in the sense that it tells the author all he needs to know to prepare a potentially acceptable new version. Furthermore, the editor should ensure that the critique is consistent. Authors quite rightly object to confronting new or changing criticisms when they submit their revision in response to the initial review. They also object to receiving contradictory advice from the reviewers. It is the editor's job to review the reviewers' comments, making sure they are consistent and reasonable and that they adequately guide the author in preparing a revision.

What are the editor's responsibilities in dealing with suspected fraud or other serious misconduct? I have already said that fraud is difficult to detect through the review process, but it does occasionally happen that reviewers or editors are led to suspect malfeasance through irregularities in a manuscript. When serious allegations or suspicions of fraud are raised, the editor should contact the appropriate authorities at the institution where the author works and leave the investigation to them. The editor cannot and should not be a detective or a judge. He has neither the authority nor the resources to conduct investigations or hold hearings. The editor does, however, have a responsibility to make sure that the investigation is settled as soon as possible. If the inquiry involves a manuscript under consideration, the review process should be suspended until the issue is resolved. If the manuscript in question has already been published, and the author is found guilty of fraud, the editor must be prepared to publish a prompt retraction. Fortunately, such retractions are required only rarely. They are quite different from the corrections of minor typographical or factual error which editors publish often, always with the author's approval and often at his request.

The peer review system I have described has come under criticism recently for being subjective, susceptible to error and bias, and ineffective at screening out fraud. (5) These complaints are to a degree true, although beside the point. No doubt efforts at improvement should be made, but the basic system enjoys wide acceptance simply because no one has ever come up with anything better. Infallibility is not to be expected, but so long as author, reviewer, and editor conscientiously do their best the results will usually be acceptable. References (1) "The Journal's Peer-Review Process," New England Journal of Medicine 321 (1989), 83739. (2) Arnold S. Relman, "Fraud in Science: Causes and Remedies," Scientific American April 1989),126. (3) Konrad Bloch and Arnold S. Relman, "Preventing Fraud," Science 245 (1989), 1436. Marcia Angell and Arnold S. Relman, "Redundant Publication," New England Journal of Medicine 321 (1989),1212-14. (5) Arnold S. Relman and Marcia Angell, "How Good is Peer Review?," New England Journal of Medicine 321 (1989), 827-29. Arnold S. Relman is editor-in-chief of the New England Journal of Medicine.
COPYRIGHT 1990 Hastings Center
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1990 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Relman, Arnold S.
Publication:The Hastings Center Report
Date:May 1, 1990
Previous Article:The offensive-defensive distinction in military biological research.
Next Article:Introduction. The troubled middle in medias res.

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters