Factors associated with scientific production citations in dentistry: Zero-inflated negative binomial regression and hurdle modelling

Data collection

The data for the present investigation were retrieved from the WoS database in August 2022; the search criteria was “dentistry”, and the search was limited to the latest articles published in journals. The total number of articles was 110,269, with 9,034 published in 2022. The files were downloaded in Excel to extract study variables such as the total number of citations (TCitations), number of pages (NPages), year of publication (Year), number of years between the time of publication and 2022 (PYears), number of authors (Nauthors), and number of references (Nreferences); the journal’s publisher was also downloaded as a variable. The name of the publisher was recorded if it was one of the top 10 in a list that was developed later, and the rest were classified into the “other” category.

The bibliometric information downloaded contained the study variables, but it was necessary to resort to text separation functions with the Excel function “Text in columns” to quantify these variables. The work was carried out by two specialists of regression models.

Data analysis

To analyse the obtained information, the methodological design involved two phases: first, statistics were used to evaluate scientific production and extract important indicators, such as the main publishers, the annual production, as well as evaluate the scientific impact based on the previously obtained variables.

The annual scientific production on dentistry in the WoS between 2010 and 2022 was used to unveil the top 10 publishers, whose average production from 2010 to 2021 was compared using the analysis of variance (ANOVA) and Tukey’s test in the Statistical Package for Social Sciences (SPSS) in version 28. Likewise, for the same period, the annual trend in the number of publications on dentistry was evaluated using discontinuity regression models, a method applied to dental visit copayments.³⁸ These models showed that scientific production increased in any given year with some discontinuity in the trend. When the associated parameter was not significant, it was reduced to a regression model by sections. The modelling was performed in Excel.

The impact of the publications was measured by the number of citations the articles received or by the occurrences of noncitation. The program used to select these factors was Stata 16, and the z test was used to count the models with the significance set at p < 0.05. The analysis of the impact included 76,641 of the 110,269 articles, which were downloaded for the period from 2010 to August 2022 and covered the 10 publishers in the ranking. The program used to estimate the models was Stata 16. Overdispersion in the Poisson model was evaluated using the Lagrange multiplier test; moreover, the standard negative binomial regression model was evaluated through the likelihood ratio (LR) test for the α parameter (if α = 0, there is no overdispersion).³⁹

The counting models used were the zero-inflated negative binomial regression model and the hurdle model, which allowed us to analyse both the number of citations and noncitations for each article.^36,37 The dependent variables were the total number of citations and noncitations per article, and the independent variables were the number of pages, year of publication, number of authors, and number of references in each article. The editors working within the top 10 journals in which the articles were published constituted a categorical independent variable, and the BioMed Central (BMC) editorial board was used as a reference. The choice of a zero-inflated model versus a standard model was made using the Vuong test, and the Akaike criteria (AIC) and Bayesian (BIC) approach were used to choose the best model.³⁹

Subsequently, in the methodological design, there was a second phase of bibliometric analysis, and once the statistical analysis of the main journals was advanced, the evaluation of the impact and of the annual scientific productivity focused mainly on the materialisation of the analysis of the keywords derived from the scientific activity.

During this process, an exclusion criterion that allows the content analysis of the main scientific articles that are available in the field was chosen so that, of the total number of articles selected for the first phase in the methodological design, 1,000 articles were used as a filter. Moreover, at the time the present research was conducted, these articles accounted for the highest total number of citations and were considered the main references in the field.

Once the extraction of the thematic analysis of the main articles was carried out, the main network was analysed to determine the co-occurrence of keywords, allowing for an understanding of the elements of thematic associativity. We also conducted an analysis of the validity and frequency of the keywords so that consolidated or increasingly used keywords were extracted as well as the keywords emerging within the research field. Finally, an analysis of the thematic evolution was carried out based on the main keywords for each research year based on the selected articles.

To execute the analysis of the main network and determine the co-occurrence of keywords in the scientific body, the free access software VOSviewer was used, while the analysis of the increasingly used and emerging words as well as the thematic evolution was carried out using the office automation tool Microsoft Excel. This phase in the methodological design was completed collectively by all of the authors of the present research with the purpose of reducing biases in the results and jointly treating the disagreements found until convergence was attained. To increase the level of replicability and further detail the methodology, Figure 1 presents a summary of the methodological design used in the research.

Figure 1. Summary of the methodological design used in this research.

Tcitations: total of citations, Npages: number of pages, Pyears: publication years, Nauthors: number of authors, Nreferences: number of references.

Figure 1 Alt Text. The figure provides a concise overview of the methodological design employed in the research, encompassing the analysis of co-occurring keywords using VOSviewer, tracking emerging words and thematic evolution through Microsoft Excel®, and collaborative author involvement for minimizing biases and achieving consensus.

Phase 1 – Statistical evaluation of scientific production

In accordance with the nature of the methodological design, Phase 1 of the results is presented in this section. This phase involved a rigorous analysis of the scientific output published by each publisher as well as this output evolved within the time window selected for the extraction of articles. The analysis also included the factors associated with the impact of scientific output.

Evolution of scientific production by publisher

Figure 2 shows the evolution of scientific production in dentistry during the period 2010-2021 through polynomial models of discontinuous regression and regression models according to the selection in the top 10 main publishers.

Figure 2. Discontinuity regression models for scientific production.

Source: Published by publishers in the field of dentistry.

Figure 2 Alt Text. Figure illustrates the scientific production trend in dentistry from 2010 to 2021 using polynomial models of discontinuous regression and regression models for the top 10 main publishers.

The WoS database is a platform composed of a large collection of databases from all disciplines of knowledge organised by the British-American company Clarivate Analytics.^40,41 In 2016, the scientific production in the WoS changed to an estimated 2,377 articles (t = 14.1, p = 0.000). During the first period—2010-2016—the growth curve was concave, with the number of articles barely exceeding 8,000 per year; however, the growth curve was convex for the period between 2017 and 2021, with more than 10,000 articles/year and R2 = 99.9%. The scientific citation indexing service is based on the online subscription Book Citation Index (BKCI), which is part of WoS and is maintained by Clarivate Analytics. The BKCI provides analytic services for academic and scientific research data and was, in a limited manner, part of the WoS Core Collection for 14 years until 2017. Since 2018, it has become a perpetually maintained database.⁴² In 2018, the complementary tool Kopernio, an extension for browsers that saves time and allows direct access to the best articles available in PDF format, was acquired by Clarivate Analytics, revolutionising access to articles from anywhere in the world.⁴³

The curves corresponding to Wiley’s activities are concave in each section. In the first section, the production is practically stable without a linear trend (p = 0.211 > 0.05) or quadratic trend (p = 0.211 > 0.05). However, in the second section, the linear trend increases (p = 0.029 < 0.05), and the quadratic trend is not evident (p = 0.690 > 0.05) and R2 = 92.7%.

Like that of the WoS, the curve corresponding to Elsevier’s activities is concave and convex, showing a discontinuity of 550 articles/year (t = 2.85, p = 0.029) and R2 = 94.3%. The activities of Quintessence Publishing are shown to have slightly changed during the 2015-2022 period compared to the 2010-2014 period, with R2 = 83.8%. In this regard, a linear model would have presented an increasing annual trend of approximately 13 articles/year (t = 4.11, p = 0.002 < 0.05) and R2 = 62.9%.

The regression models for scientific production published in the WoS and Elsevier are discontinuous regression models with leaps in 2017, and those corresponding to Wiley and Quintessence Publishing are reduced to regression models by sections.

The curve for the Churchill Livingstone publishing house shows a difference in the number of publications between the periods 2010-2014 and 2015-2021, ascending in the first and descending in the second. The number of publications by Springer Nature can be divided into the 2010-2016 and 2017-2021 periods; both periods exhibit positive growth in those numbers but changes are more intense in the second period. The number of articles published by MOSBY-ELSEVIER declines in 2012 and experiences a slow recovery thereafter. Activities in the other top 10 publishers fluctuate, with a stable number of publications or small increases throughout the study period. It is important to note that the number of articles to be published in 2022 by these publishers is predicted to be lower than expected, as shown in the trend analysis. The models used for the number of publications produced by these publishers were not estimated.

Scientific output published by publishers

The annual scientific production on dentistry published in journals made up the top 10 categories during the period 2010-2022, as shown in Table 1. The number of published articles increased from 6,850 in 2010 to 12,126 in 2021 at a rate of 440 articles/year, with great changes between 2016 and 2017, when the number of published articles reached 2,330 articles over one year. The trend in the number of articles published by publishers, in some cases, shows very similar patterns. As of 2022, some journals still had not published related publications.

The quantity of published journal articles is merely one of numerous factors taken into account when assessing publishing companies such as Wiley in rankings, as illustrated in this study. This parameter, while vital, is only a component of a larger set of metrics and criteria utilized to evaluate the academic and scientific significance, influence, and importance of publishers. Publishers strive to enhance their image by increasing the number of publications and ensuring their quality and relevance. Besides, several other factors, such as article citation frequency, impact on the scientific community, and journal accessibility, also determine their rankings.

Factors associated with the impact of scientific output

In general, the average number of citations per article was 13.4 ± 23.79, with a maximum of 1,621 citations; noncitations reached 12.08%. On the one hand, citations are the basis of the SCImago Journal Rank (SJR) index that ranks journals, and noncitations also represent an indicator for the quality of the researchers, articles, journals, and now publishers. On the other hand, the average number of pages was 7.61 ± 2.84, the average number of authors was 5.19 ± 2.59, the average number of references was 31.63 ± 16.15. These results are also important for authors to consider when producing scientific articles in dentistry.

The number of citations for the articles published in journals owned by the top 10 publishers and the factors associated with these articles are shown in Table 2.

In addition, in the UK, for example, the UK Research and Innovation (UKRI) suggests that impact factors should not be used to indicate quality and should be ignored, as should other journal metrics.⁴⁴

In terms of the average number of citations, Sage Publications, Inc. stands out with 23.2 ± 45.9 citations and a maximum of 1,040; however, results show 13.6% noncitations.

Quintessence Publishing Co., Inc. shows the maximum number of citations (1,621), followed by Springer Nature, with 1,407 citations. However, the former also shows the second highest noncitation rate (15.5%) after BMC (15.8%) as well as the maximum number of pages (124 pages) and references (612) per article. The maximum number of authors was found in an article published by Elsevier (131 authors).

Modelling the impact of scientific production on dentistry

On the one hand, Poisson’s model assumes that the mean is equal to the variance, but Table 1 indicates that the standard deviation (square root of the variance) is greater than the mean of the total number of citations per article. The presence of overdispersion is assumed to be evident. The presence of overdispersion was identified using the Lagrange multiplier test in the Poisson regression model (LR value = 15263275, p = 0.000 < 0.05) and confirmed by the likelihood ratio test in the negative binomial regression model (chibar2(01) = 8.8e+05, p = 0.000 < 0.05). The presentation of the results estimated for these models is omitted. However, these findings justify the discarding of Poisson’s regression model since resorting to alternative models was necessary.

On the other hand, according to Table 2, the percentages of noncitations for the articles published in the top 10 journals are within the range of 6.0-15.8%, suggesting a high occurrence of noncitations. Because the average number of citations is 9.4 citations per article, the probability of an article not being cited is 0.008%; this probability becomes very small (practically zero) as the average number of citations increases. In these circumstances, zero-inflated models are justified.

The zero-inflated negative binomial regression model and negative binomial hurdle regression model estimated for citations and noncitations are shown in Tables 3 and 4, respectively. The models do not show the results corresponding to the BMC publisher because that publisher constitutes the reference category; moreover, the regression coefficients establish the differences between BMC and the other publishers.

With respect to citations, the negative binomial regression model (Table 3) indicates that the number of pages (p < 0.001), the number of years since the article was published (p < 0.001), the number of authors (p < 0.001), and the number of references used (p < 0.001) contribute favourably to increasing the number of citations per article since the regression coefficients are positive. Regarding the publishers, in comparison with BMC, the publishers with the highest number of articles being cited were Wiley (p < 0.001), Elsevier (p < 0.001), Mosby-Elsevier (p < 0.001), and Sage Publications, Inc. (p < 0.001) but also Springer Nature (p < 0.05); moreover, those with the fewest citations were Quintessence Publishing Co., Inc. (p < 0.001) and Wb Saunders Co-Elsevier Inc. (p < 0.001). These results reflect that the factors have a favourable effect and that the differences between the publishers seeking to achieve a higher number of citations improved the h-index among dental researchers when adjusting for the other factors.

In reference to the zero-inflated model, that is, the model without citations, the factors under study contributed to the decrease in the number of publications, which is explained by the negative regression coefficient.

Some journals are known to limit the number of authors per article, especially when the topic covered in the articles does not justify the presence of numerous authors. Each publisher makes decisions in this regard.

The negative zero-inflated binomial regression model, on the one hand, confirms the presence of overdispersion by testing the likelihood ratio (chibar2(01) = 8.1e+05 and p = 0.000 < 0.05); on the other hand, the information criteria provide AIC = 531 580.4 and BIC = 531 848.5.

The hurdle regression model, which is presented in Table 4, also establishes the positive effect of these factors on the impact of scientific publications compared to the zero-inflated negative binomial regression model in both the direction and the magnitude of the effect. Likewise, the hurdle model confirms the differences in the number of article citations in the different journals compared to BMC; the model also shows where there are no differences. The similarity of the effect of the factors and the differences between the publishers are also maintained with respect to the percentage of citations, with the exception of the logistic model that did not show differences in the number of citations in Elsevier’s journals compared to those in of BMC (p > 0.05). The confirmation of the factors through this model shows that the factors under study are as important in evaluating both the number of citations and noncitations.

In the evaluation of the goodness of fit for the negative binomial regression model, Hurdle provides the values AIC = 531 390.6 and BIC = 531 658.8 as information criteria. It is important to note that when estimating this model, it is necessary to create artificial variables to represent the publishers.

Based on the assumption that a model with lower AIC and BIC values is better, Hurdle’s model is slightly more suitable than the negative zero-inflated binomial regression model; moreover, adducing a criterion of parsimony to decide otherwise may limit the choice of which model should be adopted in articles in dentistry. However, the findings show that both models will be useful when selecting robust factors associated with citations and noncitations in the dental field.

Between 2010 and 2021, there were differences among the publishers’ average output, according to ANOVA (F = 107.191, p = 0.000). Wiley was the publisher with the most publications (1,938 ± 256 articles/year), followed by Elsevier (1,221 ± 257 articles/year); there were significant differences between these major publishers and other publishers, according to Tukey’s test.

With a portfolio of 1,600 journals in various scientific categories, Wiley is the world’s leading publisher. Elsevier’s journals lead in the field of medicine and health sciences.⁴⁵ These articles published by these publishers differ in the average number of times they are cited (F = 136.03, p = 0.000) as well as in the number of pages (F = 616.8, p = 0.000), the number of years since publication (F = 420.39, p = 0.000), number of authors (F = 286.59, p = 0.000), and number of references used (F = 382.64, p = 0.000).

The publishers for which they articles show a lower percentage of noncitations than BMC were Wiley (p < 0.001), Springer Nature (p < 0.001), Mosby-Elsevier (p < 0.001), Sage Publications Inc. (p < 0.001), Elsevier (p < 0.05), Churchill Livingstone (p < 0.05), and Pergamon-Elsevier Science Ltd. (p < 0.05). In contrast, articles published by Quintessence Publishing Co., Inc., exhibit a higher percentage of noncitations (p < 0.001), a result that is adjusted for the effect of other factors.

Phase 2 – Analysis of the association, evolution, and trends of the main studies

Based on the nature of the methodological design, results from Phase 2 are presented here. These results provide a rigorous analysis of the scientific output published by publishers; the results also show trends in the scientific output published by publishers within the time window selected for the extraction of articles and based on the factors associated with the impact of scientific production.

The materialisation of the bibliometric analysis also allowed for an exhaustive analysis of the keywords associated with the scientific literature on dentistry so that the output covered in the literature, its associations, and the scientific trends represented can be understood. Therefore, Figure 3 presents the main keyword co-occurrence network in dentistry research that can be found in the WoS database; this figure allows us to identify relational and associative factors for keywords through clusters for an understanding of thematic affinities.⁴⁶

Figure 3. Grouping of thematic clusters.

Figure 3 Alt Text. The Figure depicts the main keyword co-occurrence network in dentistry research from the WoS database, revealing thematic relationships and trends based on rigorous analysis of scientific output and impacts in Phase 2 of the methodological design.

The main thematic cluster in red shows that “dental education” is a central concept. This concept is associated mainly with other keywords such as “paediatric dentistry”, “curriculum”, and “undergraduate”, which add important elements to the analysis; for example, it includes the idea of learning paediatric dentistry through cutting-edge educational models such as the flipped classroom⁴⁷ and virtual reality simulations.⁴⁸ Moreover, this approach allows for a rigorous analysis of the basic components that must be understood in the curricula of paediatric dentistry, as studied by Kaur et al.⁴⁹ in the Australian educational context.

The green thematic cluster is the second most relevant in terms of dentistry scientific production. On the one hand, results show that the concept of “paediatric dentistry” is also associated with other terms such as “children”, “oral health”, and “dental care”. These terms have generally been associated with shaping the scientific literature in terms of the dental needs of children who have special health needs, as described in studies that have analysed the dental characteristics of paediatric oncology patients, for example in Brazil.^50,51

On the other hand, there is an association between the green cluster and the yellow thematic cluster, which accounts for concepts such as “evidence-based dentistry” and “knowledge”.

Indeed, academia has expanded the knowledge, attitudes, behaviours, and practices of oral health professionals based on evidence-based dentistry. According to Rawat et al.,⁵² this approach has improved the clinical experience of dental professionals. The approach has allowed researchers to improve their decision-making in relation to treatment safety; it has also increased treatments’ success rate. However, other authors have suggested that this success may be due to the development of robust and specialised educational programs that use evidence-based dentistry to allow dentists to improve their knowledge and skills.⁵³

In turn, this cluster also presents an important association with the purple thematic cluster, which accounts for overall essential concepts of health today, such as “stress” and “anxiety”; indeed, students of dentistry experience great psychoemotional burdens and high rates of anxiety. These concepts affect the dental system, including behaviours such as bruxism or cheek biting, which are important oral health problems.^54,55 This issue is of interest not only to dental students but also to general dentistry practitioners and specialists in paediatric dentistry, as shown in the context of Iran.⁵⁶

Finally, the blue thematic cluster includes important concepts such as “pandemic”, “infection control”, “endodontics”, and “infection”, which have become relevant in the scientific literature on dentistry in recent years. This cluster aids in the analysis of specific situations in the context of the COVID-19 pandemic; for example, it helps in assessing the risk mental health workers have had of contracting the virus, showing that it has been considerably low because of intensive triage and comprehensive preventive measures⁵⁷; moreover, the cluster shows the effectiveness of blended learning in dentistry, specifically in endodontics, in the pandemic-induced context of isolation.⁵⁸

Subsequently, the analysis of the conceptual evolution of the scientific literature on dentistry is shown in Figure 4, which presents the most frequent and investigated keyword in each year during the time window included in the database. In that sense, it is clear that the trends in dentistry research at the beginning of the previous decade already revealed the interest of academics, researchers, and oral health professionals for paediatric dentistry, with “paediatric” being the most researched concept in 2010, time in which important advances were made on the topic. For example, in the study by Poletto and Junior,⁵⁹ one of the key review articles in the field, the authors concluded that the previous literature provided little conclusive scientific evidence, specifically in the Brazilian scientific literature, a context in which the prevalence of alveolar bone loss was also studied in children considered healthy and treated in private paediatric dentistry clinics.⁶⁰ Moreover, it is evident that this concept has dominated the literature since that year, constituting the most researched term in 2012 and 2019, which further indicates the relevance of the topic.

Figure 4. Evolution of scientific approaches.

Figure 4 Alt Text. The figure showcases the conceptual evolution of dentistry's scientific literature, highlighting the most frequently explored keyword for each year within the specified database timeframe.

Moreover, dental education has been a central issue in the research field, and it was the most investigated concept for the year 2015, when important contributions were made. For example, Musa, Bernabé and Gallagher⁶¹ studied the main motivations of Malaysian students for studying dentistry based on psychometric methodologies and the robust statistics of a confirmatory factor analysis. In addition, dental education has been the area of interest of the greatest number of studies in 2022, where the topic has been addressed in terms of state-of-the-art technologies, such as the implementation of standardised patient methodologies that allow students to develop communication skills, deeper diagnostic reasoning, and critical thinking.⁶²

In conjunction, it is evident that for 2020 and 2021, the priority in research was associated with the COVID-19 pandemic, and the term “pandemic” was the most studied concept in the scientific literature on dentistry. Some studies covered aspects associated with fears, eating habits, and perceptions of oral health,⁶³ and general evaluations of the impact of the pandemic on dentistry were made in specific contexts such as the Czech Republic, as evidenced in Schmidt et al.⁶⁴ These results contribute to the analysis of and debate on communication management during the current pandemic, a time during which dentistry has clearly emerged as a field in which research must be developed in consideration of international collaboration. The results for the last three years allow us to infer that the pandemic is an aspect that can propose new directions for research in this field.

Finally, terms of keyword analysis for the scientific activity in dentistry that has been indexed in the WoS database are shown in Figure 5. From a Cartesian plane, this figure compares the frequency of keywords with the year in which they were used as a measure of validity. Therefore, there are a total of four quadrants: quadrant I includes the most frequent and current concepts; quadrant II plots the least frequent and most current concepts; quadrant III includes the least frequent and the least current concepts; and finally, quadrant IV shows the most frequent concepts that are less current in the scientific literature.

Figure 5. Validity and frequency of keywords.

Figure 5 Alt Text. This figure displays a keyword analysis of indexed scientific activity in dentistry from the WoS database. Using a Cartesian plane, the figure illustrates the frequency of keywords against the year of usage, categorizing concepts into four quadrants based on frequency and recency. Quadrants I-IV depict varying combinations of frequency and currency in the scientific literature.

Therefore, quadrant IV, by positioning the concepts most frequently used for years showing a low average use, represents the terms whose use has declined in the research field; these terms only include “oral health”, which was previously key in stratifying the population for preventive dental care programs.⁶⁵

In a less prominent position, quadrant III groups the current concepts used the least frequently in the research field. This quadrant provides a general panorama for concepts that are not very prominent in the immediate future of dental research, noting terms such as “prevention”, “forensic”, “ethics”, “public health”, “dental caries”, and “restorative”. This phenomenon may indicate that researchers have already explored these topics in great depth or that there has been a constant evolution in the areas and/or the terms addressed; this scenario is reflected in the use of “forensic” in dentistry, which is currently related to the study of dental DNA, involving the use of prosthetic devices in recognition procedures, or to the issue of identification from the oral microbiota.

On the other hand, quadrant II includes those keywords that, although they are among the least frequently used in dentistry research, are among the most current, being categorised as emerging keywords. Among these main emerging keywords in the scientific literature on dentistry, there are research methodologies such as “bibliometrics”, which has been used in studies such as Paschoal et al.’s,⁶⁶ one of the most cited articles on the application of lasers in dentistry; also emerging is “infection control”, which in recent years was investigated by Kızılcı et al.⁶⁷ in relation to the level of knowledge that dental students should have regarding the factors that influenced the spread of the COVID-19 pandemic; and “digital” has also emerged, referring to the implementation of new technologies such as artificial intelligence in the dental context to improve medical diagnoses and patient care.⁶⁸ Emerging keywords are related to new technologies and new conditions, although they can also reflect the continuous development of an existing technology. The term “digital” can be associated with different concepts that involve not only artificial intelligence but automation, virtual reality and augmented reality, and 3D printing and robotic dentistry, among others.

Likewise, in terms of the specific analysis of the concept of the COVID-19 pandemic, future research must not explicitly focus on the analysis of that disease but understand that the post-pandemic context is already being discussed. Therefore, greater prominence will be given to the themes derived from this historical context, such as the application of new technologies, which were recently mentioned, and the response to the Industry 4.0 context or the Fourth Industrial Revolution.

Finally, quadrant I, which relates the most frequent and current use of concepts, is considered to account for the growing use of concepts on the subject; this quadrant includes the concepts that are the most prominent and relevant to the scientific literature on dentistry. It also includes the term “dental students”, who have been studied from different perspectives. For example, the effectiveness of e-learning classes in contexts such as Pakistan has been analysed⁶⁹ in a study in which the authors drew conclusions about the level of dissatisfaction generally perceived among students, revealing problems related to the use of new technologies. In addition, the keyword “education” was identified, where important contributions to dental practice have been generated, as evidenced in studies that have evaluated the quality of the educational system’s services for dental students based on a validated psychometric model such as SERVQUAL and on variables such as educational sector, country, gender, and year.⁷⁰ Finally, the concept of “paediatric” has emerged, referring to paediatric dentistry in which new approaches have made it possible to understand the role of new technologies such as cone-beam computed tomography; studying factors, including the degree of awareness; and knowledge of correct use.⁷¹ The scientific production centered on students is focused on their skills and experiences that contribute to problem solving and lifelong learning; this research has focused on students’ performance in building meaning based on previous experiences and new information.

Strengths and limitations

A total of 110,269 articles published in WoS journals were reviewed. Although this volume is large, it must be recognised that this study constitutes only a partial vision because other databases have not been included. On the one hand, the analyses carried out include different numbers of articles, which could confuse readers; however, it was strictly necessary, as the annual scientific production selected could not include the year 2022, which is not published until the third of that month. The factor analysis did include all of the studies, and it was evident that the 9,034 articles published in 2022 contained current bibliometric information. On the other hand, comparing journals was expected for the analysis of impact factors; however, editorials were considered for technical reasons, and the journals were part of more than one scientific category. Moreover, when a good article was not accepted in a journal based on field or category, suggestions for other journals were received from the same publisher. Finally, the bibliometric analysis includes 1,000 more recent articles because these contain information on emerging issues from the point of view of journals and therefore publishers. Perhaps these results constitute valid reasons for the authors, but not for the readers.

The study includes modelling in the bibliometric analysis. First, for a demonstration through statistical tests rather than subjective appraisals, the great growth in scientific production in dentistry is shown through discontinuous regression models. Changes in the timing of production are shown through regression models by sections, and these models are adaptable to biometric analyses on emerging issues in various scientific areas or categories. Moreover, this study retrieves factors related to the positive impact (citations) or lack thereof (noncitation) of articles published in dentistry. Authors and journals may take these results into consideration to improve their bibliometric metrics and thereby increase the prestige of the editors while guiding them in making decisions about editorial policies. Finally, bibliometric analysis provides information on emerging issues that authors should consider regarding the acceptance of their articles in WoS journals using classical and current methodologies.

Additionally, although it is true that the factors associated with the citations of scientific production in dentistry are similar to those used for other health disciplines, we recognise that some factors can be differentiated, such as those base research topics, since dentistry focuses on diagnosis, the treatment and prevention of diseases and the conditions of the dental system. Other health disciplines may focus on different areas of the human body or methodological aspects of research. Although there is a wide variety of research methodologies addressed in health disciplines, some may be more common in dentistry. The existence of specialised magazines that publish scientific articles have framed dentistry. Moreover, some health disciplines have superior means of dissemination for research results compared to dentistry, which affects the output in that field.