Mapping of global scientific research in comorbidity and multimorbidity: A cross-sectional analysis

Background The management of comorbidity and multimorbidity poses major challenges to health services around the world. Analysis of scientific research in comorbidity and multimorbidity is limited in the biomedical literature. This study aimed to map global scientific research in comorbidity and multimorbidity to understand the maturity and growth of the area during the past decades. Methods and findings This was a cross-sectional analysis of the Web of Science. Searches were run from inception until November 8, 2016. We included research articles or reviews with no restrictions by language or publication date. Data abstraction was done by one researcher. A process of standardization was conducted by two researchers to unify different terms and grammatical variants and to remove typographical, transcription, and/or indexing errors. All potential discrepancies were resolved via discussion. Descriptive analyses were conducted (including the number of papers, citations, signatures, most prolific authors, countries, journals and keywords). Network analyses of collaborations between countries and co-words were presented. During the period 1970–2016, 85994 papers (64.0% in 2010–2016) were published in 3500 journals. There was wide diversity in the specialty of the journals, with psychiatry (16558 papers; 19.3%), surgery (9570 papers; 11.1%), clinical neurology (9275 papers; 10.8%), and general and internal medicine (7622 papers; 8.9%) the most common. PLOS One (1223 papers; 1.4%), the Journal of Affective Disorders (1154 papers; 1.3%), the Journal of Clinical Psychiatry (727 papers; 0.8%), the Journal of the American Geriatrics Society (634 papers; 0.7%) and Obesity Surgery (588 papers; 0.7%) published the largest number of papers. 168 countries were involved in the production of papers. The global productivity ranking was headed by the United States (37624 papers), followed by the United Kingdom (7355 papers), Germany (6899 papers) and Canada (5706 papers). Twenty authors who published 100 or more papers were identified; the most prolific authors were affiliated with Harvard Medical School, State University of New York Upstate Medical University, National Taiwan Normal University and China Medical University. The 50 most cited papers (“citation classics” with at least 1000 citations) were published in 20 journals, led by JAMA Psychiatry (11 papers) and JAMA (10 papers). The most cited papers provided contributions focusing on methodological aspects (e.g. Charlson Comorbidity Index, Elixhauser Comorbidity Index, APACHE prognostic system), but also important studies on chronic diseases (e.g. epidemiology of mental disorders and its correlates by the U.S. National Comorbidity Survey, Fried’s frailty phenotype or the management of obesity). Conclusions Ours is the first analysis of global scientific research in comorbidity and multimorbidity. Scientific production in the field is increasing worldwide with research leadership of Western countries, most notably, the United States.


Methods and findings
This was a cross-sectional analysis of the Web of Science. Searches were run from inception until November 8, 2016. We included research articles or reviews with no restrictions by language or publication date. Data abstraction was done by one researcher. A process of standardization was conducted by two researchers to unify different terms and grammatical variants and to remove typographical, transcription, and/or indexing errors. All potential discrepancies were resolved via discussion. Descriptive analyses were conducted (including the number of papers, citations, signatures, most prolific authors, countries, journals and keywords). Network analyses of collaborations between countries and co-words were presented. During the period 1970-2016, 85994 papers (64.0% in 2010-2016) were published in 3500 journals. There was wide diversity in the specialty of the journals, with psychiatry (16558 papers; 19.3%), surgery (9570 papers; 11.1%), clinical neurology (9275 papers; 10.8%), and general and internal medicine (7622 papers; 8.9%) the most common.

Introduction
Over the last three to four decades, substantial progress has been made toward reducing mortality and extending life expectancy worldwide [1,2]. Although health seems to have improved globally, more people than ever are spending more time with functional health loss and disability [3,4]. In many countries and regions, the management of multiple chronic diseases in a given patient at the same time (the so-called, "comorbidity" or "multimorbidity") poses major challenges to health services [5][6][7][8][9]. People with two or more chronic (physical or mental) diseases are more likely to have poor health outcomes, more complex clinical management and increased healthcare costs [8,9].
Analysis of scientific research in comorbidity and multimorbidity is limited in the biomedical literature [10][11][12][13][14][15][16]. For example, Fortin et al. [12] previously investigated the characteristics of the publications on multimorbidity (or comorbidity) and compared the number of publications on it with the number of publications on three common chronic conditions (asthma, hypertension, and diabetes). A restricted search of MEDLINE in 2002 identified 353 papers on multimorbidity and comorbidity for the period 1990-2002. The number and diversity of articles were both insufficient to provide relevant data to inform evidence-based care of people affected by multiple chronic conditions [12].
The scientific landscape has changed considerably in the subsequent years, including the launch of important initiatives for the clinical management of multiple chronic diseases [17][18][19][20], but also the proliferation of open-access journals to disseminate research findings [5,7,[21][22][23][24][25]. Considering research is needed to increase knowledge in a changing research area, this study aimed to map global scientific research in comorbidity and multimorbidity to understand maturity and growth during the past decades.

Search strategy
We conducted a cross-sectional analysis of the Web of Science, Science Citation Index-Expanded (SCI-E) database, from inception to November 8 th 2016. The Web of Science has been considered the world's leading taxonomic reference for citation analysis and prior to 2004, the only data source on citations available [26]. The search strategy for this study was designed by two senior health information specialists (AA-A, RA-B) and a clinical epidemiologist (FC-L), based on a previously published strategy [13]. The search strategy was constructed by using a combination of the following terms related to comorbidity and multimorbidity (see Box 1 for terminology): comorbidit Ã OR co-morbidit Ã OR multimorbidit Ã OR multi-morbidit Ã OR multidisease Ã OR multi-disease Ã OR multipatholog Ã OR multi-patholog Ã OR polimorbidit Ã OR poli-morbidit Ã OR polipatholog Ã OR poli-patholog Ã OR pluripatholog Ã OR pluri-patholog Ã (full strategy is available in S1 Table). We included two types of papers: research articles or reviews on comorbidity or multimorbidity of any type (physical or mental). Meeting abstracts, proceedings paper (journals, book-based), editorials, book chapters, corrections, retracted publications and other items (e.g. notes, news, etc. . .) were excluded. No restrictions in languages or publication date were applied to the database search.

Box 1. Terminology
The terms of "comorbidity" and "multimorbidity" are often used interchangeably. Many possible definitions and interpretations of the concepts of "comorbidity" and "multimorbidity" have been reported in the biomedical literature [9][10][11]. For example, Valderas and colleagues [9] reviewed the definitions of "comorbidity (and multimorbidity)" and their relationship to related constructs. A brief overview of common terms follows.

Comorbidity.
A widely accepted definition of "comorbidity" is the occurrence or the existence of any distinct additional medical condition to an index disease [31]. In general, the role of coexisting conditions is of less importance and one does not assume an interaction between the multiple conditions. The nature of the conditions that co-occur have variously included (physical or mental) diseases, disorders, conditions, illnesses, or health problems. Comorbidity was first included as a MeSH term in 1989 [9,10].
Multimorbidity. Most authors define "multimorbidity" as the co-occurrence of two or more medical conditions in an individual without any reference to an index disease [6,[9][10][11]. Therefore, in multimorbidity, no index disease is defined and all conditions (or "morbidities") are regarded of equal importance. Multimorbidity constitutes a more generic, patient-centered concept, whereas comorbidity is an index disease-based concept [11]. At present, no MeSH term exists for multimorbidity.

Data extraction
For each included paper, data on the year of publication, the journal title, subject category, keywords, and the authors' names, institutional affiliation(s), and country was downloaded online through the SCI-E from the Web of Science by one researcher (AA-A) in November 2016. A second researcher (FC-L) verified the data to minimize potential information errors. The SCI-E platform is a database that contains all the above information, including the full addresses of all authors of every paper. We also used the SCI-E to determine the extent to which each paper had been cited in the scientific peer-review literature using the "times cited" number (that is, the number of times a publication has been cited by other publications). A process of standardization was conducted by two researchers to bring together the different names of an author or country, and keywords. Specifically, one researcher (AA-A) checked the names by which an individual author appeared in two or more different forms (for example, "Ronald C. Kessler" or "Ronald Kessler" or "Ron Kessler"), using coincidence in that author's place(s) of work as the basic criterion for normalization (for example, Harvard University, United States), and a second researcher (FC-L) verified data. We used both ''author keywords" and ''keyword plus," which are automatically assigned by the Web of Science from the titles of the references of the articles because this approach has proven to be highly effective in representing the conceptual content of articles. To ensure consistency in the data, one researcher (RA-B) corrected keywords unifying grammatical variants and using only one keyword developed names of the same concept (for example, "diabetes mellitus" or "diabetes" or "adult diabetes" or "diabetes type 2" or "type 2 juvenile diabetes"). In addition, the same researcher (RA-B) removed typographical, transcription and/or indexing errors, and a second researcher (FC-L) verified data. All potential discrepancies were resolved via discussion. All these data were entered into a Microsoft Access1 (Microsoft, Seattle, WA, United States) database.

Data analysis
In this paper, we analyzed data including the number of papers, citations, signatures of authors, collaboration index (which is the mean number of author's signatures per paper), countries, journals and keywords. Data were summarized as frequencies and percentages for categorical items. We have presented in tables the most prolific authors and countries (> 100 papers), and the most cited papers (>1000 citations). We have presented network graphs (or diagrams) to represent data visualization of the structure of the most intense scientific collaboration between countries applying a threshold of 50 papers in collaboration. In order to depict the frequency of the most frequently used keywords, a word cloud was created using Wordle (http://www. wordle.net/), which is free-software that generates "word clouds" from text that the user provides and places more emphasis on words that appear with greater frequency in the source text. We identified the most frequently used keywords per journal subject category. We also presented the "co-words network" of keywords representing the co-occurrence phenomenon of highly frequent words in the papers. The co-words network reflects the relation among multiple terms, and so is effective in mapping the associations between keywords in textual data [27]. We used Pajek [28], a software package for large network analysis that is free for non-commercial use, to construct network graphs. PRISMA checklist [29,30] (http://www.prisma-statement.org/) guided the reporting of the present analysis (and is available in S1 Checklist).

Results
A total of 85994 papers (76350 articles and 9644 reviews) were identified and included in the analyses (Fig 1). Table 1 details the general characteristics of the papers.

Publication trend
The number of papers increased exponentially over the study period (Fig 2). Approximately two-thirds of the papers have been published since 2010. The first paper was published in 1970 by Prof. Alvan R. Feinstein [31] providing the seminal definition of comorbidity referring to "any distinct clinical entity that has co-existed or that may occur during the clinical course of a patient who has the index disease under study." Journals and subject categories 3500 journals published 85994 papers. 596 (17.0%) journals published only one paper, 344 (9.8%) journals published two, 220 (6.3%) journals published three, and 2340 (66.8%) published four or more papers. PLOS One (n = 1223; 1.4%) and the Journal of Affective Disorders (n = 1154; 1.3%) published the largest number of papers, followed by the Journal of Clinical Psychiatry (n = 727; 0.8%), the Journal of the American Geriatrics Society (n = 634; 0.7%) and Obesity Surgery (n = 588; 0.7%). Most papers were classified in one (n = 50354; 58.6%) or two (n = 52748; 30.7%) journal's subject categories. There was wide diversity in journal's subject categories, with psychiatry, surgery, clinical neurology, and general and internal medicine the most common (Table 1).

Authors and countries
Most papers were written by 4 or more authors (72.5%; n = 62311) and only 5.5% (n = 4730) of papers were written by one author. The first authors of the papers were based most commonly Mapping of global scientific research in comorbidity and multimorbidity: A cross-sectional analysis in North America and Europe; first authors from the United States were responsible for 38.6% (n = 33171) of the papers (Table 1). We identified 20 authors who published 100 or more papers (  Overall, 168 countries worldwide were involved in the sample of papers. The productivity ranking for countries with respect to the number of papers (Table 3)

Most cited papers
Overall, included papers received 1.9 million citations, of which 40.9% citations (n = 808817) corresponded to 3596 (4.2%) papers with at least 100 citations. The most cited papers by number of citations ("citation classics" with at least 1000 citations) are listed in Table 5. All the 50 most cited papers were published in English. These most cited articles were published in 20 journals, led by the Archives of General Psychiatry (now, renamed JAMA Psychiatry) with 11 papers and followed by the Journal of the American Medical Association (JAMA) with 8 papers. The list of most cited papers (Table 4) contains contributions dealing with methodological aspects, but also important epidemiological studies on chronic non-communicable diseases, comorbidity and/or multimorbidity. Some of the methodological papers present the most commonly used measures of comorbidities in health services and outcomes research: the "Charlson Comorbidity Index" [32,33] and its various adaptations (papers number-1, number-4, number-17 and number-20 in Table 5) and the "Elixhauser Comorbidity Index" [34] (paper number-10). These comorbidity index measures capture the "comorbidity burden" that   Mapping of global scientific research in comorbidity and multimorbidity: A cross-sectional analysis  Table 5); and the development of the widely-used "Acute Physiology and Chronic Health Evaluation (APACHE) prognostic system" [50] (paper number-14 in Table 5) to quantify the severity of illness in the intensive care units.

Discussion
In this cross-sectional analysis, we analyzed the global scientific research in comorbidity and multimorbidity for the period 1970-2016. We have identified the most productive investigators and countries, most common subjects and keywords, most prolific journals and "citation classics" in comorbidity and multimorbidity based on publications in multiple specialties and disciplines. The most striking results are the increasing number of published articles in recent years, with approximately two-thirds of the papers published since 2010. To the best of our knowledge, this is the first comprehensive global mapping analysis of scientific publications in comorbidity and multimorbidity. This analysis complements and expands the perspective of Mapping of global scientific research in comorbidity and multimorbidity: A cross-sectional analysis previous studies that analyzed some characteristics of articles in comorbidity [10,12], the diversity of terms used in the literature referring to the presence of multiple concurrent diseases [10,12,51], or reviews on the implications and the understanding of research needs and treatment impact [52,53].
In line with previous research in other areas [54][55][56][57], the global productivity of scientific papers is dominated by the United States (as a central hub of knowledge), followed by other nodes in Western Europe (such as the United Kingdom, Germany and Italy) and Canada. The large number of publications on comorbidity and multimorbidity from these countries reflects Mapping of global scientific research in comorbidity and multimorbidity: A cross-sectional analysis the importance that Western societies devote to research as the basis for socio-economic and technological development, but also reflects the interest in understanding and addressing important challenges of population aging and increased complexity of chronicity. Ageing of the world's population is increasing the number of people living with sequelae of multiple diseases, with an increasing trend in low-income countries [1][2][3][4]. As might be expected, the scientific community captured is centered on a nucleus of scientists and researchers from academia, medical and health research centers from North America and Western Europe, but also from Australia and Taiwan (Republic of China). Specifically, the most intense global collaborations took place between authors and institutions from the United States, the United Kingdom and Canada. Perhaps, the very limited participation of low and middle income-based researchers and institutions in research on comorbidity and multimorbidity could warrant further pragmatic action given that the epidemiological transition (e.g. replacement of infectious diseases by chronic diseases) imposes more constraints to deal with the burden of multiple chronic diseases in a poor environment characterized by ill-health systems [3,4,[58][59][60]. Papers on comorbidity and multimorbidity were published most often in journals devoted to neuropsychiatry and neurosciences. Psychiatry has become one of the fastest growing medical disciplines [61]. In fact, the publication activity and interest of comorbidity and multimorbidity in people with mental disorders seems to be increasing [62,63]. Our analysis revealed that nearly 20% of all scientific production was published in journals belonging to psychiatry and mental health. This large relative productivity in psychiatry may be explained by the important role of comorbidity and its implications for theories of etiology, prevention and treatment of mental disorders [63]. Within psychiatry, comorbidity has been traditionally  Mapping of global scientific research in comorbidity and multimorbidity: A cross-sectional analysis used to refer to the overlap of two or more psychiatric disorders [64]. Similarly, comorbidity (and multimorbidity) between mental disorders and substance use disorders [65][66][67], cardiovascular diseases [68][69][70], cancer [71,72] or other chronic disorders [6,73] has gained prominence within the past few decades. Our analyses suggest that other medical disciplines with a large number of papers on comorbidity and multimorbidity, including surgery [74,75], clinical neurology [76][77][78], general and internal medicine [3,4,79,80], cardiology [81] and oncology [82], focus on those conditions with a high global burden of disease. The subject analysis has revealed that the keywords' prioritization in comorbidity and multimorbidity depends on the addressed subject area. For example, "Depression" is the most commonly used keyword in the subject categories of Psychiatry, Clinical Neurology, Neurosciences, Psychology; but also in General and Internal Medicine, Geriatrics and Gerontology, Pharmacology, Endocrinology, Public, Environmental and Occupational Health, and Health Care Sciences and Services. Depression is a common mental disorder that occurs in people of all ages across all world regions and represents a leading cause of disease burden [3,4,78]. Despite existing evidence of the effectiveness of multiple interventions, traditional approaches to the management of the depressive disorders (such as medication alone and brief psychotherapy) have contributed to large treatment gaps [83][84][85][86]. In this respect, the complex pathogenesis implicates factors of diverse nature that should be considered in research. For example, integrating the management of depressive disorders with other common mental disorders (e.g. anxiety disorders and bipolar disorder) or other chronic conditions (e.g. cancer, diabetes and diseases of the cardiorespiratory system) through transdiagnostic interventions [83].
The topic analysis of the most cited papers ("citation classics") allowed us to determine which topics have attracted the most interest in the research on comorbidity and multimorbidity. These include landmark methodological developments in measuring comorbidity (such as Charlson's index, Elixhauser's index and their modifications) [32-34] and descriptive epidemiological studies measuring the burden of comorbidity [35-40], among others. However, important knowledge gaps in comorbidity and multimorbidity remain. The limitations of clinical practice guidelines and treatment for single diseases are well recognized in the biomedical literature, along with the call to make better use of the best evidence base [87,88]. Clinical trials are usually conducted in homogeneous populations, which prevents us from knowing whether treatment effects in people with multiple chronic diseases are equivalent to those in patients with single diseases [89,90]. The evidence base for interventions to improve outcomes for people with multimorbidity therefore remains limited; however, emerging evidence is being Mapping of global scientific research in comorbidity and multimorbidity: A cross-sectional analysis generated to support disease management policies in primary care and community settings [53]. The consideration of people with multimorbidity is essential in future study design and evaluations of health services and technologies. To be of value, it is important that research includes the evaluation of the benefits of multiple approaches for multiple coexisting diseases (e.g. patient-, family-and population-centered), and that generalizability and applicability problems be explicitly addressed [87][88][89][90][91].
There are several limitations to our study. We characterized knowledge structures generated by papers included in the Web of Science database, integrating subject categories of journals, keywords of papers and network analyses. However, these methods represent a scoping approach which could be complemented further by more detailed analyses, for example analyzing the content and reporting quality of papers in evidence syntheses (including systematic reviews of the literature [92]). We only analyzed research articles and review articles. Undoubtedly, there are other important reports (e.g. health policy reports, meeting abstracts and letters/correspondence [93]) that also merit consideration in global debates and discussions in comorbidity and multimorbidity. The validity of keywords mapping and the results of the coword analyses depend on the definitions of words chosen to conceptualize the papers by the authors or database indexers to categorize papers. The growing interest in comorbidity and multimorbidity by health care providers has resulted in more research on these issues, which may have led to a proliferation of different terms for the same concepts. For example, the traditional (and widely accepted) term "comorbidity" is associated with high volume of papers but may lack specificity, whereas the more recently introduced term of "multimorbidity" is associated with low number of papers (see S1 Table). As Almirall and Fortin stated "[t]he use of clearly defined terms in the literature is recommended until a general consensus on the terminology of multiple coexistent diseases is reached" across multiple disciplines [51]. Given the dynamic nature of the field, it will be interesting to see whether the growth trend remains in the coming years, and how the characteristics of the field changes of time (e.g. by means of longitudinal network analyses).

Conclusion
The global analysis presented in this study provides compelling evidence of the scientific growth of research on comorbidity and multimorbidity. Scientific research in this field is increasingly published in biomedical journals, with research leadership of Western countries, most notably, the United States. This study contributes to a better understanding in this challenging field and identifies the main areas of research, the publication sources chosen for their scientific dissemination and the major scientific leaders. Advances in several subjects and research areas will allow for use of new theories and models to fundamentally changes in the management of people with multiple chronic diseases.