Thèses de doctorat à la faculté de médecine de Tunis de 2015 à 2017: Fréquence des publications scientifiques et facteurs associés

RESUME

Introduction: La publication d'une thèse en médecine est un moyen pour augmenter la visibilité des travaux de recherche. Objectif: Déterminer la fréquence de la publication des thèses à la Faculté de Médecine de Tunis (FMT), et identifier les facteurs qui lui sont associés. Méthodes: Étude transversale rétrospective s’intéressant aux thèses inscrites de 2015 à 2017. Toutes les publications issues des thèses soutenues ont été recueillies en interrogeant les bases de données Scopus et Google scholar, jusqu’au mois d’avril 2022. Une régression logistique binaire a été réalisée pour déterminer les facteurs associés à la publication. Les Odds ratios ajustés (ORa) ont été présentés avec un intervalle de confiance à 95%. Résultats: Sur 878 thèses soutenues, (11,8%; n=104) ont été publiées au moins une fois. Sur un total de 130 publications, 90 (69,2%) ont intéressé des revues indexées dans Scopus, avec un classement Sciamgo (SJR) moyen des revues scientifiques de 0.70. La langue de publications était l’anglais dans 73,1% des cas. Le délai médian entre la soutenance et la première publication était de 15 mois. En analyse multivariée, les facteurs associés à la publication étaient le statut résident du candidat (ORa=2,35 [1,2-4,7]) et le grade assistant de l’encadrant (ORa=2,48 [1,1-5,6]). Conclusion: Comparativement au nombre de thèses soutenues, la fréquence des publications issues des thèses était relativement faible. Ceci nous incite à renforcer l’accompagnement des doctorants afin d’optimiser leur engagement dans la recherche et de promouvoir la publication scientifique à la FMT.

1. INTRODUCTION

Initiated in response to a knowledge gap in a medical field, doctoral thesis elaboration represents an opportunity to acquire new skills in research methodology and to develop a critical mind in relation to other research activities [ 1]. Although requiring intense work, medical theses are generally read by few scientists, as often available in local university libraries but not in open access international databases [2]. The dissemination of thesis results to the scientific community is the cornerstone of medical progress. In fact, thesis publication is a valuable way to ensure knowledge dissemination to a broader audience, and to increase the research visibility and impact [3]. Besides, conducting research, writing, and publishing, are considered as the pathway to securing academic promotions, obtaining funding for research endeavors, and enhancing the overall rankings of institutions [4]. During the last years, the number of publications has increased exponentially, which is mostly explained by the scientific community growth, the increasing availability of research funding, technology advancement, the rise of digital publishing and open access journals [5]. Moreover, the top medical universities encourage their students to publish their medical theses as scientific papers and they provide guidance and resources to support them in this process, including workshops, writing centers, and access to research databases [6]. Unfortunately, publishing in developing countries faces several challenges such as requirement of a good to high level of proficiency in english language and important financial investment due to higher publication fees [7 , 8]. In Tunisia, Tunis El Manar University, of which the Faculty of Medicine of Tunis (FMT) is a part, contributes significantly to the university's overall scientific production. In this context, bibliometric studies are important in evaluating the research output of the faculty, assessing trends over years, and identifying areas for improvement. Through this study, we aimed at providing a focus on scientific publication at the FMT, by estimating the thesis-related publication rate, and identifying associated factors to publication.

2. METHODS

This was a retrospective, cross-sectional study, conducted at the FMT. We included medical doctoral thesis registered over two years (January 2015- January 2017), including all specialties (as listed in the Official Journal of the Tunisian Republic) [9]. We excluded theses that have been canceled after registration (n=69) and theses identified as duplicates (n=4) based on the registry of thesis department, using the candidate’s name and thesis title. All data related to the candidate’s and supervisor’s characteristics, presence or not of co-supervision, thesis content (including study design, number of cases and key words, thesis language, overall quality of abstracts), as well as main delays (thesis registration to thesis defense, thesis defense to thesis first publication), were collected. Abstracts of defended theses were retrieved from the FMT library's database. Based on predefined criteria elaborated by the thesis committee of the FMT, a 14-item scale allowing an objective evaluation of abstracts’ quality was used (Table 1). An additive score was calculated by assigning one point for each of the 14 items, and a threshold value of 11 was considered to define a good quality abstract.

Thesis-related publications were collated by scanning Google scholar and Scopus databases, up to April 2022. Were considered, all articles and conference papers written in french and/or english languages. The candidate and supervisor names were introduced (in its different phonetics), and only publications related to the main thesis topic were considered. Furthermore, we collected data related to the number of publication(s), the title of the scientific journal chosen for publication, the journal indexation or not in Scopus database, the 2018-Scopus Journal Ranking (SJR) calculated from citations over the preceding three years [10], and the journal category (generalist Vs specialized), defined as follow: “Generalist journals are those publishing articles from a wide range of scientific disciplines and covering diverse areas of research. Thus, highly regarded in the scientific community and having high impact factors. Specialist journals, are those focusing on a specific area of science, and publishing research articles related to that area. These journals have a more targeted readership and are often considered important in their specific field of research” [11].For data analysis, simple frequencies and percentages were calculated for qualitative variables. For quantitative variables, we tested the distribution normality using the Shapiro-Wilk test. In univariate analysis, percentages comparison was done using Pearson's Chi 2(ﬡ2) test or Fisher's exact test if validity conditions were not verified. The association between explanatory and dependent variable, was quantified with crude Odds Ratios (crude OR) presented with 95% confidence interval (95% CI). Then, explanatory variables with a significance level less than 20% in univariate analysis were introduced into the

final model of binary logistic regression (stepwise top-down Wald method) [12]. After adjusting for the potential confounding factors, the strength of association was quantified with Adjusted Odds Ratios (AOR) presented with 95% CI. All statistical analysis was performed with SPSS software ver. 23.0 (IBM. Chicago. IL. USA), and a p-value less than 5% was considered to conclude to test significance.All data collection process was carried out after obtaining the FMT dean approval. Data analysis was processed anonymously and used only for research purposes.

3. RESULTS

Study population characteristics

During the study period, 921 theses were registered, and out of these, 878 were successfully defended, representing a completion rate of 95.3%. Health Care Facilities (HCF) that produced the largest number of theses, were those with the largest hospital capacity, and involving most of the university teachers: Charles Nicolle hospital (n=133, 15.2%), La Rabta hospital (n=115, 13.1%), and the Principal Military Hospital of Tunis (n=98, 11.1%). The study population characteristics are depicted in Table 2.

Overview on abstracts quality

A total of 742 abstracts (available at the FMT library's database), were examined. All abstracts were written according to the IMRaC (Introduction, Methods, Results, and Conclusion) structure. Out of a total of 14 points, the mean overall score of abstracts was 10.8 ± 1.2. The abstracts quality was considered “good” for 64.9% of abstracts.

Publication rates and patterns

Out of 878 defended theses, 11.8% (95% CI: 9.8-14.1) were published at least once (n=104); and the publication was in English in 73.1% of cases. The median time between the thesis defense and the first scientific publication was 15 months, Interquartile Range (IQR) (5-32). No significant difference in delays was found according to candidate status (p=0.8), supervisor grade (p=0.3) and specialty type (fundamental, medical, surgical) (p=0.6). Out of 130 publications in total, the candidate’s name appeared in the authors list in 93.8% of publications (n=122), and was listed as first author in 76.9% of cases (n=100). The share of publications interesting Scopus-indexed journals was 69.2%; 95% CI (60.9-76.7), with a mean SJR equal to 0.70. “Tunisie Medicale” journal ranked first with 26 publications, (28%) followed by the Pan African Medical Journal (5.4%), Endocrine (4.4%), Revue Neurologique (2.3%), and Nephrology Dialysis Transplantation (2.3%). Moreover, the share of publications in specialist journals focusing on a specific area of science, was 54.6% (n=71). In 77.4% of cases, these publications appeared in Scopus-indexed journals. The Table 3 listed the specialist journals titles and depicted the number of publications in each journal.

Associated factors to publication

In multivariable analysis, factors that were significantly associated to publication, were the resident status of the candidate AOR=2.1[1.1-4.1] and the assistant professor grade of the thesis supervisor (AOR=2.6 [1.2-5.8]), Table 4.

4. DISCUSSION

Including 878 defended theses, (11.8%) were published at least once, and out of 130 publications in total, 90 (69.2%) interested Scopus-indexed journals, with a mean SJR of 0.70. The median time between the thesis defense and the first scientific publication was 15 months. The publication was in English in 73.1% of cases. In multivariate analysis, factors that were significantly associated to publication were the “resident status” of the doctoral student (adjusted OR=2.3[1.1-4.7]) and the “assistant professor grade” of the thesis supervisor (adjusted OR=2.4 [1.1-5.6]). To our knowledge, this study is the first to identify associated factors to thesis-related publication at the FMT, considering all specialties combined and after adjustment on several confounding factors. To cover as many publications as possible, all thesis-related publications were collated by scanning Scopus and Google scholar databases. Google scholar includes online early articles, english and non-english publications in indexed and non-indexed journals; and is still considered as the most comprehensive source compared to Web of Science and Scopus [13]. In our study, limitations related to the lack of data collection regarding length of thesis and the availability of external funding for thesis project over the study period are important to mention. Regarding the language of dissertation and delays to first publication, our results were consistent with those reported in Morocco (Rabat University) showing that 99.1% of registered thesis between 2011 and 2021, were written in french. A difference regarding the publication language was however observed, as half of the papers (49.4%) were written in french (vs 26.9% in our study) [14]. Zemni et al, reported that the rate of english medical publications in Tunisia have known

a significant increase these last years, going from 30% to 54% , then reaching 68% in 2014 [15]. Our study revealed a low rate of publication compared to the number of defended theses at the FMT. In fact, a certain stagnation over time is noted when compared to previous publication rates of 13.4% and 17.3% at the FMT [16- 17]. In a previous study based on a thorough examination of scientific medical literature focused on the Greater Maghreb, authors reported also the gap between the number of theses and the number of publications; and confirmed the scarcity of medical scientific publications despite the global upward trend during the last decades [18]. Similarly, low publication rates have been reported in studies conducted in other developing countries, confirming the difficulty of bringing to publication a work that often constitutes the first contact with research for a medical student. In Peru, the publication rate in biomedical-indexed journals was (17.6%), and (5.8%) were published in PubMed-indexed journals [19]. Results from two medical universities in Croatia showed that the publication rates in Medline-indexed journals were 13% and 14% (Rijeka and Zagreb universities, respectively) [20]. The significant gap in the amount of scientific output in the medical field between developed and developing countries has been emphasized in literature. In New Zeeland, the publication rate of theses produced by medical students was 32.7% [21]. Besides, the analysis of the global profile of biomedical research productivity showed that in terms of continents, North America had the highest number of biomedical publications per million population per year (341.3); followed by Australia and Oceania (288.3), Europe (136.8), Asia (12.8), South America and Caribbean region (10.8), and Africa (3.5) [22]. This could be explained by the encouragement of research and scientific production in developed countries, the availability of funding resources to conduct the research and to cover article processing charges expenses, and the availability of collaborative research opportunities. In Australia for example, universities receive extra funding based on their academic publication rates and academic promotion is difficult without a good publication record [ 23]. Furthermore, poor publication rates in developing countries may be also explained by brain-drain to developed countries and their contribution in scientific production in these countries where enhanced support for scientific research is provided [24, 25]. Regarding the mean SJR when publication interested Scopus-indexed journals, similar estimates have been reported by Touissi et al in Morocco with a mean SJR of 0.69 [14]. A recent review of medical journals showed that in the 2017-Scopus list of medical journals, the mean SJR was 0.82. Therefore, our results are not far from international standards in medical field. However, this figure varies depending on the countries and across specialties [ 26-27].For associated factors to publication, other researchers reported that the thesis author's status, the supervisor's grade, and the formulation of clear recommendations by the end of the thesis, were associated factors to the publication of preventive medicine theses at the FMT [28]. Others reported that a shorter thesis length, a more experienced thesis supervisor, the student’s academic performance and the relevance of the research theme, were associated to thesis publication [29-31]. In conclusion, this study showed that at the FMT, factors that encourage publication are related to the significance of scientific publications in academic promotion. However, taking into consideration the importance for medical doctors to continuously stay updated on recent advancements in medicine and critically evaluate evidence for better implications in patients’ management; clinical expertise is no longer enough and improved engagement in research and publication is highly recommended [32]. Hence, there is a need for an increased awareness among medical students regarding this matter, even if publication is related to only negative results [33-34]. Besides, implementation and/or continuous reinforcement of writing courses, writing support groups, active engagement in collaborative research endeavors, and regular participation in research funding application opportunities are of crucial importance for an effective contribution to scientific progress in the medical field [ 23, 35-36 ].