Lifestyle risk behaviors of university students : a bibliometric analysis

Objective: Examine the evolution of publications on the lifestyle risk behaviors of university students. Methods: Bibliometric research was carried out on using 9011  articles retrieved from the SciELO and Web of Science database. The analyses included, the trend of the production, the most productive countries, journals, institutions, and the relevant information was extracted based on frequency of co-occurrence of keywords between 2009 and 2019 using the bibliometric software, namely VantagePoint, VOSviewer, NetDraw and UCINET. Results: The study revealed a significant concentration of publications between 2014 and 2018 in North American countries, especially the United States. However, the most productive institution was the University of São Paulo (Brazil). Stress, physical inactivity, obesity, and smoking were the risk behaviors most used as keywords in the articles, and these have a strong relationship with other keywords subsets related to mental health, forms of treatment, the study population and lifestyle behaviors. Conclusion: The scientific map of lifestyle risk behaviors among university students was supported by exhaustive research. It was possible by the establishment of research networks between the various centers of knowledge production, especially with the American researchers.

Traditionally, identification of health risks has focused on middle aged and older adults. However, the high prevalence of health risk in young adults (Santos et al., 2015;Zajac-Gawlak et al., 2016), a population that includes university students (Cruz-Sánchez et al., 2016;Macedo et al., 2020;Morawiec et al., 2016;Torquato et al., 2016) has garnered the attention of health professionals and researchers as a modifiable behavior that amenable to early intervention to prevent illness and avoidable deaths.
In recent years, the number of publications on risk behaviors in the university population has grown more than 100%. Knowing the research trends and characteristics of these publications is important to identifying gaps in knowledge and to assessing how themes have evolved in the production of science in this area (Avena & Barbosa, 2017;Giménez-Espert & Padro-Gascó, 2019).
Bibliometry or literature bibliometric analysis is an important field of information science that offers a method for making inferences about bibliographic production. Through statistical analyses, bibliometry and bibliometric indicators quantify information about the scientific production of an area or certain topics (Ravelli et al., 2009;Zeleznik & Kokol, 2017).
Currently, bibliometric studies are increasingly used for the quantification and analysis of scientific knowledge in a wide range of fields, including health (Avena & Barbosa, 2017;Giménez-Espert & Padro-Gascó, 2019). The interpretation of bibliometric indicators is essential to support decision-making on funding strategies to support the best return in Indicators of bibliometric production can contribute to the management of Graduate Programs (Maciel, Faria, Milanez, & Lança, 2018) and to academic success by mapping the most cited articles, the most productive institutions, the preferred journals (Ravelli et al., 2009;Zeleznik & Kokol, 2017), current topic trends, growth in knowledge (Avena & Barbosa, 2017), as well as identifying relationships between subjects and areas of knowledge (Iqbal et al., 2019;Shi, Miao & Si, 2019).
According to a previous survey of the scientific literature, bibliometric analysis and mapping methods have not yet been used to analyze the literature on lifestyle risk behaviors, which constitutes an important research domain with university students.
This study highlights the usability of bibliometric analysis to understand the scope of publications on this theme with university students, as well as the characteristics and the trend of publications, besides serving as a guide or example for other studies. Thus, the objective of this study is to examine the evolution of publications on the lifestyle risk behaviors of university students.

Materials and Methods
This study refers to the national and international bibliographic production on the lifestyle risk behaviors of university students, between 2009 and May 2019. The study is a bibliographic, descriptive, and quantitative approach research.
The bibliographic metric analysis included the identification of the number of publications per year, countries, institutions with the largest number of publications, journals with the most articles, in addition to mapping the co-occurrence of the keywords.

Database
The Web of Science (WOS) and the Scientific Electronic Library Online (SciELO) database were used as a source of information for the elaboration of indicators of scientific production of the lifestyle risk behaviors of university students in its wide and varied number of scientific records.
The WOS is is one the world-leading scientific citation search and competing analytical information databases. Also, It is an increasingly significant scientific instrument across countries/regions and knowledge domains, being the database dataset for large-scale Development, v. 9, n. 10, e1699108624, 2020 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v9i10.8624 6 data-intensive studies with thousands of published academic studies over the past 20 years (Li, Rollins, & Yan, 2018).
The SciELO provides academic literature in sciences, social sciences, arts, and humanities, published in the main open-access journals in Latin America, Spain, Portugal, the Caribbean, and South Africa, in English, Spanish and Portuguese languages.
When connected to the multidisciplinary citations on the Web of Science™ platform, the SciELO Citation Index offers a full citation network to connect with superior accurately and confidently to the highest quality multidisciplinary research. Therefore, researchers from more and more and knowledge domains are involved in the use of the WOS and SciELO.

Search strategy
A specific search strategy was developed using the keywords (single words or key words) related to lifestyle risk behaviors, indexed in the Health Sciences Descriptors (DeCS) to locate, and retrieve the articles from the databases.
After establishing the sets of primary keywords, that is, the keywords that correspond to the central concept of the document, the wildcard character was defined, represented by the asterisk (*), which was used in searches to locate plurals and variants of words. Moreover, the terms were associated with the Boolean operators "and", "or" and "not"(DeCS, 2016), as shown in Table 1.  (((((diet* or "nutrition status" or "alcohol drinking" or "alcohol drinking in college" or smoking* or "tobacco smoking" or "cigarette smoking" or obesity* or overweight* or "body mass index" or exercise* or "physical activity" or hypertension* or "blood pressure" or stress* or "diabetes mellitus" or dyslipidemia*) and (student* and university*)) not ("middle school" or adolescent* or child* or teenager*) 09/05/2019 11.587 Filter = full papers available on-line 9.125 Deleting replicas 9.011 Source: Research data.
The inclusion criteria adopted for sample selection were publications between 2009 and 2019, and complete articles that were available online. Repeated publications, book chapters, reports, review articles, editorial, abstracts, theses and dissertations were excluded.

Elaboration and analysis of indicators
The 11,587 publications were downloaded from the database as text file format (.txt) and exported to VantagePoint software® (version 5.0). VantagePoint is a powerful data analysis tool that can organize, clear, and analyze the imported data to provide lists, statistical maps and matrices that identify production indicators. This software can quickly clarify questions such as who, what, when and where the publications occurred, transforming the data into information and knowledge (Dudziak, Fausto, & Costa, 2014;VantagePoint, 2016).
After verifying and deleting repeated publications, the 9,011 articles were analyzed as lists, the matrices were built in VantagePoint and transferred to Microsoft Excel 2007 to support the preparation of basic indicators, presented as tables and graphs.
For network analysis, UCINET software (version 6.0) was used, allowing the quantitative analysis of scientific collaboration networks through two-dimensional or threedimensional graphs. After this phase of organization and systematization of networks, the data were sent to the NetDraw software (version 2.10) for visual analysis of the map. NetDraw is integrated into the UCINET package and is used to maximize the perception of relationships between the variables observed in collaboration networks (Borgatti, Everett, & Freeman, 2002).
Network maps were also developed with the support of VOSviewer software (Visualization of Similarities version 1.6.11). This software has been widely used in bibliometry and allows analyzing large-scale data sets and buildings complex networks according to the connection strength. The analysis of the keywords was performed based on all the keywords that the authors used in the abstracts of the articles. The standard value of ten occurrences was considered as a cut-off point for inclusion in the co-occurrence network. The word font size and circle size describe the number of documents related to each key word, and VOSviewer allows grouping the terms into clusters to facilitate the interpretation of cooccurrences.

Results
The results of the analysis of the 9,011 articles are presented below, highlighting the production of 10 key words, countries, institutions and means of dissemination with the highest number of publications associated with the theme (i.e. lifestyle risks in university Research, Society and Development, v. 9, n. 10, e1699108624, 2020 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v9i10.8624 students). This presentation of results allows a better mapping of science, thus facilitating the comparison between those considered "Top 10" within the study area.    The USA, along with Australia and Spain, accounted for more than 50% (n = 4,505) of world production. Also, in these three countries, it was possible to observe that 2018 was the most productive in number of publications for the USA (n=302) and Australia (n=99) . Spain had its highest number of publications (n=113) recorded in 2017, data not shown in Figure 2.

Geographic location of publications
Brazil ranked fourth among the ten countries with publications on the risk behaviors of university students, contributing about 6.2% (n =556) of total production. It is worth mentioning that Brazil was the only representative of South America present in the list of top 10 countries with the highest number of publications (Figure 2).

Collaboration networks of publications
Collaborations among the Top 10 countries with most publications (2009-2019) The network analyses revealed a strong network of international collaboration (i.e. largest number of "nodes" or the size of connections, which represents a greater number of publications) between the USA in partnerships with researchers from abroad ( Figure 3).

Concentration of publications in Top 10 international institutions
The articles on lifestyle risk behaviors of university students between 2009 and 2019 were published by researchers in different countries and higher education institutions.  and Federal University of Santa Catarina (n=54). These two institutions reached about a quarter of the publications of the ten most productive institutions together (Table 2).

Journals with the highest number of publications
Top 10 international journals Among the 2,912 journals that published the 9,011 articles on risk behaviors in the university population, Nutricion Hospitalaria published the most articles (n=145), followed by the Journal of American College Health (n=120) and Plos One (n=98). The table 3 shows that the most productive journals published a total of 366 articles in the past 11 years. Despite having a lower impact factor, Nutricion Hospitalaria was the journal that published more manuscripts on the subject investigated. However, the Top 10 journals publishing articles on this topic did not include any Brazilian journal.

Top 10 keywords
This study identifies the most common keywords used by the authors in the analysis of the 9,011 articles. The keyword "university student" was present in more than 597 publications and the keyword "student" in 587 of the publications. The authors used some synonyms such as college student (280) and medical students (243) to refer to the study population or sample. In characterizing the study sample, the keyword "medical students" (n=243) ranked eighth and identified a specific professional discipline (Figure 4). Finally, a diverse range of keywords were used in these published articles identifying graduate courses in specific areas (e.g., medicine) and courses in various areas of higher education (e.g. university students and students). Among these, risk factors such as stress, physical activity, depression, obesity, anxiety, and smoking stood out as the focus of the studies. Upon analyzing the distribution of keywords, using the VOSviewer software, an important network of co-occurrence of the keywords sets was visualized ( Figure 5). The keywords are clustered into groups by colors that represent the following terms: 1) Red (n=89): mental health and stress; 2) Green (n=87) study population, university students from various programs; 3) blue (n=64), use of substances such as alcohol and tobacco; and yellow (n=63) physical activity, overweight, diet/nutrition, with an emphasis on the first two. Figure 6. Density visualization of co-occurrence of keywords in publications.
Source: VOSviewer Figure 6 shows the mapping of the co-occurrence density of the keywords following the trend identified in Figure 5. Higher density was found mainly among keywords related to mental health and forms of treatment (stress, depression, mindfulness), the study population (university students, students, college students), and lifestyle risk behaviors (alcohol beverage, smoking and obesity).

Discussion and Final Considerations
The results of this research showed that there was a significant increase in the number of publications on the lifestyle risk behaviors of university students in the period from 2009 to 2019. In total, 9,011 articles were identified and analyzed.
The growing number of studies with this population suggests an awareness that during University years students undergo lifestyle changes that may favor the emergence of one or more risky behaviors (Santos et al., 2015;Torquato et al., 2016). Comparing the number of articles published over the years reveals a significant increase in the number of publications in 2018, both internationally and nationally.
As for geographic location, the number of publications was higher in developed countries, especially the USA, Australia, and Spain. The USA invests significantly more in science and technology than any other countries and is considered a world leader in several areas of knowledge (Iqbal et al., 2019;Shi et al., 2019) that include consolidated studies on lifestyle risk behaviors, such as the Framingham Heart Study and the Atherosclerosis Risk in Communities Study (Duncan, Chor, Aquino, & Bensenor, 2012).
Brazil is a developing countries which is showing significant growth in scientific publications on this topic in recent years (Maciel et al., 2018;Sidone, Haddad, & Mena-Chalco, 2016). Data from a Clariviate Analytics report entitled Research in Brazil, indicates that the country is increasing the number of research publications worldwide due to changes in research policy and funding (Cross, Thomson, & Sinclair, 2017). In addition, according to the 2018 report entitled National Strategy of Science, Technology and Innovation, the Brazilian government has been increasing investment in research development since 1990, which is a central strategy for increasing scientific, economic, and social development.
Nevertheless, significant cuts in research by the current government seems likely to affect future scientific contributions in many areas.
Another finding is that despite the significant number of Brazilian articles, the majority were collaborations with other national institutions. International collaboration was identified with only five countries (i.e., Spain, England, Canada, Germany, and the USA).
Cultural and geographic factors may explain the low standard of internationalization of Brazilian productions. According to Sidone et al. (2016), Portuguese is a relevant language barrier that hinders the dissemination of knowledge and the formation of international collaborative networks.
Networks of collaboration across institutions and countries are critical to intellectual exchange; science is increasingly collaborative in construction of knowledge (Sidone et al., 2016). In general, the more diverse the research collaborations, the greater the productivity, visibility and scientific impact (Lovakov & Agadullina, 2019). Collaborative networks generate a greater dissemination of knowledge in different geographical spaces. Looking forward, Brazilian researchers need to strengthen and expand international partnerships to ensure the contributions of Brazilian science are recognized internationally across the scientific community (Uvacsek et al., 2014). In the last decade, the search strategy shows growth in the numbers of studies using keywords related to risk behaviors in university students. These results also point out that the greatest co-occurrence of keywords was related to the field of mental health, especially the keywords stress, present in more than 518 articles and considered the third most used by the authors. Studies suggest that the main sources of stress found in the university routine are related to academic demands (e.g. number of courses, reading, testing and assignments) leaving little time for leisure, family and friends (Brito, Gordia, & Quadros, 2016;Leiva, Martínez, & Celis-Morales, 2015).
In addition to stress, disorders associated with mental health, such as depression and anxiety, appeared as keywords of 12.5% more of the 9,011 articles. Thus, these keywords point to the high frequency of studies on the mental health of university students. Strategies for prevention and control of these disorders while in academic programs, such as the implementation of mental health care services in the university fields, are needed (Rotenstein et al., 2016;Torquato et al., 2016).
Based on analysis of study samples, a majority focused on medical students.
According to global estimates, there are about 2,409 medical schools in more than 180 countries, with the highest concentrations (representing 40% worldwide) in India, the United States, China, Brazil, Japan, Mexico, Pakistan and Russia (Rigby & Gururaja, 2017). Medical programs have clinical and academic requirements that can contribute to stress and exhaustion. In 2016, a meta-analysis reported that a higher prevalence of depression or depressive symptoms (27.2%) and suicidal ideation (11.1%) than the general population (Rotenstein et al., 2016).
The research analysis of the keywords identified other common lifestyle risk behaviors of students which affect cardiovascular health (e.g., physical inactivity, obesity, and smoking). These lifestyle risk behaviors are a main cause of premature mortality from heart disease among adults and young people, especially among students (Brito et al., 2016).
Studies conducted with university students state that physical inactivity is prevalent regardless of nationality-Saudi Arabia ( In international collaboration networks, there was a predominance of researchers with English as their first language who work in the same geographic region. The most used keywords were related to the study population (e.g., university students), followed by risk behaviors (e.g., stress, physical inactivity, obesity, alcohol consumption and smoking). In addition to stress, there was a concentration of publications with keywords related to mental health, such as depression and anxiety.
However, the bibliometric analysis showed the necessity to stimulating new studies and scientific production about the lifestyle among university students. In addition, the results suggest an expansion of the search in other databases and the deepening of the analyzes, in new studies, with different countries that illustrate and confirm the main risk behaviors related both to the lifestyle of university students and to mental health during the training period.
This study also presents some limitations. The accuracy of the SciELO and the main collection of the Web of Science platforms were used to obtain information about the publications, including data from countries, institutions, and journals. The Web of Science data were not verified with other databases due to human resources limitations. Another Despite these limitations, the value of this study is in illustrating the trend of publications on lifestyle risk behaviors, highlighting the countries, institutions, and journals.
Also, the study confirming the main network of co-occurrence of the keyword on lifestyle risk behaviors among university students.