New journal selection for quantitative survey of infectious disease research: application for Asian trend analysis

In this study, we applied the Scopus™ database [33] as a source to select journals on infectious disease research (infectious disease journals). This is a new abstract and citation database of the scientific literature provided by Elsevier B.V., which includes over 16,000 peer-reviewed journals.

On the basis of the Scopus™ database (registered in January 2008), we screened infectious disease journals, using English keywords directly linked to disease control in detection, prevention, diagnosis and medical care (A-E). The keywords intended to select journals specifically focusing on infectious disease (A), corresponding to general or categorical infectious diseases (B), in the field of clinical microbiology (C), regarding the development of medicine (D), and intended for overall technology development for disease control (E). Additionally, we selected related journals in the field of public health on the basis of our experience (F).

In parallel, we screened non-English journals, using Japanese, Chinese, French, German, Italian, Spanish, and Turkish keywords corresponding to the keywords in A-E (details shown in the legend to Figure 1). To screen Korean journals, we used English keywords, because almost all journal titles (89 of 91 journals) were registered in English, or both English and Korean (in Roman letters) in the Scopus™ database. In the survey with English and non-English keywords, we introduced an approach based on both partial matching (for a search of journal titles that contain the keywords) and complete matching (for a search of titles that perfectly matches the keywords) to capture journal titles involving inflected forms of the keywords.

After screening the journals in Step 2, we finally selected journals in which original or review articles were included in the PubMed database provided by the U.S. National Library of Medicine [34] from 2006 (articles published in 2007 can be seen in the database) because we emphasized the further usability of our survey method: the PubMed database is freely accessed and widely used, and the selected journals have continued in print.

For quantitative analysis of infectious disease research articles of Asian origin, we surveyed the number of research articles in the above selected journals that were produced in 11 Asian countries: Japan, China, India, Taiwan, Korea, Singapore, Malaysia, Indonesia, Vietnam, Thailand, and the Philippines. We also surveyed the number of the world total as a reference.

We took advantage of the 'Limits' function of the PubMed database, which contains tags for limiting the journal name ([Jour]), affiliation of author ([ad]), publication date ([PPDAT] for print date and [EPDAT] for electronic publication date) and publication type ([pt]) [35]. We targeted original articles and reviews as research articles (hereafter, research articles mean original articles and reviews); the former was considered as an indicator of research activity and the latter as an appreciation of research results. Since we considered that highly valued scientists were given more opportunities to write reviews, meaning that their research results had attracted a good opinion and had relatively good qualities, we targeted not only original research articles, but also reviews. Based on the thought that the number of reviews might be indicative of research quality, it was surveyed separately from the number of original articles.

Concerning the publication date, we prioritized the print date for journals that had both print and electronic versions; for example, we used the following text to search for research articles published on 'AIDS' during 1998-2006 and whose first author was in Japan: AIDS [Jour] AND journal article [pt] AND Japan [ad] AND 1998:2006 [PPDAT].

Additionally, we compared the number of research articles registered in the selected journals to those in the SCI Infectious Disease Category during the period, to ascertain whether the newly selected journals were appropriate to survey a wide range of articles in the field of infectious disease research. We surveyed articles registered in the SCI category in a manner similar to those in the above selected journals.

In order to evaluate socioeconomic factors associated with Asian publications, we weighted the number of research articles registered in the newly selected journals during 1998-2006 according to the population and the gross domestic product (GDP) of each country. We obtained annual data for the population and GDP from the online database of the United Nations [36] for Asian countries. Regarding Taiwan, we used the data from the Taiwan Statistical Data Book 2008 published by the Taiwan's Council for Economic Planning and Development [37]. Using the non-parametric correlation statistical test (Spearman's Rank Correlation test), the numbers of research articles were analyzed in relation to the population and GDP. Statistical analyses were performed using PASW Statistics (version 17.0; SPSS Japan Inc., Tokyo, Japan).

To screen infectious disease journals (Step 2 of Figure 1), 264 candidates were selected (see the actual journals in Additional file 1), of which 240 were selected by English key words and 24 by non-English words. The 264 journals were published in 30 countries and written in 12 languages: English, Japanese, Chinese, Korean, French, German, Italian, Spanish, Turkish, Polish, Russian, and Croatian (see 'Language used' in Additional file 1). Subsequently, we selected 100 of the 264 journals, based on the usability of the PubMed database (Step 3 of Figure 1, see Table 1 and gray labeled journals in Additional file 1). The 100 journals were published in 18 countries and written in English or 7 non-English languages: Japanese, Chinese, French, German, Spanish, Turkish, and Russian (Additional file 2). Forty-eight of the 100 journals matched the journals in the SCI Infectious Disease Category, except for 8 journals (the journals were outside our analysis condition, see details in the legend to Additional file 2). The remaining 52 journals were originally selected in this study and included 15 Asian journals, which consisted of 3 journals in Japanese and 2 in Chinese, 7 in English, and 3 in both English and Japanese or Chinese. The breakdown of the 100 journals corresponding to categories A to F in Figure 1 was: 21 in A, 35 in B, 16 in C, 17 in D, 2 in E, and 3 in F (94 journals). Six of the 100 journals belonged to 2 categories: 2 in A and B, 2 in B and C, 1 in B and D, and 1 in C and D.

In order to ascertain whether the 100 newly selected journals could survey a wide range of infectious disease research articles, we compared the 100 journals and the journals of the SCI Infectious Disease Category from the viewpoint of the difference in the actual number of Asian and worldwide research articles, and the proportion of Asian research articles relative to the world total. Regarding the actual number of research articles, a survey of the 100 selected journals showed more articles than the SCI Infectious Disease Category. The 100 journals published 14,156 research articles (original articles and reviews) of Asian origin and 118,158 of the world total, whereas the journals in the SCI category published 4,621 of Asian origin and 66,518 of the world total during 1998-2006. Regarding the total number of original articles, the 100 journals published about 3-fold of the Asian articles (the actual numbers registered in the 100 journals and the SCI category were 13,452 and 4,412, hereinafter, as described in this paragraph) and 1.8-fold of the worldwide articles (105,308 and 58,424) (Additional file 3 and 4). For the number of reviews, the 100 journals published about 3.4-fold of the Asian articles (705 and 209) and 1.5-fold of the worldwide articles (12,850 and 8,094). Based on the number of original articles published each year from 1998 to 2006, a survey of the 100 journals showed about 2.9- to 3.3-fold (Asian countries) and 1.7- to 1.9-fold (world) higher than in the SCI category (Figure 2 and 3, Additional file 3 and 4). Regarding reviews, the survey of the 100 journals was about 2.4- to 6.9-fold (Asian countries) and 1.5- to 1.7-fold (world) higher than in the SCI category.

Concerning the proportion of Asian articles relative to the world total, it was revealed that a survey of the 100 journals showed a consistently higher percentage than the SCI Infectious Disease Category during 1998-2006. The total of Asian research articles accounted for 12% of the world total in the survey of 100 journals (actual numbers of Asian and worldwide research articles were 14,156 and 118,158, hereinafter, as described in this paragraph) (Additional file 3) and 6.9% in the survey of SCI Infectious Disease Category (4,621 and 66,518) (Additional file 4). Each year during the study period, the proportion of original articles of Asian origin relative to the world total was about 8.6-14.2% in the 100 journals and 4.7-9.3% in the SCI category, and that of reviews of Asian origin was about 4.2-6.9% in the 100 journals and 1.0-3.9% in the SCI category (Additional file 3 and 4).

Subsequently, we surveyed the number of research articles of Asian origin on a country-by-country basis, using the 100 selected journals. It was revealed that Japan had the highest percentage of articles in Asian countries, followed by China, India and Taiwan in descending order during 1998-2006 (Table 2). Additionally, Japan, India and Taiwan produced relatively large numbers of reviews in comparison with other countries (see the column 'Relative to the total number of reviews (RV) of 11 Asian countries' in Table 2). On the other hand, Singapore, Thailand and the Philippines had a high ratio of reviews in their domestic total number of research articles regardless of the relatively low ratio of reviews by Asian countries (see the column 'Relative to the domestic total number' in Table 2).

Looking at the change in the number of research articles from 1998-2006, China, Taiwan, and India showed high increase rates of original articles; the numbers in 2006 were more than about 18-fold (China), 3.9-fold (Taiwan), and 3.5-fold (India) than in 1998 (Figure 4, see the data corresponding to each country in Additional file 3). The number of original articles produced in Japan was stable (the number in 2006 was about 1.2-fold that in 1998), regardless of having the greatest total number of articles by Asian countries. For Korea, original articles increased markedly (under 30 articles in 1998 to over 150 in 2006), and articles produced by other countries, Singapore, Malaysia, Indonesia, Vietnam, Thailand, and the Philippines, increased slightly or were stable. Concerning the number of reviews produced in 11 Asian countries, no noticeable increase was revealed during the study period.

As a further analysis of Asian publications, the number of research articles was compared to the population and GDP. For the population, the ratio of the number of original articles to the population of individual Asian countries showed a median value of 0.2 publications/1 million population/year (range, 0.1-6.9) during 1998-2006. Using population-adjusted ratios, Singapore (median value of 6.9) and Taiwan (6.2) were most productive (Figure 5). The ratio of the number of reviews showed a median value of < 0.1 publications/1 million population/year (range, 0-0.5). When adjusting the production of reviews according to population, Singapore ranked first (0.5), followed by Japan (0.4) and Taiwan (0.3). There was no statistically significant correlation between the average population and the number of original articles (Spearman's correlation coefficient = 0.255, p = 0.450) or reviews (Spearman's correlation coefficient = 0.055, p = 0.873) in 11 Asian countries during 1998-2006.

Regarding the ratio of the number of original articles to the GDP, the median value was 16.6 publications/100 billion GDP/year (range, 6.4-48.3). For GDP-adjusted ratios, Taiwan (48.3) and India (42.9) were highly productive (Figure 6). The ratio of the number of reviews showed a median value of 0.6 publications/1 million population/year (range, 0-3.3). When adjusting production of reviews according to the GDP, Vietnam (3.3), Singapore (2.5) and Taiwan (2.4) were most productive. There was a statistical correlation between the average GDP and the number of original articles (Spearman's correlation coefficient = 0.864, p = 0.001) or reviews (Spearman's correlation coefficient = 0.691, p = 0.019) in Asian countries.