The Unesco conference will note the growing strength of Asian biomedical science. Grant Lewison reports
The practice of counting the numbers of scientific papers produced by a country has become a statistical exercise comparable with the estimation of gross domestic product. There are expert bean counters who worry about the details of methodology in both exercises, and policy people who fret over the interpretation of the figures. Within countries, scientific outputs are increasingly being used in the allocation of scarce research funds, on the "Matthew principle" of giving more to him who hath already.
Comparisons among countries can reveal national strengths and weaknesses and may suggest changes of policy. This is particularly so in biomedicine, because all countries need to provide health care for their populations, and the commercial exploitation of biological discoveries offers the possibility of economic growth and employment.
Most international comparisons of scientific output depend on the compilation of the Science Citation Index by the United States-based Institute for Scientific Information in Philadelphia. But although national counts of papers in all science (conventionally limited to peer-reviewed articles, notes and reviews that represent original scientific work) are easily determined, outputs in biomedicine depend on the definition used for the field. I have adopted a suggestion first made in 1993 by Ringer de Bruin and Henk Moed in Leiden to use meaningful words (often contracted) in the addresses of the papers as indicators of their subject matter. For biomedicine, the list of such words is very long and includes departmental specialities such as cancer or heart, types of institution such as St-*-Hosp, national organisations such as MRC or NIH, and pharmaceuticals companies such as Glaxo* or Smithkline*. The procedure successfully picks out biomedical papers from multidisciplinary journals such as Nature.
The analysis was limited to 1993-98 and to 48 countries. Only selected results are given here. The first striking observation is that countries vary greatly in relative commitment to biomedicine as a percentage of all their scientific publications (see table below). Scandinavian countries have the highest percentage of their papers in biomedicine, about 60 per cent, together with Kenya, and the United Kingdom has 54 per cent. By contrast, Russia, Ukraine, Belarus and China have fewer than 20 per cent. The mean is just under 50 per cent, but many countries do much less than this.
The next result to emerge is the amount of biomedical research published. Clearly, large countries will have more scientists, but it has been known for some time that scientific output correlates better with GDP than with population size. There is an almost linear relationship and quite good correlation when the points for the different countries are plotted on logarithmic scales. Countries whose output lies about the trendline include the United States, the UK and Sweden, but also former communist states such as Hungary and Slovakia and developing countries such as Kenya. Comparisons with population size (in 1994) show a more extreme variation in output.
The table shows the growth in biomedical output over the six-year period for some selected countries. It is expressed as the ratio of output in 1996-98 to that in 1993-95. The fastest-growing outputs are those of East Asian countries - South Korea, Taiwan and China - with Romania rising rapidly from a very low level. The Philippines, Indonesia, Singapore and Malaysia are also growing at above average rates. The next group comprises the three big Latin American countries, Brazil, Mexico and Argentina, which are growing much faster than Venezuela, Peru and Chile. In Japan, output is growing rapidly (ratio 1:15), and it overtook that of the UK in 1998 to become the world's second largest.
Although US output is growing, albeit slowly, it has declined in relative terms from 42.6 per cent to 39.5 per cent of the world total. The absolute outputs of some countries are declining: Nigeria, Bulgaria and the countries of the former Soviet Union (Belarus, Russia and Ukraine).
These tables may please some winners and prompt the losers to try harder, but do they matter? Do they reflect the true situation, based as they are on a single database whose selection of journals is dominated by Anglophone mainstream scientific attitudes? Would not medical advances be better communicated in local journals and languages if they are to influence health-care provision?
The significance of the tables is that they are now used increasingly by politicians and senior officials to vindicate their policies or to argue for an increase in the funding of research because of an apparent shortfall in output.
In biomedicine, the justification for research expenditure in sparsely populated countries is not so much the expectation that great discoveries will be made, but rather that the national research capability will enable rapid assimilation of advances made elsewhere. These cannot simply be practised after reading papers. Hands-on experience of the techniques is usually needed.
Do these results reflect the true situation? This is always hard to answer because perspectives on science depend a lot on one's viewpoint. With regard to growth, it is clear from most anecdotal accounts that science in the countries of the former Soviet Union is mostly in retreat. On the other hand, the increase in output in East Asia reflects the substantially increased commitments to science in these countries, which see it as an important ingredient of national economic growth.
The one surprise from the analysis is the low growth in output from the US in view of its buoyant pharmaceuticals and biotechnology industries and the greatly increased resources made available recently to the National Institutes of Health, although the latter may not yet have been translated into more publications.
Another factor stimulating growth in output, as measured in the SCI, is the increasing tendency for scientists to be judged by peer-view committees for whom publications in prestigious English-language journals are the normal unit of currency. A number of countries are reforming their R&D support system to provide more competition in this way in the hope of raising standards.
Output is also being affected by the rise in international collaboration. Normally, this takes place with English as a common language, and with a goal of publishing in an international journal. As a result, papers recorded in the SCI are being used more internationally as a measure of scientists' production, which in turn influences their publication practices.
The counts of SCI biomedical papers still betray the biases inherent in the selection of journals, but their trends will increasingly signify the effects of changes in the underlying research process taking place. They provide a window on the world of biomedical research, but their limitations must also be understood.
Grant Lewison is senior policy analyst at the Wellcome Trust, London.