A descriptive analysis of authorship within medical journals, 1995-2005.

Introduction: The emphasis on publications for promotion in academic medicine would lead one to the theory that authorship numbers would increase proportionally with this emphasis. To investigate authorship trends across a number of periodicals, we performed a descriptive study comparing two full years of published articles spaced ten years apart from five medical journals.

Methods: Physician reviewers each reviewed all articles of one medical journal for the 1995 and 2005 publication years. Reviewed journals included Academic Emergency Medicine (AEM), Annals of Emergency Medicine (AnnEM), Annals of Internal Medicine (AIM), Journal of Trauma (JT), and New England Journal of Medicine (NEJM). Data collected for each article were number of authors, ordinal number of the corresponding author, type of study described, whether the described study was a multicenter trial, whether authorship listed included a "study group," and whether any author was also an editor of the journal.

Results: A total of 2927 articles were published in the five journals in 1995, and of these, 1401 (47.9%) were analyzed after the exclusion criteria had been applied; for 2005 a total of 3630 articles were published and of these, 1351 (37.2%) were included in the analysis. Across all five journals the mean number of authors per article increased from 4.66 to 5.73 between 1995 and 2005 (P < 0.0001), and four of the five journals individually had statistically significant increases in the number of authors per article. More articles had a journal editor as an author in 2005 (increased from 7.8% to 11.0%, P = 0.004), though no single journal had a statistically significant increase.

Conclusion: We describe a trend of increasing mean authors, editorial authorship, study groups, and multicenter trials over time with fewer solo authors now publishing original research or case reports. The academic medical community must pursue an authorship requirement consensus to assure that a standard of contribution for all authors on a given paper is met.

Key Words: periodicals, publishing, authorship, clinical trials

**********

Among Promotion and Tenure committees (PTC), an academic physician's record of publications is an important factor in decision making regarding retention as a faculty member and promotion to higher levels of professorship. However, little formal attention has been paid to trends in the assignment of authorship. Many methods for deciding authorship order based upon contribution have been proposed and used, but few papers have been written describing the current characteristics and authorship trends among major medical journals. (1,2) The emphasis on publications for promotion would lead one to the theory that authorship numbers would increase proportionally with this emphasis. A previous study of obstetrics and gynecology literature has demonstrated this to be true in the literature of that specialty. (3) To investigate authorship trends across a number of medical specialties, we performed a descriptive study comparing two full years of published articles spaced ten years apart from five major medical journals.

Methods

Five physician reviewers each reviewed all articles of one major medical journal for the 1995 and 2005 publication years, using a standardized data collection spreadsheet. Reviewed journals included Academic Emergency Medicine (AEM), Annals of Emergency Medicine (AnnEM), Annals of Internal Medicine (AIM), Journal of Trauma (JT), and New England Journal of Medicine (NEJM). Article types excluded from further analysis were position statements, editorials, poems and creative writing, commentaries, and any correspondence regarding previously published articles. Data collected for each article were number of authors, ordinal number of the corresponding author, type of study described (categorized as one of the following: prospective-interventional, prospective-observational, retrospective, retrospective-database, cadaveric, animal, simulation/modeling, survey/questionnaire, review article, case report/series, visual diagnosis, or other), whether the described study was a multicenter trial, whether authorship listed or included a "study group," and whether any author was also an editor of the journal. The differences in mean authorship number between the two years were analyzed using nonpaired t tests. Comparative analysis of other data was performed using [chi square] tests. Statistical significance was defined as P < 0.05.

Results

A total of 2927 articles were published in the five journals in 1995, and of these, 1401 (47.9%) were analyzed after the exclusion criteria had been applied; for 2005, a total of 3630 articles were published and of these, 1351 (37.2%) were included in analysis. Across all five journals, the mean number of authors per article increased from 4.66 to 5.73 between 1995 and 2005 (P < 0.0001), and four of the five journals individually had a statistically significant increase in the number of authors per article. The greatest mean increase was found in NEJM (5.00-6.81, P < 0.0001). Histograms of article counts by number of authors for each journal are shown in the Figure. For all journals combined, the proportion of solo author articles decreased from 8.1% to 4.5% (P < 0.001), and papers with more than eight authors increased from 11.4% to 16.7% (P < 0.001).

The first author was the corresponding author in the majority of papers in 1995 (67.5%) and 2005 (61.1%); the second author was corresponding in 8.6% and 7.4% of papers for the respective years. The last author (in articles with three or more authors) was corresponding for 11.6% and 12.6% of articles respectively, and 10.9% and 14.9% of articles did not list a corresponding author.

Across all journals, there was a decrease in the number of prospective articles (34.6% to 30.1%, P = 0.010), and a decrease in the number of case reports (21.6% to 17.4%, P = 0.006). Otherwise, the types of articles published were not significantly different between the two years. Publication types by journal and year are shown in the Table.

Across all journals, more articles had a journal editor as an author in 2005 (increased from 7.8% to 11.0%, P = 0.004), although no single journal had a statistically significant increase. The percentage of studies that listed or included "study groups" of authors increased from 4.6% to 8.6% (P < 0.001), with the largest absolute individual increase occurring in NEJM (9.4% to 17.5%, P < 0.001), and the largest relative increase occurring in AnnEM (0.5% to 4.6%, P = 0.013). The percentage of articles that were from multicenter trials increased from 9.0% to 12.2% (P = 0.006), with an increase from 11.7% to 20.6% (P < 0.001) in NEJM, and an increase from 4.2% to 10.0% (P = 0.029) in AnnEM. For all other individual journals, no significant change in percentage of "study groups" or multicenter trials was detected. When the NEJM data were reanalyzed with all multicenter and study group trials excluded, the mean number of authors in both years and the difference in means decreased significantly, and the increase in authors observed was no longer statistically significant (1995: n = 407, mean = 4.22; 2005: n = 356, mean = 4.59, P = 0.171).

Discussion

A trend toward an increase in author count was seen among all journals between the years studied, with solo authorship becoming rare. It is worthwhile to consider whether the trend, although statistically significant, is also "clinically" or intellectually significant. The mean increase in authors across the five journals over ten years was approximately one author per article. Some might argue that one more author per article makes little difference, and should be an acceptable fluctuation. This increase, however, accounts for approximately 23% of the mean number of authors in 1995. Were the observed trend to hold over the history of medical literature, it would be expected that solo authorship would be found in the majority of articles published 50 to 60 years ago, and that 20 years from now, articles will have an average of 8 to 9 authors. Given minimal change in the types of articles presented, as we observed, few would argue that a trend of this magnitude is not "intellectually" significant. To consider this in different terms, for many clinical parameters, a change of 23% would likely be considered clinically significant. However, for intellectual or academic parameters such as this one, the significance of such a change is ultimately subject to the judgment of the individual reader, as no demonstrable scientific or clinical outcomes are associated with the change.

As for why the observed trend exists, there are a number of plausible explanations. It could be argued that with increased emphasis on evidence-based practice, journals are under increasing pressure to publish larger, often multicenter trials. Multicenter and "study group" trials appear to account for a significant proportion of the increase in authorship in NEJM, for instance. However, for much longer than the past ten years, the periodical has tended to be host to articles from some of the largest studies every year, so it is reasonable to wonder whether the trials themselves are becoming larger or simply the number of authors per trial is increasing. It should be noted that the proportion of prospective trials, and specifically prospective-interventional trials, did not increase in any journal studied. Were the trend in author numbers simply due to pressure to publish more material for use in evidence-based practice, it would be expected that these types of trials would be found to have increased in number.

[FIGURE 1 OMITTED]

Another possible explanation is that academic institutions are creating pressure on faculty to publish more articles, and thus more faculty members are being listed as authors on articles published. Although some efforts have been made in recent years to increase the PTC recognition of clinical teaching, (4,5); publication achievement still appears to be perceived as critically important for academic advancement by most academic physicians. This perception of PTC emphasis upon scholarly publishing may or may not be congruent with reality. Honorary authorship, inclusion of an author who has not contributed significantly to study conception and design, acquisition of data, analysis and interpretation of data, drafting or revision of the article, still appears to be quite common, with between 4 and 60% of articles in one study containing at least one honorary author. (6)

Editorial publication within the editors' own journal has become more frequent as well, again without a clear explanation. Arguably, most journal editors are respected and productive members of their medical specialty, and might well be expected to demonstrate excellent academic productivity. Many major journals have blinded review of submissions, though lead and associate editors may often know who the key authors of these manuscripts are. Reviewer blinding has been demonstrated in at least one journal to be partially effective at best, with successful blinding occurring only about 50% of the time. (7)

There are a number of limitations of this study. The greatest is the limited number of journals included. This raises the question of whether our findings generalize to the medical literature as a whole. The cause for this limitation is simply the amount of resources required to review large volumes of articles. The journals we selected were a few of those most commonly cited in publications and presentations done by members of our department, and were not intended to represent a comprehensive sampling of the literature. Given ample physician-reviewer time, our methodology could be applied to any number of journals, and a comprehensive sampling could be achieved. We think there is a strong possibility that our findings would generalize, as prior, similarly limited studies yielding similar results. (3) Similarly, our observed trend could be examined in greater detail if resources were available to review a greater number of years of each journal. The data would be somewhat more difficult to gather for publications before 1995, as this was one of the first years for which all the journals we studied made all of their articles available online; the use of a web browser to view the articles appeared to us to be more efficient than turning pages. Another potential limitation is that of bias by the reviewers. Although our study is subject to all forms of bias common to literature reviews, we hope to minimize these by including all articles in each year, and by making our data collection such that the only datum gathered which is subject to individual interpretation is the type of study. As many articles are not clearly from one major class of study or another, the reviewers were at times responsible for judging in which class the article best fit. Since the same reviewer reviewed the 1995 and the 2005 literature for each journal, the judgments on types of studies should be relatively consistent within each journal between the two time points.

A variety of authorship standards exist, both to set standards for authorship and to help guide authorship order. Influential criteria for authorship developed by the International Committee of Medical Journal Editors has been developed (8); however, adoption of these criteria by medical specialty organizations and journals has not been universal. Indeed, recent calls to make this adoption more pervasive have been echoed. (9) No clear consensus on either of these topics has been reached, although many journals are now requiring contribution statements. Without further guidance from individual journals and the academic medical community, further increases in mean author numbers, "honorary authors," and editorial authors of noneditorials is likely to continue.

We describe a trend of increasing authors, editorial authorship, study groups, and multicenter trials over time with fewer solo authors now publishing original research or case reports. The academic medical community must pursue an authorship requirement consensus to assure that a standard of contribution for all authors is met.

References

1. Bhandari M, Einhorn TA, Swiontkowski MF, et al. Who did what? (Mis) perceptions about authors' contributions to scientific articles based on order of authorship. J Bone Joint Surg Am. 2003;85-A:1605-1609.

2. Gaeta TJ. Authorship: "Law" and order. Acad Emerg Med 1999;6:297-301.

3. Khan KS, Nwosu CR, Khan SF, et al. A controlled analysis of authorship trends over two decades. Am J Obstet Gynecol 1999;181:503-507.

4. Carey RM, Wheby MS, Reynolds RE. Evaluating faculty clinical excellence in the academic health sciences center. Acad Med 1993;68:813-817.

5. Atasoylu AA, Wright SM, Beasley BW, et al. Promotion criteria for clinician-educators. J Gen Intern Med 2003;18:711-716.

6. Bates T, Anic A, Marusic M, et al. Authorship criteria and disclosure of contributions: comparison of 3 general medical journals with different author contribution forms. JAMA 2004;292:86-88.

7. Yankauer A. How blind is blind review? Am J Public Health 1991;81:843-845.

8. International Committee of Medical Journal Editors (ICMJE). International Committee of Medical Journal Editors (ICMJE): Uniform Requirements for Manuscripts Submitted to Biomedical Journals: writing and editing for biomedical publication. Haematologica 2004;89:264.

9. Jones AH. Can authorship policies help prevent scientific misconduct? What role for scientific societies? Sci Eng Ethics 2003;9:243-256.

Marc E. Levsky, MD, Alex Rosin, MD, Troy P. Coon, MD, William L. Enslow, DO, and Michael A. Miller, MD

From the Department of Emergency Medicine, C. R. Darnall Army Medical Center, Fort Hood, TX; and the Department of Emergency Medicine, Dwight D. Eisenhower Army Medical Center, Augusta, GA.

Reprint requests to Marc E. Levsky, MD, Department of Emergency Medicine, C.R. Darnall Army Medical Center, Fort Hood, TX 76544. Email: [e-mail] 

The opinions expressed in this article are those of the authors, and do not necessarily reflect those of the United States Army or the Department of Defense.

Accepted September 8, 2006.

RELATED ARTICLE: Key Points

* Among the five journals studied, the mean number of authors on each article increased by approximately one, or 23%, from 1995 to 2005.

* In 2005, editorial authorship of noneditorials had increased, as had the number of articles written by study groups, and the number of articles that were the product of multicenter trials.

* The categories of studies published did not change significantly between 1995 and 2005.

* The cause of these trends is unknown, but may involve increasing pressure on academic physicians to publish articles.

Table. Types of article by journal and year

                            AEM                         AnnEM
Article Types        1995        2005     P        1995        2005

animal             14  11.4%   10   6.4%  0.137   17   7.9%    3   2.3%
cadaver             0   0.0%    1   0.6%           0   0.0%    0   0.0%
case report        15  12.2%   13   8.3%  0.278   42  19.5%   10   7.7%
prosp-interv       10   8.1%   19  12.1%  0.279   27  12.6%   22  16.9%
prosp-observ       35  28.5%   44  28.0%  0.937   56  26.0%   22  16.9%
retro/database      5   4.1%   21  13.4%  0.008    9   4.2%   22  16.9%
retrospective      16  13.0%   12   7.6%  0.138   33  15.3%    3   2.3%
review             11   8.9%   10   6.4%  0.417   16   7.4%   14  10.8%
simulation          0   0.0%    1   0.6%           0   0.0%    1   0.8%
survey             17  13.8%   20  12.7%  0.791    9   4.2%    7   5.4%
visual diagnosis    0   0.0%    0   0.0%           0   0.0%   23  17.7%
other               0   0.0%    6   3.8%           6   2.8%    3   2.3%
total             123         157                215         130

                  AnnEM            AIM                     JT
Article Types     P          1995        2005     P        1995

animal             0.031    1   0.5%    0   0.0%          50  12.3%
cadaver                     0   0.0%    0   0.0%           5   1.2%
case report        0.003   14   7.4%   24  16.8%  0.008  130  32.1%
prosp-interv       0.26    47  25.0%   30  21.0%  0.391   24   5.9%
prosp-observ       0.05    46  24.5%   20  14.0%  0.018   56  13.8%
retro/database    <0.001    7   3.7%   15  10.5%  0.014   27   6.7%
retrospective     <0.001   32  17.0%   11   7.7%  0.012   95  23.5%
review             0.288   33  17.6%   38  26.6%  0.048   11   2.7%
simulation                  0   0.0%    0   0.0%           1   0.2%
survey             0.608    0   0.0%    0   0.0%           5   1.2%
visual diagnosis            0   0.0%    0   0.0%           0   0.0%
other              0.785    8   4.3%    5   3.5%  0.725    1   0.2%
total                     188         143                405

                         JT                NEJM
Article Types       2005      P        1995        2005

animal             59  13.1%  0.747    0   0.0%    0   0.0%
cadaver            12   2.7%  0.135    1   0.2%    0   0.0%
case report        98  21.7%  0.001  101  21.5%   90  19.1%
prosp-interv       32   7.1%  0.49    98  20.9%  101  21.5%
prosp-observ       58  12.9%  0.678   86  18.3%   58  12.3%
retro/database     49  10.9%  0.031   20   4.3%   17   3.6%
retrospective      95  21.1%  0.4     47  10.0%   42   8.9%
review             20   4.4%  0.179   59  12.6%   58  12.3%
simulation          4   0.9%  0.22     1   0.2%    4   0.9%
survey             12   2.7%  0.135    0   0.0%    3   0.6%
visual diagnosis   12   2.7%          54  11.5%   96  20.4%
other               0   0.0%           3   0.6%    1   0.2%
total             451                470         470

AEM, Academic Emergency Medicine; Ann Emerg Med, Annals of Emergency
Medicine; Ann IM, Annals of Internal Medicine; J Trauma, Journal of
Trauma; NEJM, New England Journal of Medicine; prosp-intern; prospective
interventional; prosp-observ, prospective observational; retro-database,
retrospective database.