Bibliometric approach: increase in multi-authored scientific production.

Researchers tend to organize in collaborative patterns. Understanding the essential features of cooperative practices through bibliometric indicators that analyze the scientific production of a discipline or area of knowledge can lead to a better comprehension of the social environment in which research develops. Our study has confirmed the clear trend, also observed by previous research, of increasing scientific production on psoriasis, particularly after the year 2000. Pavlovsky et al. [10] speculate that this phenomenon could be rooted in the development of new biological treatments for psoriasis that do not displace traditional treatments, or perhaps in the existence of a high level of scientific evidence. This latter possibility is supported by Price’s theory on exponential growth in scientific literature; applying those principles to psoriasis research reveals that the field is in a mature stage and has contributed important advances in knowledge and treatment for the disease [21]. Indeed, one of the main paths for knowledge generation is through clinical trials, and this document type constituted no less than 10.6% of the analyzed documents published in 1994–2003. This figure stands in contrast to the 0.75% of clinical trial papers published on retinoblastoma [22] and the 2% published on leishmaniasis [23]. In the case of psoriasis, not only has the number of documents increased (at a rate of 97.3% between the decades 1994–2003 and 2004–2013), but also the number of authors (110.2% increase) and authorships (145%).

The co-author mean is an indicator of the general degree of collaboration in the area, and it has not stopped growing over the last several decades, a fact that can be explained by the benefits of collaboration, the mobility of researchers, and the spectacular development of communications systems [7,24]. On the other hand, the literature has also warned that author lists may be inflated due to the phenomenon of unjustified hyper-authorship, which stems in part from pressure to publish and which, according to some studies, now represents a significant proportion of authorships in the biomedical area [25,26]. With regard to the effects of collaboration at a co-author level, some references have associated the increase in co-authorships with a higher degree of acceptance among scientific journals as well as an increase in productivity or citation rates, especially when it comes to international collaborations [25, 27–31].

The degree of multi-authorship is a useful indicator in analyzing the evolution of the quantity of authors involved in the development of the studies and in the dissemination of the results, shedding light on the research effort mobilized for the performance of studies and the drafting of papers. In our study, the indicator reveals a noticeable increase in the number of papers signed by many different authors, another inherent feature of scientific collaboration. This fact responds to the rise in interdisciplinary research, which utilizes the knowledge of authors from a variety of different areas to explore complex problems from a multidisciplinary perspective [32]. In clinical research, multicenter studies and clinical trials may also contribute to the increase in multi-authorship; this type of research is increasingly common due to the level of evidence that it provides and the potential to translate the results to practice [33,34]. Other factors can also be mentioned, such as institutional policies that aim to foster collaboration as a way to raise the profile of their researchers. This practice responds to evidence showing that multi-authored manuscripts are more likely to be accepted in journals with high impact factors or in the core journals of the field, raising the potential for high citation rates [35]. The formation of multidisciplinary and multicenter groups—increasingly broad and at the same time specialized—is therefore one of the key elements that ensures the progress of knowledge and the translation of research results.

Our study suggests that there is great human potential in the field of psoriasis research, in the form of a stable community of researchers that lends continuity to the topic. First of all, there are an important number of big producers (authors with over 9 papers)—2.85% of the authors, which is well above the values seen in other disciplines that are comparable in terms of scientific production. For example, big producers only account for 1.45% of authors in the area of alcohol research [36]. A high number of continuants (authors who published before and after the years under study) was also observed. These aspects are essential in the establishment of collaborative ties and the formation of consolidated research clusters, both of which favor the efficient development of research and the integration of new members [37,38].

Social Network Analysis: an approach to structural and cooperative research relationships.

Social network analysis opens the door to new perspectives for the study and characterization of scientific collaboration, highlighting the network’s global features and identifying the skeleton of the network—the most active research components—by analyzing the interactions and the role played by the main authors within them. The identification of reference researchers in a field through network analysis can be useful in determining the most successful cooperative practices and in identifying potential collaborators. With regard to the average degree, this measure facilitates an analysis into the number of different co-workers of the researchers into the research community, determining if collaboration is more or less extensive, while the distribution of the links (considering different thresholds of co-authorship) provides insight into the extent to which authors have established stable links with other researchers or have participated in more sporadic collaborations. Both of these parameters mark the evolution of the breadth (number of collaborators) and the intensity of the collaborations (frequency of co-authorships with a single researcher).

With regard to the breadth of collaboration, the fact that the average number of co-authors has doubled in the past 20 years is striking, even if (as observed in previous studies), this collaboration is concentrated among sporadic partnerships. Indeed, stable and consolidated collaborators are still few in number, usually limited to just one co-author or a limited group of linked co-authors [39,40]. This suggests that the increase in sporadic collaborators may respond to factors such as a larger presence of occasional authors who contribute to the specialty without fully dedicating their work to it, or the participation of authors that provide concrete technical or instrumental support, for example, experts in specific methodologies, lab technicians or statisticians. The existence of a high number of transient authors is common in all scientific disciplines, although if this number rises too high, it could be an obstacle to the consolidation of the discipline or reflect the absence of a strong intellectual base of researchers dedicated to the field [41]. Gutiérrez-Vela et al. [42] interpreted this to be the case in the area of regenerative periodontal surgery, where 79.6% of all authors had only participated on one document. In telepathology, a recent and relatively unconsolidated field, 76.2% of authors are also transient [43]. In any case, both of these values are significantly higher than the 69.3% of transient authors observed in the present study.

The maximum frequency of co-authorships that we observed is far above those observed in other, less consolidated areas, such as health management [44], and approximate the values seen in the study of diseases such as leishmaniasis or in areas such as oncology [40,45]. The maintenance of strong ties (i.e., repeated co-authorships) has been associated with better scientific performance than a series of weak ties with many different researchers [46,47], which constitutes an additional facet from bibliometric analyses that attests to the importance of big producers for a scientific discipline.

The percentage of authors that belong to the largest component reflects the degree of integration of the researchers within the whole of the scientific community under study. In this sense, the value that we observed in the last decade of study (73%) is very similar to that calculated for the research network on HIV and HPV (70%) [48], although somewhat lower that observed in the networks focusing on leishmaniasis (79.7%), Chagas (84.1%), and coronary heart disease (95%) [39,40,49], which suggests the existence of a wide consensus with regard to the topics covered, but at the same time a few specific topic areas within the field of research that are covered by independent groups without ties to other researchers. Conversely, the percentage of isolate authors corresponds to the authors who do not work collaboratively; this value has gradually fallen to the marginal values seen today.

With regard to the evolution of the giant component and the modularity, two aspects are particularly significant: first, the considerable increase in the number of communities identified in the decades 1984–1993 and 2004–2013, which almost double those observed in the preceding periods. Second, the stability of the 1994–2003 period compared to the previous decade is noteworthy, as the number of communities identified barely increases, despite the considerable increase in authors making up the giant component; this period seems to be characterized by the reinforcement of existing communities rather than by the appearance of new communities. The communities identified can be interpreted as homogeneous research clusters that share a common interest in a given study topic or line of research. It is worth noting that in the study carried out and in relation to the communities identified, the evolution in their number does not evolve in a similar way to the increase in authors that make up the network or in the collaborative links that connect them, suggesting the importance that the prominent milestones and key discoveries have in driving the generation of new knowledge in a discipline, as well as the interest in modularity as a measure to characterize this aspect. The increase in the communities identified in the 1984–1993 decade coincides with the first milestone in the treatment of psoriasis, the advent of phototherapy (with or without psoralen) and immunomodulatory drugs such as cyclosporine or methotrexate, among others [50–52]. From 2000 on, it is worth noting the introduction of anti-tumor necrosis factor (anti-TNF) alpha therapies, which block the biologic activity of psoriasis. The first clinical trial was published in 2001, testing the efficacy of infliximab, and from the year 2003, a number of other published clinical trials focused on other anti-TNF agents, including etanercept or adalimumab. More recently, other biological drugs, such as ustekinumab, ixekizumab, tofacitinib, secukinumab and briakinumab, have targeted immunological mediators of the disease [53].

The average distance is a measure that can be associated with how fast information and innovations circulate in the network. The high values of this indicator observed in the decades 1974–1983 (8.3) and 1984–1993 (8.1) stand out, and can be interpreted as correlating to a nascent stage in the development of the network, with distant research clusters linked by few interconnections. Over the next two decades, the average distance decreases significantly, to values similar to those seen in other scientific co-authorship networks [54]. Thus, it appears that the psoriasis network has matured and multiplied its collaborative ties, which facilitates a faster and more effective dissemination of information, methods, and innovations within the field’s research community [55].

The clustering coefficient quantifies the degree of connection maintained between adjacent nodes for each node in the network, determining the extent to which they tend to bunch together to form cohesive groups, with a high density of ties among them. The high clustering coefficients observed, together with the average degree values observed, establish the psoriasis network as a small world network, a configuration that scientific co-authorship networks assume [16,46,56]. One of the most notable practical implications of this kind of network is the existence of a few key nodes, which make it possible to get to any other node in the network with a minimum number of intermediaries, despite the fact that most nodes are not directly connected. These key nodes are considered to be some of the most important scientific agents, favoring network cohesion and the advancement of knowledge. Another defining feature of small world networks is the likelihood that two nodes that are connected indirectly through an intermediary will eventually establish a direct tie. This aspect is predictive of the network’s evolution: an author A may collaborate separately with authors B and C, and it is probable that the latter two will end up establishing a collaborative link.

The authors represented in the co-authorship network (2004–2013) were identified by applying a high collaboration threshold to identify the stable and consolidated collaborative relationships. The resulting clusters can therefore be considered the skeleton that articulates all research in the area [49] and supports the rest of the network, which may be much larger, including transient authors, newcomers and researchers with less intense collaborative ties that do not meet the threshold imposed for the construction of the network. According to diverse studies, consolidated research groups perform better and achieve the highest degrees of citation, so it is important to establish the conditions that favor the creation of these groups and clusters [57].

Most of the sub-components of the largest component, along with many other research clusters identified, present centralized, star-shaped topologies, with a single investigator occupying a prominent position that brings together other authors. This figure suggests that the most common way that research groups are organized depends on the existence of a principal author, who acts as its leader. According to network theory research, these authors are characterized by their collaborative ties with a large number of researchers [58], their status as highly productive authors [59], and the lowest geodesic distances registered in the network, which confers on them a leadership role due to their capacity to interact more directly and rapidly with other investigators [55]. Thus, these are researchers of reference, who are responsible for articulating the development of research in the area and facilitating wider collaborative relationships, the cohesion of scientific community, and the integration of new authors in the research groups. This collective reflects the hierarchical structure of research development; indeed, one study on co-authorships in the area of chemical engineering found that many clusters represented the collaborating authors of a full professor, usually the department chairman [60].

In other cases, the research clusters appear as highly cohesive structures with a high degree of connectivity among the authors, none of which stands out from the rest. This is the case of the cluster at the top of Fig 3, which is among the clusters with the highest number of average co-authors per paper and which also stands out for the large number of papers signed by many collaborators. This probably reflects an entirely different model for organizing researchers in the production of knowledge, less stratified and more democratic and distributive. Alternatively, the pattern may be attributable to factors such as research specialization, with the participation of numerous investigators, or a markedly multidisciplinary approach, as noted by previous studies [61]. In any case, some papers have warned that structures that are excessively closed off or homogenous, with few external links to other clusters, could constitute an obstacle to the integration of new researchers and innovations, and hence to the advancement of knowledge [62,63].

It is also worth noting that with regard to the constructed networks, some research clusters have peripheral or lateral ties, which probably represent different stages of development within the group. Newer authors may be assuming leadership positions alongside more consolidated researchers, or new research nuclei could be emerging. Future lines of research should deepen the analysis of the topologies and structures in the co-authorship networks, through qualitative methodologies or network analyses based on block-modeling, in order to determine to what extent they respond to different cooperative patterns or behaviors [64–66] and which organizational forms are most effective in generating new knowledge [67].

With regard to the measures of centrality of the nodes, the highest ranked researchers in betweenness and closeness are those who occupy central positions in the sub-components they belong to, which confirms the interest in the combined use of both indicators to identify the most prominent authors, who lead their research cluster and play an important role in the network. Some authors also stand out for their role as a bridge, connecting different research clusters; these investigators are incredibly important to their field because they facilitate the dissemination of information and new ideas, the application of research methodologies and the exchange of resources [61,68].