The FASEB Journal Life Sciences Forum

Research contribution of different world regions in the top 50 biomedical journals (19952002)
Elpidoforos S. Soteriades,*, Evangelos S. Rosmarakis,* Konstantinos Paraschakis,* and Matthew E. Falagas,*,,1
*Alfa Institute of Biomedical Sciences (AIBS), Athens, Greece; Cyprus International Institute for the Environment and Public Health in association with Harvard School of Public Health, Nicosia, Cyprus; and Department of Medicine, Tufts University School of Medicine, Boston, Massachusetts, USA
ABSTRACT

We evaluated all articles published by different world regions in the top 50 biomedical journals in the database of the Journal Citation ReportsInstitute for Scientic Information for the period between 1995 and 2002. The world was divided into 9 regions [United States of America (the U.S.), Western Europe, Japan, Canada, Asia, Oceania, Latin America, and the Caribbean, Eastern Europe, and Africa] based on a combination of geographic, economic and scientic criteria. The number of articles published by each region, the mean impact factor, and the product of the above two parameters were our main indicators. The above numbers were also adjusted for population size, gross national income per capita of each region, and other factors. Articles published from the U.S. made up about two-thirds of all scientic papers published in the top 50 biomedical journals between 1995 and 2002. Western Europe contributed approximately a quarter of the published papers while the remaining one-tenth of articles came from the rest of the world. Canada, however, ranked second when number of articles was adjusted for population size. The U.S. is by far the highest-ranking country/region in publications in the top 50 biomedical journals even after adjusting for population size, gross national product, and other factors. Canada and Western Europe share the second place while the rest of the world is far behind. Soteriades, E. S., Rosmarakis, E. S., Paraschakis, K., Falagas, M. E. Research contribution of different world regions in the top 50 biomedical journals (19952002). FASEB J. 20, 29 34 (2005) Key Words: research productivity world regions top journals publications quality

reviewed journals indexed in worldwide accepted electronic databases. The evaluation process is also guided by certain generally accepted criteria such as the assigned impact factor of each journal (3). It is widely known that publication of research articles in biomedical journals has grown exponentially in the past few decades (4, 5). Several reasons have contributed to the ever-increasing number of journals and scientic papers being published either in hard copies, electronically, or both, and indexed in electronic databases around the world. Technological advances in computer science, communication, the growth of biotechnology, the spread of computerized information, and the availability of personal computers in combination with the development of the worldwide web, constitute a few important factors that have offered new opportunities in recent years and facilitate the performance of biomedical research by individuals or networks of scientists. In several studies, scientists have reported on the participation of different countries in research in several disciplines (6 9). However, little is known about the global ranking of different world regions regarding articles published in the highest-ranking biomedical journals (4). To evaluate the contribution of individual countries, different geographic areas, and entire continents in the production of research papers presented in the top 50 biomedical journals of the Journal Citation Reports (JCR) database (10) of the Institute for Scientic Information (ISI) between 1995 and 2002, we conducted a study of all relevant published articles in the PubMed database (11). We also evaluated the research productivity of each region adjusted for funds spent on research and development, as well as the number of scientists involved. MATERIALS AND METHODS
The methodology used has been described elsewhere (9). Briey, research productivity was evaluated by dividing the Correspondence: Alfa Institute of Biomedical Sciences (AIBS), 9 Neapoleos St., 15123 Marousi, Greece. E-mail: matthew.falagas@tufts.edu doi: 10.1096/fj.05-4711lsf
29
1

Scientific research is recognized as one of the cornerstones of economic growth and development in the developed world and appears to be gaining momentum for the economic development in other less developed areas (1). Investment in research is also thought to be a powerful tool for governments struggling to compete in the current environment of globalization (2). The scientic community operates within an overarching paradigm, which places a signicant weight in peer0892-6638/06/0020-0029 FASEB

TABLE 1. Number of research articles published by different world regions in the top 50 biomedical journals between 1995 and 2002
Year 1995 1996 1997 1998

World regions USA Western Europe Japan Canada Asia (excluding Japan) Oceania Latin America and Caribbean Eastern Europe Africa Total

n (%) 9,020 (61.2) 4,058 (27.5) 664 (4.5) 477 (3.2) 180 (1.2) 235 (1.6) 51 (0.3) 35 (0.2) 27 (0.2) 14,747 (100)

8,317 (60.5) 3,835 (27.9) 600 (4.4) 503 (3.7) 190 (1.4) 204 (1.5) 42 (0.3) 34 (0.2) 28 (0.2) 13,753 (100)

7,923 (58.2) 4,051 (29.8) 604 (4.4) 525 (3.9) 194 (1.4) 204 (1.5) 47 (0.3) 27 (0.2) 32 (0.2) 13,607 (100)

8,805 (61.8) 3,840 (27.0) 565 (4.0) 456 (3.2) 242 (1.7) 228 (1.6) 46 (0.3) 41 (0.3) 21 (0.1) 14,244 (100)

TABLE 2. Number of research articles published by different world regions in 26 journals that were ranked in the top 50 biomedical journals during the whole study period (1995 and 2002)
Year 1995 1996 1997 1998

World Regions USA Western Europe Japan Canada Asia (excluding Japan) Oceania Latin America and Caribbean Eastern Europe Africa Total

n (%) 3,921 (67.9) 1,336 (23.1) 230 (4.0) 153 (2.6) 55 (0.9) 49 (0.8) 16 (0.3) 11 (0.2) 5 (0.1) 5,776 (100)

4,060 (67.7) 1,328 (22.1) 239 (4.0) 194 (3.2) 59 (1.0) 72 (1.2) 18 (0.3) 20 (0.3) 10 (0.2) 6,000 (100)

3,801 (65.4) 1,411 (24.3) 245 (4.2) 196 (3.4) 55 (0.9) 76 (1.3) 18 (0.3) 9 (0.1) 5 (0.1) 5,816 (100)

3,911 (64.8) 1,435 (23.8) 286 (4.7) 203 (3.4) 79 (1.3) 83 (1.4) 18 (0.3) 11 (0.2) 6 (0.1) 6,032 (100)

world into different regions, identifying all articles published during the study period (19952002), and documenting the country of origin for each rst author. The number of published articles was considered as an index of quantity of research productivity. The mean impact factor of the published articles was used as a quality indicator. Finally, the product of the number of articles multiplied by the impact factor of the journal, for the year studied, was considered a combined indicator of the quantity and quality of research productivity. The sum of the above products from all journals, for each world region within a year, was documented as a total product of research for each particular region. World regions For the purpose of our study, the world was divided into nine regions based on a combination of geographic, economic, and scientic criteria (12): the United States of America (the U.S.), Western Europe, Japan, Canada, Asia, Oceania, Latin America and the Caribbean, Eastern Europe, and Africa. Puerto Rico and the Virgin Islands were evaluated together with the U.S. Greenland was assigned to Western Europe and Japan was studied as a separate region from the rest of Asia. All former socialist countries of Europe were included in the category of Eastern Europe. Journals All journals, which were ranked as the top 50 journals in the Journal Citation Reports for 19952002 at each year, were
30 Vol. 20 January 2006

included in our study. Several journals were not ranked in the top 50 category for certain years of the study period. To evaluate any signicant differences among the journals, which were not always included in the top-ranking list, we performed similar analyses restricting our sample size to 26 journals that were ranked in the top 50 category for the entire study period (19952002). Search procedures A phrase consisting of four parts joined together by the so-called Boolean operators (i.e., AND, OR, and NOT) was used in our search in the PubMed database. Each search was limited to a specic year using the Limits function, which is incorporated in the search engine. We used data only on original articles and reviews, excluding publication types such as letters, editorials, and news reports. This was accomplished by selecting publications with the characterization journal article [pt] in the search elds of the database (pt designates publication type). All countries of each region were included in the rst parenthesis of the search phrase. In the second parenthesis, after the word NOT, certain addresses were excluded in order to avoid double counting. Subsequently, the results of our search (the number of articles produced by each world region in a specic journal within a year) were summed up. Further conrmation was accomplished by summing up the articles retrieved from our search for all different world regions in a specic journal and comparing them to the actual total number of articles pubSOTERIADES ET AL.

The FASEB Journal

TABLE 1. (continued)
Year 1999 2000 2001 2002 Total

8,464 (60.8) 3,874 (27.8) 535 (3.8) 410 (2.9) 259 (1.9) 222 (1.6) 63 (0.4) 46 (0.3) 36 (0.3) 13,909 (100)

8,921 (61.4) 3,929 (27.0) 637 (4.4) 463 (3.2) 231 (1.6) 209 (1.4) 66 (0.4) 55 (0.4) 12 (0.1) 14,523 (100)

6,690 (57.6) 3,589 (30.9) 440 (3.8) 401 (3.4) 194 (1.7) 194 (1.7) 38 (0.3) 36 (0.3) 35 (0.3) 11,617 (100)

6,555 (58.7) 3,243 (29.1) 433 (3.9) 353 (3.2) 220 (2.0) 213 (1.9) 50 (0.4) 42 (0.4) 48 (0.4) 11,157 (100)

64,695 (60.1) 30,419 (28.3) 4,478 (4.2) 3,588 (3.3) 1,710 (1.6) 1,709 (1.6) 403 (0.4) 316 (0.3) 239 (0.2) 107,557 (100)

TABLE 2. (continued)

Year 1999 2000 2001 2002 Total

3,920 (65.0) 1,391 (23.1) 282 (4.7) 201 (3.3) 103 (1.7) 84 (1.4) 19 (0.3) 21 (0.3) 7 (0.1) 6,028 (100)

4,164 (61.3) 1,814 (26.7) 305 (4.5) 233 (3.4) 94 (1.4) 111 (1.6) 33 (0.5) 27 (0.4) 8 (0.1) 6,789 (100)

4,203 (61.4) 1,859 (27.2) 281 (4.1) 224 (3.3) 108 (1.6) 108 (1.6) 27 (0.4) 25 (0.4) 10 (0.1) 6,845 (100)

4,101 (61.1) 1,797 (26.8) 302 (4.5) 211 (3.1) 122 (1.8) 104 (1.5) 28 (0.4) 33 (0.4) 12 (0.2) 6,710 (100)

32,081 (64.2) 12,371 (24.7) 2,170 (4.3) 1,615 (3.2) 675 (1.4) 687 (1.4) 177 (0.4) 157 (0.3) 63 (0.1) 49,996 (100)

lished in the same journal for a specic year. The total number was obtained from PubMed without using any address limits. Using the above methodology we were able to cross-examine missed or unretrieved addresses. The above scenario was occasionally useful in cases of articles with no address registered or articles where only the afliated institution or the city (not the country) was recorded. If less than 5% of the total articles of a specic journal during a year had missing addresses, those articles were excluded. On the other hand, if more than 5% of the total articles of a specic journal had missing addresses, we checked the authors address by looking at other articles of the same author within the same year. To include addresses where only cities or areas were registered, we expanded our search criteria, including search phrases with big cities or capitals (e.g., Munchen, London, or Moscow) and all individual states of the U.S. Regional economic and scientic resources To further evaluate the research productivity of each region, adjusted for other factors, we used relevant World Development Indicators (13) from the online databases of the World Bank. The research productivity of different world regions (estimated by the total product) was adjusted (divided by) for the total population, the gross domestic product (GDP) in standard 1995 U.S. dollars, the gross national income (GNI) per capita (Atlas method), the percentage of GDP spent on

research and development (R&D) and the number of researchers per million population for each region for which data were available.

RESULTS A total of 107,557 articles indexed in PubMed were published in the top 50 biomedical journals around the world for the period 19952002, and were included in our study with an annual average of 13,444 articles. The annual averages for each geographic region studied were 8087, 3802, 560, 448, 214, 214, 50, 39, and 30 for the U.S., Western Europe, Japan, Canada, Asia, Oceania, Latin America, Eastern Europe, and Africa, respectively. Table 1 presents the total production of articles by each world region and the relative contribution of each region to the total number of articles published in the top 50 biomedical journals every year. Consistently, 6 of every 10 articles published every year during the 8-year period were from the U.S., and another 3 of 10 originated in Western Europe. Only 1 of 10 articles was being published from the rest of the world. The U.S., by far, ranked rst every year, Western Europe second,
31

GEOGRAPHY OF RESEARCH IN THE TOP 50 BIOMEDICAL JOURNALS

TABLE 3 Number of research articles multiplied by their impact factor published by different world regions in the top 50 biomedical journals between 1995 and 2002
Year 1995 1996 1997 1998 1999 2000 2001 2002 Total

World Regions USA Western Europe Japan Canada Asia (excluding Japan) Oceania Latin America and Caribbean Eastern Europe Africa Total

140,841 60,939 9,847 7,099 2,695 3,350 747 490 432 226,440

144,358 62,589 10,089 8,365 3,119 3,386 840 700 526 233,972

129,468 62,549 8,957 7,903 2,796 3,140 758 426 520 216,517

139,989 59,186 8,792 7,371 3,451 3,326 670 518 325 223,628

139,986 59,574 8,649 6,567 3,911 3,327 889 719 538 224,160

155,354 67,664 10,883 8,065 3,872 3,729 1,156 950 220 251,893

134,084 67,002 8,795 7,635 3,605 3,683 778 729 595 226,906

141,151 66,854 9,469 7,248 4,528 4,151 1,059 948 855 236,263

1,125,231 506,357 75,481 60,253 27,977 28,092 6,897 5,480 4,011 1,839,779

Japan third, Canada fourth, Asia and Oceania were sharing fth place; the rest of the world regions had lower rankings. Despite several changes occurring every year in the list of the top 50 ranked journals, the ranking of different regions showed no notable changes. The total production of articles of all world regions, for the 26 journals that were always ranked in the top 50 category during the entire study period is presented in Table 2. There were no signicant differences with respect to the contribution of each region between the two tables. There was a small increase in the contribution of the U.S. compared with a corresponding decrease in the production of Western Europe and the other world regions. A similar picture is seen in Table 3, where we present the product of the number of published articles multiplied by their corresponding impact factor. The U.S. continues to lead with a signicant distance from the second-ranking region (Western Europe). Japan, Canada, Oceania, and Asia have much lower outcomes, and the other three regions are very far behind. However, there were no important changes over time in either the numbers of articles produced every year from each region or the product of the articles multiplied by their impact factor. An interesting nding is seen in Fig. 1, representing the percentages of published articles over time from each region, adjusted by their corresponding population size. After the adjustment, the U.S. continued to lead, however, Canada ranked second, surpassing Western Europe, and Oceania ranked fourth above Japan, while the ranking of the other regions was not signicantly altered. In Fig. 2 we present the association between the gross national income per capita (GNIPC) in U.S. dollars and the number of articles published from each region, adjusted for population size. In this gure, the U.S. is dominating the world since it has the second highest GNIPC and by far the highest production of research. Canada has a relatively low GNIPC, but ranks second in the number of articles published per million population. Western Europe follows closely with Canada, and
32 Vol. 20 January 2006

Oceania has a very interesting outcome. Although it shows the lowest GNIPC among the developed nations, it produces the fourth highest number of research articles per million of population. Japan, although with the highest GNIPC among all regions, appears to have the lowest research productivity among the developed nations. Asia, Latin America and the Caribbean, Eastern Europe, and Africa have all very low GNIPC and research production. When we examined the funds spent on research compared with the number of articles produced, Canada ranks rst with 37.8 articles per billion of U.S. dollars spent, the U.S. ranks second with 36.0 articles, Western Europe third with 19.4 articles, and Japan

Figure 1. Annual percentages of articles contributed by different world regions in the top 50 biomedical journals adjusted for population size of each region during the 8-year study period. The different colored lines indicate different world regions.
SOTERIADES ET AL.

The FASEB Journal

Figure 2. The number of articles per million population published by each world region in association with the gross national income per capita (GNIPC) of each region in U.S. dollars. Different colors represent different regions. Each region has 8 small squares representing each year of the study period.

fourth with only 3.4 articles per billion of U.S. dollars spent in research. A different ranking, however, was seen when we looked at the number of articles produced per 1000 scientists-years of research. The U.S. reclaimed its rst place with 7.2 articles, Canada second with 5.4 articles, Western Europe third with 4.3 articles, and Japan fourth with 0.9 articles per 1000 scientistsyears spent on research.

DISCUSSION The number of scientic publications in the top 50 biomedical journals along with their corresponding impact factor may be considered, based on current standards, a strong indicator of the quantity and quality of research productivity around the world. The results of our investigation are not surprising as they point out that the U.S. is by far the leading power in the eld, as measured by almost all raw and/or adjusted indicators. However, it is impressive that two-thirds of the published articles in the top-ranking journals originated in the U.S. Although Western Europe ranks second in terms of number of scientic articles and their corresponding impact factor, Canada represents a surprisingly positive example when evaluated by adjusted indicators such as the number of articles adjusted for funds spent on research or expressed per 1000 scientists-years of research. Although Japan appears to lack behind in adjusted indicators compared with the other

developed regions, this may represent an indirect language effect. As mentioned, the U.S. not only ranks rst but also contributes about two-thirds of all published articles in the top 50 biomedical journals as opposed to its contribution in specic disciplines, which has a lower range (from 35 to 45%) with respect to the worlds research output (4). Western Europe, mostly represented by the European Union, is lagging behind by all measures of comparisons although it occupies the second place in most of the above-examined indicators. European contribution in the worlds top publications is one-third compared with the U.S. European Union countries would need to increase their investment and perhaps improve their efciency and effectiveness in integrating a unied research environment that will facilitate the development of networks and institutions capable of competing in top-notch research elds. Similar recommendations may also apply to other world regions that aspire to compete with the leading power in scientic publications. Our study also provides information on the relative contribution of different world regions in the topquality research produced around the world. The adjusted indicators presented allow for comparisons of the research productivity between areas of the world with diverse population size, economic status, and funding for different research priorities. In addition, our data may be used as baseline information in evaluating the return of investment on research in the future for different world regions. Previous studies have reported on similar factors related to the geographical distribution of medical publications (14 16). Our ndings fall within the boundaries of previous reports in terms of ranking of research contributions of different regions/countries. Our study has several limitations. First, we used JCR criteria to identify the top 50 biomedical journals included in the study every year. Articles published in non-JCR-cited journals were not included, although we recognize that they contribute to scientic research. The above is specically important for regions with authors, who are not native speakers of the English language (i.e., Japan, Asia and Eastern Europe), and may publish their ndings in journals of their own language (17). Furthermore, the impact factor has often been criticized as a tool for evaluating the quality of scientic research (18 21). However, we believe that the impact factor represents one of the best-known methods of quality evaluation for biomedical journals (3, 22). In addition, we used PubMed, an easily accessible and widely used database. Nevertheless some scientic articles are not included in this database and so were not evaluated in the current investigation. In PubMed, only the address of the rst author is presented, so the fact that a study may be the result of a multinational cooperation is not taken into account. Our search criteria were not perfect in retrieving the addresses of all articles; therefore, the data collection process may
33

GEOGRAPHY OF RESEARCH IN THE TOP 50 BIOMEDICAL JOURNALS

have not been comprehensive, although we adjusted our procedures in order to avoid double counting of articles. As a result, we assume that the number of missed articles did not signicantly affect our results. We used our own criteria to divide the world into different regions. We think that our categorization takes into account geographic and economic criteria that may be replicated by other researchers. Finally, not all journals were ranked in the top 50 category during the entire study period. However, our results were not altered when we restricted the analysis to 26 journals, which were always ranked in the top 50 list. In conclusion, we evaluated the contribution of different world regions in the top 50 biomedical journals during an 8-year study period. Our ndings represent a tool for governments as well as global agencies and organizations to assess different regions in terms of research productivity. Such evaluations may also be used to allocate resources and create incentives for those regions that are lacking behind. Furthermore, they may be used as baseline criteria for future comparisons over time.
M.E.F. developed the idea and nalized the methodology. E.S.R. collected the data. K.P. did the statistical analysis. E.S.S. wrote the original draft. All co-authors reviewed and made contributions to the nal manuscript. E.S.S. and M.E.F. are guarantors.

6. 7. 8.

9.

10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22.

REFERENCES
1. 2. 3. 4. 5. Saravia, N. G., and Miranda, J. F. (2004) Plumbing the brain drain. Bull. World Health Org. 82, 608 615 Benner, M. (2004) Catching up in pharmaceuticals: government policies and the rise of genomics. Aust. Health Rev. 28, 161170 Gareld, E. (1955) Citation indexes for science; a new dimension in documentation through association of ideas. Science 122, 108 111 Rahman, M., and Fukui, T. (2003) Biomedical publication global prole and trend. Public Health 117, 274 280 Vergidis, P. I., Karavasiou, A. I., Paraschakis, K., Bliziotis, I., and Falagas, M. E. (2005) A bibliometric analysis of global trends of

research productivity in Microbiology. Eur. J. Clin. Microbiol. Infect. Dis. In press Mela, G. S., Martinoli, C., Poggi, E., and Derchi, L. E. (2003) Radiological research in Europe: a bibliometric study. Eur. Radiol. 13, 657 662 Ugolini, D., and Mela, G. S. (2003) Oncological research overview in the European Union. A 5-year survey. Eur. J. Cancer 39, 1888 1894 Keiser, J., Utzinger, J., Tanner, M., and Singer, B. H. (2004) Representation of authors and editors from countries with different human development indexes in the leading literature on tropical medicine: survey of current evidence. Br. Med. J. 328, 1229 1232 Rosmarakis, E. S., Vergidis, P. I., Soteriades, E. S., Paraschakis, K., Papastamataki, P. A., and Falagas, M. E. (2005) Estimates of global production in cardiovascular diseases research. Int. J. Cardiol. 100, 443 449 Institute for Scientic Information SCI: Science Citation IndexJournal Citation Reports, 1996 2000. (2004) The Institute for Scientic Information, Philadelphia National Library of Medicine Index Medicus database (PubMed). Bethesda, Maryland, 2004 United Nations Statistical Yearbook 42nd issue, CD-Rom Edition. 2004. World Development Indicators 2002, CD-Rom Edition, The World Bank. 2004. Benzer, A., Pomaroli, A., Hauffe, H., and Schmutzhard, E. (1993) Geographical analysis of medical publications in 1990. Lancet 341, 247 Heer, L., Tempfer, C., and Kainz, C. (1999) Geography of biomedical publications in the European Union, 1990 98. Lancet 353, 1856 Thompson, D. F. (1999) Geography of U.S. biomedical publications, 1990 to 1997. N. Engl. J. Med. 340, 817 818 Coates, R., Sturgeon, B., Bohannan, J., and Pasini, E. (2002) Language and publication in Cardiovascular Research articles. Cardiovasc. Res. 53, 279 285 Seglen, P. O. (1997) Why the impact factor of journals should not be used for evaluating research. Br. Med. J. 314, 498 502 Gallagher, E. J., and Barnaby, D. P. (1998) Evidence of methodologic bias in the derivation of the Science Citation Index impact factor. Ann. Emerg. Med. 31, 83 86 Neuberger, J., and Counsell, C. (2002) Impact factors: uses and abuses. Eur. J. Gastroenterol. Hepatol. 14, 209 211 Whitehouse, G. H. (2002) Impact factors: facts and myths. Eur. Radiol. 12, 715717 Luukkonen, T. (1990) Bibliometrics and evaluation of research performance. Ann. Med. 22, 145150 Received for publication July 13, 2005. Accepted for publication October 5, 2005.

34

Vol. 20

January 2006

The FASEB Journal

SOTERIADES ET AL.