Research integrity among PhD students within clinical research at the University of Southern Denmark

Responsible conduct of research (RCR) is the basis for the credibility of all research being carried out. Research misconduct is most often defined as falsification, fabrication or plagiarism [1, 2]. A recent report from an

expert committee on the Danish system that handles

research misconduct produced a number of recommendations for changes [3]. Among these changes, a revision was recommended for the definition of research misconduct so that it resembles the American definition.

The amendment of the Danish act on research misconduct came into force on July 1, 2017 [4]. The Danish definition of research misconduct is now (English translation from The Danish Ministry of Higher Education and Science): "Fabrication, falsification, and plagiarism which has been committed willfully or with gross negligence in planning, performing, or reporting of research. Fabrication: Undisclosed construction of data or substitution with fictitious data. Falsification: Manipulation of research material, equipment or process as well as changing or omitting data or results making the research misleading. Plagiarism: Appropriation of others' ideas, processes, results, texts or specific terms without rightful crediting" [5].

The law defines the grey zone between research misconduct and RCR, also referred to as questionable research practice (QRP), as: "Breaches of current standards on RCR, including those of the Danish code of conduct, and other applicable institutional, national and international practices and guidelines on research integrity" [5].

The true frequency of research misconduct may be very difficult to estimate. Meta-analyses of survey data report that a pooled weighted average of 1.97% of researchers admitted to having fabricated or falsified data or results at least once [6], and a pooled estimate of 1.7% reported having committed plagiarism [7]. QRP seems to occur much more frequently with up to 33.7% admitting to having engaged in such practices [6].

Early career researchers have a crucial role in forming future research environments. Several survey studies of medical PhD students' knowledge, attitude and experiences related to research integrity have been carried out in recent years in Norway and Sweden [8-11]. Similar insights have not yet been obtained in Denmark. The Faculty of Health at University of Denmark (SDU) introduced a mandatory course in RCR for PhD students in 2014. The course was introduced to educate young researchers in responsible research and thereby mitigate potential research ethical conflicts. We therefore undertook a survey of knowledge of, attitudes to and experiences with research misconduct among PhD students within the field of clinical research at SDU.

METHODS

The questionnaire combined questionnaires initially developed in Sweden [8] and the USA [12], and previously used in Swedish and Norwegian studies [9-11]. The paper-based, English questionnaire was converted into an online, electronic questionnaire (SurveyXact), with a few linguistic adaptions to enhance comprehension. The questionnaire was distributed to all PhD students affiliated with either of the two clinical departments at the Faculty of Health Sciences, SDU (as of 30 May 2017; n = 330). In order to obtain the most valid picture of the present state of knowledge, attitudes and practices among the PhD students, we included all PhD students in the study, and not just newly enrolled PhD students.

An invitation to participate, along with a link to the questionnaire, was sent out by e-mail on 2 June 2017, with two weeks for completion. Two follow-up e-mails were sent to non-respondents. A recommendation to participate was also included in department staff newsletters.

Participation was voluntary. The mail invitation informed recipients that data from the survey would only be processed statistically and would not be traced back to individuals. According to Danish law, questionnaire surveys are not recorded with the Health Research Ethics Committee.

Results are presented based on descriptive statistics. Chi-squared tests were used on categorical variables to identify any dependence between background variables and responses. For continuous variables,

t tests were applied. 5% was used as the significance

level.

Trial registration: not relevant.

RESULTS

A total of 329 PhD students successfully received the invitation to participate. In all, 165 completed the questionnaire in full or in part, yielding an overall response rate of 50% (similar response rates within each of the participating departments: 50% and 51%). Table 1

presents the characteristics of the respondents, the majority holding a master's degree from Denmark, being female and doing clinical research. Length of enrolment shows an equal division between first-, second-, and third-year PhD students. This is in line with current enrolment. Likewise, the gender split of the data does not give rise to specific concerns (63% of the PhD students enrolled at the Faculty of Health Sciences are female [13]).

The PhD students’ knowledge of research misconduct practices is summarised in Table 2.

Attitudes towards research misconduct are shown in Table 3.

Experiences with research misconduct are summarised

in Table 4.

Results show that many PhD students have heard about research misconduct behaviour (Table 2): 18-34% report to have heard, within the past year, about researchers who have plagiarised, falsified or fabricated data, or plagiarised publications. No one reports this to occur in their own department. The knowledge of written department policies is limited (Table 2). The highest knowledge levels occur for policies regarding

application for funds or use of funds (24% and 35%). Overall, however, large proportions reply "I am uncertain" to the questions worded "Does your department have a written policy about …" (40-57%).

Nearly all find it unacceptable to report experimental data without actually having conducted the

experiment, or to alter such data to improve the significance of experimental results (94-99%) (Table 3). Likewise, taking credit for the words, writing, data or ideas generated by someone else is considered unacceptable (97-99%), as is falsifying or fabricating data to speed up publication (97%). The large majority also finds it inappropriate to selectively omit contradictory results (90%) or to try out different methods of analysis until a statistically significant result is found (84%).

Apparently, writing grant publications is regarded a different business, not subject to exactly the same behavioural norms as research; only about half of the respondents regard it as important to report data as truthfully in a grant application as in a publication.

Most agree to having an ethical obligation to act if they discover that someone is committing research misconduct (88%). Slightly fewer would be willing to act personally by reporting for instance a co-worker or a

supervisor (71-74%). Slightly less than half (46%) think that co-authors must equally share the blame if a publication is produced on fabricated data, but only about one in five (23%) think that all co-authors should receive the same punishment.

Results show small but statistically significant differences in responses based on gender. Female PhD students agreed significantly more to it being inappropriate to alter experimental data (p = 0.045), or to take credit for the words or writings of others (p = 0.026), or for their ideas (p = 0.028). Likewise, female PhD students agreed more that all must equally share the blame and punishment if fabricated data are discovered in a paper (p = 0.019 and p = 0.02).

None of the participating PhD students have – or admit to having – committed research misconduct themselves in terms of fabricating, falsifying or plagiarising data, or plagiarising publications (Table 4). A few state that they have felt under pressure to either falsify data (1%) or present results in a misleading way (3%). The real pressure on the PhD students concerns authorships: One in five (22%) state having felt an unethical pressure (within the past 12 months), when it comes to either the inclusion or the order of authors. One in ten replies ‘I am uncertain’ to this item (10%, as opposed to 1-3% in remaining questions in the section): thus, a total of 32% think they may have been under unethical pressure regarding the inclusion or order of authors. Perceptions of unethical pressure regarding design/method, analysis and results also exist, but are not

nearly as prevalent (2-4%). In summary, only very few state that they have been affected by the consequences of research misconduct (1%).

Results show perceived pressure concerning authorships to differ with length of enrolment (p = 0.039). Among third year PhD students, 30% state to have been under pressure with regard to inclusion or order of authors, compared to only 16% and 19% among first and second year PhD students.

DISCUSSION

None of the respondents reported having fabricated,

falsified or plagiarised data or publications (Table 4), or knew of anyone in their department who had done so (Table 2). Pressure to falsify data or present data in a misleading way was reported by a small number of respondents (Table 4), indicating that albeit it seems to occur at a low frequency, research misconduct involving PhD students is taking place.

Most were aware of their ethical obligation to act on research misconduct actions of others, but showed limited willingness to actually take action if a co-worker or supervisor commits research misconduct (Table 3). This may be explained by the unequal power relationship between the PhD student and his or her supervisor, which may keep some from such an overt act as reporting. Furthermore, misconduct may involve the PhD student personally, for instance through co-authoring of a flawed publication.

Interestingly, less than half agreed that co-authors should equally share the blame if a publication is produced on fabricated data and even fewer (23%; Table 3) agreed that all co-authors should get the same punishment. This indicates that co-authoring is not equated with equal responsibility. This is in line with the authorship guidelines of The Danish Code of Conduct for Research Integrity [14]. Inspired by the internationally acknowledged and commonly used authorship guidelines by The International Committee of Medical Journal Editors (ICMJE; the Vancouver guidelines) [15], these guidelines state that all authors are responsible for the content of the publication. The Danish guidelines do, however, also state that the responsibility should be assessed with due account of area of expertise, level of experience and seniority (as well as other relevant factors).

A disturbing 22% of respondents state that they have been exposed to unethical pressure concerning

authorships (Table 4). Who is allocated authorships and the position in the author list of a publication is crucial to researchers, and reflects the intensely competitive environment of much medical research where being first author and being an author of many publications meet the current reward mechanisms in research (money and recognition). Perceived unethical pressure may also be due to poor matching of expectations regarding contributions prior to a research project; neither the Danish Code of Conduct for Research Integrity nor the ICMJE guidelines for authorship include specific descriptions or instructions for the ordering of authors on research publications [14, 15].

The perceived pressure on authorships is considerably higher than in the studies conducted in Sweden and Norway where the same questionnaire was used [8-11]. The studies are not completely comparable though: While we have included all enrolled PhD students in the study population, the studies in Sweden and Norway covered only newly enrolled PhD students. Social pressure regarding authorships is likely to be higher in later phases of the PhD process, which is when the PhD student typically starts to more actively publish.

Knowledge of existing written department policies on research misconduct and QRP was limited (Table 2). This indicates that written department policies are somewhat invisible, at least to PhD students. Department policies may also be unknown or invisible to the supervisors and other senior researchers who interact and collaborate with the PhD students. This may contribute to the pressure regarding authorship perceived by the PhD students, as well as to other uncertainties.

The small study population is a limitation since the study is based on PhD students enrolled in only one of the Danish universities. The question is, however, how generalisable our study results actually are? Larger

studies are needed to clarify if our findings – such as the high fraction perceiving unethical pressure on authorships – reveal a common picture among PhD students in Danish research. Another limitation may be in the individual’s understanding of the questionnaire items. Here, we have been very cautious to present definitions in the clearest way possible. We cannot exclude, though, from the existing questionnaire, that the identified gender differences are rooted in a possible response bias (towards agreeing more with items) since the statements of the original questionnaire are generally not reversely formulated.

CONCLUSIONS

The study indicates that research misconduct and QRP involving PhD students are taking place in Danish research, the main issue being related to authorships.

As a means to limit research misconduct and QRP, the next generation of researchers at least needs to be able to perform research on good grounding. Methodological courses and institutional courses on RCR for PhD students may be a useful tool to this end. Likewise, clear, visible and operable institutional policies will support researchers not wanting to act in wrongful ways. Furthermore, we recommend that opportunities for open ethical dialogue among staff be provided to foster healthy research environments and strengthen researchers’ abilities to deal in responsible ways with the dilemmas and grey zones that may occur in research projects, despite regulations and written guidelines.

Correspondence: Lone Bredahl Jensen. E-mail: lbredahl@bib.sdu.dk

Accepted: 2 February 2018

Conflicts of interest: none. Disclosure forms provided by the authors are available with the full text of this article at www.danmedj.dk.

Acknowledgements: Thanks to Bjørn Hofmann, Norwegian University of Science and Technology, and Søren Holm, University and Manchester and University of Oslo, for sharing their questionnaire, and for kindly answering questions relating to the survey.