Trends and topics in eye disease research in PubMed from 2010 to 2014

Bibliographic search

The search for papers to be included in this study was carried out on January 16th 2015, using the PubMed database (http://www.ncbi.nlm.nih.gov/pubmed), developed by the National Center for Biotechnology Information (NCBI) at the National Library of Medicine (NLM). PubMed was chosen because it is the most widely used in medicine (Falagas et al., 2008). The search strategy was the following: eye diseases [MH] AND journal article [PT] AND 2010:2014 [DP] where MH stands for “Medical Subject Headings,” DP “Date of Publication,” and PT “Publication Type.” “Journal Article” includes the following publication types: journal articles, introductory journal articles, and reviews. MeSH is the National Library of Medicine’s controlled vocabulary thesaurus (http://www.nlm.nih.gov/pubs/factsheets/mesh.html) and consists of sets of terms named “MeSH terms” arranged in a hierarchical structure (called a MeSH tree) with more specific terms arranged beneath less specific terms. In the indexing process, 2 to 50 MeSH terms are assigned to PubMed documents by a team of trained experts (indexers) to properly identify the content, and indexers always index as specifically as possible (MEDLINE Indexing Online Training Course, 2015). It is worth mentioning that it is not obligatory for author’s keywords to be taken into account by indexers to determine MeSH terms of an article (Boudry, 2015, personal data). The MeSH term “Eye Diseases,” the definition of which in the NLM-controlled vocabulary thesaurus is “Diseases affecting the eye,” was chosen because it covers all diseases of the eye when used in a PubMed search. Indeed, when using a MeSH term in a PubMed search, articles that carry the specified MeSH term are retrieved, but so are all of the articles that carry any of the more specific MeSH terms located beneath it in the tree structure. Thus, searching with the MeSH term “Eye Diseases” automatically includes all the MeSH terms located beneath “Eye Diseases” corresponding to diseases of the eye in general.

Analysis of MeSH terms

As done in other studies (Ugolini et al., 2007, 2010; Ramos, González-Alcaide & Bolaños-Pizarro, 2013; Neri et al., 2015), an analysis of the MeSH terms used by PubMed indexers to classify the articles was done by calculating their frequency in the retrieved articles. Check tags, i.e. MeSH terms obligatorily used by indexers to describe recurrent patterns in medical articles, were excluded from this frequency analysis because of their compulsory and recurrent character. They are shown in Supplemental Table S1. The co-occurrence is the simultaneous association of two identical MeSH terms in different articles. The frequency and the percentage of co-occurrence of the MeSH terms located beneath the MeSH term “Eye Diseases” in articles has been calculated. Likewise, the frequency of all other MeSH terms not situated beneath the MeSH term “Eye Diseases” retrieved has also been calculated.

Analysis of growth of literature

The average yearly growth rate of the literature related to eye diseases was calculated as the mean percentage of annual growth rate for the period studied using the equation: Annual Growth Rate = Current Year Total Number of Articles – Previous Year Total Number of Articles/Previous Year Total Number of Articles (Deshazo, Lavallie & Wolf, 2009). The average yearly growth rate was also calculated for the whole PubMed database for the period 2010–2014.

Bradford’s law

Bradford’s law has been used extensively in the information science literature to describe the dispersion of articles in any scientific field (Goffman & Warren, 1969), or to identify “core journals” of serial titles (Deshazo, Lavallie & Wolf, 2009; Venable et al., 2014). Bradford’s law states that “If scientific journals are arranged in order of decreasing productivity of articles on a given subject, they may be divided into a nucleus of periodicals more particularly devoted to the subject and several groups or zones containing the same number of articles as the nucleus.” To identify the core journals containing articles dedicated to eye diseases, Bradford’s law was applied by dividing the journals ranked according to publication output frequency into three groups with each group representing approximately the same number of articles.

Analysis of journals and language of publication

The 2013 Journal Citation Reports (JCR) (Thomson Reuters, New York, USA) was used for Impact Factor (IF) determination and for the ranking of the journal in the category “Ophthalmology” of the JCR. Language of publication was determined using the “Language” field for all articles retrieved. The percentage of papers published in English in the entire PubMed database between 2010 and 2014 was also calculated using the following search strategy: Medline [sb] OR publisher [sb] OR pubstatusaheadofprint AND 2010:2014 [DP] AND journal article [PT] where sb means “subset.”

Quantitative and qualitative analysis of publication ouput

A quantitative evaluation of the scientific production related to eye diseases was done by evaluating the total number of articles published for each country. To implement a qualitative evaluation, each article was related to the corresponding journal manually in the JCR 2013 to determine the Five-Year Impact Factor (5-Y IF). If a journal did not have an 5-Y IF, the value zero was assigned to the article. For each country, the cumulative 5-Y IF was calculated as the sum of the 5-Y IF of all articles. The mean Five-Year Impact Factor (m5-Y IF) was calculated as the cumulative 5-Y IF divided by the number of articles.

For the quantitative and the qualitative evaluation, we calculated the number of publications and the cumulative and m5-Y IF according to countries and continents.

The country of affiliation of the first author was determined by the address in the “Affiliation” field. When the country was absent in the address, it was determined from the city or email address using an Internet search engine if necessary. If the name of a country no longer existed (e.g. Yugoslavia), the city was used to find the name of the existing country in 2015. England, Scotland, Northern Ireland, and Wales were grouped into the United Kingdom. Hong Kong was considered as part of China.

The GDP is the market value of all officially recognized final goods and services produced within a country in a given period. GDP per capita is often considered an indicator of a country’s standard of living (World Bank Group: http://www.worldbank.org/). For each country and each continent, the 5-Y IF GDP index (cumulative 5-Y IF per 1 billion US dollars of GDP) and the 5-Y IF population index (cumulative 5-Y IF per million inhabitants) were calculated using the World Development Indicators from the online databases of the World Bank (http://data.worldbank.org/). The 5-Y IF GDP and the 5-Y IF population index were calculated using the mean GDP and mean number of inhabitants from 2010 to 2013 (data for 2014 were not available). Countries were clustered by their continent according to the United Nations classification (United Nations Statistics Division- Standard Country and Area Codes Classifications: http://unstats.un.org/unsd/methods/m49/m49regin.htm).

Journals

The search in PubMed resulted in a total of 62,123 articles over the period 2010 to 2014. We calculated that the average growth rate of literature related to eye diseases was 4.27%. We also calculated that the average growth rate for the whole PubMed database was 6.59% for the same period. Thus, the difference in growth rates between the growth rate of eye diseases and all scientific production in the PubMed database is equal to −2.32 (4.27 − 6.59), which indicates a moderate interest of the scientific community for eye diseases for the studied period, and a stabilization of publication rate (Michon & Tummers, 2009).

The 62,123 articles found in the present study were published in 3,368 different journals over the period 2010 to 2014.

Fifty-four journals of 58 belonging to the category “Ophthalmology” in the JCR were found in the 3,368 journals retrieved. These 54 journals published 29,183 articles of the 62,123 (46.98%) that we found in the present study.

According to Bradford’s law, the total number of articles compared to the number of journals in order of decreasing productivity was divided into 3 zones containing approximatively the same number of articles. The first third of the total number of articles (n = 20,779) represented the journals (n = 19; 0.56%) that published between 3,619 and 589 articles (Table 1). The middle third (n = 20,637) corresponds to the journals (n = 118; 3.5%) that contained an average number of articles (between 290 and 579 articles), and the last third (n = 20,707) includes the “long tail” of journals (n = 3,231; 95.9%) that published the fewest articles (less than 114). It is important to note that 1,428 journals (42.4%) published only 1 or 2 articles related to eye diseases over the 5-year study period.

Languages of publication

Thirty-one different languages were identified in the 62,123 articles retrieved. The five predominant languages were English (n = 55,829; 89.87%), German (n = 1,401; 2.26%), French (n = 1,132; 1.82%), Chinese (n = 927; 1.49%), and Japanese (n = 641; 1.03%). All other languages amounted to less than 1%. The percentage of articles in English in the entire PubMed database was 93.49% for the same period.

MeSH terms for eye diseases

The MeSH term “Eye Diseases” is located beneath the MeSH term “Diseases” in the MeSH tree. Twenty-seven more specific MeSH terms situated just one level beneath “Eye Diseases” were found in the MeSH tree (Fig. 1A). The MeSH tree beneath “Eye Diseases” includes 6 levels. In total, 243 MeSH terms located beneath “Eye Diseases” were found in these 6 levels. As an example, the MeSH tree for the MeSH term “Chorioretinitis” located six levels beneath “Eye Diseases” is described in Fig. 1B.

Analyzing the 62,123 articles, we found 853,182 MeSH terms, which represent an average of 13.73 (SD = 4.98) MeSH terms per article. Among these 853,182 MeSH terms, 14,689 different MeSH terms were identified: 243 were located beneath “Eye Diseases” and 14,446 were not located beneath “Eye Diseases.” The 20 most frequent MeSH terms located in all the levels beneath the MeSH term “Eye Diseases,” and for each MeSH term, the 5 most co-occurring MeSH terms in articles are presented in Table 2 (for information on the 21st to 50th see Supplemental Table S2).

Similarly, Table 3 describes the 20 most frequent MeSH terms retrieved in articles and not located beneath “Eye Diseases” (for information on the 21st to 50th see Supplemental Table S3). Among these MeSH terms not located beneath “Eye Diseases,” note that some are related to methods of investigation (Tomography, Optical Coherence, Magnetic, Resonance Imaging; Fluorescein Angiography) or type of studies (Retrospective Studies, Follow-Up Studies, Prospective Studies, Cross-Sectional Studies, Case-Control Studies).

Geographical distribution and socioeconomic factors

The determination of the country of the first author was possible for 59,060 articles (95.07%). One hundred thirty-two different countries were identified. As shown in Table 4A, the qualitative evaluation with the m5-Y IF by continents led to a new ranking compared with the absolute production (total number of publications): North America and Oceania initially 3^rd and 4^th, were respectively first and second when considering m5-Y IF. Geographic distribution by continent of publication was also different after adjustment for 5-Y IF population index and 5-Y IF GDP index (Tables 4A and 4C).

When analyzing countries, the United States was the most absolute productive country for total number of publications (Table 5A). Among the most productive countries, 15 of 20 were classified as developed countries (according to the definition of the United Nations). Qualitative evaluation using m5-Y IF highlighted the quality of the scientific production related to eye diseases of some countries (i.e. United kingdom, the Netherlands, and Singapore). After adjustment for population size (Table 5B), the United States was downgraded to 9^th place, and China disappeared from the top 20 countries. The ranking positions changed overall in favor of smaller countries with a low population size, with the appearance of Iceland, Switzerland, Denmark, and New Zealand in the top 10. When normalized by GDP (Table 5C), 10 of the top 20 countries in terms of absolute production (number of articles produced), disappeared (i.e. China, Japan, Germany). Furthermore, the ranking positions changed in favor of some developing countries (i.e. Malawi, Guatemala, Singapore ). Nine countries appeared in the top 25 countries for the three parameters studied: the United States, United Kingdom, the Republic of Korea, Spain, Australia, the Netherlands, Switzerland, Israel and Singapore.