Honorary and ghost authorship

What qualifies a person to be an author of a research paper? Is it simply a matter of enlisting a patient in a clinical trial, or does authorship require a particular level of participation in planning and executing a study? Should there be a “substantial” contribution to the writing? It may be that satisfactory answers to the complex questions around authorship are not feasible, desirable, or even possible. However, most would agree that pursuing clarity of contributions to a scientific article is a worthy goal. Journal editors serve the medical community best when readers can judge accurately how the work was done and communicated. Accurate identification of authors is the first step in creating transparency of contributions.

In the linked study (doi:10.1136/bmj.d6128), Wislar and colleagues compare the prevalence of honorary (guest) and ghost authors in six leading medical journals in 2008 and 1996.1 The authors used the International Committee of Medical Journal Editors (ICMJE) criteria to define honorary and ghost authorship.2 The results showed that inappropriate authorship declined from 29% of articles in 1996 to only 21% in 2008, an improvement from previous studies cited by the authors, which were also based on the ICMJE criteria but lacked the breadth of the authors’ 1996 and 2008 studies. However, this proportion of inappropriate authorship is a concern to institutions and journals responsible for integrity in scientific reporting. Despite some limitations highlighted by the authors, a nagging question remains: are the ICMJE authorship criteria the right standards with which to accurately gauge and enforce “appropriate” authorship?

The journal Neurology has wrestled with this question since several studies written by hired professional writers prompted development of an authorship policy that departs from the ICMJE criteria and encourages greater transparency and disclosure.3 The ICMJE criteria leave “substantial contributions” incompletely defined and, moreover, require three conditions to be met (box 1). However, in practice as the numbers of contributing specialists increase author groups tend to ignore the ICMJE criteria, adding “honorary authors” when contributors do not fulfil all three criteria. Thus the author byline may not have accurate information about individual contributions to a study or may not reflect important contributions from people who were not named.

Neurology sought to strengthen the ICMJE’s contributor model and avoid honorary and ghost authorship altogether by identifying all individuals who contributed to the study, who wrote the reports, and who paid for the research and reporting. This policy is codified in Neurology’s information for authors (www.neurology.org/site/misc/auth2.xhtml) and outlined in box 2. Any one of the three Neurology criteria defines an intellectual contribution and requires identification as an author. All authors must acknowledge all versions of the manuscript and sign statements about their specific contributions and disclosures via the manuscript tracking system, and these are subsequently published. All authors therefore take public responsibility for their particular contributions. To ensure integrity, authors are expected to be as careful in assessing appropriateness of authorship as they are in reporting research results. Those who do not qualify as authors by one of our three criteria are listed with their contributions as co-investigators or contributors in an appendix or in the acknowledgments section.

This policy requires that any medical writer who wrote the first draft or responded to the reviewers’ comments be included in the author byline and make full disclosure. The first draft is the intellectual framework for revisions, and any responses to reviewers’ comments are intellectual contributions that ought to be transparent to editors and readers. Identification of professional writers as authors is transparent, fair, and anti-discriminatory: credit is given where credit is due. Authorship recognises a professional writer’s accomplishments in scientific reporting and promotes the writer’s professionalism. Some investigators and medical writers have objected to this requirement,4 although others have agreed.5

Neurology’s authorship policy is a beginning in the effort to increase transparency in authorship. Editors are unlikely to have sufficient resources to validate all authorship claims or conflicts of interest, however.6 By aiming for transparency, we hope to move closer to the target of greater integrity in scientific publishing, so that readers can evaluate the results in the context of knowing how each author or investigator contributed to the study and who subsidised and influenced the research and writing. Although sound scientific rigour is the best guard against bias, greater disclosure contributes.

Wislar and colleagues’ study may catalyse further studies and prompt journals, especially medical journals, to re-evaluate their authorship policies and consider adopting tighter criteria that can better serve researchers and readers. The study soundly shows whether leading journals actually adhere to the ICJME policies. As research becomes more collaborative and complex, the challenges to transparency in authorship and disclosure become greater. The ICMJE encourages journals to list contributors to foster greater transparency and author accountability, but most journals do not state contributorship criteria in their instructions for authors or publish contributions.7 One exception is the journal Blood, which bases authorship on one or more contributions from the ICMJE authorship criteria (http://bloodjournal.hematologylibrary.org/site/misc/ifora.xhtml). A minority, such as the BMJ, require authors to qualify according to the ICMJE criteria and also list authors’ and others’ contributions, with details about who planned, conducted, or reported the study (http://resources.bmj.com/bmj/authors/article-submission/authorship-contributorship). Are any current authorship policies, even Neurology’s, which we believe are the most specific, sufficient? Perhaps not. Further work is needed to assess whether greater definition of roles and conflicts of interest substantially change the prevalence of inappropriate authorship.