What Are the Factors Determining Authorship and the Order of the Authors' Names?: Title and subTitle BreakA Study Among Authors of the Nederlands Tijdschrift voor Geneeskunde (Dutch Journal of Medicine) FREE

EDITORS are familiar with the definition of authorship of the International Committee of Medical Journal Editors (ICMJE).^{1 - 2} Although these rules are enforced by the editorial boards of more than 500 biomedical journals, including the Nederlands Tijdschrift voor Geneeskunde (NTVG, the Dutch Journal of Medicine), the rules seem to be little known among researchers and authors,³ and do not always seem to be followed.^{4 - 5} We investigated how authorship was established for articles in the NTVG.

Authors (N=450) of all original articles (N=115) published in the NTVG in 1995 were mailed a questionnaire enclosed with a copy of the first page of their article. Anonymity was guaranteed and a small monetary compensation was offered. A reminder was mailed after 1 month. Articles with 1 or 2 authors were excluded because we did not expect to find many problems with authorship in these articles. The NTVG does not publish articles with more than 6 authors.

The questionnaire included 23 questions and was designed to determine whether and to what extent the ICMJE criteria had been applied in practice, regardless of whether the authors were familiar with these criteria. The authors were asked to score themselves and their coauthors on their contributions to study design, material, collection of data, statistics, and writing, using only plus, minus, or question mark symbols. The questionnaire also contained open-ended questions to determine whether authors were familiar with the ICMJE criteria for authorship. The main subject of each question is listed in Table 1.

To determine whether an author perceived his or her own contributions differently than the coauthors perceived them, the number of positive answers an author gave himself or herself was totaled and compared with the mean total score the author had received from each of the coauthors. The difference could vary between 22 and−22 (question 23 [no contribution] was omitted from analysis).

Authors were considered to satisfy the ICMJE criteria if they answered affirmatively (+) to questions 1 and 2 (idea and design) or question 14 (analysis), as well as question 16 (writing), 17 (rewriting), or 18 (critical reading), and question 19 (approval). An author was considered to have not fulfilled the criteria if at least 1 of their coauthors scored the author negatively (−) for question 1 or 2 and question 14, or questions 16, 17, and 18, or question 19.

The data were loaded into EpiInfo, Version 5.1 (World Health Organization, Geneva, Switzerland), and processed with SAS, Version 6.12 (SAS Institute Inc, Cary, NC). The data input had an error percentage of less than 1%.

Of the 450 questionnaires, 362 (80.4%) were returned and 352 (78.2%) could be analyzed. Contributions the authors considered themselves to have made are shown in Table 1.

The discrepancy scores had a Gaussian distribution with the top score at 2 (ie, most authors thought that they had contributed to 2 items more than their coauthors thought); the fifth and 95th percentiles scored−4 and 8, respectively.

The ICMJE criteria were fulfilled by 224 (63.6%) of 352 authors (according to their own scores), regardless of whether they were familiar with them; 79% of the first authors fulfilled the criteria and 58% of other authors fulfilled them. However, in 46 (21%) of 220 authors who fulfilled the criteria according to their own scores, more than 50% of the coauthors reported that the author did not fulfill the criteria. (In the other 4 cases, no coauthors replied.)

Of the 352 total respondents, 128 did not fulfill the ICMJE criteria according to their own scores. The most important contributions of these authors were critical reading of the manuscript (77%), providing the patients (63%), approving the final version of the paper (58%), collecting data (53%), and patient care (51%).

Most authors (202 [59.8%] of 338) stated they did not know the ICMJE criteria: 61% of first, 44% of the last, and 66% of the intermediate authors. In 70% of the articles, at least 1 author indicated that he or she knew the criteria.

Most respondents reported that there had been problems in determining authorship and the authors' order. A particular source of dispute was that clinical work alone is insufficient for authorship.

Although our investigation has a number of flaws (eg, the questions can be interpreted in various ways, an appeal was being made to the memory of the participants after a considerable length of time), we believe that the response rate of 80.4% and the small discrepancy between the respondents' and their coauthors' answers allow a good interpretation of the results.

Intellectual contributions (the idea for and design of the investigation and the critical reading, rewriting, and approval of the manuscript) ranked highest among the 7 ICMJE criteria for authorship in NTVG articles. However, more than 50% of the respondents had contributed to or performed the practical work of the investigation (eg, providing patients or research material, carrying out a pilot study, collecting the data). Four authors indicated that they had made no contributions whatsoever.

The answers showed that 63.6% of the authors fulfilled the ICMJE criteria, regardless of familiarity with them (59.8% of the respondents were unfamiliar with them). Their coauthors felt that 21% did not fulfill the criteria. It may be that the ICMJE criteria are logical or the reflection of good social behavior, but regardless, they were little known as such.

Many authors considered the rules to be too strict. The biggest problem was the failure to appreciate clinical work. According to many clinicians, provision and care of the patients are sufficient criteria for authorship. Because the NTVG is a medical journal, most authors are clinicians, and their opinions most likely biased this investigation (eg, in contrast with statisticians).

Our investigation confirms that the ICMJE criteria are insufficiently known. However, many authors appear to apply them implicitly. Confusion regarding authorship could be reduced by making the criteria more widely known.