A Comparison of Authorship Policies at Top-Ranked Peer-Reviewed Biomedical Journals FREE

In a highly competitive scientific environment, authorship decisions are important. Including authors who do not meet authorship criteria dilutes the merits of other authors and may lead to inappropriate academic advancement and have a corrupting and discouraging influence on research.^1- 2 To ensure the honesty of the scientific process, the International Committee of Medical Journal Editors (ICMJE) defined 3 criteria for authorship eligibility, which, taken together, are indicative of personal effort and accountability.³ More than 500 biomedical journals have requested listing themselves as subscribers to ICMJE Uniform Requirements (URM) for Manuscripts Submitted to Biomedical Journals. However, the prevalence of authorship policies and criteria and authorship definitions vary widely, with many journals having implemented no criteria.

We studied high-impact, peer-reviewed journals to assess these variations.

We performed a cross-sectional study in 135 peer-reviewed biomedical journals from 35 publishers, including the 15 top-rated journals in 9 Journal Citation Reports (JCR) medical categories rated according to the 2009 JCR Impact Factor (IF) (eTable 1).⁴ We included English-language journals publishing (1) original research and other types of contents (all-content journals) and (2) all contents except original research (review journals). Guidelines or instructions for authors and manuscript submission available on journal Web sites were reviewed independently by 2 authors (J.M.P. and C.H.) using a standard form (eAppendix) in December 2010.

We coded journal authorship criteria (if any) using 12 criteria largely based on Archives of Internal Medicine and JAMA authorship criteria as a baseline (eAppendix). Criteria were coded into 8 categories according to compliance with the 3 ICMJE authorship criteria: categories 1 through 6 correspond to criteria 1 to 6 in the eAppendix; category 7, criteria 7 to 10 in the eAppendix; and category 8, none. The eAppendix and eTable 2 show the information collected for each journal.

The number of journals publishing contributorship disclosures was determined by review of 10 randomly selected articles from each journal published between July and December 2010.

We analyzed ICMJE criteria required according to different variables including the 4 publishers with the most journals.

The median IF was 6.1 (interquartile range, 4.4-9.8). No association was observed between the IF and journal requirement of none or 1 or more criteria.

Three criteria were required by 51.4% of all-content journals vs 21.8% of review journals (P = .005). There was a significant association between criteria requirement and URM endorsement (P < .001): of the 50 journals subscribing to ICMJE requirements, 18% required no criteria, compared with 51.7% of the 85 that did not (Table). Each criterion was required by approximately 50% of journals (eTable 3).

There were significant differences in the number of criteria required between the 4 publishers analyzed (P = .008) (Table). Of the 60 journals including 3 criteria, the largest number belonged to infectious diseases (n = 11) and medicine (general and internal) (n = 10) (eTable 4). There was a significant association between subject category and criteria requirement (P = .01). Only 28 (20.7%) of journals, and no review journal, included published contributorship disclosures (eTable 5). There was a significant association between URM endorsement and journal category (P = .001) but not between endorsement and journal region (eTable 6).

Scientific journals themselves, together with ICMJE institutional support, are the prime movers in formulating authorship guidelines and regulations.⁵ This review of top-ranked journals' authorship policies suggests that much work remains to be done and that policies and criteria vary widely. Although there was no significant association between ICMJE criteria requirement and the IF, only 44.4% of journals required all criteria, while 39.3% required none. Worryingly, although the relationship between criteria requirement and URM endorsement was significant, only 68% of ICMJE journals required all criteria and 18% required none.

Journals use different authorship and contribution policies, statements, and forms; place ICMJE or their own authorship criteria in different Web site sections; and require submission of authorship forms at different stages. This results in misunderstanding and confusion and does little to prevent undeserving authorship.⁶ Furthermore, some publishers' Web sites post authorship statements and policies that do not parallel their journals' polices.⁷

It is unclear why 32% of ICMJE journals do not require all 3 criteria: possibly, editors have decided to use their own. In an international survey, 55% of 74 health care editors professed unawareness of ICMJE URM, while 21% were aware but had never used them.⁸

We found significant differences in the number of criteria required by the 4 leading publishers with, strikingly, only 30% of journals from Elsevier, the world's largest biomedical publisher, requiring all criteria (50% requiring none). The significant association between medical category and criteria requirement probably reflects the differences in URM endorsement between categories.

Contributorship statements, required by some journals at manuscript submission or acceptance to guide editorial decision making, do not totally resolve the difficulties inherent in honorary or ghost authorship but appear to have had an impact on some journals.^6,9 However, only 20.7% of journals, mainly from general and internal medicine, included contributorship statements in published articles, possibly owing to a lack of information on contributorship from the ICMJE and many journals.

No review journal required contributorship disclosures, and only 21.8% required all 3 ICMJE criteria compared with 51.4% of all-content journals. These differences may suggest more-relaxed editorial attitudes by review journals normally containing commissioned articles with fewer authors, making unwarranted authorship less probable. However, ICMJE authorship definitions do not distinguish between original and nonoriginal research and the Archives and other leading journals publish contributorship statements for clinical, systematic, and narrative reviews.

The study limitations include the cross-sectional design and selection of journals.

Clear authorship criteria are essential to ensure rigorous scientific inquiry and help readers judge which authors are making what types of contributions. Journals not posting authorship statements nor adhering to ICMJE policies are serving science badly because, without their support, the numbers of unmerited authors are unlikely to diminish significantly. Biomedical journals should pursue consensus on authorship definitions more vigorously, ensuring uniform adoption of ICMJE criteria, either current or updated to include quantitative measures, for original articles and reviews.

The ICMJE guidelines should be revised to create a standardized contributorship disclosure form to be published by as many journals as possible. Publishers should avoid duplication of standardized authorship policies and discrepancies with their journals.