Discussion
Our data suggest that the respondents were largely unaware of internationally accepted authorship criteria. Most postgraduate medical trainees felt it should be an area of focus for their respective College of Medicine to detail their own authorship policy and also to provide a means of resolving authorship disputes. The rates of perceived and ICMJE-defined honorary authorship are of concern and require further study. We attempted to quantify the rate of respondents who did not perceive the inclusion of an honorary author (unperceived honorary authorship) and found this rate was high.
There needs to be a paradigm shift from the current system, where enforcement of ethical authorship practices is shifted away from journals editors. Instruction on the topic should be provided to medical trainees throughout medical school and continued throughout postgraduate training and enforced through accreditation standards. Furthermore, a process should be outlined to resolve authorship disputes which is available to faculty and medical trainees. Undertaking these measures may lead to an improvement in awareness of the topic and encourage researchers to have an open discussion on the authorship prior to the commencement of a research project. Having a framework for each coauthor's responsibilities may encourage those involved in the project to adhere to them, and this framework can also be used to resolve authorship conflicts which may arise during or after the completion of a research project. It is hoped that this paper will encourage medical trainees to advocate for more training and support on ethical authorship practices and that those in positions of power will support these changes. Ultimately, there may be instances where the right decision was made to include a coauthor who did not meet the ICMJE authorship criteria, but this decision cannot be made unless researchers are fully aware of the criteria and have a support system in place to resolve an authorship dispute if they decide to exclude a coauthor.
It is of interest to compare the rates of honorary authorship obtained in our paper with those reported in the literature. The five most recent papers on honorary authorship in the medical field used a similar methodology to ours and reported rates of perceived honorary authorship of 18.0–33.4%, which is slightly below that reported in our paper (38.1%). Four of these studies also reported a rate of ICMJE-defined honorary authorship of 50.3–57.6%; the rate reported in our paper (57.3%) is at the upper range of these studies. It is important to note that these papers listed non-authorship tasks and defined the rate as the proportion of respondents who noted at least one coauthor who only completed one or more of these tasks. We decided to directly ask respondents which of the three original criteria coauthors did not perform, and also included only performing the statistical analysis to avoid any ambiguity (Box 1). These five papers also looked at the correlation of factors in relation to the rates of perceived and ICMJE-defined honorary authorship and found several associations. We did not perform this analysis in our study. It does appear that the rates of honorary authorship extracted from our data are in keeping with those previously reported.
The manner in which the rate of honorary authorship is determined in a study is also important to consider. Marusic et al investigated the effect of the structure of the question used to extract the rate of honorary authorship (open-ended, categorical and instructional). They found that the instructional format had a lower rate than the open-ended and categorical rates, and proposed that the most likely reason for this discrepancy was the desire of the corresponding authors to conform to socially (editorially) desirable answers. Thus, our rate of perceived honorary authorship could underestimate the true rate of honorary authorship, as the manner in which the question was posed may have discouraged respondents from answering in the affirmative in cases where honorary authorship existed. However, caution must be exercised when using the ICMJE criteria as the gold standard for defining authorship. When researchers who were largely unaware of the ICMJE criteria were introduced to them, they largely disagreed that all three of the older criteria needed to be followed, and that fulfilling just one should merit authorship. It could be reasonable to assume that the true rate of perceived honorary authorship in our study lies somewhere between our reported rates of perceived and ICMJE-defined honorary authorship, as some respondents may have falsely responded in the negative to question 14, and others who responded in the negative to question 14 and in the affirmative to questions 17–20 or 22 disagreed with the ICMJE authorship criteria. Although some respondents may have disagreed with the ICMJE criteria in theory, the alarmingly high rate of unawareness of the criteria (90.1%) is a cause for concern. It would be interesting to see if the proportion of perceived honorary authorship increased in future studies after making respondents aware of the ICMJE authorship criteria.
It appears clear that honorary authorship is an issue, and one of the central components of this issue in medical trainees is a lack of awareness of the topic in general and the ICMJE authorship criteria. Our respondents almost unanimously indicated that institutions should provide instruction and support on this issue. The effect of authorship instruction on the agreement of medical students with the ICMJE criteria was investigated by Hren et al, who found that both groups given and not given instruction on authorship recognised the first two ICMJE criteria as important but only the group given authorship instruction considered the third criterion to be important (Box 1). This finding is consistent with previous literature, where the third criterion was found to be the least important to researchers who were also largely unaware of the ICMJE authorship criteria. Hren et al also investigated whether instruction about formal authorship criteria affected medical students' decisions about authorship dilemmas and found no difference between the groups, but did note that those given instruction on authorship attempted to use the ICMJE criteria to make their decision. These findings suggest that trainees are open to using the ICMJE criteria for authorship, but more extensive training may be required on the topic for them to use the criteria properly.
Medical students surveyed on the topic by Karani et al suggested various means of providing this training, including self-directed modules, discussion guides and structured didactic sessions. The authors of this study went on to suggest the creation of a multiformat curriculum to which medical trainees could be exposed throughout their training. This adds to the growing argument of moving the responsibility of authorship policing away from journal editors, as the use of research contribution declaration forms by journals has been ineffective in addressing unethical authorship practices.
One of the limitations of this study was the low response rate. However, it is much the same as that reported in similarly designed studies on the topic (15.8–36.8%). An older study on the topic performed by coauthors affiliated with the American Medical Association obtained a response rate of 70.3%. Second, we presented data from one postgraduate medical training centre, which limits the general applicability of our findings. The findings of this study would probably not differ from other training centres in Canada or the USA as they follow the same accreditation standards. Of note, a previous unpublished iteration of this study was attempted across all English speaking medical schools in Canada. This study was limited by very low response rates to the online survey and seven centres declined to participate in the study. Of the seven centres that declined to participate, three provided us with the policy used to make this decision and the other four declined to explain their decision. In addition, the accuracy of the responses was susceptible to recall bias, our survey was not validated and our respondents were skewed to lower years of training, which may have had an unknown effect on our findings.
Despite the significant weaknesses of this survey-based study, our results support the need to change the medical education system to include support and education about ethical authorship practices. In addition, we hope this paper encourages researchers to study this topic in medical trainees. However, before embarking on further research on the topic, a validated survey must be created to ensure the results can be properly interpreted. Furthermore, encouraging involvement of multiple training centres and attempting to administer questionnaires in person may further improve the applicability of the results. To date, authorship policy has focused on original research manuscripts but, as further research is undertaken on this issue and policy changes are contemplated, it would be advantageous to recognise that different authorship guidelines are needed for the multitude of manuscript types now accepted by journals (eg, case reports, clinical reviews, systematic reviews).