Our study showed that authors, when asked about authorship in a non-instructional way, i.e. without reference to the ICMJE criteria as a standard in biomedicine , mostly declared contributions that could be matched to the first two ICMJE criteria (executing research and writing the manuscript), but not to the third ICMJE criterion (approving of the final manuscript version). The authors also most often used the wording of the ICMJE definition to describe their contributions. The declarations from 13% of the authors could not be matched to any ICMJE criterion. The fraction of such authors decreased with the increasing number of authors and their remoteness from the first byline position.
Our study is limited by its cross-sectional design, but our findings are strengthened by a high response rate, since filling out authorship forms is mandatory for manuscript processing in most science journals. Generalizability of the findings to other medical journals and research and academic settings can also be questioned as the study was performed in a single journal. However, the results are consistent with the previous finding from authorship studies in our journal, which included both survey and randomized study designs [9–12], as well as with studies from other journals or academic settings [3, 4, 6–8].
Despite the fact that we asked an open-ended question and did not provide instruction on ICMJE definition as accepted authorship criteria in biomedicine, most of the reported contributions matched those described in the ICMJE definition of authorship. However, only 15.6% of the authors whose contributions could be matched to ICMJE definition satisfied all three ICMJE criteria. Further 38.6% declared contributions exclusively to the first two ICMJE criteria (research and writing). As authors made this declaration on a signed statement after manuscript submission, it can be assumed that they gave approval to the manuscript submitted to the journal. If their signature is then taken as a fulfillment of the third ICMJE criterion, the overall fraction of deserving authorship according to the ICMJE increases to 54.2%, which is similar to the results of our previous studies (range from 39% to 75%) that were based on the ICMJE definition and which had different study designs [9–12]. Approval of the manuscript can be regarded as something that is outside of the creative effort of researchers, and it can even be impossible to obtain in cases when one of the researches dies before the final version of the manuscript is finished. Today many journals require contact e-mails from all of the listed authors, subsequently informing them that the corresponding author had submitted an article in their name. Consequently, this makes the final approval a procedural requirement, and not necessarily a criterion for authorship contribution.
The lack of regard for final approval as a criterion for deserved authorship observed in this study, which had a cross-sectional design and did not refer to the ICMJE criteria, confirms the results of our previous study where we showed in a randomized study design that the “final approval of the article” was an inherently different category from other contributions and that it should be considered rather as an administrative requirement similar to signing of a copyright transfer . Furthermore, our recent analysis of journals from different research fields, including social sciences and humanities, also demonstrated that authorship definitions by journals, publisher and professional organizations or associations mostly addressed research and/or writing as contributions necessary for authorship . In our study, authors who declared a single contribution that could be matched to an ICMJE criterion, declared research contribution more often (in 31.8% cases) than writing contribution (in 9.4% cases).
The “fractionation” of authorship into more contribution categories was evident in multi-authored articles, where the number of contributions declared increased in manuscripts with more than 8 authors and specifically in those where authors declared ICMJE matching contributions. Taken together with the finding that overall, non-ICMJE matching contributions were more frequent in manuscripts with few authors (1–3 per byline), this indicates that authors from smaller research collaborations do not see the need to elaborate on their contributions, as their role in research presented is clear. In larger collaborative groups, contribution declaration seems to require a coordinated effort, particularization and careful distribution of contributions. Our study was not designed to check whether each individual author really filled in the form (although each was separately signed by individual authors), but the finding that manuscripts with more than 6 authors had more declaration forms with identical wording of the declarations indicates that filling in the forms could either be a centralized effort to formally satisfy journal’s requirements or that authors simply copied from each other, without engaging in truthful elaboration of their contributions. Different behavior in authorship declaration with increasing number of authors on manuscripts can also be related to the current publishing practices in biomedicine, where the number of authors was not perceived as an important issue for academic performance, in contrast to the position on the byline and the journal’s impact factor, explaining at least in part the increase in the number of authors per publication .
Single authors or authors of manuscripts with 2 to 4 authors had most ICMJE non-matching contribution statements. These authors usually stated that they made a significant contribution, without any specification, or they disclosed their professional expertise in the field. These statements, however, cannot be taken to imply undeserved authorship; they rather suggest differences in perceptions of the authors to the established criteria for authorship and the means of their reporting.
Despite the fact that our authorship forms had a required line where the authors had to sign their name if they agreed to be listed as an author in the submitted manuscript; a small number of authors, who answered our authorship question with only a “Yes”, most likely perceived that question as just another required confirmation of their authorship. Perception of authorship declarations as a form of external check-list necessary for manuscript submission is also supported by the finding that almost a third of the respondents in our study did not use full or partial sentences to a question that required an answer in the form of a sentence. Providing a list of contributions instead of a sentence may be the result of authors’ experience and familiarity with the prevalent practice in biomedical journals to formulate their contribution declarations as checklists.