Authorship and author order can have important implications.^1-3Consider the recent extreme example of Dr. Eric Poehlman, who admitted in court documents to falsifying numerous federal grant applications and the data on which many of his manuscripts were based and is now facing up to 5 years in prison and $250,000 in fines.⁴The question of what consequences the coauthors should face arises. Furthermore, authorship and authorship order (eg, first vs second) have significant implications in academia and are used in determining academic achievement for the purposes of promotion, allocated research time, and funding.

According to the International Committee of Medical Journal Editors (ICMJE), the designation of “author” should be based only on “1) substantial contributions to conception and design, or acquisition of data, or analysis and interpretation of data; 2) drafting the article or revising it critically for important intellectual content; and 3) final approval of the version to be published.”⁵All three conditions must be met. Furthermore, “acquisition of funding, the collection of data, or general supervision of the research group, by themselves, do not justify authorship.”⁵

Rennie and colleagues published a commentary in the Journal of the American Medical Association (JAMA) that noted the increasing decline of meaning and utility of the concept of “author” owing to increasing complexity in relationships between coauthors and more specialized contributions and evidenced by an increase in the number of coauthors per manuscript.⁶The authors concluded that a concept of “contributors” rather than coauthors had become necessary for readers to more accurately assess credit and assess accountability, both crucial components of medical literature. Furthermore, Rennie and colleagues argued for the necessity of a guarantor for the entirety of the manuscript and for the data on which the manuscript is based. The notion of public and published specific contributions per author had thus become center stage, whereas previously relative contribution seemed to have had more importance.

In an effort to ensure true authorship, TheLancet (Lancet) began to require signatory statements of manuscript and research contribution,⁷followed by the British Medical Journal (BMJ),⁸ Radiology,⁹and the American Journal of Public Health.¹⁰The initial experience was very positive.¹¹A large number of medical journals now require authorship contribution forms prior to publication. Each author must declare and satisfy a minimum number of contributions and conditions to merit coauthorship.

This brings up the issue of author order. Perhaps it is possible to differentiate the “true” or “deserving” authors based on author order. With the exception of a minority of journals that list authors alphabetically, most list authors according to author-rated level of contribution. It is our impression that in most medical cultures, the first author contributed more than the second author, and so forth, whereas the author in the last position tends to be the “senior” author or “principal investigator” and therefore made a substantial contribution. However, the decision about author order may actually have much to do with author seniority rather than contribution.

The objective of this study was to assess the relationship between an author's place in the author list and the type of contribution reported by that author and to develop a proposal to modify the practice of author order. We included articles from JAMA, BMJ, the Canadian Medical Association Journal (CMAJ), and Lancet, each of which has a slightly different method by which authorship contribution information is detailed.

Results

A total of 109, 62, 106, and 94 articles were analyzed from JAMA (2001-2003), CMAJ (2001-2003), BMJ (1998-2000), and Lancet (1998-2000), respectively. Contributions for 2,586 authors were recorded (371 for each of first, second, and last authors and 1,473 for middle authors). The kappa values for agreement on studies that were read by two raters were high for all contribution categories, ranging from 0.92 to 1.00.

Figure 1 shows the percentage of authors in each position who participated in the various categories of contribution. Visual examination reveals several patterns. The first authors had the greatest participation in most categories and were generally responsible for planning the study, performing the study, including data acquisition and analysis, and drafting the manuscript. Authors in the last position were generally responsible for planning the study, critically revising the manuscript, and supervising. The second authors, and in particular the corresponding authors, played dominant roles in most aspects of the research, comparable to that of the first author (Figure 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Author contribution of original research manuscripts published in JAMA (2001-2003), Lancet (1998-2000), CMAJ (2001-2003), and BMJ (1998-2000).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Contributions of corresponding authors (all journals combined; n = 370).

Combining data from all journals, few authors specified a contribution to “data management” or “obtaining funding” or reported that they were the “principal investigator or guarantor” (see Figure 1).

Statistical analysis showed that some of the differences in the levels of contribution by author order shown in Figure 1 were statistically significant. For three of the categories (interpretation of data or statistical analyses, drafting the manuscript, and supervision), there is a high probability (> 95%) that all author positions have different contribution frequencies. Each of the other categories has at least one pair of author positions with similar levels of contribution frequency (ie, a 95% confidence interval for an odds ratio that crosses 1). This includes the comparisons of second with last author for conception and design all author positions for data management; first with second author and second with middle author for data acquisition; second with middle author and middle with last author for manuscript revision; first with last author for obtaining funding; all author positions for administrative, technical, or material support; and second with last author for principal investigator or guarantor.

Authors publishing in JAMA were most likely to fulfill the ICMJE criteria for authorship, whereas authors publishing in Lancet were least likely (Table 1). Middle authors were least likely to fulfill authorship criteria (see Table 1). Combining data from the four journals (JAMA, Lancet, CMAJ, and BMJ), 93.5% of corresponding authors (n = 370) fulfilled ICMJE criteria 1 and 2 for authorship.

Figure 2 shows the pattern of contribution of corresponding authors. Corresponding authors were most commonly in the first author position (76.8%, n = 284), followed by the last position (13.5%, n = 50), second position (7.6%, n = 28), and middle position (2.2%, n = 8). One article had two corresponding authors, whereas two articles had none.

Discussion

In this study, we were able to demonstrate two findings. First, author order correlates with a pattern of contribution indicating that the first, second, and last authors had the highest levels of participation in most areas. Second, according to data published by the authors themselves, a significant minority of authors do not merit the designation “authorship” according to the ICMJE criteria.

JAMA authors demonstrated the highest proportion of authors following the ICMJE guidelines. This is a result of the method of eliciting participation using a specific checklist. The JAMA checklist outlines how many categories have to be designated for someone to be considered an author, thereby ensuring higher compliance. This may or may not reflect a true higher rate of participation. This also represents a limitation of using the data collected to make interjournal comparisons as some journals had more extensive contribution bylines than others.

It was also clear that the majority of articles published did not indicate any author responsible for obtaining funding (see Figure 1). This would imply that the majority of published clinical research did not require any funding. This seemed odd to us. The corresponding author, who deals with the journal editors and readers, was most commonly the first or last author.

It was not too long ago that the idea of “contributorship” gained momentum, wherein all authors would descriptively detail the parts of the study for which they were responsible.^6-8,13A great deal of hope was placed on this new proposal, and it was incorporated to varying degrees by many medical journals thereafter. It was clear, however, that its effect would have to be monitored and its utility demonstrated,²given that it created yet one more step in what was already often a laborious process of manuscript submission.

Based on our results, we would like to put forth a new proposal that may clarify the accountability of authors, as well as alleviate the issue of honorary authorship. We propose that journals eliminate the traditional method by which authors are listed, in which the first and the last positions are generally (but not always) viewed as the most significant, followed by the second position, third position, fourth position, and so forth. Instead, this method should be replaced with a system in which there are only three designations: primary author(s), senior or supervisory author(s), and contributing author(s). For each designation, more than one author should be allowed, if deemed appropriate by the study investigators.

Primary authors are those who (1) plan the study, (2) execute the study, (3) write the manuscript (ie, are primarily responsible for the drafting and redrafting of the manuscript), and (4) guarantee the accuracy of the data or manuscript. Senior or supervisory authors are those who (1) plan the study, (2) substantially revise the manuscript, (3) supervise, and (4) guarantee the accuracy of the data or manuscript. Finally, contributing authors are those who (1) follow the ICMJE authorship criteria for authorship but (2) do not fulfill the criteria for either the primary author or senior or supervisory author (Table 2). Authors should fulfill all criteria of their respective author category to receive that designation.

Primary and senior or supervisory authors should be responsible for guaranteeing the accuracy of the data on which the manuscript is based; to the best of their knowledge ensure that there was no fraud or error; and be able to explain the results of the entire manuscript. Contributing authors should not be responsible for having the ability to explain the results of the entire manuscript or for guaranteeing the accuracy of all of the data. We would therefore still recommend the requirement of signed declarations of authorship for all authors and of guarantor only for the primary and senior or supervisory authors. By doing so, primary and senior authors would understand that they would all be held accountable for any error or fraud contained in the paper.

The definitions of these three authorship positions still should follow the ICMJE authorship guidelines. Manuscripts may list primary author(s) and contributing author(s), without any senior or supervisory author(s), if this was indicated.

There are several reasons why we believe that our proposed system is an improvement over the current method of listing authors. First, it would standardize the patterns of contribution that seem to be currently followed by authors as observed in this study and in that way would help new or inexperienced authors in providing guidance about what author order means. For example, it would standardize the position of the senior author, whereas now the senior author might sometimes be the second and at other times the last author. Second, it would allow a clearer gradation of authors' relative contribution. Third, by allowing more than one primary author and more than one senior or supervisory author, there is much greater flexibility in assigning relative rank. We are certain that there are many cases in which the decision about which single author to list in the first and last positions leads to strife and potentially arbitrary choices. Fourth, this system would help clarify situations in which there are only two authors as to whether the second author is more of a primary author or more of a senior author. Fifth, this system would clarify which authors are accountable for error and fraud and which are not. The current method of author listing is vague in this regard. Finally, this system will make it easier for those who evaluate relative authors' contributions during processes such as promotion without having to retrieve each paper to examine the author's listed contributions. Many investigators currently annotate their curriculum vitae outlining their role in the paper. A lack of standardization and unusual designations (eg, problems such as declaring yourself to be a senior author when you are in the middle of the pack) would be greatly aided by our proposed system.

There has been somewhat of an analogous reformation occurring in the world of research grant applications. For example, the National Institutes of Health recently sought “input and advice from the scientific community on various concepts associated with permitting more than one Principal Investigator… to be associated with an NIH funded grant, contract, or cooperative agreement” and is now allowing more than one principal investigator from some studies.¹⁴This was due to recognition that research has become increasingly complex, often involving collaboration within a multidisciplinary team of researchers.

Of course, if there are advantages to our proposed change, there must also be disadvantages. Some of these include deciding how to order the lists of first and senior authors if there is more than one, the potential to exacerbate the problem of honorary authors in the contributory category, and the potential to dilute the credit given to those who are now first and last authors. As well, there would need to be discussion regarding how journal indexers would handle the new requirements, which this proposal does not address.

Our proposed authorship system would be a great improvement over the current approach by clearly indicating the relative role and contribution of each listed author, without the ambiguity that exists with the current system of authorship. We recognize that this proposal would ask the scientific community to change a standard practice, similar to the introduction of the standardized abstract about 10 years ago, but believe that the benefits in terms of clarity and flexibility outweigh the inconvenience. Furthermore, the presented proposal is not intended to be definitive and may be refined by other researchers, journal editors, and authors. The proposal is intended to provoke and kick-start a discussion on the topic.