Electronic cigarettes: Emerging trends and research hotspots

INTRODUCTION Research on electronic cigarettes is an emerging field, with the number of articles in this field noted to have grown exponentially over recent years. We used a bibliometric analysis method (co-word analysis) to analyze the emerging trends and research hotspots in this field. METHODS Publication data on electronic cigarettes from 2010 to 2018 were retrieved and downloaded from the PubMed database. Theme trends and knowledge structures were analyzed on the relevant research fields of electronic cigarettes by using a biclustering analysis, strategic diagram analysis, and social network analysis methods. Research hotspots were extracted and compared from three periods. RESULTS Core topics that have continuously develop between the years 2010 and 2018 include: tobacco use cessation devices; tobacco products; tobacco use cessation devices/adverse effects; smoking prevention and adverse effects; electronic nicotine delivery systems/economics; and public health. Some currently undeveloped topics that could be considered as new future research directions include: tobacco use disorder/therapy; tobacco use disorder/epidemiology; students/psychology; students/statistics and numerical data; adolescent behavior/psychology; nicotine/toxicity; nicotinic agonists/administration and dosage; and electronic nicotine delivery systems/legislation and jurisprudence. CONCLUSIONS Results suggest that some currently immature topics in strategic coordinates and emerging hotspots in social network graphs can be used as future research directions.


INTRODUCTION
An electronic cigarette (e-cigarette) refers to a cigarette consisting of a battery, an evaporation heating device, and a tobacco tube containing a liquid smoking product. The nicotine-containing tobacco liquid can be turned into vapor by nebulization for the user to inhale 1 . There is growing evidence that, even though e-cigarettes may produce fewer toxic substances than traditional cigarettes, they may still pose health risks to smokers and people around them 2,3 . The long-term effects of e-cigarettes on health are not currently apparent. Moreover, there is insufficient evidence to show that such products may assist people to quit traditional cigarettes 4 .
A large number of recent studies have shown that, from a toxicological standpoint, e-cigarettes as a substitute for nicotine, compared to traditional smoking, may aid in improving public health 5,6 . However, e-cigarettes may contribute to adverse reactions in the respiratory system 7,8 , cardiovascular system 9 , liver 10 , and nervous system 11 . Ever since e-cigarettes became popular among adolescents, the number of studies on e-cigarettes has increased 12,13 . In recent years, researchers have become increasingly concerned about tobacco-use disorders 14 . Some studies have shown that comprehensive interventions are needed to help protect adolescents' mental health 15,16 . With the development of bibliometrics comes its widespread use in health topics. However, there is only a limited amount of bibliometric analyses that focus on tobacco, and we have found only two that focused on e-cigarettes 17,18 . This study used a co-word analysis instead of the co-citation analysis applied by the above two articles, to examine trends in e-cigarette research.

Data collection and bibliographic matrix setup
This study used the Bibliographic Item Co-Occurrence Matrix Builder (BICOMB) to extract data for the three periods 2010-2012, 2013-2015 and 2016-2018, from PubMed including journals, countries, authors, and major MeSH (Medical Subject Headings) terms/ subheadings from which were calculated the number of high-frequency major MeSH terms/subheadings, using the Donohue equation 19 : T=(1+√(1+8i) / 2 where i is the number of major MeSH terms/ subheadings that have occurred only once. BICOMB was then used to generate the term-source article and co-occurrence matrix as the data basis for our subsequent bibliometric analysis. Ethical approval was not necessary according to the Ethics Committee, Medical University of Lodz, as this was not an experimental study.

Biclustering analysis
The current study used a biclustering analysis of the high-frequency major MeSH terms/subheadings and PubMed unique identifiers on e-cigarette related publications retrieved from PubMed 20 . Biclustering analysis was conducted to classify the major MeSH terms/subheadings based on the term-source article matrix. In this work, we use word groupings of high-frequency major MeSH terms/subheadings as keywords. The gCLUTO software was used to create a clustering mountain visualization and to construct a visual matrix using the repeated bisection method.
In a clustering mountain visualization, each peak represents a cluster and each peak position, volume, height and color correspond to the relevant cluster data, enabling an easy, intuitive observation of the structure of each research hotspot in the field of e-cigarettes. The color of the mountain peaks is proportional to the standard deviation within the class. Red represents low standard deviation and blue represents a high standard deviation, but only the color of the peak is significant. The distance between the peaks expresses the similarity between clusters. This result proposes and analyzes research priorities in related fields. GraphPad 5 software was used to create a strategic diagram to analyze the mutual influence within and between domains of e-cigarettes. The strategy diagram has two axes: the y-axis represents density, indicating the ability to maintain and develop itself; and the x-axis represents the centrality, indicating the degree of interaction between the indicated field and others 21 . Based on the high-frequency significant MeSH terms co-occurrence matrix and biclustering analysis, we can calculate the centrality and density of each cluster to describe the internal connection and interaction between clusters in a research field. Density is the closeness of the keywords in each cluster, which indicates the ability of the cluster to maintain and develop itself, and centrality is the closeness between the keywords of each cluster and of other clusters, which indicates the interaction effect. The two axes produce four quadrants with major MeSH terms/subheadings assigned to different quadrants based on the results of the biclustering analysis. By comparing the strategic diagrams of the three study time periods, the evolution of a field of e-cigarettes can be estimated.

Social network analysis
The current study used Ucinet 6.0 (Analytic Technologies Co., KY, USA) to construct a social network analysis (SNA) in order to further analyze the knowledge structure of the e-cigarette field. Applying NetDraw 2.084, the major MeSH terms/ subheadings network was able to be presented in a 2D map. The nodes represent the major MeSH terms/subheadings, and the links represent their co-occurrence frequencies. Each significant MeSH term/ subheading can be evaluated by three parameters, namely, degree, closeness, and betweenness. Degree refers to the number of other nodes directly connected to a node. The higher the degree centrality of a node, the more critical it is to indicate its position in the network. Betweenness indicates the number of shortest paths through a node. The more times a node acts as an intermediary node, the higher is its significance in evaluating the importance of a node in the network. Closeness reflects the closeness between a node and other nodes in the network. It is the sum of the reciprocal of the shortest distances from a specific node to all other nodes in the network. This means that the higher the closeness, the shorter is the distance from this node to other nodes in the network. It is worth noting that the betweenness index was chosen as an evaluation index for the in-depth study of the e-cigarette field.

Characteristics of e-cigarette related publications
We retrieved and analyzed 54, 968, and 2406 publications for each period, respectively. The number of articles related to e-cigarettes has increased from 8 in 2010 to 988 in 2018, nearly 120 times during the past nine years. Table 1 lists the top ten countries by number of publications, while Table 2 lists the top ten journals (by number of articles), for the three periods. We suggest that these represent the core changes in the field of e-cigarettes over the past nine years. England was the country with the most published articles in the three periods, followed by the United States. China ranked tenth in the first period, but had no special contribution in the other two time periods.
In the first time period (

Research hotspots in MeSH term clusters
According to the publications retrieved for the three time periods, there were respectively 15, 26 and 49 high-frequency major MeSH terms/subheadings (Supplementary file, Table S1) with a total frequency of occurrence of 49.6%, 49.5% and 50.4%, respectively. We considered these as the research hotspots for the three time periods. A biclustering analysis leads to a division of MeSH terms into 4, 3 and 5 clusters for each period, respectively (Table 3). Biclustering analysis results of the high-frequency major MeSH terms/subheadings in the field of e-cigarettes for the Table 1 Table 3. The size of a signal node represents the number of major MeSH terms/subheadings involved in each cluster.

Theme trends of e-cigarettes
Callon 22 explains the meaning of strategic diagrams (as shown in Figure 1). Knowledge structure of e-cigarettes As shown in Figure 2, the three SNA diagrams are constructed by the three indicators: degree, betweenness, and closeness; where betweenness is the index. The size of the nodes is proportional to the betweenness of these major MeSH terms/subheadings, while the thickness of the lines represents the cooccurrence frequency of MeSH terms pairs. As

DISCUSSION
As our knowledge of e-cigarettes continues to increase, the amount of related research also continues to grow and e-cigarette research has become an emerging field. Using a biclustering analysis, strategic diagrams, and social network analysis diagrams, we analyzed, in detail, the evolution of thematic trends and knowledge structures in the field of e-cigarettes over the past nine years. This is the first time co-word analysis was used to analyze trends in this field.
The current study examines e-cigarette publications, globally, by comparing three time periods in the past nine years (2010-2018). Over that period of time, the number of publications related to e-cigarettes has grown rapidly, with England and United States leading the way with the most published articles. Nicotine & Tobacco Research, and Tobacco Control, are the journals that have published the most articles, however the current study only analyzed the number of e-cigarette articles published in a journal, which is greatly affected by the total number of articles published in a journal. We note that although some journals have a smaller total volume of articles, they also publish in the field of e-cigarettes and have considerable influence, such as Tobacco Induced Diseases, and Tobacco Prevention and Cessation etc.
Strategic diagrams were used to analyze the theme trends of publications of three time periods. In the first time period (2010-2012), only cluster 0 is in quadrant I, which includes Nicotine/administration & dosage, and Smoking Cessation/methods. We consider these two themes to be research hotspots in the field of e-cigarettes during this period. The research in this period suggests that there is a need for more effective drugs to help smokers quit smoking. During this period, some important research focused on smoking cessation methods and compared various methods including nicotine replacement therapy (NRT), bupropion and varenicline, and nortriptyline and clonidine 23 . Clusters 1, 2 and 3 are in quadrant III. Research in these clusters focuses on: Smoking Prevention and adverse effects; Tobacco Use Cessation Devices/adverse effects; Tobacco Products; Tobacco Use Cessation Devices; and Smoking Cessation/legislation & jurisprudence. These topics are immature and are located in the periphery of e-cigarette research during this period. They may gradually shift to a central and/or mature position, if further research is undertaken. Studies have focused on adverse effects of smoking, and some have now been widely recognized, e.g. smoking cigarettes is the strongest risk factor for chronic obstructive pulmonary disease (COPD) 24 and remains the primary risk factor also for lung cancer 25 . Legislators are increasingly recognizing the dangers of tobacco products as well as the importance of tobacco control and protecting the public from the harm of tobacco 26 . With the development of e-cigarettes, there is increased concern about their safety as cessation devices, for which there is significant debate; however others have noted that they may provide a potential to quit smoking 27 .
In the second time period (2013-2015), Tobacco Use Cessation Devices, and Tobacco Products, were noted as two themes gradually maturing. In the last period (2016-2018), researchers had high hopes for the role of Tobacco Use Cessation Devices in smoking cessation. In this period, research was more extensive, for example, with the aim to apply large cross-sectional surveys to assess the effectiveness of e-cigarettes as an aid to smoking cessation 28 , or to quantify how smokers evaluate the attributes of e-cigarettes 29 . In addition to paying attention to the harm of Tobacco Products, it was also found that e-cigarettes also have the potential to do harm because they contain nicotine, which is addictive and can cause adverse reactions 30 . Therefore, both tobacco products and e-cigarettes should be treated with caution 31 40 . People are paying more attention to the connection between smoking and psychology, particularly the impact of e-cigarettes on youth psychology. In addition, from a social perspective, people also attach great importance to the legislation and jurisprudence governing e-cigarettes 41 . Three SNA diagrams were made according to the high-frequency MeSH terms/subheadings. In these three time periods, 4, 10 and 16 major MeSH terms/ subheadings, respectively, had a high degree of centrality. Nicotine/administration & dosage in the first time period, and Electronic Nicotine Delivery Systems/statistics & numerical data in the second and third periods are at the center of the SNA diagrams and have the greatest number of direct connections to other nodes, suggesting the most significant impact during each period.
In addition, in the second period, there are nine MeSH terms at the edge of the network that are new and immature. Among these, Nicotine/analysis, Nicotine/adverse effects, and Public Health became developed in the third period. We can consider these MeSH terms as emerging hotspots in the second period (2013)(2014)(2015)

Strengths and limitations
To the best of our knowledge, the current study is the first to use a co-word analysis method to perform a comprehensive analysis of e-cigarette publications. The e-cigarette field is constantly evolving, and there will be more in-depth research in the future. It is our contention that the emerging hot issues mentioned above can guide clinicians and researchers to develop new projects in the e-cigarette area. At the same time, this research has certain limitations. The first is that we only searched for journals, excluding comments and other types of literature, and perhaps missed some research hotspots. Secondly, co-word analyzes high-frequency MeSH terms only, which may affect the results of the cluster analysis as our results are based on the number of articles published in each medium and not the impact of each article. Moreover, in the future, we could use a variety of databases for analysis, such as Cochrane, Embase, clinical trials.gov, some guidelines could also be searched, as well as manual searching for grey literature.

CONCLUSIONS
We applied a biclustering analysis, strategic diagram and SNA methods, to analyze high-frequency MeSH terms, and conducted a co-word analysis on the field of e-cigarettes.