Evolution of coauthorship networks : worldwide scientifi c production on leishmaniasis

Introduction: Collaboration is one of the defi ning features of contemporary scientifi c research, and it is particularly important with regard to neglected diseases that primarily affect developing countries. Methods: The present study has identifi ed publications on leishmaniasis in the Medline database from 1945 to 2010, analyzing them according to bibliometric indicators and statistics from social network analysis. Examining aspects such as scientifi c production, diachronic evolution, and collaboration and confi guration of the research groups in the fi eld, we have considered the different types of Leishmania studied and the institutional affi liation and nationality of the authors. Results: Seven-hundred and thirty-fi ve authors participate in 154 prominent research clusters or groups. Although the most predominant and consolidated collaborations are characterized by members from the same country studying the same type of Leishmania, there are also notable links between authors from different countries or who study different clinical strains of the disease. Brazil took the lead in this research, with numerous Brazilian researchers heading different clusters in the center of the collaboration network. Investigators from the USA, India, and European countries, such as France, Spain, the United Kingdom, and Italy, also stand out within the network. Conclusions: Research should be fostered in countries such as Bangladesh, Nepal, Sudan, and Ethiopia, where there is a high prevalence of different forms of the disease but limited research development with reference authors integrated into the collaboration networks.

Leishmaniasis is a parasitic zoonosis estimated to affect 15 million people worldwide, with nearly two million new cases reported every year.Despite these fi gures, leishmaniasis is also considered a neglected disease 1 .The World Health Organization (WHO) has promoted different initiatives to foster the development of new methods of prevention, diagnosis, and treatment 2 ; scientifi c collaboration is an essential part of that process, particularly in developing countries in which there are few resources dedicated to research 3 .
Since the early 20 th century, bibliometrics has been shaped as a discipline that uses the quantitative analysis of scientifi c publications to evaluate the size, growth, and distribution of scientifi c bibliography.Similarly, the fi eld examines scientifi c agents that produce, transmit, and use the body of research, analyzing their visibility and impact in subsequent literature 4 .Smith 5 reviewed the main bibliometric studies that have analyzed the area of tropical medicine throughout the past several decades; these studies are generally focused on the study of scientifi c production, citation, or the impact factor of the journals in the fi eld.Recently, different papers have examined the scientifi c production specifi c to leishmaniasis [6][7][8] .All of these studies report an increase in both scientifi c production on leishmaniasis and collaboration among institutions and countries.However, the collaboration among authors and the formation of research groups in the fi eld have not been evaluated, and no studies have analyzed whether research collaboration is primarily based on the study of the same type or of different clinical strains of leishmaniasis (where different interests would exist to obtain results for the control and management of the disease).In this sense, the study of scientifi c coauthorship through social network analysis allows for a more precise analysis of the collaboration structures within a discipline or area of knowledge than studies based on bibliometric indicators alone.
The aim of the present study is to analyze collaboration among researchers with regard to the global scientifi c production on leishmaniasis considering the different forms of the disease and studying the diachronic evolution of the formation of research groups in the fi eld.

METHODS
The study was carried out via a bibliographic search on the PubMed platform, which was used to identify indexed documents with the keywords or medical subject headings (MeSH) González-Alcaide G et al -Worldwide scientifi c production on leishmaniasis

RESULTS
Leishmania and leishmaniasis published between 1945 and 2010.The bibliographic information from the collected registries was introduced into a locally stored, relational database to carry out a bibliographic normalization or homogenization of data and the quantitative analysis.The normalization process was conducted to consolidate variations of author names.Such consolidation is often necessary because an author may be identifi ed by one fi rst name or more than one, initials, or a full name or due to the use of hyphens to separate names.The criterion followed in this homogenization process was the occurrence of the institutional signature associated with the variations of names and surnames.
We identifi ed the primary form of Leishmania studied by researchers by quantifying the MeSH descriptors assigned to the documents as well as their country.The authors were assigned to the form of Leishmania cited most frequently as a MeSH term in the documents published (cutaneous leishmaniasis, diffuse cutaneous leishmaniasis, and visceral leishmaniasis).
With the information obtained, we carried out a social network analysis, applying it to the coauthorships of the scientifi c publications and using the documents published on the Medline database as the unit of analysis.Bibexcel software (developed by Olle Persson, Inforsk, Umeã University, Sweden) was used to process the bibliographic information 9 , whereas the statistical calculations and visual representation of the networks were performed with the social network analysis program Pajek 10 .
Below, we delineate the indicators and concepts used in the present study.

Collaboration index
A widely used bibliometric indicator that establishes the average number of authors who have participated in the group of analyzed documents.

Network
A graphic representation (graph) comprised of a group of nodes or vertices (authors of scientifi c publications) and links between the nodes, which represent coauthorships or joint signatures from one or more scientifi c publications.

Network average degree
The average number of collaborators per author.

Collaboration threshold
The degree of intensity of the links established between the nodes.The threshold or collaboration intensity is higher for a greater number of papers signed by two given authors.Graphically, the collaboration threshold has been expressed in the networks using a thicker line representing the links.

Component
A group of nodes that are interconnected directly or through intermediaries.

Size of the largest component or giant component
Number of interconnected authors that comprise the largest component, considering all links and without applying any collaboration threshold.

Isolated researchers
Researchers who are not connected with any other researcher.

Network density
The proportion between the number of links established in the network and the maximum number of links that are theoretically possible.

Average distance
The average number of intermediaries between linked nodes.

Research clusters or research groups
Scientists working together in partnership to realize shared goals.In the present study, these clusters were identifi ed by applying a high collaboration threshold to identify the stable and consolidated collaboration relationships, that is, the groups of authors that maintain a higher degree of interconnectedness and therefore cohesion and differentiation compared to other groups.

Cutpoint or articulation point
Nodes through which all paths or geodesics pass, allowing linkage between other nodes; if these points were eliminated, parts of the network would become disconnected, increasing the number of components and reducing their size.Thus, these authors play an essential role (key players) as intermediaries in ensuring connectivity and communication between different parts of the network, preventing them from becoming isolated.
We identifi ed 20,780 documents on leishmaniasis published between 1945 and 2010, signed by 35,588 authors.The diachronic evolution displays constant growth in the number of documents published and researchers who have participated in studies on leishmaniasis (Table 1).
With regard to the collaboration indicators, we observed sustained increases in both the mean number of signatures on the papers and the average number of author collaborators (Figure 1).
The social network analysis statistics from the coauthorship network of scientifi c publications on leishmaniasis indicate that the size of the largest component has gradually increased according to the increase in the number of researchers that comprise the network.The relationships maintained between authors have also multiplied, with 80.9% of all authors integrated in the largest component in the 2001-2010 period.Inversely, the number of isolates has been progressively decreasing, falling to less than 1% in the last decade studied.The network density has also declined as its size has increased.Finally, the average distance between the researchers making up the network grew progressively, reaching 6.01 in 1981-1990, before decreasing slightly in the last two decades (Table 1).
The scientifi c production and number of collaborators of the 25 most productive authors (>99 papers) are also shown (Table 2).Most of the group (n=17) initiated their scientifi c TABLE 1 -Indicators of production and collaboration in scientifi c publications on leishmaniasis (1945-2010).

Indicator
Before 1951 1951-1960 1961-1970 1971-1980  1981-1990 1991 activity prior to the 1981-1990 period, whereas the remainder (n=8) did so during that period, and all have remained active until the present.In contrast, some researchers have shown a decline in scientifi c production in the most recent period.All researchers have had sporadic collaborations with a high number of authors, but they only have a small group of regular collaborators with whom they publish a signifi cant proportion of their scientifi c production.In that sense, the 25 most productive authors have published 10 or more papers with only 0.7%-8.97% of their total collaborators.
We have identifi ed 1,567 components comprised of two or more authors who are connected by at least one coauthorship.Seven-hundred and thirty-fi ve authors comprise 154 research clusters, which constitute important research foci whose authors maintain close relationships with each other; indeed, the links between the members are derived from the fact that they have González-Alcaide G et al -Worldwide scientifi c production on leishmaniasis coauthored at least 10 papers.The visual representation of these clusters (Figure 2) illustrates the rank that the 101 most productive authors (>49 publications) hold as well as the type of leishmaniasis that they normally study and their country of origin.The largest cluster has 85 authors.The fi ve clusters that follow in size bring together 36, 26, 20, 19, and 18 authors.Eight others comprise 10-14 authors, whereas the remaining 140 have two to nine authors.Most of the 101 most productive authors occupy a prominent position in the clusters in which they participate, drawing together other less productive authors or serving as important cutpoints between different groups of authors.
The 735 authors identifi ed belong to 49 countries; Brazilians predominate (n=140), followed by American (n=105), French (n=72), Indian (n=55), Spanish (n=44), British n=38), Italian (n=25), Swiss (n=24), and German (n=20) authors.In general, the predominant links in the clusters are between authors who study the same type of leishmaniasis and are from the same country.However, some inter-Leishmania connections stand out.In terms of international collaborations, the links that some American researchers have with Brazilian and Indian colleagues are also worth noting.
At an institutional level, three Brazilian centers stand out as having the greatest number of affi liated authors on the subject of leishmaniasis: the Oswaldo Cruz Foundation (n=45), the Federal University of Bahia (n=15), and the University of São Paulo (n=12).
The next positions are held by the Rajendra Memorial Research Institute of Medical Sciences of India (n=11) and the Tehran University of Medical Sciences of Iran (n=10).In the West, University of Montpellier of France (n=8) and Ohio State University of the USA (n=7) lead the ranking.The latter of these countries is characterized by highly dispersed scientifi c production among its universities, with 105 authors who are distributed among 67 different institutions, as opposed to the 36, 33, and 23 centers identifi ed in France, Brazil, and India, respectively.
In the same way that research groups tend to be clustered in a single country, many authors of the same cluster belong to a single institution.However, there are outstanding collaboration links between the main institutions involved in tropical diseases research and training (e.g., in the largest cluster between the Federal University of Bahia and the University of Brasilia in Brazil; the CSIR-Central Drug Research Institute and the Balaji Utthan Sansthan in India; and the Institute of Tropical Medicine and the University of Antwerp in Belgium).
With regard to the formation of the research clusters, we devised a graphic representation (Figure 3

DISCUSSION
Although all areas of knowledge have experienced noticeable growth in recent years, diverse studies have drawn attention to the even higher rates reported in scientifi c publications on tropical diseases 11 , particularly leishmaniasis [6][7][8] .Some reasons may include the impulse shown by WHO and the new interest in neglected diseases through initiatives such as the TDR Programme, the Special Programme for Research and Training in Tropical Diseases started in the mid-1970s, the changes in global knowledge, and some outstanding developments and papers that have raised interest in research on the topic 12 .
The progressive growth in collaboration over the last few decades is one of the defi ning features of the evolution of science, which helps to explain the increasingly specialized nature of knowledge and the need for an interdisciplinary or multidisciplinary approach to complex problems.This strategy also allows for easier access to technologies and equipment, better management of processes, and a more positive impact with regard to the quality of the results obtained.Keiser and Utzinger 11 identifi ed a signifi cant increase in the average number of authors per paper in publications on tropical medicine between 1952 and 2002, a rise which has continued to the present day.This general trend also applies to specifi c tropical diseases, such as Chagas disease 13 .The present study, in addition to confi rming this same trend in relation to leishmaniasis, has allowed us to describe other aspects of the scientifi c collaboration in the fi eld, such as the fact that in last several decades, the number of different people with whom researchers collaborate has signifi cantly increased.Indeed, the average degree tripled from 1971-1980 (3.4) to 2001-2010 (10.6).This trend, together with the fact that the percentage of authors integrated in the largest component has been gradually increasing over time, even as the number of isolates that have not established collaborative links has been decreasing, is a refl ection of the higher degree of interconnection observed among scientists and the importance that cooperative research has been acquired over the last few decades; this method is now the principal and defi ning feature that characterizes the social organization of researchers in their quest to generate new knowledge 14 .The percentage of authors with scientifi c publications on leishmaniasis and integrated in the largest component throughout the period (79.7%) is similar to that observed in the coauthorship network observed in the work performed on Chagas disease, where this value was 84.1% 13 .Even higher fi gures have been observed in some fi elds, where up to 95% of authors are included 15 .
The increase in scientifi c collaboration has been favored by other factors apart from those already mentioned; the process of globalization has been an important driver of this tendency, facilitating researcher mobility, access to information sources, communication, and the development of Information Technology (IT) tools for cooperative work.These conditions have all favored the creation of collaborative links between researchers that are expressed through coauthorships in scientifi c publications 16 .Inversely, the integration of researchers in the largest component has also resulted in a gradual reduction in the network density, a refl ection of the multiplication of research hubs at a global level and the formation of small-world networks, to which the coauthorship networks adapt.
Another of the defi ning features of the scientifi c collaboration networks is the fact that stable collaborative relationships can only be maintained with a small number of researchers, whereas most collaborations occur only once (this is the case for 83.1% of the coauthorships identifi ed in the present study).Unlike the number of total collaborators, this feature is constant throughout the entire period analyzed, with values ranging from 81% to 92%.
The fact that 47% of the clusters or groups identified comprise only two authors is concurrent with the value observed by Yu et al. 15 in the coauthorship networks considering coronary heart disease.Although some clusters are more numerous, many of the authors comprising these clusters are rather disconnected from each other, separated by a great number of intermediaries.It is possible to distinguish different subgroups of authors who present a higher degree of cohesion and density in their relationships, suggesting that these nuclei of stable collaborators constitute the customary structures that articulate the scientifi c research, independently of the sporadic collaborations that exist with other authors.
The diachronic evolution of the coauthorship networks on leishmaniasis coincides with that observed by Lee et al. 17 in physics; these authors identifi ed three phases: the emergence of small components (nuclearization); their gradual integration into a giant component; and the multiplication of the links in that giant component, generating a dense and complex web of interrelationships.In line with what we have observed in the present study, this process is largely facilitated by the establishment of collaborative links between the most productive authors or between those authors who occupy a role of leadership within their respective components.These researchers bring together a large number of authors, so when they collaborate with each other, they increase the size of the components that they lead.This phenomenon is in line with the preferential attachment model of growth proposed by Barabasi and Albert 18 , in which the nodes with a higher number of relationships have a higher probability of connecting to new nodes.
In addition to graphic representations that facilitate the visualization and analysis of collaboration structures in an area of knowledge, social network analysis can be used to identify key agents in a given fi eld of study, enabling the cohesion or facilitating the connectivity and the information fl ows in the network.Borgatti 19 identifi ed two types of key players: those who stand out for their position as agents that bring together a large group of authors, maintaining the cohesion of the network and enabling the interconnection among its members by means of direct links or shortest pathways; and authors who act as a bridge, enabling the interconnection of different parts of the network and impeding its disintegration into separate and smaller components.In the main component of the leishmaniasis network, Edgar M. Carvalho, Aldina María Prado Barral, Philip D. Marsden, Steven G. Reed, Jean Claude Dujardin, and Shyam Sundar Sharma would fi t into the fi rst group, whereas E. Alejandro Llanos Cuentas would be the paradigmatic example of the second group.
The clusters or research groups identifi ed have a marked national character, and they generally investigate a single clinical strain of leishmaniasis, although occasionally some authors or research groups in the same country study different clinical forms of the disease.The collaboration among authors who investigate different clinical forms of leishmaniasis is limited, presumably because the pathologies require different therapeutic responses.However, some collaborative relationships between authors with different leishmaniasis specialties do exist, possibly due to migration of the vector or the presence of animal reservoirs, which has led to an extension of studies on this disease to places where few or no cases have been reported historically 20 .These collaborations may also focus on aspects such as epidemiology, the vector or the reservoir, giving rise to collaborations between entomologists, biologists, medical doctors, epidemiologists, and specialists in public health, who may specialize in one or various forms of leishmaniasis.
The observation by Al-Mutawaket et al. 6 , who observed an absolute dominion of the USA in terms of international collaboration, can be qualifi ed based on the results obtained in our analysis of the coauthorship networks.Although we confi rm that the USA boasts more international ties, Brazil is shown to play a leadership role in the research on the topic, with three outstanding hubs in the center of the network, dominated by Brazilian authors, as well as a notable presence in the largest component.The preeminence of these groups and the presence of numerous Brazilian investigators in the identifi ed research groups stems from the fact that leishmaniasis is an important public health problem in Brazil, where incidence of the disease in all of its forms is higher than in the remainder of the continent.Thus, Brazil is developing its research potential in the fi eld, making leishmaniasis-as well as other neglected diseases -a priority on the national health research agenda [21][22][23] .
With regard to the collaboration between researchers from different countries, several factors come into play in the pronounced degree of collaboration between the USA and Brazil.In addition to the development of the scientifi c systems in both countries, there is the geographic proximity between them (relative to other countries, such as India) and the institutional links developed through academic exchange programs or grants, as researchers who return from the USA or Europe generally maintain their collaborative ties or initiate other common projects with their colleagues from abroad 24 .To understand the collaborations among researchers, some authors have drawn attention to the importance of homophily, that is, sharing certain attributes, such as age, sex, academic or professional rank, or amount of similar experience 25 .Previous experiences also shape relationships; decisive factors that contribute to stable collaborative links include compatibility of working styles, the creation of a trustful working climate, and the existence of common interests 16 .
The collaboration among European countries is favored by several circumstances, including the incidence of scientifi c policies that aim to foster research in all areas of knowledge, the traditional leadership of these countries -together with the USA -at a scientifi c level, and their interest in protecting European populations from tropical diseases whose emergence on the continent has been precipitated by migratory and touristic movements.Interest in research on leishmaniasis has also been attributed to the fact that it constitutes and excellent parasitic model 6 .
The fact that more than 90% of all cases of visceral leishmaniasis are concentrated in six countries (India, Bangladesh, Sudan, South Sudan, Brazil, and Ethiopia) 1 contrasts with the limited development of research on the topic in these countries, with the exceptions of India and Brazil, as shown by the results of our analysis of research groups.In addition to this scarce research, Keiser and Utzinger 11 have signaled the limited presence of authors from developing countries in positions of leadership and prestige in scientifi c publications in the fi eld of tropical medicine (fi rst and last signature on author lists), a situation which would have refl ected the end of scientifi c colonization, a situation in which researchers in wealthy countries and those on the vanguard of scientifi c development were the ones to set priorities and lines of research for developing countries, in favor of a situation of scientifi c subordination.Although collaboration with more developed countries can constitute an effective mechanism to stimulate and foster research in less developed nations (a fact which has given rise over the years to numerous initiatives to promote North-South collaborations) 3 , the ultimate aim to avoid this type of subordination should be to create stable research structures in developing countries and to develop South-South collaborations through the joint development of research initiatives, which are particularly important in the case of health problems that directly affect these countries 21 .Regional reference centers and existing collaboration networks have an important role to play in promoting research, establishing collaborative alliances between other emerging countries, and integrating researchers from these countries into international research networks 26 .
Attention has also been drawn to the importance of developing initiatives that attract private interests, particularly pharmaceutical companies, as the lack of commercial interest from this sector constitutes one of the main barriers to the development of drugs to treat neglected diseases 23,26 .In this respect, the formation of multidisciplinary research networks and the establishment of collaborative alliances between the public sector and industry should be fostered 26 .
Finally, other authors have also noted that to tackle neglected diseases, it is essential to consider the economic, political, social, cultural, and environmental context in which they appear, promoting research in social science disciplines, such as anthropology, health economics, medical sociology, social policy, and social epidemiology, among others.In disease control, contributions from these fi elds can be as important as biomedical research focused on the study of the infectious agent and the corresponding therapeutic response 27 .
Any study that aims to analyze collaboration through coauthorships in scientifi c publications must assume certain limitations given that the fact that two authors signing a paper does not necessarily mean that a real collaboration has been established between them or that all collaborations have been faithfully refl ected in the author list.Moreover, other phenomena or customs practiced at the time of determining the author list might distort these analyses, such as unjustifi ed hyperauthorships 28 .In addition, other phenomena, such as the high mobility of researchers, may affect the results presented and should be considered.However, this type of study can be of great use, offering an approach to better understand the markedly cooperative organizational and social context in which scientifi c knowledge is generated.

TABLE 2 -
Distribution by decade of the number of collaborators corresponding to the most productive researchers (>99 papers) on leishmaniasis(1945-2010).cols docs cols cols cols docs cols cols cols docs cols cols cols docs cols cols cols docs cols cols cols Author n of published documents; cols: number of different authors with whom they have co-authored the documents.
) that illustrates the evolution by decade (1981-2010) of the authors belonging to the largest component from the 2001-2010 period.Applying a collaboration threshold of four or more coauthorships, we also indicate the position occupied by the 25 most productive authors to facilitate the visualization of the networks and focus

FIGURE 2 -
FIGURE 2 -Main research hubs in scientifi c publications on leishmaniasis (>9 coauthorships).Shapes: ellipse (leishmaniasis visceral), box (leishmaniasis cutaneous), triangle (leishmaniasis mucocutaneous), and diamond (several forms).Colors: blue (Brazil), yellow (USA), green (France), red (India), dark blue (United Kingdom), pink (other European countries), white (Canada), orange (other Near East and Asian countries), brown (African countries), black (Australia), and grey (Peru).The size of the nodes is proportional to the number of papers published and the thickness of the links to the intensity of collaboration (number of coauthored papers).

FIGURE 3 -
FIGURE 3 -Evolution by decade of the components integrated in the 2001-2010 period in the giant component.