DANTE'S INFERNO: SEVEN DEADLY SINS IN SCIENTIFIC PUBLISHING AND HOW TO AVOID THEM THOMAS F. BABOR AND THOMAS MCGOVERN Relinquish all hope, ye who enter here. Dante Alighieri, Inferno, Canto III, 9 More than 700 years ago, Dante Alighieri (1265-1321) wrote an epic poem about a man's journey through the afterworlds of Hell, Purgatory and Heaven. In his Divine Comedy, he catalogued the vices and virtues of people who had passed into those spiritual domains, in part to provide a valuable insight to us, the living. Dante described hell as a very unhappy and inhospitable place that had different levels ranging from the blazing inferno of the eternally damned to a rather benign area, called the First Circle, which was reserved for worthy individuals who were born before the world was redeemed and therefore could not enter the gates of heaven. Within this general metaphor, this chapter will take the reader on an educational journey through the various levels of scientific misconduct. Our purpose is not to scare the fear of God into the gentle hearts of our readers. Rather, like Dante's journey through the netherworld, it is important for us to see the mortal consequences of scientific misconduct so that we can learn how to avoid them. Box 1 shows the seven types of misconduct this chapter explores. In addition to describing these various 'sins' and the people who commit them, we also discuss their relative seriousness, the punishments that can result, and how to prevent these kinds of problems before they arise. In the chapter following this one, we discuss the same issues within a framework of ethical decision-making, using case studies to illustrate each topic. The first issue is carelessness, exemplified by citation bias, understating the accomplishments or findings of others, and neglecting to reference findings that an informed reader would need to know to interpret the author's conclusions. In its most benign form, this problem consists of a failure to read and understand the articles one cites. A more serious offence is the distortion of others' work so that their ideas or findings support a preconceived point of view that the author is trying to advance. Carelessness can also be manifested in poor management or inaccurate presentation of data. The second ethical issue is redundant publication, which occurs when two or more papers share any of the same data without full cross-referencing. The third issue we consider is unfair or irresponsible authorship. According to the Committee on Publication Ethics and similar organizations, all persons named as authors should have made a major contribution to the work, not just a token contribution. 155 Failure to declare a conflict of interest is the fourth ethical issue considered in this chapter. A conflict of interest is a situation or relationship in which professional, personal, or financial considerations could be seen by a fair-minded person as potentially in conflict with the researcher's or author's independence of judgment. The fifth ethical violation is the failure to conform to minimum standards of protection for animal subjects or human research participants. The latter includes confidentiality of patient records and other data, informed consent, and proper explanation of the risks of research participation. Abiding by standards set by national and institutional boards for the protection of animal or human subjects is an important aspect of research under this rubric. BOX 1 THE CIRCLES OF HELL IN RELATION TO THE SINS AND PUNISHMENTS OF THOSE WHO ENGAGE IN PUBLISHING MISCONDUCT CIRCLE SIN EXAMPLES PUNISHMENTS CIRCLE SIN EXAMPLES PUNISHMENTS 1 Carelessness Citation bias, Request for correction, understatement, letter to editor negligence 2 Redundant publication Same tables or literature Rejection of manuscript, reported without noting Copyright infringement prior source 3 Unfair authorship credit Failure to include eligible Angry colleagues, authors; honorary authors complaints to editor or employer 4 Undeclared conflict of Failure to cite funding Letter to editor, public interest source apology 5 Human/ animal subject No ethical approval Rejection of manuscript, violations notification of employer 6 Plagiarism Reproducing others' work Retraction of manuscript,, or ideas as one's own notification of employer 7 Scientific fraud Fabrication or Retraction of manuscript, falsification of data; notification of employer, misappropriation of publication ban others’ ideas or plans given in confidence Plagiarism is the sixth issue. Plagiarism ranges from the unreferenced use of others' published and unpublished ideas, including research grant applications, to submission under 'new' authorship of a complete paper. The final level is scientific fraud. This form of misconduct consists of the deliberate fabrication of data or the alteration of findings to make a study more credible and acceptable for publication. How prevalent are these various ethical problems among scientists? In studying the publication practices of researchers, it has been found not only that negative results are less likely to be published but that even positive findings are withheld if this is perceived as beneficial for the authors. In one study, 20% of researchers reported delaying publication of results for their own advantage (Blumenthal et al., 1997). It is likely that these and even more serious issues are often hidden from the eyes of busy editors and reviewers. For example, editors and reviewers are unlikely to detect scientific fraud in the normal editorial process because data fabrication can be easily hidden in lab records and computer files that are inaccessible during the review process. Skilled reviewers are more likely to detect plagiarism and citation bias, but there is a general suspicion that the cases of identified and provable misconduct are the tip of an iceberg that needs to be melted rather than remaining submerged. In the following sections of this chapter, each of these ethical improprieties is discussed in terms of its relative importance, possible consequences, and strategies for avoidance. Box 2 provides definitions of the various types of ethical problems discussed in the chapter. NEGLIGENT CARELESSNESS AND CITATION BIAS The First Circle of Hell described in Box 1 is reserved for minor forms of negligent carelessness and citation bias that are likely to mislead readers and distort the value of scientific research. Perhaps the most benign and most prevalent form of ethical impropriety, negligent carelessness, is characterized by such deficiencies as a failure to adequately review the literature on a topic, lack of candour or completeness in describing one's research methods, or presentation of data that are based on faulty statistical analyses. A related problem occurs when an author cites articles taken from other reports or from published abstracts without having read the primary sources. A more serious form of carelessness in scientific writing is citation bias. One variety of this bias is the selective citation of only those articles that support a particular point of view, ignoring or understating the importance of articles that contradict that viewpoint. The intention to deceive others makes this practice especially reprehensible. Another form of citation bias is selective citation to enhance one's reputation, epitomized by self-citation. Box 2 DEFINITIONS OF TERMS REFERRING TO VARIOUS FORMS OF SCIENTIFIC MISCONDUCT Citation bias A form of carelessness that ranges from a rather benign failure to read the articles one is citing to the distortion of the meaning of others' work. Copyright The legal right granted to an author, publisher, or distributor to exclusive publication, production, sale, or distribution of a scientific work. Divided publication Information from a single research study is divided for publication in two or more papers. Also called "salami science.” Duplicate publication Re-publication of the same information in two different places Fabrication Presenting data in a research report that have not been obtained in the manner or by the methods described in the report. Fractionally divided publication Reporting in a single paper only a fraction of the data that have been or will be reported in their entirety in another paper Misappropriation Illicitly presenting or using in one's own name an original research idea, plan or finding disclosed in confidence. Misrepresentation (falsification) of findings Altering or presenting original findings in a way that distorts the result in a scientifically unjustified way, or by omitting results or data pertinent to conclusions. Partial repetitive publication Repeatedly publishing parts of the same information in modified form. Plagiarism To present someone else's manuscript, article, text, or idea as one's own. Repetitive publication Publishing the same information two or more times (e.g., in journal articles and book chapters). Self-plagiarism To copy and present one's own text or article without properly attributing its original source. Unethical authorship Authorship which violates the principle that all persons named as authors should have made a major contribution to the work reported and be prepared to take public responsibility for it. CONSEQUENCES If the effect of these practices is to mislead or misinform the reader, then they are considered a mild form of scientific misconduct, especially if they only occur at the drafting stage when they are often detected by observant colleagues or reviewers who are likely to request a more balanced literature review, or the correction of obvious mistakes. In some cases an editor may reject an otherwise acceptable manuscript if reviewers raise questions about the author's objectivity or intellectual sloppiness. The consequences could be more serious if carelessness or citation bias is detected only after the article is published. If readers of a published article detect a statistical mistake, a clear bias in the formulation of a research question or the selective reporting of the literature, they may write letters to the editor pointing out the problem. Editors in turn may ask for corrections to the text or the data analyses, which are subsequently published as a special note to readers. Beyond these embarrassing consequences, failure to cite relevant studies and bias in the interpretation of previous research are likely to create a negative impression of the author among his or her colleagues. Another negative consequence may be opprobrium and damage of reputation experienced by the institution with which the author is affiliated. PREVENTION The best way to avoid these problems is to follow appropriate citation practices, conduct a thorough review of the literature, read all of the articles you cite, present research findings accurately, and interpret them objectively. Authors who collaborate on multi-authored papers have a special responsibility to read all drafts of a manuscript with extreme care to make sure that these problems are detected during the early stages of the publication process. Even when several authors divide responsibility for writing different sections of a research report, authors should always check each others' work. REDUNDANT PUBLICATION Authors wishing to reach the widest possible audience, or a variety of specific audiences, may seek to report a single definable body of research in more than one paper, in repeated reports of the same work, in fractional reports, or in reports in more than one language (Huth 1986). But there are also less noble motives for redundant publication, including the desire for multiple publications to enhance one's reputation. Redundant publication occurs when two or more papers, without full cross-reference, share the same data (Committee on Publication Ethics 2001). According to the Editorial Policy Committee of the Council of Biology Editors, repetitive publication refers to "publication of essentially the same study more than once without clear notification of editors, reviewers, and other appropriate parties" (see Jerrells 2001). In general, journal editors expect authors to ensure that no significant part of the submitted material has been published previously and that the article is not concurrently being considered by another journal. Repetitive publication has become such a serious problem that many journals now require authors to state in writing whether the data have been previously reported in part or in whole. As indicated in Box 2, a number of different terms have been used to describe this phenomenon. Although there are some important differences among prior, duplicate, repetitive, fragmented, and redundant publication, they are all part of a common problem. Redundant publication and its variants consume valuable resources that otherwise might be devoted to other authors who are publishing original data or ideas. Because of limited journal space, the publication of one person's paper means that another's paper will be rejected. If there are questions about the extent of the overlap between two articles, editors and reviewers need to take extra time to review several publications to determine the extent of redundancy and whether it violates any copyright agreements. Regardless of whether the repetition occurs with data or ideas (e.g., repetitive review articles), the information from duplicated sources is sometimes inadvertently cited in a way that implies that the findings or conclusions are independent of each other, when in fact they are based on the same source. Without proper identification, meta-analyses and review papers may come to biased conclusions because the effect of a given finding is multiplied or distorted. As Huth (1986) has noted, some types of repetitive publication are legitimate and should not be considered scientific misconduct. This is particularly the case in the publications associated with large datasets that involve multiple investigators across many sites. Often, the collaborating investigators have included measures related to a particular hypothesis or methodology, which could and should be reported in separate articles even though the article presents the same subjects, methods, procedures and even some of the same data as other articles. Such publications may be intended to highlight the relevance of particular clinical findings for a particular audience, especially if they have been first published in a technical journal that did not permit the reporting of particular findings or the discussion of clinical implications. Another acceptable variant is publication of the same article, often in its entirety, in two different languages. It is also acceptable to re-publish ideas, data, or review findings when journal editors or book editors request that a popular author write a topical review or commentary for their publication. As long as the author tells the editor about previously published material and cites all relevant reports in the commissioned paper, it is generally acceptable to re-publish or update such material. A special case of redundant publication is 'self-plagiarism'. According to Griffin (1991), this occurs when an author re-uses text from a previously published paper in a way that fails to give proper acknowledgement to its source and its owner. By 'owner' is meant the person or organization that owns the copyright (see Box 2 for definition), which is often the publisher of the previous version of the borrowed text, not the original author. This problem typically occurs when authors re-use text from a literature review or the Methods section of an article either without changing the wording or by quoting the original text. Unlike the re-use or re-publication of original data, self-plagiarism is something that is more the result of laziness than dishonesty. It can also be a form of self-aggrandizement. CONSEQUENCES If a duplicate publication constitutes a copyright infringement, it may result in a reprimand for the author, a retraction of the article, or an apology to the journal editors and the publishers involved. If editors are embarrassed by the need to publish a retraction, they may adopt policies and regulations that place an added burden on honest authors who follow the rules. And when instances of scientific misconduct like this are reported to the public, they diminish the reputation of scientists and their work. In general, an author is not allowed to re-use previously published material when the rights have been assigned to the publisher, which occurs in most instances of scientific journal publications. Reprinting more than one or two sentences verbatim without proper attribution may constitute a violation of copyright and could result in legal sanctions, although this rarely occurs in cases of minor copyright violations. Editors are unlikely to consider small amounts of 'borrowing' to be a major problem, but if an observant reviewer detects widespread self-plagiarism, the editor may reject the article. Nevertheless, the more that authors re-use text without proper quotation or attribution, the more they risk adverse consequences from editors and publishers, ranging from a reprimand to legal action for copyright violation. PREVENTION Authors of overlapping papers would be seriously remiss in failing to cite their previously published work (see Jerrells 2001 for a discussion of this problem). When there is any possibility of repetitive publication, authors must notify editors to explain the connection between the current paper and its predecessors. Ideally, the author should submit all related publications to the editor along with an explanation of the potential overlap and the reasons for the new report. Second, all versions of related papers must contain appropriate citations and complete references to the related papers so that readers and editors can evaluate the implications of the repetition and overlap. This includes citing illustrations or tables reprinted or adapted from other journals. The publication of the same paper in two different languages is acceptable when the editors of both journals agree to it and when the second version cites the previous version as the primary publication. Regarding self-plagiarism, set off short quotations from a previously published article in quotation marks and cite the original version. Permission must be requested from the publisher or other copyright holder when large sections are reproduced. When there is a need to repeat the information contained in a previously published literature review or a Methods section, the best solution is to change some of the wording in each sentence and to refer the reader to relevant sources for previously published material (e.g., 'As discussed in our previous report [Bloggins 1899]'). UNFAIR AUTHORSHIP Authorship of a scientific report refers not only to the writing of a manuscript, but also to the origin of a writing project, any experimentation or other research connected with it, and the substantive kinds of work that led up to it. 161 According to the Committee on Publication Ethics 2001) and other practice codes, all persons named as authors should have made a major contribution to the work reported and be prepared to take public responsibility for its contents (in proportion to the credit they claim on the author list). An editorial (Editor, 1982, p. 613) in the Annals of Internal Medicine defines relevant terms as follows: Responsibility means the ability and willingness to defend the content of the paper if it is challenged by readers. Public means that authors are willing to carry out this responsibility in a published defense, such as a signed letter to the editor; private defense in private correspondence would not reach the scientific public. Content means not simply packages of data but also the conceptual framework on which they are hung: the justification for a study or clinical observations; the basis for the study design; methods for collection of valid data: the analysis and interpretation of the data; and the logic that led to the conclusions. There are a number of ways in which authorship decisions can result in ethical improprieties. First, some persons who have made significant contributions to an article may not receive sufficient credit, or may receive no credit at all. This occurs when an article is drafted without the knowledge or consent of someone who made a substantive contribution earlier in the process. It also occurs when a decision to list the order of contributions is not made fairly with the full agreement of the co-authors, as when a major contributor is listed after a minor contributor to enhance the ego or career of the minor contributor. Another instance of inappropriate credit occurs when a coauthor, such as a science writer, is not listed because the research group might be embarrassed to admit that someone else wrote the paper, such as a science writer hired by a drug company to expedite the publication of favorable findings. This is called 'ghost' authorship because the author's identity is unknown to those who read the article. A second type of authorship problem arises when some persons are listed as co-authors even though they made no substantive contribution to the article or the research. A common example is the practice of listing the head of a department or a research centre director, often at the end of the authors' list, a custom known as gratuitous, honorary or gift authorship. Again, in the light of this practice, one must question the ethical climate in research settings that allow such behaviour to occur. Ethical guidelines, appropriately crafted and implemented, might deter such transgressions. Between these two extremes, there are a number of related infractions, such as the failure to give proper recognition to a person's contribution by listing him or her inappropriately low in the author list, or the tendency to award co-authorship for minor contributions based on personal or political considerations. Our purpose here is to discuss the seriousness and consequences of this type of misconduct, and to summarize the steps that can be taken to prevent its occurrence. CONSEQUENCES Authorship credits may be one of the most contentious issues in scientific publishing. At the level of collaborating research groups, the consequences range from hurt feelings to formal complaints made to a scientist's unit director or institutional authority. In between these extremes there are likely to be recriminations, perceptions of unfairness, and poisoned working relationships, which could damage the reputations of some of the parties involved. When instances of unfair authorship credit are detected, the editor's response could range from the rejection of a pending manuscript to the call for a correction to a published paper. But these matters rarely come to the attention of editors unless there is a case of scientific fraud, where co-authors might claim that they were not sufficiently involved in the writing of the paper to detect the fabrication in the first place. PREVENTION How can authors best deal with ethical issues related to authorship? Early agreement on the precise roles of the contributors and collaborators, and on matters of authorship and publication, is advised. The lead author should periodically review the status of authorship credits within a designated working group by having open discussions of substantive contributions with all prospective collaborators. In order to avoid disputes, distribute and discuss authorship guidelines with all potential collaborators on a manuscript. An open discussion, related to authorship, should be on the academic agenda of research centres. Involving an institutional ethics committee in drawing up institutional guidelines might also be helpful. Open and ongoing conversation about these issues, combined with institutional policies, is the best way to avoid problems. UNDECLARED CONFLICT OF INTEREST A conflict of interest is a situation or relationship in which professional, personal, or financial considerations could be seen by a fair-minded person as potentially in conflict with independence of judgment (ISAJE 1997). One way to determine whether a conflict of interest exists is to ask the question: If the situation or relationship were revealed to the editor or the reader only after the paper was published, would it make a reasonable person feel misled or deceived? Conflict of interest is not in itself wrongdoing. It is the failure to declare real or potential conflicts to an editor, one's coauthors, and the readers of a paper that constitutes scientific misconduct, to the extent that potential conflicts are very important in the evaluation of any piece of scientific work. There are three levels of conflict of interest: real, apparent, and potential. A real conflict of interest means that the author, or the administrative unit with which the author has an employment relationship, has a financial or other interest that could unduly influence the author's position with respect to the subject matter being considered. An apparent conflict of interest exists when an interest would not necessarily influence the author but could prompt others to question the author's objectivity. A potential conflict of interest exists when there is an interest or relationship that would make a reasonable person uncertain or suspicious whether or not it should be reported. Conflicts of interest can be financial, personal, political, or academic. Financial interests include employment, research funding, stock or share ownership, payment for lectures or travel, consultancies, and company support for staff (Committee on Publication Ethics 2001). These kinds of conflict are most often discussed in ethics codes and reports on research integrity because they are easier to document and quantify. In addition to financial and commercial conflicts of interest, a conflict may be personal, political, or academic. Personal conflicts might include a vendetta against another researcher who is disliked by the author of an article. Political conflicts exist when researchers distort their findings or interpretation to conform to 'politically correct' ideas or ideology. Academic conflicts include the attempt to validate 'pet' theories supporting one's own ideas. These kinds of conflict are difficult to detect, but they should nevertheless be considered by authors when evaluating their own work. Authors in the past have been provided with little guidance in evaluating and responding appropriately to issues of conflict of interest. The existence of compliance offices in research settings is helpful, but these institutions of themselves will not solve the problem. Researchers need appropriate training about the ethical dimensions involved as well as opportunities for ongoing dialogue and conversation, in order to develop informed individuals and groups (Institute of Medicine 2002). The potential for conflict of interest in the is enhanced by any relationship or funding connected with commercial or industrial sources like the tobacco industry, the food industry, for-profit health care systems, private hospitals, the pharmaceutical industry, and 'social aspect organizations' that receive their primary support from these sources. For example, in the search for drugs that may be used to treat chronic diseases like diabetes, scientists involved in research on a particular pharmaceutical product may have financial ties with companies that have a business interest in that product. Some industries like the tobacco companies have funded researchers to conduct studies of the health consequences of their products. Sometimes the industry funds studies directly; at other times it funds studies indirectly through 'social aspect' organizations legal firms that receive their support from industry sources. In addition to research funding, industry ties can include paid consultancies, conference presentations, stockholding, advisory board membership, and patent holding. Two major questions regarding the need for conflict of interest policies and precautions are whether industry funding affects the quality and eventual publication of research, and whether the effect is deleterious. Bias toward 'positive' results may exist even among articles where financial ties to industry are disclosed (Cho 1998). For example, pharmaceutical industry-supported drug studies are significantly more likely to report 'positive' findings (i.e., the manufacturer-associated drug better than placebo) than nonindustry funded studies (Stelfox et al. 1998). Another risk is publication bias, in which industry-favourable studies are more likely to get published than unfavourable ones. There are several possible mechanisms to explain how conflicts, especially those connected with industry ties, may lead to publication bias (see Cho 1998). One is suppression of publication, whereby negative findings are not published because either the author fears loss of funding from industry sponsors or the industry itself imposes restrictions on publication. Another mechanism is self-selection or industry selection of researchers who are more likely to get positive results. CONSEQUENCES One consequence of competing financial interests is the possible limitation of publication options. Although most journals do not ban publication of articles because of their authors' financial interests, some journals have now begun to prohibit authors of editorials and review articles from publishing if the author has a substantial financial interest in the product discussed in the editorial or review (Relman 1990). This policy does not apply to authors of scientific reports that present original data. Undeclared conflicts of interest, when detected, may have serious consequences, such as the rejection of a pending article, the retraction of a published article or the author's need to publish an apology. A more subtle effect of real or apparent conflict of interest is the perception by one's scientific colleagues that one's scientific work is biased because of a personal or financial interest. PREVENTION According to Loue (2000), the best way to avoid problems associated with potential conflict of interest is self-elimination from participation in potentially conflicting activities.. Arrangements with industry can be particularly problematic when the researcher is asked to sign a restrictive contract regarding the ownership of data, the sponsor's control of the data, and the investigator's right to publish it. Many academic institutions have rules governing financial support for faculty activities. These rules describe when faculty must disclose particular interests, and when they must divest themselves of particular financial interests. Conflict of Interest Committees, when they operate as part of Ethical Review Committees, are a part of institutional compliance oversight, and hold promise in this respect. In addition, many professional societies and associations, such as the Association of American Medical Colleges (AAMC), the American Association of Universities (AAU), and the Council of Government Relations (COGR), have developed policies on conflicts of interest. Authors should pay close attention to the guidelines issued by these committees and associations. The scientific community has issued warnings about the advisability of accepting any funding from several industries, and has suggested rigorous adherence to voluntary ethical codes when such funding is accepted. Even when these guidelines have been followed appropriately, however, each author should declare to the editor any real, potential or apparent conflict of interest with respect to his involvement in a particular publication. Each author should declare conflicts between (a) commercial entities and the participant personally; and (b) commercial entities and the administrative unit with which the participant has an employment relationship. Authors should also declare sources of funding for a study, review, or other publication in a way that can be clearly understood by the reader. A footnote or an acknowledgement is the most appropriate mechanism. Describe funding sources in sufficient detail so that an average reader can recognize potential conflicts of interest. If a funding source is a social aspect organization with an ambiguous name such as 'The Tobacco and Health Fund', the reader should be informed that, for example, the organization is supported by a group of tobacco companies. Disclosure alone will not necessarily eliminate publication bias. Researchers who are serious about avoiding even the appearance of conflict of interest are advised to dilute the conflicting relationship by getting funding from both industry and non-industry sources, and by refusing to sign industry agreements that do not guarantee the researcher's right to publish the results regardless of the study's outcome. Other management strategies include avoiding additional financial ties that are not absolutely necessary to the pursuit of the research, such as the acceptance of advisory board memberships, stock options, or consulting fees from companies sponsoring research (Cho et al. 2001). HUMAN/ANIMAL SUBJECTS VIOLATIONS Research involving human and animal subjects has been conducted for over a century. During this period regulations governing human and animal experimentation have developed into a very complex set of procedures that are typically governed by appointed committees located at institutions involved in biomedical research. These procedures include ethical review of research protocols, safety monitoring of animals and human research participants, and informed consent requirements for human participants. These procedures were developed out of concern for the rights of research participants following a series of well-publicized medical experiments in which human subjects were exposed to harmful agents or had effective treatments withheld without their knowledge or consent (Loue 2000). It has now become customary, if not mandatory, to submit proposed research for independent review by an Ethical Research Committee to determine its ethical acceptability from the perspective of the local community and the researcher's institution (Institute of Medicine 2003). Such boards focus primarily on the protection of research participants by assuring that the study's procedures minimize risks of unwarranted harm to participants. It is becoming increasingly recognized that scientific journals have an important role to play in the protection of human and animal research subjects. Journals are responsible for the dissemination of research findings. They "are obligated to publish research that meets high ethical standards […] for which the authors have attested to their compliance with regulatory ethical standards" (Institute of Medicine 2003, p. 205). Most journals that publish research dealing with human research participants have implemented policies requiring authors to certify compliance with ethical informed consent procedures. CONSEQUENCES Failure to follow recommended or required journal procedures regarding protection of human and animal research subjects could have several important consequences. Although most journals do not ban publication of articles because they have not been submitted for ethical review, some journals now require authors to state whether their research conforms to the minimum standards outlined in the Helsinki Declaration. In particular, social and behavioural research such as survey studies and research involving archival records may not require stringent informed consent procedures. However, it would be an error to rely on this perception: surveys, on occasion, have resulted in significant harm to individuals and to institutions. It is safer to submit all research for institutional review, letting the committee decide whether the researcher is exempt or not. Failure to obtain ethical approvals or informed consent from research participants may lead an editor to question the purpose and value of the research, and could result in a decision not to send the manuscript out for review or, when the failure is detected during peer review, to decline the manuscript. Another consequence could be the notification of an official from the author's institution. PREVENTION It is always wise to mention both in the cover letter to the editor and in the text of a submitted manuscript that the researchers have followed appropriate ethical review procedures. If there are any questions regarding the applicability of human subjects requirements, these should be raised with the editor in the cover letter or in a telephone call or email message prior to submission of a manuscript. Often these questions can be resolved by consulting the journal's instructions to authors or website. The International Committee of Medical Journal Editors has provided the following guidance regarding ethical issues: