Abstract
Keywords
Despite an explosion of interest in psychiatric ethics generally (e.g. [1, 2]), ethical aspects of publishing in psychiatry have received scant attention. The huge expansion of psychiatric literature and radically new methods of disseminating information [3] make it desirable to consider ethical problems in this context. Our aims are to examine these problems, using cases on public record to highlight the damage that can ensue, and to map out strategies for their resolution. We have conducted a comprehensive search of medical databases (Index Medicus/MEDLINE, EMBASE/Excerpta Medica and PsycLIT) and relevant websites (e.g. [4–8]), and conferred with experts in the fields of ethics, research, publishing and law.
Historical perspective
Awareness of ethical dilemmas in publishing has only evolved since the late 1970s. One response then was the formation of the International Committee of Medical Journal Editors (ICMJE), the Vancouver Group, which, inter alia, began to highlight the ethical dimension of publishing [9]. Other bodies have followed more recently. For example, in the 1990s the Committee on Publication Ethics (COPE) was established in the UK by a group of leading medical editors [10] and the World Association of Medical Editors was launched in Italy [7]; one aim of the latter group is to ‘achieve the highest level of ethical medical journalism’ [7]. Some publishers and specialist medical journals (e.g. in anaesthetics, cardiology, endocrinology and gastroenterology) have united to tackle ethical problems and to release corresponding position statements [11, 12]. International congresses on biomedical publishing have taken place in 1989, 1993 and 1997 [13–15] with ethical themes broached at each of them.
Ethical problems
Ethical problems can occur in any phase or aspect of publishing. Redundant publication, plagiarism, publication of fraudulent or inhumane research, and issues concerning authorship, use of pejorative language, conflict of interest and bias inherent in the publishing process have garnered recent attention. Much of this literature pertains to what might be termed ‘publishing misconduct’. Surprisingly, recorded cases of such misconduct in psychiatry are few. For example, among all journals indexed by Medline, 261 papers in the period 1983–1999 were retracted, of which only two were psychiatric. Whether this paucity in psychiatry reflects a truly low rate and/or a failure to recognise or report publishing misconduct is unknown. We now deal with each major issue in turn.
Redundant publication
This practice is variously termed fragmented, duplicate, multiple and ‘salami’ publication, and publishing in ‘least publishable units’ [16]. Its essence is producing more articles from ‘data’ than is objectively necessary. Redundant publication is undesirable in that it inflates curriculum vitae deceptively, squanders journal space and may distort meta-analyses. On the other hand, it may be warranted in, for example, republishing a ‘classic’ paper or a journal article as a book chapter, or an article in a journal whose readers may not have ready access to the original [17], provided there is acknowledgement of the source material. In these situations, ‘redundant publication’ is a misnomer.
We have identified four cases of redundant publication in psychiatry that have attracted editorial comment. One of these concerned two papers, with much overlapping data, on postpartum psychiatric illness [18–21]. In two further cases, relating to antidepressant trials [22–29], identical (or near-identical) papers were published by a research group, without the editors' knowledge.
In redundant publication, an author usually duplicates partly or wholly what is already published, submitted or in preparation. The fourth case of redundant publication in the psychiatric literature concerns authors failing to declare discrepant material. In 1978 a paper in the New England Journal of Medicine (NEJM) reported that platelet monoamine oxidase levels were lower in schizophrenic patients compared with controls [30]. A correspondent [31] noted that a publication around the same time in the American Journal of Psychiatry [32], having two authors in common with the NEJM paper, contradicted the NEJM findings. The latter's contributors defended their failure to refer to the conflicting data given their policy ‘not to discuss unpublished data in a published paper’ [33]; censure by the NEJM editor followed [34].
Redundant publication illustrates how a lapse in ethical publishing tends to violate more than one principle. An author's integrity may be questioned in that similar (or discrepant) work is not declared when material is submitted for publication. Steinbok [35] comments that, in the context of ‘distributive justice’, redundant publication wastes journal space, clutters databases and abuses referees', editors', readers' and indexers' time. Harm may also be done to a journal's reputation, the host institution and unsuspecting co-authors.
Plagiarism
The psychiatric literature has not been spared in relation to plagiarism. We have come across four cases. A professor of psychiatry resigned following his detection as a plagiariser of large sections of four review articles in the 1960s and 1970s [36, 37]. It was apparently his practice, in talks, to quote extensively from colleagues' publications, but to omit appropriate attribution when he converted the addresses into articles [36]. In another case, a postgraduate student plagiarised an entire paper on the differential diagnosis of dementia [38, 39]. In a third case, a paper on repressed memories [40] included a passage virtually identical to paragraphs from a book published the previous year [41]; an apology was subsequently offered [42]. In the fourth case, a section of an article on insulin binding in anorexia nervosa [43] was copied by a referee of the paper [44]; the plagiarist, on exposure, returned to his native country, his supervisor forfeited an academic appointment, and the person whose work had been copied, embittered by the experience, abandoned research altogether [45, 46].
Publication of fraudulent research
Reporting falsified data is another serious form of publishing misconduct. The most notorious case in psychiatry was an American researcher [37, 47] who published papers in prestigious journals (e.g.[48]) which purported to show that, in mentally retarded children, tranquilisers were less effective and produced more side-effects than stimulants and also decreased the children's response to behavioural treatment. The studies shaped attitudes towards the use of psychotropics in this group [49], even leading some American States to alter management policy [47]. In fact, the studies had never been carried out. The perpetrator's court sentence included 250 h of community service and 5 years of probation. He was also required to repay grant money and prohibited from professional practice for several years. Unsuspecting co-authors were harmed. As one remarked [50], ‘In retrospect, every hour that I spent collaborating [with him] was worse than time wasted. Our interactions have tarnished my reputation, and caused me pain’. The case is a landmark in that it was probably the first with major implications for patients and where criminal prosecution ensued [37]. One ‘positive’ of the unsavoury affair was a dramatic decline in citations to the offending articles once the articles were exposed, suggesting that researchers can purge the literature of falsified work [47].
Publishing inhumane research
It is far from clear that a life-saving discovery made by sadistic scientists applying grotesquely immoral procedures should never be communicated to the world. However, we suggest that inhumane research ought not be published, irrespective of its scientific merits. The Declaration of Helsinki on biomedical research includes a statement that ‘reports of experimentation not in accordance with the principle [in this Declaration] should not be accepted for publication’ [51]. In any event, the quality of inhumane studies is often poor [52], rendering them unethical from a second standpoint.
Authorship and its vagaries
Researchers regard authorship as a reward for the ‘blood, sweat and tears’ of their toil. Given this value, the essence of authorship has been keenly debated and its potential abuse noted. What in fact constitutes authorship? The Vancouver Group [9] suggests that it should be based on a ‘substantial contribution’ in each of the following: (i) conception and design, or analysis and interpretation, of data; (ii) drafting the article, or revising it critically for its intellectual content; and (iii) approval of the version to be submitted. These elements have not been uniformly accepted, even by members of the Group itself. The British Medical Journal (BMJ), for example, argues that the criteria have ‘serious flaws’ in not advocating the need to declare who has contributed what to a study [8]. The BMJ has thus contemplated replacing authorship with a system of ‘contributorship’ in which, at the end of a paper, details are provided of who did what in planning, conducting and reporting the work. Whether ‘contributorship’ will be adopted widely and be preferable to ‘authorship’ are open questions.
Although controversy prevails about the nature of authorship, when it should be unequivocally withheld is usually straightforward. Thus, ‘guest’ (also known as ‘honorary’ or ‘gift’), ‘planted’ and ‘ghost’ authorship are all considered taboo. A guest author is a person (commonly a head of department) who does not fulfil authorship requirements, but is knowingly listed as an author. Planted authorship refers to authorship conferred without knowledge of the person involved. In ghost authorship a person with a right to authorship has declined that position.
The order of authorship warrants brief mention. Firstnamed authorship generally carries most prestige. Readers may also believe that he or she has assumed greater responsibility than co-authors for the published work. This premise was quashed, however, a few years ago when an article on dementia had to be retracted [53, 54] due to unsound methodology and the host university subsequently apportioned blame to all listed authors.
Entitlement to, and ordering of, authorship is best determined at the outset of any work, although the decisions may alter with time. Public airing of grievances and perceived injustices are generally not in a profession's interests. In psychiatry, fortunately, few such difficulties have occurred. Nevertheless, correspondence in Nature in 1995 [55, 56] revealed the possible oversight of an author of an article [57] on the genetics of schizophrenia.
Sensitive use of language
In writing in psychiatry, it is vital to use language sensitively and responsibly. Appellations like ‘psychotics’ (rather than ‘people with psychosis’), ‘schizophrenics’, ‘neurotics’ and ‘alcoholics’ are obviously pejorative. Invented terms may unfairly lay blame and become entrenched. ‘Schizophrenogenic mother’, coined by Frieda Fromm-Reichmann in the 1940s [58] and popular until the 1970s [59, 60], and ‘refrigerator parent’, which had currency in relation to autism [61], are notorious illustrations. Terms may harm, often unintentionally, in other ways. For example, the expression ‘adolescent turmoil’ for many years probably contributed to teenagers in need of treatment not receiving it, as they were ‘negotiating a normal stage of development’. The need to write sensitively does not negate creativity nor impede new ways of thinking. But authors should exercise restraint so that advantages outweigh any possible harm.
Conflict of interest
A conflict of interest prevails when participants in the peer review or other part of the publication process have ties to activities that could inappropriately influence their judgement [62]. Conflicts include financial affiliation with industry, close personal relationships and academic rivalry. At face value, it would seem desirable that conflicts of interest are declared so that readers can determine for themselves the effect on the material they read. To date, many journals have confined themselves to asking for disclosure of financial interests, a request not well received in all quarters. One commentator views this as the ‘new McCarthyism’ in science [63]. Another has argued that there is no evidence to indicate financial interest impairs an author's capacity to be objective, nor data to suggest financial ties are more telling than, for example, personal rivalry [64].
Inherent bias in the publishing process
Inextricably linked to conflicts of interest is the dilemma of inherent bias in the publication process. For example, editors' capacity to serve authors and readers may be subject to the bias of their preferred subjects for publication and directions in which they wish their journals to head. Similarly, referees' abilities to be fair to authors may be compromised by their theoretical orientation and own ambition to publish in the area. These biases may manifest in subtle ways, for example, referees, perhaps unconsciously, may procrastinate in returning assessments or modify their own research on the basis of a paper they have reviewed. Whether such bias can be lessened by the strategies we suggest below is unclear.
Factors contributing to publishing misconduct
Identifying factors contributing to publishing misconduct may suggest ways to curb it. Relevant literature relates more to research than publishing misconduct per se, and is not specific to psychiatry. Nevertheless, two offending groups are identifiable [37, 65]. The first are typically young, bright men studying a biological subject, publishing frenetically, working in prestigious institutions governed by a ‘publish or perish’ ethos, and who are unable to access supervision readily. Other factors promoting unethical behaviour in this group are an expectation to generate ‘positive’ data, the declining status of replication studies (which might uncover fraud) [66], and linking of publication output with academic self-esteem, tenure, promotion and securing of grants. The second group consists of relatively isolated professionals, mostly clinicians recruited by pharmaceutical companies as investigators in drug trials, who fabricate data and are then detected by astute research monitors. For example, a UK psychiatrist concocted data in an antidepressant trial and was consequently deregistered [37]. A psychiatrist in the USA who similarly fabricated results in a drug study was disqualified as an investigator by the Food and Drug Administration [46]. Fortunately, misconduct was uncovered prior to publication in both cases.
It is salutary to point out that all psychiatrists who involve themselves in some aspect of publishing are ‘at risk’ given a spectrum which covers less grave violations such as excessive self-citation, probable lack of awareness of appropriate ethical conduct, and ubiquitous institutional demands.
The way forward: possible strategies to promote ethical publishing
No single measure will ensure full awareness of ethical issues in publishing or guarantee that ethical standards are maintained or, where necessary, enhanced. Neither are immediate solutions at hand for day-to-day quandaries (e.g. sorting out authorship order). We offer the following nine suggestions – not in any hierarchy, although some may be more pertinent in particular circumstances.
Developing ethical guidelines
Comprehensive guidelines on publishing ethics are few. The most detailed are those of the American Medical Association [67]. Ethical codes of professional organisations either ignore publication entirely [68, 69] or subsume it under the rubric of research [70]. Psychiatric colleges and associations could well imitate their psychological counterparts [71, 72] which distinguish publishing from research in their ethics codes.
Teaching publishing ethics and scientific probity
Teaching publishing ethics in psychiatry is unusual. By contrast, undergraduate and postgraduate research students learn much about statistics and research methods. This gap and that in the continuing education of all psychiatrists needs tackling. Encouragingly, teaching about ‘scientific dishonesty’ and its underpinnings has been introduced for physicians and PhD students in Denmark [73] and, more recently, for doctoral students at the University of Sydney. The curriculum for such teaching should cover the topics we have raised in this article. Incorporating concrete examples of these issues together with students' active participation would seem to us to be desirable features of the teaching.
Journals should attend to publishing ethics
Psychiatric journals could adopt several initiatives. First, ‘Instructions for Authors’ could be extended to encompass ethical position statements. Second, as ethical issues arise for a journal, they should be addressed directly through editorials. Third, the profile of publishing ethics could be raised by encouraging articles. Fourth, journals could add a standard letter to authors when papers are submitted, and one to referees when requesting assessments, reminding them of ethical responsibilities. Fifth, authors acting improperly or teetering on the edge (sometimes unknowingly), could be counselled by the editor and offered, if appropriate, the means to gracefully retreat from predicaments; there should also be provision for assisting an ‘ill’ member of the publishing community. Finally, appointing a ‘journal ombudsman’ to resolve disputes, as the Lancet has done [74], may be advantageous.
Forming consortia
Just as journals in other medical specialties are forming consortia to address aspects of publishing ethics, so may it be worthwhile for psychiatric journals to do likewise. Editors of psychiatric journals would no doubt also benefit through representation on the committees of various editors' organisations.
Learned institutions discouraging wasteful publication
At some universities (e.g. Harvard, the University of California), the publication records of candidates for tenure or promotion are evaluated on the basis of a limited number of papers (e.g. up to 10) rather than the total corpus [75]. One cited reason [75] is to reduce redundant publication and other unethical publishing practices. The innovation has not been widely implemented or evaluated. A proposal to restrict an author to a maximum number of papers each year may seem Draconian, but at least raises the basic question of quality versus quantity and how quality is optimally rated (we do not underestimate the complexity of judging quality).
Harnessing information technology to promote ethical standards
Developments in information technology can enhance ethical publishing, and have already proven valuable. For example, a journal website (e.g. [8]) may contain information on publishing ethics, otherwise cumbersome to convey in every issue. As databases become readily accessible, they will be able to monitor plagiarism and redundant publication more effectively [76]. Innovative methods of refereeing are under trial, as evidenced by the Medical Journal of Australia's use of the Internet to facilitate open, reader-assisted peer review [77]. Of course, ethical repercussions, predictable and unforseen, arise with such innovative measures. Thus, for instance, Internet-based peer review may enhance transparency but, necessarily, impinges on confidentiality of the process and may lead to plagiarism.
Detection and reporting of misconduct
All those with publishing responsibility or an interest in psychiatric publishing are duty-bound to report suspected misconduct. Editors may have been loath, hitherto, to involve themselves in these delicate matters. As a former editor of the Journal of Clinical Investigation quipped [78], ‘We are the JCI, not the FBI’ or as Freedman asserted in relation to the Archives of General Psychiatry [79], ‘Our ultimate purpose… is not to catch a thief, but to apprehend useful knowledge’. While a journal's primary function is clearly not to investigate fraudulent practice, minimum duties arguably include reporting serious misconduct and issuing retraction notices or apologies following its discovery.
How are improper practices to be detected? Sophisticated statistical ways to identify fraudulent data exist [80]. The necessary mathematics is sometimes simple. For example, scepticism prevailed in Australia when a psychiatrist purportedly assembled a data set on 8000 schizophrenic patients showing a relationship between the disorder and season of birth [37]. To collect these data, the researcher would have had to examine 16 000 files, far exceeding the number of files (3000) held by the study centre's records department [37]. The psychiatrist's unethical conduct was thus readily detected, leading to his deregistration.
A vital caveat is that ‘whistleblowers’ act with integrity and not out of malice. They need to be protected from possible adverse consequences. The student who notified authorities about the professor of psychiatry who plagiarised expressed fear about what lay ahead [81]: ‘… people say that I did a good thing, but the bigname places may regard me as a trouble-maker’.
Imposing penalties
As with any ethical problem, it would be preferable to promote desirable behaviour rather than punish miscreants. Regrettably, however, serious publishing misconduct has to be punished to prevent further harm and to deter the offender from recidivism and others from following suit. Severity of the penalty depends partly on the nature of the ‘offence’; options include reprimand, deregistration, job loss, ineligibility for research grants, a ban from publishing in the journal in which the misconduct occurred, excluding the author from a group of journals, and circulating a ‘blacklist’ of discredited persons among journal editors. Of course, legal liability of any action calls for diligent consideration. Journals themselves should perhaps be liable too; it has been advocated, for example, that journals tardy in retracting papers should be removed from Medline [82], surely a death-knell.
Further research
Any process requiring ethical conduct must itself be open to continuing scrutiny. A quarter of a century ago, it was pointed out that ‘material often obtained in the most rigorous kind of biomedical experimentation is rated as publishable or not by a system that has rarely been subjected to any analysis’ [83]. The situation has changed, with conferences and journal columns [84–86] devoted to research into medical publishing. Its quality is no doubt improving; for example, a 1998 congress on peer review included several randomised controlled trials on the utility of blinded assessment [15]. There is ample scope for more work of this ilk. Furthermore, although much research on publishing ethics has appeared in general medical journals, psychiatry is seldom covered and findings from other branches of medicine may not always generalise to our discipline.
Conclusions
Psychiatrists face a challenge to familiarise themselves with, adhere to and encourage ethical publishing without stifling initiative and creativity. A new ethos is called for. Faraday's directive to ‘work/finish/publish’ [76] should perhaps no longer be heeded unthinkingly. While publishing is a prima facie duty [87], we are still bound to take stock and raise new questions: ‘Is this material worth publishing?’ (vs ‘Is there a paper in this?’) and ‘How few papers can we get this project down to?’ (vs ‘How many papers can we get out of this?’). ‘Givens’ of academic institutions, like the pressure to ‘publish or perish’, are under scrutiny. Whether readers can be merely passive recipients of what is offered to them by journals is arguable. We are obliged to join our colleagues in the rest of medicine by contributing to the promotion of publishing ethics and pondering over remedies being promulgated.
Footnotes
Acknowledgements
We thank Gordon Parker, Joseph Rey, John Ellard, Harry Minas, Helen Herrman, George Halasz, Julian Suvalescu, Sarah Bird, Steve Ash, the State and Territory medical boards and the staff of Concord Hospital library for their assistance.