Let the divining ROD do it

Graham Dawson and Grant Lewison describe how their Research Outputs Database can show who is funding what where. Two recent articles in The THES have drawn attention to sources of research funding that are often insufficiently appreciated.

"Fair exchange" and its accompanying vignettes of three individual universities (Edinburgh, Liverpool and Middlesex) on June 21 described how some universities were benefiting from European Union schemes. On May 3, "Oxbridge wins donation race" provided data on how universities and Oxbridge colleges were fund-raising individually, albeit on a far smaller scale than many United States colleges.

United Kingdom universities have shown a remarkable ability to obtain research funding from a variety of sources in the face of declining support from government.

But there is little data on the extent to which this support has generated outputs, and which sources are the most productive of scientific benefit.

Similarly, it can be extremely difficult for any individual funding body to determine what its grants have achieved and how it compares with other sources of finance.

The Wellcome Trust, as a large source of biomedical research funding for UK universities, was also in this position. It decided in 1993 to set up a database of biomedical research papers in which the sources of funding for each one were determined by inspection of the original paper and recorded for analysis purposes.

The Research Outputs Database (ROD) comprises seven years of data running from 1988 to 1994 and records about 25,000 papers per year. Since the data is of use not only to the trust but also to other funding bodies and to research performers such as universities and hospitals, the project was set up as a "club" with external members such as the Medical Research Council and other medical charities.

For the first time it became possible to investigate in detail who is funding research and generating research papers in particular places (each paper in the ROD has its correct UK postcode) or on particular subjects.

To investigate the latter, we have defined some 20 or so subfields of biomedicine, such as genetics of renal medicine, by means of "filters" which selectively select papers both from relevant specialist journals and in other journals if they have certain title keywords.

This is important: in most subfields, two thirds of the papers are not in relevant specialist journals and they are usually the more influential ones, at least as measured by citations.

The identification of the postcode area (the first one or two letters of the postcode, for example, G for Glasgow, BS for Bristol) in the addresses on the papers allows maps to be prepared showing the geographical distribution of outputs, either by scientific subfield or by funding body.

We analysed the UK's 120 postcode areas according to the numbers of biomedical papers funded by the European Union during the seven-year period.

Three areas stand out with over 8 per cent of the EU-funded UK total: OX (Oxford), with almost 12 per cent; WC (London, mostly University College London) and CB (Cambridge). EH (Edinburgh), which was highlighted in The THES feature, is in the next group with almost 6 per cent of EU-funded UK biomedical papers, as is NR (Norwich), but the latter's presence is not primarily due to the University of East Anglia but to the John Innes laboratory of the Biotechnology and Biological Sciences Research Council.

Overall, UK EU-funded biomedical papers have been increasing at an annual rate of 24 per cent in recent years. For some postcode areas the growth has been much stronger even than this, notably: WC: plus 45 per cent; RG (Reading): plus 50 per cent; and LE (Leicester): plus 70 per cent).

We also ranked the 21 leading postcode areas in terms of EU-funded biomedical papers with their annual growth rates, and the ratio of the percentages of EU-funded papers to all biomedical papers, which represents the amount of EU support relative to their biomedical research capability. Postcode areas for which this ratio exceeds one are doing relatively well out of EU programmes.

The high ratios for NR and RG reflect the emphasis in these programmes on research relevant to agriculture, such as BIOTECH (the third EU biotechnology programme), ECLAIR (agro-industrial research) and FLAIR (food-linked agro-industrial research).

Until recently the medical programmes have been largely coordinated (the Commission has only paid for travel and administration, not research costs) rather than cost-shared (in universities, marginal costs and a small allowance for overheads are paid).

Consequently the big London medical schools in SE (Kings, UMDS); EC (Bart's); SW (St George's) and W (Imperial/ St Mary's/ Charing Cross & Westminster/RPMS) have not been able to obtain so much EU funding for their work.

However, WC (including University College London) has been relatively more successful as it works in many areas of science outside clinical medicine.

Sources of funding in the ROD are categorised by type as well as by name; this allows aggregation for purposes of analysis. The types include charities, foundations, government agencies, industry and "mixed", which we use for academic bodies that are both endowed from past beneficiaries and currently raising money as described in The THES article.

For the leading 21 postcode areas, we calculated how many and what proportion of their university biomedical papers are funded by the local university using its private funds.

Cambridge is easily top in terms of papers funded this way (215) but Reading (52 papers) has the highest percentage (nearly 8 per cent overall).

To put the Cambridge University figure in perspective, there were 1,099 papers from the university funded by UK charities during the seven-year period and 1,037 funded by the Wellcome Trust.

Surprisingly, the University of Oxford provides support for fewer papers than do the universities of Edinburgh and Glasgow, although the number has doubled over the seven-year period, as it has at Cambridge and Edinburgh.

Relative to their output, the Scottish universities in the five main postcode areas (Aberdeen, Dundee, Edinburgh, Glasgow and KY (St Andrews)) give a lot of support to their own biomedical research activity and are all in the top ten UK postcode areas in this respect.

In Oxford and Cambridge the individual colleges have separate endowments and some of them support fellowships or studentships in biomedical research.

In terms of biomedical outputs for both universities, where papers acknowledging either a college or the university have been identified, Green College, Oxford, a recent foundation, emerges as the leading college in terms of support for biomedical research.

The colleges in total supported 136 papers, more than the university, although this may change in the future as a result of the university's successful appeal for funds.

In Cambridge, St John's College is seen to be easily the leader, but all the colleges together supported only 149 papers, many fewer than were funded by the university.

This shows how some of the funding sources described in recent press articles have led to research outputs. However, there are many other potentially fascinating analyses that can be performed using the ROD, such as the role of industry in funding academic research and that of locally-based hospital charities in the support of medical research within the NHS.

ROD provides a new dimension in the use of analysis of research outputs to inform and guide research policy and is expected to provide a wealth of data for this purpose in coming years.

Graham Dawson is research assistant, Grant Lewison is senior policy analyst, Policy Research in Science and Medicine (Prism), the Wellcome Trust.