Changing authorship system might be counterproductive

See editorial by Richard Smith

Editor—The problem of authorship in science1 should be set in the wider context of debates that have raged in literary criticism since the early 1970s. In the work of Derrida most notably, the traditional concept of authorship with its implications of individualism and authority over the interpretation of textual meaning has been overthrown in theory, if not entirely in practice.2 Authorship is a political problem; it involves staking and maintaining territorial rights, colonisation, and empire building. In this it fairly accurately reflects the power game that is involved in the conduct of science itself. The sociology of scientific knowledge shows the actual practice of science to be remote from the received image of selfless dedication to the pursuit of disinterested knowledge.3 4 It is, rather, a highly politicised sector of the economy in which the first priority is to obtain the funds necessary to establishing and maintaining those research programmes on which so many scientists' careers, at all levels of seniority, depend.

This is not a criticism, merely an observation. Hence I think it would be a mistake to conclude that the authorship system has broken down and needs radical revision.5 If the present system reflects the structural and dynamic power relations that constitute scientific communities, and if these relations are the inevitable basis of the institutionalisation and conduct of science, then the only reasonable justification for change would be to represent accurately a radically revised basis of science itself. As far as I can tell, this revision is impracticable because power relations will always be the essence of the generation of scientific knowledge. A depoliticised authorship system implies a depoliticised science, which implies its total detachment from the state, industry, and society. Even if such a science were possible it would lose what tenuous significance it has for the world outside its own institutions.

Would it not be better, therefore, to leave the present authorship system to continue its present slow evolution in so far as it accurately reflects the real power relations in science? What is really required, I suggest, is to educate scientific and wider communities into the art of reading scientific literature from a different perspective which more accurately reflects the power paradigm of its production. I do not think that by manipulating the authorship system one could effect any important change in the conduct of science itself. On the contrary: the most likely outcome would be to add another layer of obscurity to conceal its essentially political nature.

System acknowledging roles of contributors is best

Editor—Richard Smith has invited comment on the issue of authorship.1 I think it would be an excellent idea to scrap the concept of authorship as we now know it and to opt for a system in which the role of various contributors is acknowledged. I do not believe that the issue of ultimate responsibility is a major problem. Indeed, the descriptive system would facilitate the identification of the person ultimately responsible. In the great majority of my nearly 200 papers in refereed journals I have acted as initiator, raiser of funds, supervisor, more often than not writer, and certainly the person accepting ultimate responsibility. For more than 20 years, however, I have had the responsibility of supervising research students, who have invariably been the first author. I frequently encounter senior colleagues who have not been aware of my involvement with such work. Our studies relating to the health consequences of vegetarianism and diet and diabetes are particularly good examples. I would regard these two issues as being among my main contributions during my entire research career, yet my name has appeared second or last on the papers. There has been no clear distinction between my role and that of, say, the statistician or dietitian, whose role may have been important but certainly was not instrumental in the study.

I appreciate that such a fundamental change would undermine citation indexes, but perhaps these would then be replaced by more appropriate systems of academic credit. From an immediate practical point of view it would simply be necessary to obtain agreement relating to descriptors, which I hope would go beyond “contributors” and “guarantors” as suggested in Smith's editorial.

Question of authorship concerns everyone in training grades

Editor—While many benefits make up the carrot of having work published, the sticks held up before potential authors should be public accountability for “their” words and the availability of accurately produced data.1 Recent episodes have highlighted the seriousness of failure to uphold these standards.2 Perhaps the number of authors on a work should be limited, although this would bring its own problems. Certainly some journals' practice of listing only the first three authors in the references goes some way towards this. Another alternative, which was not particularly favoured in Raj Bhopal and colleagues' study,3 is the limiting of citations on one's curriculum vitae, yet this would probably have attractions in a different group, such as trainees. For many or perhaps all of us in the training grades, appearing in the literature and thus authorship is a game we have to play to get on. There are no other ways of distinguishing between equally well qualified people on paper, and personal recommendation is now perceived as politically incorrect. The “vexed question of authorship” is of concern to all of us in the training grades, but one question remains: was having nine authors for this study3 deliberate?

Coworkers should be named

Editor—Medical and scientific research papers are always the fruit of collaboration of many workers at different levels, and even when eight or more authors are named, many remain unsung. Before publication some papers have been revised by anonymous referees or partly rewritten by medical editors for clarity, relevance, logic, and conciseness. Librarians, pharmacists, statisticians, etc may have given helpful advice. The work of data collection may have occupied nurses, interviewers, technicians in special departments (radiology, biochemistry, etc)—work for which they were routinely employed or did as an extra. Backing it will have been finance officers, secretaries, porters. In cinema and television the credits roll to name not only the creative artists, the sound and lighting technicians, and the set carpenters but even the location caterers, the accountant, and the insurers. Why not do the same in journals?1

Some team members want authorship for five minutes' personal glory, or to further their careers, because bureaucrats find “authorship counting” an easy assessment of individuals and departments. But readers (and publishers) do not want a space-occupying meaningless list of names—they want to know who chiefly is responsible for the planning, coordination, analysis, and presentation of the work; who is leader of the group or department; and what part of the work other specialists (named) have contributed. These are separate matters.

Authorship should be strictly limited to a maximum of four people, who take responsibility for everything said and done—that is, they are satisfied with the validity of the data provided by their specialist coworkers. In the abstract of the paper, after the conclusions, there should be an additional section, headed “Coworkers.” In this section will appear the coworkers' names and specialisms and what they contributed to the project, and the journal's editors will decide where to draw the line in accepting some or all for publication. The head of the research group can be named in parentheses after the name of the clinic from which the work comes (as has been the practice in German journals), or in the coworkers section. People who do extra work of the sort for which they are already employed might get monetary reward or the equivalent of Air Miles for their pains, but that is not a matter for journals.

Work done by junior researchers gives rise to problems

Editor—We think that Richard Smith's ideas for crediting work done and for having guarantors of integrity fit nicely with the way that we do things, because our work is necessarily done by a team.1 Each member brings different talents and skills.

It is the recognition of the necessary spadework (often done by junior researchers learning their trade) that vexes us most. The “diggers” are often in a position neither to give a formal presentation nor to conceive or design a study, because of lack of experience. Despite making an enormous contribution they might fall short of the criteria for authorship if the Vancouver guidelines were followed to the letter. To disenfranchise them would be unfair; credits would get round this problem.

We now hold a “public” read through of the final versions of papers leaving this unit, in the presence of all involved. This ensures not only that everyone involved approves the final version but also (we hope) a more readable and understandable paper. We also believe that it helps us to comply with the spirit of the Vancouver guidelines.

Bhopal and colleagues' suggested method of ordering authors wouldn't work

Editor—Although admirable in intent, R S Bhopal and colleagues' suggestion that the order of authors on research papers should be decided by aggregating individual preferences is unworkable in practice.1

Consider five authors—a, b, c, d, and e—who respectively think that the order of their colleagues' names should be (b, c, e, d); (d, e, c, a); (a, b, d, e); (e, a, b, c); and (b, d, c, a). Following the suggested method—in which first place is awarded four points, second place three points, and so on—yields: first b (13 points), then d and e (both with 10 points), then a (9 points), and finally c (8 points), so that the “democratic” ordering would be b, d=e, a, c.

Even if the problem of what to do about the equal scores of d and e is ignored, the result is problematic because, according to the individual preferences, majorities exist for a being before b (both c and d are in favour of this, and only e favours the reverse) and c being before a (this is favoured by both b and e whereas d is against it); neither of these preferences is reflected in the group ordering. As the voting scheme gives no indication of the intensity of the authors' opinions about their choice of ranking, the group ordering is little more than arbitrary.

This is not a special case—political economists have been troubled by such “voting paradoxes” for over 200 years.2 The economist Kenneth Arrow won the 1972 Nobel prize in economics largely for proving that, in general, there is no fair and logical way of aggregating individual preferences into a single, collective ordering.3 This makes any sort of voting scheme for deciding sequences of authors impracticable. Fortunately, no such difficulties were encountered in deciding the order of authorship for this letter.

New authorship practices are needed in developing countries

Editor—Richard Smith expresses his dissatisfaction with the authorship paradigm and supports a radical solution: abolish the concept of authorship and replace it by a descriptive system (like film credits).1 A study by Raj Bhopal and colleagues showed that, among researchers in a British medical faculty, guidelines on authorship were inadequate and often not adhered to.2

Hitherto, the debate has focused on the publication of research from developed countries. The issue, however, is of equal interest to the developing world. After reading Bhopal and colleagues' paper, I solicited views about the rules concerning authorship from 10 Tanzanian researchers at the National Institute for Medical Research in Mwanza. As in Britain,2 a range of practical and intellectual contributions were considered grounds for authorship. Eight respondents considered gift authorship to be a problem; four thought it was common, and the others had no idea how common it was. Seven researchers said that gift authorship should be banned. Problems with or conflicts about authorship were commonly reported, such as exclusion from the list of authors when authorship was thought to be deserved (6) and inclusion when it was not merited (3). Eight researchers liked the idea of a film credit system, mostly because of its transparency and frankness. The two others did not like it because they thought that it was too cumbersome and would not support existing reference systems and curriculum vitae.

Current systems of authorship in large studies in developing countries vary. Gift authorship or inappropriate ranking of authors seems to be common and to be part of development aid in some projects. The researchers themselves, however, prefer to be acknowledged for what they have actually done rather than be part of a politically correct process of allocating authorship.

I agree with Smith that it is time for a paradigm shift. In the context of developing countries the current methods of allocating authorship are obscure and paternalistic. Even though the name of the researcher in the host country ranks high on the list of authors or appears in many publications, he or she may not be taken seriously, simply because senior international researchers are also listed or because the researcher is one of several fellow researchers listed from that country who barely contributed to the study. Current authorship practices may contribute to further underdevelopment of research and researchers in developing countries and fail to acknowledge the researchers appropriately for their actual and essential contributions. As a first step, all articles should have a footnote specifying the contributions of every author.

Authorship is influenced by power and departmental politics

Editor—As an undergraduate, I researched and analysed 33 000 postmortem records in a university institute. For the study I won a university award. The institute wanted to use and publish my findings without my name, arguing that the original materials were the property of the institute and that it paid for my work by giving a cheque with the award. I was so incensed that I refused to take the award (and the money) and wrote to the rector of the university, forbidding it to use my research findings. The reply was swift. The university too forbade me to publish my findings because the original material belonged to a university institute which did not consent to it. Seventeen years on, the study has still not been published.

In clinical research, data derived from patients belong to consultant X or Professor Y, on the grounds that the doctor is in overall charge of the patients' treatment. The custom of this “possession” is so ingrained that nobody dares question it. Thus it is not possible to use clinical data without the explicit consent of the consultant in charge of the patient. Requesting this usually brings a request from the consultant to see the draft of the paper. Because he or she usually makes some suggestions one can expect the explicit or implicit request for the consultant to be listed as a coauthor. Junior staff, who are in a vulnerable position because they depend on their seniors, will willingly or grudgingly offer them at least coauthorship before they formally request it. Very few consultants will turn down such an offer, saying that they did not participate enough to deserve it. How many of us are brave enough to tell our consultant, “Your participation in this paper does not fulfil the Vancouver criteria, but would you be kind enough to support my application for that post?” I know a consultant who encouraged his juniors to write separate papers putting each other's names on them, and of course his—all in the spirit of the Vancouver recommendations.

A further problem in Britain is that juniors in clinical medicine stay in one place for only a short time. By the time all the data have been collected, analysed, and written up to be published they have probably already moved on. Their contribution is hardly mentioned, if mentioned at all.

Fierce disputes about order of authors sometimes occur in China

Editor—I agree with Richard Smith that the issue of authorship of biomedical papers needs reappraising.1 In China only the first author is eligible for promotion, no matter what the nature of his or her position (academic, professional, or technical). So it is not surprising that many unpleasant (sometimes fierce) disputes regarding the order of the authors arise between senior and junior doctors and between doctors and technical assistants. I agree with Smith that the concept of “contributorship” is a good solution because it can reflect the concrete contribution of every participant. Accordingly, the “designer” of the paper or the “coordinator” of the “contributor team” should be indicated, to show who will be responsible for the scientific integrity of the whole paper.

Knowing who did what in studies is important

Editor—I am glad to see the topic of authorship being discussed.19 In my work with trainees in public health medicine and on continuing medical education for public health doctors in New Zealand, publications are important in the presentation of projects and in providing points for recertification. It may be difficult to establish what role a person had.

I look forward to seeing the development of a practical and realistic way of recognising and authenticating the work that is done. Its acceptance will require a shift in employers' and academic attitudes to the tasks involved in research and publication.

Without a putative contributor, would the integrity of the work change?

Editor—Richard Smith highlights the difference between novelists and scientific writers and examines the problems of crediting authorship to scientific writers.1 The concept of authorship is, I think, satisfactory. Three elements merge inseparably in the formulation of a novel: inspiration, or ideas, which are developed into characters and a plot; research; and writing. Although scientific papers are quite different, the Vancouver criteria for authorship are comparable: conception and design of a study (inspiration) or analysis and interpretation of data (research) and drafting or revising critically (writing). The problem is not the concept of authorship but that multiple “authors” may be involved in producing scientific papers and may not meet all of the criteria.

The solution is not to redefine authorship but to recognise in some other way important contributions by those who do not qualify as authors. The concept of contributorship—analogous to film credits—has the advantage of defining relative roles in collaborative, including multicentre, ventures. It would, for example, acknowledge the efforts of those whose contribution to research has been important yet whose diluted involvement or non-involvement in the conception of the study or writing the paper has precluded them from qualifying as authors. It may allow more eloquent dissection of contributions to a paper and clarify accountability. It should not, I think, necessarily replace authorship. Many genuine authors remain.

Whether authorship or contributorship is used, the struggle for inclusion remains: almost everyone would like the “glory.” Writing is never completely original. Ideas are drawn from experience, from endless encounters. Research examines the work of others. Even the process of writing evolves through various admired styles. No achievement is entirely one's own, and potential acknowledgements are limitless.

How far, then, should we go in acknowledging putative contributors? Clearly, this is not easy. We could be too dogmatic in our limits or too nebulous. It would be reassuring to think that common sense would prevail, but there is no definition of common sense. A reasonable way to decide whether a contribution is important could be to consider whether, without the putative contributor, the integrity of the work would essentially change. Potential non-contributors may become obvious and may include departmental heads, nepotists, and inquisitive colleagues.

There is, I suspect, a place for authorship and contributorship. There is doubtless a place for non-contributorship, and that is in readership.

Assessment of authorship depends on culture

Editor—Strict adherence to objectivity and high ethical standards are defining virtues of science. Though scientists share a common language and culture of science, they are also influenced by values and norms in their societies.1 Our perception that there were more Japanese than non-Japanese authors listed on research articles prompted us to investigate how culture might influence authorship criteria. We identified two qualitatively similar medical journals (criteria available on request): Circulation Research, which is published in the United States, and the Japanese Circulation Journal. We compared the number of Japanese and non-Japanese authors per article in the 1983, 1993, and 1996 issues of these journals. In each year there were 2-3 more Japanese authors per original article in the Japanese Circulation Journal than in Circulation Research (1). These trends were similar for other contributions too: there were 296 authors on 45 case reports in the Japanese Circulation Journal (mean 6.6 (range 2-12) authors/report) and 225 authors on 59 communications in Circulation Research (3.8 (1-11)).

Though these data are from a limited sample, they suggest intercultural variations in crediting authorship. The Japanese penchant for “groupism”2 and limited individual funding probably lead them to involve more people in research endeavours. Research groups in Japan possess a cohesive sense of unity and mutual reliance on the group and senior leader,3 who typically is included on every paper from the laboratory. A professor who orchestrates funding and consensus building qualifies as an author because these contributions are critical for conducting4 and publishing research. According to the requirements of the International Committee of Medical Journal Editors, such contributions qualify simply as an acknowledgement,5 but such discrimination would be likely to disrupt group harmony. Firstly, greater value seems to be placed on the act of contributing than on the value of the contribution. Secondly, scrutinising the value of contributions may strain relationships and prove counterproductive to cohesiveness in the laboratory.

These data and analysis imply that publication of research manifests the interface between the scientific method and the culture of contributing investigators. Scientific investigation is neither devoid of its own cultural milieu nor immune to the influence of the values and beliefs of investigators using scientific methods. While the criteria of the International Committee of Medical Journal Editors do not accommodate cultural variation in crediting authorship,5 they reflect the cultural background of the majority of the members and not codification of undisputed scientific principles. The movement to credit only those who deserve authorship is noble, though the assessment of legitimate authorship is a cultural, not a scientific, judgment.

Researchers' objective is to get the job done

Editor—Perhaps one explanation for the differences in how authorship is perceived by editors and researchers is these groups' different objectives. Editors, encouraged by a frisson of puritan zeal, wish to allocate responsibility for published work. Researchers, on the other hand, ultimately want to get the job done and push back frontiers. As Richard Smith indicates, modern research needs large multidisciplinary teams, and authorship, like the honours system, is a cost free way of getting many busy people to participate.1

In clinical areas the person with the drive and ideas may not have the patients or the necessary skills in molecular biology, radiology, pathology, etc; and vice versa. Students of human relationships will recognise immediately their day to day contrariness; offering authorship slices through these difficulties and permits better and more innovative research.

Surely editors should agree with the researchers' objective, which is to get the job done; all else, though important, is secondary.

Author saw fraud, misconduct, and unfairness to more junior staff

Editor—As a junior researcher who graduated six years ago, I find the issue of authorship the most frustrating of all the problems within the medical research community.1 It is frustrating because it is remediable, and the main culprits are often experienced researchers. Two of the most interesting books that I have read are Follies and Fallacies in Medicine and Fraud and Misconduct in Medical Research.2 3Both of these books document cases of unacceptable practice.

During my career I have personally experienced events that constituted fraud, misconduct, or simple unfairness to more junior staff. These were notably:

• Plagiarism: I have witnessed a senior person submit a dissertation for a higher degree in which an appreciable part of the introduction had been copied verbatim from a published paper

• Important data were manufactured on two occasions

• Junior staff were omitted from published work

• A junior researcher conceived a study and wrote the first draft and much of the modification to the proposal, but was excluded from the final submission, which received funding of £100 000. This exclusion was considered to improve the chance of funding

• Token authorship

• Token supervision of higher degrees

• Destruction of computerised data when a funding body decided to scrutinise the work more closely

• Multiple applications for funding of essentially the same proposed work.

Of the solutions to the problem of gift authorship that Richard Smith offered, the radical solution of a list of “credits” appeals most.4 Editors should require authors to describe their input and demand justification if there is any doubt about someone's inclusion. I would go further and support the idea of an agency to police research.5 Research and development of drugs undergoes this process because of the potential consequences of poor or bad research practice in the pharmaceutical industry and the large financial incentives involved. The difference between research and development of drugs and other clinical research may be a medicolegal one in that drugs can be fairly easily identified as a root problem whereas a deceptive paper that has influenced a doctor's clinical judgment is more difficult to identify.

The vast majority of people whom I have encountered and practice that I have experienced have been perfectly acceptable. The remainder should be weeded out.

Excluding authors may be impossible

Editor—I am glad that the definition of authorship is finally being addressed.1 2 The publication of original scientific articles may be a key to some recognition, but whether it necessarily leads to academic success is far from clear.

As a trainee, I was able to do a few pieces of original research, most of which got published in professional journals, including the BMJ. During these years of training I came across some people who never hesitated to give help and support for any sort of research and some who had hardly been involved in the research but demanded to be listed as an author. I am sure that my experience was not unique. I remember a particular instance, when the editor of the BMJ had a major criticism—that there were too many authors for a paper of the size that this paper was. On further analysis, according to the BMJ's recommended criteria, one could exclude four coauthors, but was it possible? No, because any such attempt could possibly lead to a misunderstanding and cracks in relationships. Instead, a path of least resistance was taken and the article was published in another, less esteemed, journal.

This reflects one side of the coin. The other side is when one gets intimately involved with all the practicalities of a piece of research but does not even get an acknowledgement in the paper, let alone inclusion in the list of authors. This experience again is perhaps not unique to me.

It is important to have a clear policy on who should be included as an author, even before the start of the research. Researchers and editors need to develop a unanimously acceptable strategy for authorship. It would be appreciated by most researchers around the world.

Number of publications given on curricula vitae should be limited

Editor—Authorship will continue to be a problem1 2 3 while the curricula vitae of both senior and junior doctors are judged on the number of publications rather than the quality of a limited number of them. Only once in my career have I been asked to limit the number of publications on an application form, and even then it was to too high a number.

I suggest that applicants for posts for preregistration house officers, senior house officers, and specialist registrars should be asked to choose their best two publications for inclusion in their curriculum vitae and to be prepared to discuss them at the interview. This would also allow interview panels to have copies of the publications available at the interview, so long as this was not at the expense of the candidate. Applicants for consultant posts should be asked to choose their best five publications and be prepared to discuss each paper, their personal contribution to it, and the impact that the paper has or will have on their specialty.

For their annual assessments, specialist registrars should be asked to discuss one of two publications that they have submitted in the previous year. For consultants, discretionary points, merit awards, and continuing medical education points should be related more to quality and relevance than to number of publications.

Multicentre trials give only a few people “authorship” but depend on the participation of many and lead to important steps forward in treatment of a wide range of conditions. Participation in such trials should be recognised.

Authorship given to a member of the team for participating in only one aspect of a project (for example, collecting data or retrieving and reviewing case notes) has some advantages, in that it can be used as an incentive to get work done quickly and finish off a project started by others. This, in some cases, is useful as it allows publication of an important finding, which may otherwise be delayed or not submitted at all. Publication counting is rife and ingrained into medical thinking; thus any changes would take a long time to filter through to medical practice.