Scientometrics analysis of research activity and collaboration patterns in Chagas cardiomyopathy

Background Chagas cardiomyopathy is a serious and common complication of Chagas disease. Methods Through bibliometric and Social Network Analysis, we examined patterns of research on Chagas cardiomyopathy, identifying the main countries, authors, research clusters, and topics addressed; and measuring the contribution of different countries. Results We found 1932 documents on Chagas cardiomyopathy in the MEDLINE database. The most common document type was ‘journal article’, accounting for 79.6% of the total (n = 1538), followed by ‘review’ (n = 217, 11.2%). The number of published records increased from 156 in 1980–1984 to 311 in 2010–2014. Only 2.5% were clinical trials. Brazil and the USA dominated the research, participating in 53.1% and 25.7%, respectively, of the documents. Other Latin American countries where Chagas is endemic contributed less, with Bolivia, where Chagas disease is most prevalent, producing only 1.8% of the papers. We observed a high rate of domestic collaboration (83.1% of the documents published in 2010–2016) and a lower but significant rate of international collaboration (32.5% in the same time period). Although clinical research dominated overall, the USA, Mexico and several countries in Europe produced a considerable body of basic research on animal models. We identified four main research clusters, focused on heart failure and dysfunction (physical symptoms, imaging techniques, treatment), and on myocarditis and parasitemia in animal models. Conclusions Research on Chagas cardiomyopathy increased over the study period. There were more clinical than basic studies, though very few of the documents were clinical trials. Brazil and the USA are currently leading the research on this subject, while some highly endemic countries, such as Bolivia, have contributed very little. Different approaches could help to redress this imbalance: encouraging researchers to conduct more clinical trials, launching international collaborations to help endemic countries contribute more, and strengthening links between basic and clinical research.

the USA are currently leading the research on this subject, while some highly endemic countries, such as Bolivia, have contributed very little. Different approaches could help to redress this imbalance: encouraging researchers to conduct more clinical trials, launching international collaborations to help endemic countries contribute more, and strengthening links between basic and clinical research.

Author summary
Scientific production on Chagas cardiomyopathy has grown considerably since the turn of the 21st century, probably reflecting the increased incidence of Chagas disease in nonendemic areas like the USA and Europe. Brazil and the USA dominate the research, but we found a very small proportion of clinical trials on Chagas cardiomyopathy and a low scientific production in several endemic countries with a high prevalence of the disease such as Colombia, Chile, Mexico and Bolivia. We observed a polarity between endemic and non-endemic countries where clinical research and basic research predominate, respectively. Different approaches could help to redress the observed imbalance of research on Chagas cardiomyopathy: encouraging researchers to conduct more clinical trials, launching international collaborations to help endemic countries contribute more, and strengthening links between basic and clinical research. It is crucial to foster translational research in order to link basic knowledge on the physiology of the disease with clinical applications in diagnosis and treatment. method based on brain natriuretic peptide levels and diastolic function could detect early cardiac involvement in Chagas disease [15], potentially overcoming the limitations of traditional diagnostic methods that may misclassify patients. Such advances are highly significant, as the condition is the most common non-ischemic cardiomyopathy and a leading cause of complications and death in Latin America [16]. It carries a risk of malignant arrhythmias, conduction disturbances, heart failure, and pulmonary and systemic embolism, killing around 4% of patients treated for the condition on an outpatient basis every year.
Given the disease burden associated with Chagas cardiomyopathy, a specific analysis of research publications and collaboration networks in this area is warranted to build on the more general bibliometric studies of Chagas disease [17][18]. Increased knowledge on research in this pathology can help to foster North-South collaborations and other research initiatives with and among endemic countries that nevertheless may have relatively little scientific development on the topic. Moreover, this type of study is useful for the research community, clarifying the main lines of research that are being developed with regard to the diagnostic methods and treatments for the disease.
In this study, by analyzing scientific papers on Chagas cardiomyopathy published in the main international scientific journals, we aimed to identify the leading researchers, the contribution of different countries to the overall research effort, the degree and nature of scientific collaboration, and the topics addressed.

Methods
This study is based on bibliometric methods and Social Network Analysis, through which we collected relevant information from the scientific records indexed in the main bibliographic databases of scientific literature.

Identifying the population of documents
We selected the body of documents for our study from the MEDLINE and Science Citation Index Expanded (SCI-Expanded) databases. MEDLINE is the main international database for health sciences, and the terminology included in its Medical Subject Headings (MeSH) thesaurus can be used to search for published documents on specific aspects of a given topic. For our search, we identified all the documents indexed in the MEDLINE database with the MeSH descriptor of "Chagas cardiomyopathy". We then restricted the results to the "article" and "review" document types and to the 1980-2016 period for the calculation of all indicators and analyses, as articles and reviews are the main document types of reference that are subjected to peer review with regard to the dissemination of research activities.
The use of the MeSH thesaurus ensures that all the documents recovered focus on the topic analyzed, as this detailed instrument for controlling terminology combines the use of a team of specially trained indexers who analyze each article and assign medical subject headings to it with automated functions to improve the indexing process [19][20][21]. To identify the documents with a primary focus on Chagas cardiomyopathy, we screened search results by hand, analyzing titles, abstracts and key words. Moreover, we classified all documents that used "humans" or "humans" and "animals" as clinical, epidemiological, or basic studies.
One limitation of MEDLINE is that until 2013, journal articles included only the address of the first author. To depict more precisely the geographical distribution of research activities and collaboration, we identified the documents from MEDLINE that were also included in SCI-Expanded (61.64% of the MEDLINE documents), as this database shows the addresses of all authors. Moreover, as SCI-Expanded is a multidisciplinary database containing the publications with the greatest visibility and impact, it can reveal the patterns of collaboration in the most internationally relevant journals and enables the analysis of the impact of the publications, based on the citations they generate.
We performed the electronic searches on 25 March 2017 using the platform Web of Science, which contains both databases. Although we did not limit our results by date of publication, it is worth noting that the term "Chagas cardiomyopathy" was not in the MeSH thesaurus until 1981.

Indicators obtained and aspects analysed
Scientific activity and collaboration. We examined the evolution in the number of records published over the years as well as their distribution by document type and journal. We also analysed the geographical distribution of scientific production by institutions and by countries, and the extent and nature of scientific collaboration. We identified, in that sense, the documents with no collaboration (signed only by authors from a single institution) along with the documents arising from domestic (authors affiliated with at least two national institutions) and international collaboration (with authors from at least two countries participating). We created a graphic network to visually depict the collaboration between countries, with the thickness of the links representing the strength of collaboration (number of documents published together). Additionally, we determined the country of the first and the corresponding author of each of the documents, estimating the relative contribution of each country based on these indicators, which are frequently associated with a larger role or a higher degree of research participation and leadership.
Characteristics of main research clusters and leading researchers. We identified the most productive authors and research clusters in the field, which we represented graphically within a co-authorship network. A research cluster can be defined as a group of investigators with a high density of interconnections, affording certain homogeneity and distinguishing them from other clusters.
To determine the most productive authors' influence in the network, we calculated their betweenness, a centrality measure widely used in Social Network Analysis to measure the degree to which a node (in this case, an author) enables connections between other nodes. We used Pajek program to calculate this indicator.
Topics addressed. To identify the specific topics addressed by researchers, we measured the frequency of MeSH used and analysed their content. We also created a co-occurrence network to identify firstly the groupings and interrelationships of the descriptors, and secondly the connections between the descriptors and the previously identified research clusters. We excluded the most generic terms (Chagas cardiomyopahy, Trypanosoma cruzi, Chagas disease, humans and animals) from the analysis to focus on the more specific descriptors.
We created the collaboration networks at country and co-author level using the Pajek program, and the MeSH co-occurrence network using VOSviewer.
Citation analysis. In order to determine the research impact in the area, we have calculated the total number of citations, the mean citations per document and the H index of the most productive institutions and author clusters identified.

Document types and evolution of number of documents published
We retrieved 1932 papers from the MEDLINE database for the whole study period. The most common document type was 'journal article', accounting for 79.6% of the total (n = 1538), followed by 'review' (n = 217, 11.2%). Other prominent document types that we identified were 'letter' (n = 127, 6.6%) and 'editorial' (n = 43, 2.2%). We also identified document types by clinical interest. In that sense, only 2.5% of the published documents were clinical trials, and 9.7% were case reports; other document types appear only occasionally (Table 1).  impact factor, subject category and ranking in the 2015 Journal Citation Reports (JCR) as well as their country and language of publication. Two of these journals were not included in the JCR because they had no impact factor. The most common subject categories among the core journals (>14 papers) were "Cardiac & Cardiovascular systems" (n = 7), "Tropical Medicine" (n = 7), "Parasitology" (n = 4), and "Medicine, general & internal" (n = 3). Nine journals were published in the USA, nine in Latin America (six in Brazil plus one each in Argentina, Mexico, and Chile) and six in Europe (one each in France, Germany, Ireland, the Netherlands, Spain, and the UK).

Collaboration
Of the 1070 documents found in SCI-Expanded, 736 (68.8%) involved domestic collaborations between different departments or institutions in a single country. International collaborations produced 278 (26%) of the documents. Table 5 shows the evolution of domestic and international collaboration from 1980 to 2016. Domestic collaborations were almost three times as prevalent as international collaborations, though both types increased progressively over the study period.
Most countries in Latin America and the United States show a high degree of domestic collaboration (54%-66% of the documents in which they participated), with the exception of Bolivia (26.3%) and Mexico (44.4%). More significant are the differences observed in relation  (Table 6).
Brazil is the country that stands out the most in terms of leading collaborative papers, as authors of this country occupy the position of first author in 59.8% of the collaborative publications in which Brazilian authors participated, and of corresponding author in 77.8%. These values are markedly higher than those seen in the United States (44% as first authors and 58.9% as corresponding authors). Germany and Spain also hold positions of leadership on these indicators, while Bolivia is at the bottom of the ranking on both ( Table 6).

Authorship and research clusters
The 1755 papers (articles and reviews) in MEDLINE were produced by 4686 authors responsible for 9858 signatures. The average number of authors per paper over the whole study period was 5.6. This number was less than 2.  Table 7 shows the ranking of the 30 most productive authors (! 24 papers) and the authors with the highest values of betweenness centrality in the co-authorship network. The most productive author was C. Mady (n = 65), followed by M. O. C. Rocha (n = 58), and F. Pileggi (n = 57). The most influential cutpoints in the networks were J. A. Marin-Neto, followed by A. C. C. Carvalho and A. L. P. Ribeiro. Thirteen of the 30 authors appear in both lists (Fig 3).

Classification of the type of research study and analysis of MeSH
The manual revision of the studies we retrieved showed that 1225 documents (69.8%) mentioned Chagas cardiomyopathy in the title or abstract, indicating that this was the central focus of the publications. There were also 336 documents (19.1%) that referred to Chagas disease and 176 (10%) to Trypanosoma cruzi in the title or abstract. While these studies focused on those subject areas, the abstracts also mentioned terms like cardiomyopathy, myositis, heart, or another word that justified their inclusion under the MeSH descriptor of Chagas cardiomyopathy. Finally, 18 documents (1%) include a generic description of cardiovascular disease in Latin America.  Table 8 lists the 100 most frequently used MeSH in documents on Chagas cardiomyopathy. The term "humans" (n = 1412) appeared twice more often than "animals" (n = 722). We assigned the "humans" descriptor to 58.9% (n = 1033) of the documents and the "animals" descriptor to 19.4% (n = 341), while we used both descriptors for 21.7% (n = 381). Of the documents reporting studies on humans, 63.9% (n = 660) were clinical studies that addressed aspects related to diagnosis or therapeutic strategies; 16.1% (n = 166) were epidemiological studies, and 20% (n = 207) reported basic research on immunological, biochemical or molecular aspects of the disease. The documents classified under the joint "human-animal" descriptor were also most commonly rooted in basic research approaches (43.3%, n = 165), while the rest were on animal models (17.6%, n = 67), clinical research studies (24.9%, n = 95) or epidemiological studies (14.2% n = 54). Overall, 43% of the documents were clinical (n = 755), 12.5% (n = 220) epidemiological, and 44.4% either animal (n = 408) or basic (n = 372) studies. The most commonly used MeSH related to human research were "Chronic Disease" (n = 420), "Electrocardiography" (n = 341), "Heart Failure" (n = 168), "Arrhythmias, cardiac" (n = 115), "Echocardiography" (n = 114), and other descriptors related to patient follow-up and prognosis assessment. Some descriptors, such as "Myocarditis" (n = 136) and "Parasitemia" (n = 69) were more common in animal-based research.
Brazil (n = 121) is the main geographic MeSH term assigned to the documents, followed at considerable distance by the United States (n = 40), Mexico (n = 30), Chile (n = 23), Argentina (n = 19) and Bolivia (n = 16).  each of the four research clusters. Cluster I has a strong clinical focus, whereas cluster II is more closely associated with research on animal models. Cluster III is the most heterogeneous of the four, with ties to both approaches. For its part, cluster IV is associated with the study of genetic aspects of the disease.
When we analysed the documents by country of publication, we found some striking differences in research foci (Table 4). Human-based research is most prominent in Brazil, with 55.3% of records containing the MeSH "humans", versus 24.3% for "animals". This trend was even more pronounced in Bolivia (73.7% of records containing "humans"), Chile (67.9%), Colombia (68.2%) and Venezuela (57.1%). In the USA, although human-based research predominated, a considerable proportion of records were related to animals (39.3% vs 33.8%, respectively). We found a similar trend in France (31.8% vs 29.5%) and Mexico (29.6% vs 18.5%). The distribution of the documents by study type confirms the predominance of the clinical approach in countries like Bolivia (57.9% of the documents) or Chile (39.3%, compared to 21.4% of documents reporting basic research). It also reflects the hegemony of the basic approach in the United States (61.1%), European countries like France (77.3%) and Spain (57.9%), and in Argentina (62.4%) and Mexico (48.1%).

Research impact
The most productive institutions generally present a higher degree of citation, with the Universidade de São Paulo the main reference center with regard to all citation indicators. The Fundação Oswaldo Cruz (FIOCRUZ) and the Universidade Federal de Minas Gerais also stand out in terms of the absolute number of citations received and the H index. Other prominent institutions include the Universidade Federal do Triângulo Mineiro, the Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET) and the University of Texas Medical Branch, which all present a high average number of citations per document despite showing a more moderate absolute production (Table 3).
With regard to the clusters of authors and their citations, cluster II shows the highest absolute degree of citation (n = 5336) and the highest H index (n = 37), with an average number of citations per document of 26.8 (SD ±34). Cluster IV, despite its lower degree of citation (n = 3167) and H index (n = 28), the average number of citations per document is greater (41.7 ±90.4). Documents published by authors of cluster I yielded a total of 3932 citations (mean 33.6 SD ±43.8; H index = 35), while publications by authors from cluster III generated 2937 (mean 23.3 SD ±25.8; H index = 32). All of the clusters present mean degrees of citation above the average observed for the whole set of documents analyzed (23 SD ±36.3), with cluster III showing the lowest levels of citation among the research groups, despite its relevance as a bridge between the other clusters. This relatively moderate impact may be a reflection of the Research activity and collaboration patterns in Chagas cardiomyopathy

Scientific activity and international collaboration
Scientific production on Chagas cardiomyopathy has grown considerably since the turn of the 21st century, probably reflecting the increased incidence of Chagas disease in non-endemic areas like the USA and Europe, and the possibility of providing new knowledge of the disease, with the introduction of treatment projects and new diagnostic tests. Another relevant factor is the increase in domestic and international collaboration, which stimulates further research [15,[22][23][24]. Our findings revealed a high degree of domestic collaboration, especially in recent years, in response to the multidisciplinary nature of this clinical entity, which is relevant to tropical medicine, parasitology, cardiology, pathology, biochemistry and immunology, infectious diseases, physiology, microbiology and public health [25][26].
We found that most research on Chagas myocardiopathy was concentrated in journals related to cardiovascular systems, tropical medicine and parasitology, which is logical because  the disease affects the heart, is caused by a parasite, and is most prevalent in tropical and subtropical regions. Brazil and other South American countries, together with the USA, produced the most papers on Chagas cardiomyopathy, for two main reasons: firstly, Chagas disease mainly affects Latin American populations; and secondly, the USA, a leader in the research of this subject, has established many collaborative ties with these countries, especially with Brazil [17][18]. Among the published documents, 9.7% were case studies and only 2.5% clinical trials, probably because measuring the efficacy of the few existing therapeutic tools is highly problematic [25,27]. Thus, pharmacological treatments and antiarrhythmic therapies for Chagas myocardiopathy are often based on results for other etiologies. Specific clinical trials are crucial for helping health professionals to better understand and manage the complication [1]. Our findings showed Brazil at the center of the collaboration network, with strong links to other countries in Central and South America (particularly the bordering Argentina, Venezuela and Colombia) and also to non-endemic countries-primarily the USA, but also some countries in Europe. The fact that Brazil and the United States are the two primary geographic MeSH terms assigned to the documents confirms the research leadership exercised by these countries, but it also reflects the high prevalence of the disease in Brazil and the fact that the USA also has vectors and documented transmission [28].
Brazil plays a central role in research on Chagas myocardiopathy because it has the largest population of all Latin American countries and also a strong research sponsorship scheme that has driven international collaboration [29]. Bolivia, on the other hand, has the highest prevalence of Chagas disease and therefore of Chagas myocardiopathy, but it produced only 1.8% of the records in our study, a reflection of its low overall health system expenditure. Strengthening cooperative networks could help to address this deficit. Accordingly, in recent years Bolivia has increased collaborative research with other countries in Latin America, North America, Europe, and even Japan [23,30].

Authorship and research clusters
We found that the average number of authors included in the published documents increased over the study period, consistent with all health sciences disciplines. This increase in scientific cooperation will doubtless have several positive consequences, and may help to integrate less developed countries into research activities, but the trend may also be associated with negative aspects that should be avoided wherever possible: processes of neocolonialism or scientific dependence, whereby the more developed countries decide on the lines and topics of research without taking into account the expectations and interests of the countries where the disease has the largest impact [31][32]. The medium-term goal of establishing collaborative links between countries with more and less economic and scientific development should be to empower these latter countries with mutually beneficial and balanced partnerships [33][34].
Among the authors who have established strong collaborative links, we found a few key players who occupy a prominent position, collaborating with several other investigators and acting as intermediaries between different research clusters. These investigators play a crucial role in driving research in the field, since they help to integrate new researchers and generate resources. Moreover, they produce a high volume of work, disseminate information and new ideas, and foster the application of research methodologies and exchange of resources [35][36].

Research topics
The fact that MEDLINE indexers use the most specific term available in the MeSH branched hierarchy [19] explains the high percentage (70%) of documents tagged as focusing on Chagas cardiomyopathy. This is a specific descriptor for Chagas disease that is only assigned to documents when the title, abstract or key words make explicit reference to it, or when the thematic focus can be deduced by statements linking Chagas or Trypanosoma cruzi with a cardiovascular concept.
The proportion of human-based research varied from 55% in Brazil to 74% in Bolivia. The predominance of human-based research in Latin American countries like Bolivia, Brazil, Chile, Colombia and Venezuela reflects the importance of applied clinical practice in Chagas disease in these settings. In contrast, higher income countries without vector-borne transmission of Trypanosoma cruzi conduct a higher proportion of animal research. The USA and France are the leaders in this respect, with 34% and 29.5% of papers in these countries containing the MeSH term "animals", which is roughly balanced with the 39% and 32% of reports related to "humans". Among the countries in Latin America with the highest scientific production, Brazil (24.3%), Argentina (23.1%) and Mexico (18.5%) conducted the highest proportion of research on animal models. Since the physiopathology of myocardial damage is poorly understood in Chagas disease, animal models are crucial for improving scientific knowledge, and high-income countries are largely the ones generating this research. Thus, their collaboration with countries where the disease is endemic in humans is crucial. Initiatives like the Special Programme for Research and Training in Tropical Diseases (TDR), an independent global program of scientific collaboration cosponsored by the United Nations Children's Fund, the United Nations Development Program, the World Bank, and the World Health Organization, could serve as a model in this respect, as it enables researchers from different countries and institutions to work together on projects related to tropical diseases [25,37]. The manual revision also showed the relevance of animal research and experimentation in the field of Chagas cardiomyopathy due to the gaps in knowledge around the etiopathogenesis of the Trypanosoma cruzi infection and the resulting myocardial damage. Indeed, analysis of the kind of animal experimentation described in the documents that also report basic research in humans shows that nearly half of the research papers on Chagas cardiomyopathy are focused on clarifying the underlying mechanisms of myocardial damage and the factors involved in this pathology.
The most commonly used MeSH related to research in humans described diagnostic imaging tests. This reflects the fact that Chagas cardiomyopathy is a chronic disease that can present as arrhythmia and even heart failure, two complications where traditional and simple diagnostic techniques such as electrocardiography are being displaced by other methods like echocardiography [38][39]. We also found MeSH terms detailing the follow-up of patients with cardiomyopathy, and the evolution of Chagas disease towards cardiomyopathy, which shows that these topics are still poorly understood [25,[40][41][42]. With regard to animal models, the related MeSH terms reflect an interest in understanding the immunoallergic and microbiological implications of Chagas disease and Chagas myocardiopathy [26,43].

Limitations
The main limitation of our study is that the databases used (MEDLINE and SCI-Expanded) do not index some of the journals published in Central and South America, particularly in some countries where the disease is endemic, such as Bolivia, Paraguay and El Salvador. This may have limited the visibility of the papers produced in these countries. We performed the analysis based on searches of MEDLINE and WoS because these are the databases of reference at a bibliometric level and with regard to the most widely disseminated literature at an international level. They also enable far more exact analyses because bibliographic data are standardized, citations of publications are recorded, and search terminology is controlled very precisely through the MeSH. Furthermore, it is possible that our search strategy, which was based on a clinical manifestation of the disease, could have resulted in a relatively modest presence of basic studies, which may be indexed under "Chagas disease" rather than "Chagas cardiomyopathy".
Finally, it is worth noting that the use of addresses may not reflect the actual nationality of authors and the geographic setting where the studies took place, due to researcher mobility, the existence of research stays in other countries, and the development of research projects in endemic, low-and middle-income countries that are nevertheless funded and directed by institutions and researchers from high-income countries. Thus, the indicators obtained from authors' institutional affiliations should be interpreted only as an approximation to the institutions that are responsible for driving the performance of the studies and publishing their results.

Conclusions
In our study we identified the authors who comprise the main international research foci on Chagas cardiomyopathy, measured the extent of collaboration between authors, and identified the topics covered. It is noteworthy that despite the important degree of research development described, less than 1% of patients are able to access treatment. This wide gap points to the need to translate research results into comprehensive public health policy that extends access to clinical services for the disease. Related to this point, the low number of studies with Bolivian authors and researchers from other countries with a high prevalence of the disease, such as Mexico or Colombia, reflects the concentration of resources and higher education institutions in larger, wealthier countries. There is thus an acute need for capacity building in research infrastructure in endemic countries, especially the development of clinical trials with funding from states, industry and civil society, among other initiatives [44].
Domestic and international collaboration plays an important role in the research on this area. Further international cooperation could help to reduce the concentration of research activities in countries with more developed scientific systems such as Brazil or the USA, and reduce the polarity observed between endemic and non-endemic countries, where clinical research and basic research predominate, respectively. It is crucial to strengthen the link between basic research, which is focused on understanding the physiology of the disease, and clinical research, which concerns diagnosis and treatment.