Bibliometric profile of the global scientific research on autism spectrum disorders

Autism spectrum disorders (ASD) represent a group of an early-onset neuro-developmental disorders that affects social interaction and communication abilities of affected people (Association AP 2013; Coleman and Gillberg 2011; Vinogradova 2015; Waterhouse 2013; Wiggins et al. 2015). ASD is an umbrella term that includes five different neuro-developmental disorders with varying degrees of severities, patterns and epidemiology. Such neuro-developmental disorders include: autism (autistic disorder), Asperger’s syndrome, pervasive developmental disorder not otherwise specified (PDD-NOS), childhood disintegrative disorder and Rett syndrome.

Autism Spectrum Disorders are universal affecting all racial, ethnic, and socioeconomic groups (Baio 2014). It is estimated that there are 52 million cases of ASDs worldwide (Baxter et al. 2015). The prevalence of ASD varies for each type of disorder and across different studies (Newschaffer et al. 2007). Data and statistics provided by Centers for Disease Control and Prevention (CDC) indicated that prevalence of ASD is increasing, however, a recent study argue against this finding and claimed that there is no clear evidence of a change in prevalence for autistic disorder or other ASDs between 1990 and 2010 (Baxter et al. 2015; Autism Spectrum Disorder, http://www.cdc.gov/ncbddd/autism/data.html).

ASD have negative economic consequences with real burden on families and governments. Furthermore, ASD have great emotional, social, and health impact on individuals with ASD and their families (Lavelle et al. 2014; Karst and Van Hecke 2012; Hayes and Watson 2013; Estes et al. 2013). A study indicated that autistic disorders accounted for more than 58 disability-adjusted life-years (DALYs) per 100,000 population while other ASDs accounted for 53 DALYs per 100,000 (Baxter et al. 2015). It has been reported that people with ASD had higher average medical expenditures than those without ASD (Shimabukuro et al. 2008). A study found that a child suffering from any kind of developmental disability imposes extra stress on his parents compared with children with no developmental disabilities (Karst and Van Hecke 2012). The stigmatization and discrimination led to the development of many social movements such as the autism rights movement (ARM) that encourages society to accept and deal with autism as a variation in functioning rather than a neurological or developmental disease to be cured by different therapeutic modalities (Jaarsma and Welin 2012).

A bibliometric profile of any medical topic allows researchers to acquire more knowledge about research trends, gives an insight into contribution of a particular country or institution to a given medical topic, and sheds more light into co-authorship and collaboration. Furthermore, bibliometric indicators such as citations and h-index can be used for university ranking, fund application and for scientific prestige among the scientific community. Many bibliometric studies have been published in various scientific disciplines and several ones were published about neurological disorders like multiple sclerosis, Parkinson’s disease and dementia (Li et al. 2008; Araujo et al. 2006; Aleixandre-Benavent et al. 2015; Gupta et al. 2013). The aim of this study was to analyze research output on ASD using a bibliographic analysis of articles indexed in Scopus.

The data in this study were synthesized using Scopus database which has many important features that facilitate bibliometric analysis. Such features include citation analysis, country and author contribution as well as source titles and productivity per year. Scopus is produced by Elsevier and covers more than 20,000 journals that have 100 % Medline coverage. Scopus offers about 20 % more coverage than Web of Science, whereas Google Scholar offers results of inconsistent accuracy (Falagas et al. 2008). Since the data for this study was obtained from electronic sources that are publicly available and not pertaining to specific patients’ data or profile, IRB ethical approval for the study was not required.

The study period was set from January 01, 2005 to December 31, 2014. All subject areas in Scopus search engine (life sciences, social sciences, health and physical sciences) were chosen. All documents with the following title words: (TITLE (autism) OR TITLE (autistic) OR TITLE (“Asperger syndrome”) OR TITLE (“Asperger’s syndrome”) OR TITLE (“Rett syndrome”) OR TITLE (“childhood disintegrative disorder”) OR TITLE (“pervasive developmental disorder”) OR TITLE (“Heller’s syndrome”) OR TITLE (“disintegrative psychosis”) OR TITLE (“cerebroatrophic hyperammonemia”), plus all articles published in autism journals were retrieved and analyzed. The key words have been selected based on the broad definition of ASD and any alternative names available for any of the syndromes within the ASD definition. These key words are present in official web site related to ASD and other recent reviews about the subject (Autism, Autistic Spectrum Disorders (ASD) and Pervasive Developmental Disorders (PDD), http://www.med.umich.edu/yourchild/topics/autism.htm). To increase the accuracy of our search, all articles published in autism journals were included in the analysis. Furthermore, documents classified as errata, or books, or book chapter or un-defined type of documents or conference papers were excluded and therefore this study is restricted to documents that are considered journal articles. All documents obtained after refining the results were transferred to Statistical Package for Social Sciences software version 20 to present the bibliometric indicators. The validity of our method was assessed by assessing the first 100 top cited documents retrieved by the method mentioned above. All retrieved documents were in the ASD domain.

The main bibliometric indicators presented in this study included type and language of the published documents, country and institutional affiliation, source/journal title in which documents were published, most productive authors, most cited articles, and collaboration patterns (Sweileh et al. 2013, 2014; Zyoud et al. 2015a, b, c, d). The growth rate of publication per specific period was calculated as follows: (difference in number of articles published during that period/number of articles published at the start of the period) × 100. Many of the bibliometric indicators were presented in rank order. Research productivity was assessed by the quantity of publications while the total number of citations was used to identify the most influential articles in the field. The Scientific Journal Ranking (SJR) of journals was used as a measure of quality of journals and was obtained from SCImago Journal and Country Ranking website. The Hirsch index (h-index) was used to assess the quantity and quality of publications per country or per institution or per author. A country or an institution will have an h-index of x when it has published x papers, each of which has been cited at least x times to date (Hirsch 2005). The research productivity of different countries was normalized using population size and national gross domestic product (GDP) retrieved from the online databases of the World Bank (Countries and Economies, http://data.worldbank.org/country).

Bibliometric maps and visualization methods were made using VOSviewer software (van Eck and Waltman 2010). Using the VOSviewer and thresholds of minimally 10 fractionally counted articles for each term, density visualization maps were generated for most frequently encountered terms in title of retrieved articles. In these maps, most frequent terms had dense colored cluster. For co-authorship analysis, a minimum number of 500 authors were selected in VOSviewer program. Authors located within or close to a large cluster are believed to have higher number of co-authors suggestive of inter and intra country collaboration.

The whole data are available through Scopus and can retrieved using the search query in methodology. Furthermore, the whole data can be sent upon request to any interested researcher. The data can be sent either as a small file containing up to 2000 articles with full information pertaining to these articles. A second form is to send the whole data, 18,000 articles, with citations information only. The policy adopted by Scopus does not all the whole information of the whole data and allow up to 2000 articles to be exported. The exported file will be in an excel format.

In the current study, we aimed to investigate the scientific research on ASD using bibliometric indicators for articles published during 2005–2014. Scopus was used to achieve the objective of this study. Scopus is considered a larger database than Web of Science and has several advantages over Medline like availability of citation analysis and many other bibliometric parameters (Falagas et al. 2008).

The worldwide growth in ASD research should be considered within the following points: there is an overall increase in various medical and biomedical research topics in the last few decades facilitated by increased number of specialized journals, advancement of molecular genetics technology, and increased governmental and non-governmental funding for unsolved mysteries of several diseases. Several publications have indicated that there is a prominent increase in neurology and psychiatry research topics like Multiple Sclerosis, Epilepsy, Schizophrenia, Anxiety, Dementia and others (Araujo et al. 2006; Aleixandre-Benavent et al. 2015; Gupta et al. 2013; Morandi et al. 2015). It should be emphasized that part of the growth of ASD research is due to the social health consequences of the ASD. Many international societies have been established to support research and to support families with ASD children. Examples of such international associations include the International Society for Autism Research and Autism Network International. Finally, the presence of articles with non-English language is also an indicator of worldwide growing interest in the scientific and social aspects of ASD.

The global increase in number of annual publications on ASD was not accompanied by an increase in inter-institutional collaboration and the number of different institutional affiliation per article with time remained approximately constant. Again, this suggests that the majority of research publications in ASD research is led by research groups affiliated with one institution. It seems that well established research centers in autism tend to interact domestically with authors within the same country because international collaboration might be of limited benefit for such well-established research centers. University College London in UK had the greatest share of publications, total citations and h-index value. Furthermore, UK had the greatest share when the number of publications was normalized with GDP. However, USA with so many active institutions ranked first with a total of 8594 articles and an h-index of 176. Of the top 10 countries, none of the countries were from Latin America or East Europe or Africa, or Middle East. It should be noted here that although the data was stratified by population size and GDP, still, the interpretation of data pertaining to country list needs to take into consideration the guidelines adopted for each country in the diagnosis of ASD and the governmental and non-governmental funding available for research centers and researchers to carry out research on ASD.

Publications from authors affiliated with USA or Japan, or UK or Canada showed dominant domestic collaboration. On the hand, publications from authors affiliated with Italy, France, Australia, Netherland, and Sweden showed dominant international country collaboration. Publications with author from Germany were split in equal between domestic and international country collaboration. International collaboration in ASD research and publication should be encouraged and emphasized given the fact that ASD affects all races and cultures all over the world. Collaboration in social, psychological and therapeutic research could be established between developed and developing countries. A bibliometric study on sleep apnea showed that the probability of citation increased by 1.23 times for each additional author, and by 2.23 times for each additional country represented; independent of time since publication, journal, or the country of the author (Huamani et al. 2015).

The top 10 cited articles revealed that more than half of the hot articles were focused on molecular genetics and cytogenetics of ASD. The other ones were related to epidemiology and pathophysiological characteristics of ASD. The genetics of autism and other ASD remain unsolved and extensive research is being carried out to understand the molecular aspects of these disorders and that is why most of genetic articles about ASD received high number of citations. No wonder that such articles were published in highly prestigious journals that are not in the specific field of autism. As expected, autism journals dominated the top 10 list of journals publishing on ASD during the study period. However, psychology journals like Journal of Child Psychology and Psychiatry and Allied Disciplines which published on psychological aspects of ASD has reached international recognition and audience with an impact factor of approximately 6.5 and an average citation per doc of 48 per article. Of a considerable interest is PLOS ONE journal which ranked 5th in publications about ASD. PLOS ONE is a multidisciplinary, open access journal, with broad scope which explains the large quantity of ASD articles in this journal. An important final note regarding highly cited articles is that it is possible that number of geneticists is greater than number of psychologists and family therapists in most countries which made genetic research occupies most of highly cited articles.

Our study showed that the number of uncited articles represents 16.23 % which is relatively higher than that reported in a similar bibliometric study about Multiple Sclerosis (Aleixandre-Benavent et al. 2015). However, the number of citations of any article varies from time to time and from one journal to another. Therefore, comparison of uncited articles from one subject to another might not be of great benefit. Regarding the most cited articles, it was about number of mutations and it association with autism. The article was published in 2007 in Science journals. The second most cited article was also about chromosomal variations and autism and was published in 2008 in American Journal of Human Genetics. It is noteworthy that the journals in which the first seven most cited articles were non-neurology, non-psychiatry journals, but ones with extremely high impact factor like Science and Nature journals.

Our study has a few limitations related to search strategy and methodology (Sweileh et al. 2013, 2014; Zyoud et al. 2015a, b, c, d). For example, our study did not include articles published in non-Scopus database. However, Scopus remain a reliable source for bibliometric studies in general. Another limitation in our study is the keywords strategy. False positive and false negative results could be obtained regardless of how accurate the search stagey was. With a total of more than 18,000 articles, we believed that false positive or negative results will be very marginal and could hardly affect the accuracy of the results. Furthermore, the use of the keywords in title search instead of title/abstract/keywords would minimize false positive and negative articles and keep non-relevant articles in the minimum tolerable number. There are several journals specialized in autism and we thought that it is logical to include all articles published in these journals regardless of keywords used in the title of articles published in these journals. One might argue against such strategy, but we thought it will be unfair not to include all articles published in these special journals. Finally, the indicators used to assess quality of publications need to be considered carefully. There is no direct measure for quality of publications. However indirect measures like citations, h-index, and IF have been suggested as indirect indicators for quality. The h-index is based on the number of citations received by a certain author, institution, journal, or country. The number of citations is complicated by self-citations which created a lot of bias in assessing quality and impact of publications by authors (Nightingale and Marshall 2013; Gaster and Gaster 2012). Therefore, names of authors listed in the manuscript might not really reflect a true rank and a deeper analysis regarding self-citations of top ranking authors need to be examined. Such bias in self-citations have been the focus of debate regarding certain authors in certain medical field including autistic disorders (Tolisano et al. 2016). A study has shown that the h-index is not a valid parameter to differentiate between assistant professor and associate professor ranks or between MD’s and PhD’s (Doja et al. 2014). A recent study has discussed the merits and demerits of various parameters used to assess quality of publications and suggested “Original Research Performance Index (ORPI)” for evaluation of an author’s original research which can minimize bias arising because of self-citations, gift authorship, inactive phase of research, and length of non-productive period in research (Saxena et al. 2013). Some authors suggested the use of three factors; IF, Eigenfactor Score (ES), and SCImago Journal rank indicator (SJR) when judging quality of the certain medical journals to overcome the several shortcomings of IF (Ramin and Sarraf Shirazi 2012). A very recent editorial in Nature described the use of IF as crude and misleading and that the best assessment of publications is not the journal metrics, rather, reading the paper itself and forming the opinion is the best judgment (Nature Publishing Group 2016).

To the best of our knowledge, this is the first bibliometric study on ASD. The results of our study showed the following characteristics regarding ASD research and publications: there is a growing interest in this topic as seen by the linear increase in the number of publications with time; there is a dominant leadership for Northern Americas and Western Europe in ASD publications; there is a wide variety of journal names in which ASD articles are published; there is a great focus on molecular genetics in ASD research as demonstrated by the title of hot articles in the field; and there is a wide variation in inter-country collaboration in ASD research among the top leadership countries like USA and UK and there is a common trend toward domestic different countries.