Barriers to publishing in biomedical journals perceived by a sample of French researchers: results of the DIAzePAM study

The main sociodemographic and occupational characteristics of respondents are presented in Table 1.

Regarding previous publications, 94% (N = 1123) of respondents had been an author of at least one article in the previous 2 years: 65% at least once as first author, and 39% as last author. 73% of authors stated that they actually participated in the writing and/or submission of their articles while 19% reported having been only investigators and 9% that their contribution has been minimal. Impact factor was the main journal selection criterion for 85% of authors. The other criteria are presented in Fig. 1.

According to the respondents (N = 1191), the main barriers to publication were lack of time to write articles (79%), limited skills in English (40%), limited skills in writing (32%) and difficulty in starting writing (35%) (Fig. 2). Female gender and working as hospital practitioner or senior registrar/hospital university assistant were significantly associated with self-reported limited skills in English (P = 0.021 and P < 0.001, respectively), self-reported limited skills in writing (P < 0.001), and with difficulty in starting writing (P < 0.001). Young age was significantly associated with self-reported limited skills in writing (P < 0.001), but not with self-reported limited skills in English (P = 0.55) or difficulty in starting writing (P = 0.2). Compared with other respondents, respondents reporting limited-skills in English, limited skills in writing, or difficulty in starting writing were less frequently first, second/third, or last author of their publications (P < 0.001) and less frequently satisfied with their number of publications (P < 0.001); they more frequently said that they would appreciate technical support for future publication (P < 0.001). Respondents with self-reported limited skills in English (P = 0.013) more frequently received support for their previous publications than the other respondents. On the contrary, no significant difference in support was observed between respondents with and without self-reported limited skills in writing (P = 0.35) or difficulty in starting writing (P = 0.32). Factors associated with self-reported limited skills in writing are presented in Table 2.

According to the respondents, the main reasons for publishing were scientific (dissemination of the results and change in practice rather than being the first to publish) and for peer recognition and career advancement (Fig. 3). Although 35% of respondents reported publishing under pressure from their institutional hierarchy, only 17% reported publishing to obtain SIGAPS points.

Seventy six percent of authors claimed they did not publish enough, mainly because of lack of time (85%). Seventy four percent estimated that writing these recent articles (from drafting the manuscript to final acceptance by the journal) took between 6 months and 2 years.

The main reasons for refusal of articles was, according to the respondents: editorial judgment of lack of originality (41%) or problems of methodology and/or results (32%), negative and inappropriate comments by reviewers (25%), poor choice of the journal by the authors (21%), and the fact that study results were of purely French origin (15%). Respectively 19% and 11% of respondents claimed that refusal was due to poor English or writing quality. Conflicts of interest were rarely mentioned (4.6%).

Most respondents (87%), and in particular those who published less than 3 articles in the previous 2 years, regardless of their position (i.e., first, second or third, or last authors), would accept technical support to overcome their difficulties in publishing. They would particularly appreciate support for English (79%) or critical reediting (63%), formatting before article submission (52%), writing responses to reviewers (47%), writing some parts of the article (41%), and/or article submission (44%). This support was expected to save time (92%), avoid some refusals (83%), or increase high-impact-factor journal submission and acceptance (75%).

For future articles, 79% of respondents claimed that they would need funding, in particular for English reediting (56%), publication (38%), Open Access (25%), or medical writer (21%) fees. According to respondents, this financial support should be covered by their institution, AP-HP, (73%), the added financial value obtained by their department from previous research activity and publications (SIGAPS/MERRI points; 56%), the upstream research project budgeting in response to a call for projects (31%), or a pharmaceutical company (11%) (Fig. 4).

Free comments were made by 256 (21%) respondents. Overall, these comments showed respondents’ dissatisfaction and/or referred back to issues raised by this questionnaire. Some barriers to publication already dealt within the questionnaire were reiterated: lack of time (27%, mostly due to clinical and administrative overload), limited skills in English (11%), and lack of funding (6%). Beyond the barriers to publication evaluated by the questionnaire, other upstream barriers which obviously impact the publication process emerged: lack of assistance for statistical analyses (19%), complexity of regulatory dossier preparation before study implementation (7%), lack of assistance in information technology (database management, supply of software such as for managing bibliographies and references) (6%), and lack of logistical assistance in the conduct of studies (e.g., clinical research technician, data collection) (6%).

A few respondents commented that some topics are difficult to publish in high impact journals (e.g., palliative care, qualitative research, pediatric surgery, environmental pathology). Some also raised the growing issue of conflicts of interest of reviewers in the refusal of articles and believed that being a non-English speaking author reduced their chances. Of the respondents who made a free comment, 18 (7%) regretted lack of information about the use of SIGAPS and that the financial value of SIGAPS/MERRI points was not allocated back to researchers (a point sometimes vehemently directed at their institution). Finally, 47 (18%) of respondents spontaneously welcomed the survey (only 1 respondent commented that the survey had no interest, but still answered all items).

Lack of time was the main barrier to publication. This result was consistent with the literature [27] and with the number of respondents who had published but thought they were not publishing enough (76% of authors), mainly because of lack of time (85%). It was also consistent with the percentage of authors claiming to have had difficulty in writing the Discussion (60%), as this section needs time to think, read the literature, and compare present and previously published results [28, 29].

The other main reasons given by more than 20% of respondents included self-reported limited skills in English or in writing, limited submission skills, and difficulty in starting writing. As expected, being young and working as hospital practitioner or senior registrar/hospital university assistant were associated with self-reported limited writing skills, but being young was not associated with self-limited skills in English. Surprisingly, being a woman was a factor associated with self-reported limited skills in writing. This possibly adds up with the other reasons leading female physicians to have a lower publishing rate than their male colleagues [30]. Unfortunately, combination of lack of time and limited skills increase difficulties. Indeed, English and writing skills make writing easier; easy writing takes less time and energy, and this decreases the risk of procrastination. It has been showed that this combination decreased the chance of publishing in a high-impact journal: respectively 78% and 70% of researchers publishing in high-impact journals had English as current official work language and spent more than 50% of their time on research versus for 35% and 27% researchers publishing in low-impact journals [31].

According to the respondents, another significant barrier to publication is negative or unoriginal results. Negative results are more difficult to publish, as there is a greater interest in publishing results that can immediately change clinical practice [32]. However, there is presently an international groundswell in favor of proper transparency and full reporting of all results so as to avoid the pitfalls of publication bias, and this movement involves different stakeholders such as Drug Agencies [33] and medical journals (www.​alltrials.​net), so that this barrier should now be a thing of the past. Knowledge of negative results may be as useful as positive results in achieving good non-deleterious patient care [34].

Last but not least, lack of funding, in particular for publication fees, was a barrier to publication for 22% of respondents. Over the past decade, the budget for publications has become a growing problem, in particular with the increasing number of Open Access journals that allow free access to readers, but not authors [35, 36]. In addition, 19% of respondents declared that they lacked funding for a medical writer or translator. According to Pavia et al., publishing in high- and low-impact journals is also dependent of financial resources [31].

On the one hand, most respondents (87%) would accept technical support to overcome their difficulties in publishing, calling into question the legend of the researchers who must do everything (research and writing) themselves [37]. On the other hand, among the 28% of authors who already called upon additional support, only 26% had used services of a medical writer. This is possibly due to the lack of funding. However, as 61% of authors who already called upon additional support had used services of a translator, another possible reason is the lack of knowledge of the job of medical writers and the confusion between the jobs of translator and medical writers. Indeed, English reediting, critical reediting, formatting/submitting and writing were the main tasks for which support was needed, and medical writers fulfill all these functions. One thing is to be a fluent writer of English, and quite another to know the socio-pragmatic features of today’s academic rhetoric. Gattrell et al. [38] recently showed that declared professional medical writing support was associated with more complete reporting of clinical results and higher quality of written English in the sample of journals they analyzed. In addition, in this study, authors who called upon additional support more frequently published at least once as first author in the previous 2 years than the other authors, suggesting that external support increased publication as first author. Thus, as is the case in Scandinavian countries, French universities and/or research centers should have an “academic/scientific writing center” with competent applied linguists well versed in the rhetoric of today’s academic/scientific discourse. Such a Center should be free of charge for all researchers. Introduction of publications officers, regular and early training in English and writing [39], and use of Writing Aid Tool (WAT) such as the online Consort-based Web tool for randomized controlled trials [40] could contribute to increase both number and value of published research articles.

In addition to technical support, 79% of respondents declared a need for funding support. Respondents thought that support could be provided by AP-HP directly (73%), or through the added financial value obtained by their department from previous publications (SIGAPS/MERRI points; 56%). In 18 (7%) free comments, respondents regretted that the financial value of SIGAPS/MERRI points was not currently clearly allocated back to researchers by AP-HP, and a lack of information about its use. Such funding support could allow researchers to pay for English reediting, publication, Open Access, or medical writers’ fees. These results suggest that the process of conversion between one type of capital (credibility) and another (material resources), known as “cycles of credit” [41], was perceived as damaged and needs to be reactivated. According to this process (and as expected by respondents), articles issued from arguments elaborated using research data contribute to increase recognition which in turn increases funding enabling investment in human and material resources necessary for more data collection and arguments constructions.

Electronic surveys avoid the need for double entry and enable questionnaire personalization (questions are automatically hidden or revealed according to previous answers) and timeliness, at low cost [42, 43]. Limited access to the Internet, overloaded e-mail inboxes and obsolete e-mail addresses are limitations to electronic surveys, and were encountered in the present study; besides overloaded inboxes and obsolete e-mail addresses, some researchers registered in the AP-HP SIGAPS database contacted us for help, finding it impossible to connect due to their institution’s firewalls. Nevertheless, as the target participants were pressed for time, the method was probably the best way to conduct the present study. Other limitations common to all questionnaires, and not specific to electronic surveys, are questionnaire size (long questionnaires are often less filled out than short ones), user-friendliness, and type of response. We paid careful attention to completion time and type of response in developing the DIAzePAM questionnaire, and ran some tests on the draft version. However, we did not include any questions on user-friendliness and so cannot know whether the questionnaire was well perceived and easy to complete.

Only volunteers responded to the questionnaire, automatically creating a bias. As no compensation for participation was offered, it can be hypothesized that researchers who perceived difficulties in preparing manuscripts for publication may have been incentivized to respond to the online survey. However, some respondents had published several articles in the previous 2 years. The decision to respond to the online survey or not may also have been affected by the fact that the survey was conducted in the name of the AP-HP. Sociodemographic data on AP-HP staff are sparse and do not target the population of researchers who publish or intend to publish, hindering comparison with the respondents to the present survey. However, based on figures for medical staff working at AP-HP in 2014 [44], 25% of assistant professors, 20% of senior registrars, 15% of professors and 6% of hospital practitioners responded to this online survey. Although hospitals practitioners were under-represented as compared with the other professionals, this result is probably consistent with the population of medical staff performing research and publishing.