European Non-Communicable Respiratory Disease Research, 2002-13: Bibliometric Study of Outputs and Funding

This study was conducted in order to map European research in chronic respiratory diseases (CRDs). It was intended to assist the European Commission and other research funders to identify gaps and overlaps in their portfolios, and to suggest ways in which they could improve the effectiveness of their support and increase the impact of the research on patient care and on the reduction of the incidence of the CRDs. Articles and reviews were identified in the Web of Science on research in six non-communicable respiratory diseases that were published in 2002–13 from 31 European countries. They represented only 0.8% of biomedical research output but these diseases accounted for 4.7% of the European disease burden, as measured by Disability-Adjusted Life Years (DALYs), so the sub-field is seriously under-researched. Europe is prominent in the sub-field and published 56% of the world total, with the UK the most productive and publishing more than France and Italy, the next two countries, combined. Asthma and Chronic Obstructive Pulmonary Disease (COPD) were the diseases with the most publications and the highest citation rates. They also received the most funding, with around two acknowledgments per paper (in 2009–13), whereas cystic fibrosis and emphysema averaged only one. Just over 37% of papers had no specific funding and depended on institutional support from universities and hospitals.


Introduction
Chronic Respiratory Diseases (CRDs) include a wide range of non-communicable conditionssuch as Chronic Obstructive Pulmonary Disease (COPD), emphysema, allergic rhinitis, asthma, pulmonary arterial hypertension and cystic fibrosis with pulmonary manifestations-which have the common characteristic of adversely affecting patient airways and lung structures [1]. They are usually distinct from communicable ones, such as tuberculosis and influenza.
Archivos de Bronconeumologia, from 1970 to 2000 with particular reference to research on smoking, looked at the bibliometric characteristics of the articles [28] but did not find any striking results. A more comprehensive study of all respiratory disease research, including that on lung cancer, showed that Finland, Canada, Spain and the UK had the greatest relative commitment to respiratory disease research expressed as a ratio of their share of world biomedical research, and that charitable funding helped the UK to score well in CF research [29]. Recently, there has been a study on respiratory research outputs from China, Hong Kong and Taiwan; the former has now overtaken both of the others in output but its research remains less well cited [30].
The research described here was carried out as part of an investigation for the European Union (EU) on the outputs of research on five non-communicable diseases (NCDs) in Europe. This was intended to reveal gaps and overlaps, and to learn more about the funding sources for this research. [The other four NCDs are cancer, cardiovascular disease (including stroke), diabetes and mental disorders, and the findings of the project on these will be published separately.] Funding for research on CRDs, as for other medical conditions, comes from four main sources: national and regional government, private-non-profit organisations (including collecting charities, endowed foundations, hospitals and universities, and voluntary non-profit associations), industry and international bodies such as the EU. This project was intended to assist them to understand the research environment in Europe, and to suggest ways in which they could be more effective in their support of research and of activities that would promote its transfer to the care of patients and public health campaigns to prevent illness through changes in behaviour. It was also intended to inform the researchers on the areas of research in need of more attention, and of which countries could provide relevant expertise to those ones wishing to improve their capability. The researchers could also use the findings to seek more funding from external funders and from internal sources.

Methodology
The bibliographic details of papers recorded in the Web of Science published by Thomson Reuters (WoS) were identified by means of a special filter. This was based on title words for the names of the respiratory diseases separated with Boolean operator "OR" ("asthma OR bronchiectasis OR cystic fibrosis OR CFTR OR COPD OR chronic obstructive respiratory disease OR emphysema OR mucoviscidosis"), and four specialist journals: "COPD-Journal of Chronic Obstructive Pulmonary Disease OR International Journal of Chronic Obstructive Pulmonary Disease OR Journal of Asthma OR Journal of Cystic Fibrosis". Papers were selected if they had one of the title words, or were in one of the specialist journals, or both. The WoS software combines the results of separate search statements so that duplicates are eliminated.
"Articles" and "reviews" (as defined in the WoS) from the 12 years, 2002-13, with an address in one or more of the 28 European Union Member States, plus Iceland, Norway and Switzerland, were downloaded to a series of files, 500 at a time. This time frame allowed an adequate time period for any recent trends to be observed. The list of countries is shown in Table 1, with their digraph International Standards Organisation (ISO2) codes. They were then transferred to an Excel spreadsheet by means of a macro written by Philip Roe of Evaluametrics Ltd, St Albans, UK. The addresses were parsed to show the fractional count of each country in each paper. (A paper with one German and two French addresses would be counted 0.33 for Germany and 0.67 for France.) Five-year citation counts were determined for the papers published in 2002-09 to allow fair comparisons between later and earlier papers since previous research in the domain by the authors have used the same fixed window of time.
However, the impact of medical research can be better gauged from its influence on clinical practice. This is difficult to judge, but a proxy indicator is the extent to which it has formed the evidence base of clinical guidelines, which are being developed and published in increasing numbers. As part of the EU mapping project, we collected these from 19 different countries, numbering 45 in total, and matched their cited references to papers in the WoS. We thereby compiled a database of 7184 references, and could compare the numbers from each European country, and their percentage of the European total, with the presence of each country in respiratory disease research. This was done on a fractional count basis, both for asthma and for COPD. Some papers would have been cited on many different guidelines; this would indicate their importance for guiding clinical practice in Europe.
Comparisons were made between the outputs of the 31 European countries in respiratory disease research (RESPI) and their wealth, as measured by their Gross Domestic Products (GDP). Several previous studies [31,32] have demonstrated that there is a close correlation between these two indicators; they are normally plotted on log-log scales because of their wide variation in values (several orders of magnitude). The best (least-squares) regression line to fit the data is based on a power-law, and appears as a straight line on log-log paper. Departures from this line can then show which countries are performing particularly well or badly, with observed outputs being compared with those expected on a Poisson distribution with one degree of freedom.
Country relative commitments (RC) to RESPI research were compared with their biomedical research output in the same years. This was based on a special address filter that was originally developed to distinguish between biomedical and other papers in multi-disciplinary journals such as Nature and Science. These were based on integer counts. For research on the individual diseases, comparisons were made with the European average, and were based on fractional counts.
Details of the funding sources were obtained from the WoS for papers from 2009-13. The list of funders given in the WoS included many false positives where companies had remunerated authors for unrelated work declared in a statement of Conflict of Interest [33]. These statements were first identified by means of another macro, and then individually read to redact the list of explicit funders. There were also implicit funders taken from the addresses-government research laboratories, charity laboratories, and commercial companies. Papers without either type of funding would have received institutional support from a university or a hospital.
Because the names of the funders were not standardised, it was necessary to give them codes so that they could be individually identified. These were in three parts: a trigraph to identify the funder, e.g., MRC = UK Medical Research Council; a digraph to show their sector and subsector (see Table 2); and their ISO digraph code. The thesaurus contained almost 12,000 individual funders, but small funding sources, of which there were very many, were given "generic" codes that simply identified their sub-sector and country. All codes were manually applied with the aid of two thesauruses which were developed from previous work; one of funders and the other of author addresses. Finally, we carried out a survey of some leading researchers in several European countries with different levels of income per caput in order to try and determine the average cost of a published paper in the five NCDs. We asked the researchers for their total annual research budgets during the last five years (2009-13) and compared their responses with their outputs of papers, fractionated according to the numbers of addresses. These results were then compared with results of earlier surveys carried out with the same methodology [34,35] that had been used to determine the cost of a biomedical research paper in other subject areas.

Disease burden
In order to put our research results in context, we sought data on the burden from these noncommunicable respiratory diseases in Europe. Data on the 31 European countries were downloaded for the year 2010 from the IMHE website and revealed that these diseases accounted, on average, for 4.7% of all Disability-Adjusted Life Years (DALYs), including 2.9% for COPD and 1.1% for asthma.
The countries suffering most from asthma are the UK (1.8% of DALYs), Ireland (1.7%), Portugal, Cyprus and Sweden (1.6%). But the "accession" countries in eastern Europe suffer much less, with the lowest burdens being in Lithuania, Bulgaria and Latvia (about 0.4%). In COPD the relative burden is greatest in Denmark (4.7%), Switzerland (4.5%) and the UK (4.2%), and least in the Baltic countries: Estonia (1.2%), Latvia (1.4%), and Lithuania and Finland (2.0%). Burden of disease data are not given by the IHME for cystic fibrosis, but surveys in European countries have been made of the incidence of the disease and reported by Farell [36]. The results for the EU27 Member States are shown in Fig 1 except for the seven countries with populations below 4 million. They can be compared with the results of surveys of the genetic mutation (DF508) that is mainly responsible for the disease [37]. Both surveys put Ireland in first place; the UK and Italy also have a high incidence of CF. However, it is much lower in eastern Europe (Poland, Bulgaria and Romania) and in the Baltic countries (data for Latvia and Lithuania not shown in Fig 1,

Respiratory research outputs and relative commitments
In the 12 years, 2002-13, there were 18,822 European RESPI papers in the WoS, representing 56% of the world total of 33,629. The European total was only 0.8% of their biomedical output of 2,442,063 papers in 2002-13. The European output increased by 75%, and the international fraction (i.e., the difference between the integer total for the 31 countries and the fractional total, divided by the former) doubled from 6.2% to 12.4%. These percentages are similar to those for the other NCDs.  Table 3 on an integer count basis. Values more than sqrt(2), less than sqrt (0.5) or less than 0.5 are shown in a different font. All of the differences between observed values and those expected on the basis of the European average are very highly significant on the Poisson distribution with one d/f except for ES, GR, and PT where p < 0.1%; for PL and IE where p < 5%, and FR, IT, FI and NO which are not statistically significant.
Ten of these 18 countries have a relative commitment to respiratory research greater than 1.41: this means that there are many non-European countries with a much lower RC to the subject area. But only two have their output in respiratory disease research above 1% of biomedical research (Sweden and the Netherlands) and their disease burden is 5% or higher, as is  Table 1. The European countries varied in their concentration on the different respiratory diseases. Data are given in Table 4 only for the three main ones: asthma (AST) with 7563 papers (40% of the total of 18,822 RESPI papers), COPD (COP) with 4763 papers (25%) and cystic fibrosis (CYF) with 3281 (17%). The countries that concentrate most on asthma are Finland and Poland, with Ireland doing relatively little. In COPD, the leaders are Spain, Greece, the Netherlands and Norway in terms of relative commitment, and in cystic fibrosis Ireland stands out, which is appropriate, given its very high incidence of the disease (see Fig 2), together with Portugal and France. Finland does very little CF research, which also accords with its low incidence of the disease.

Citation scores
European RESPI papers were slightly more highly cited than the world average. The number of citations received in the five years following publication (Actual Citation Impact, ACI) increased from 14. 3. Within Europe, the best-cited papers were those from the UK, see Table 5, which gives the mean ACI values and the percentages of the countries' papers with enough five-year citations (52) to put them in the top 5% of European papers, both on a fractional count basis. On this basis, it is not possible to give values for the s.e.m. as the citation counts are each multiplied by the fractional presence of each country in the paper.
A country's presence in the top 5% of papers is a rather more sensitive measure of impact than mean citation score (ACI) as these are the important papers that are likely to influence research and treatment. Papers on COPD were more highly cited than those in asthma and cystic fibrosis, see Table 6.

Citations on clinical guidelines
Of the 7184 references, 3744 were to the 23 clinical guidelines on COPD and 3052 were to the 20 on asthma. Table 7 shows the numbers of references (on a fractional count basis) from the 16 leading countries in each disease area, and for comparison, the numbers expected from the countries' fractional count presence in COPD and asthma research, see Table 4. The countries whose research contributes most to the evidence base of European clinical guidelines, relative to its volume, are the UK and Denmark for both asthma and COPD. In

Funding of respiratory disease research
The analysis of funding, both explicit acknowledgments and ones implicit from the addresses, shows that RESPI research does not attract much specific support, and over 37% of 2009-13 papers do not have it. The mean number of funders per paper overall was 1.98, and one paper had over 100, but there was considerable variation between countries, see Fig 3. The countries whose papers obtained the most funding, often given to their partners in international collaborations, were three Scandinavian ones (Finland, Sweden and Norway), and Denmark was in fifth place in this presentation. [Iceland would have been easily first, but published only 16 papers in 2009-13 on a fractional count basis.] Central and Eastern European countries attracted little funding except from their own governments, though Hungary was an exception, probably because of extensive international collaboration. Croatia (HR), in southeast Europe, was relatively scientifically isolated, and 30 of its 49 papers acknowledged no specific funding source. An analysis was also conducted of the leading funders in terms of fractional credit, both overall and for the leading countries, and of their sectors (government, private-non-profit, industrial and international). The results for the leading funders are in Table 8, and the chart showing how the leading countries' support is provided is in Fig 4. Estimates of annual expenditure are based on our estimates of the mean cost of a paper (see below). However, the expenditure by pharma companies is likely to be far higher as they will be carrying out intramural work that does not lead to open publications.
In this analysis, it has been assumed that national public and private-non-profit funders only support researchers in their own country, that European Union funding goes to the European countries whose addresses are on the paper in proportion to their number, and that industrial funding can go to any country. It is striking that pharmaceutical companies appear so prominently in this list, with six in the top 18.  Support from the European Union (EU) comes from a variety of sources, mostly but not exclusively the Framework Programmes. Table 9 shows the countries that receive the most support (fractional count of papers), and also the ones where the EU support is proportionately the greatest (Latvia, Slovakia and the Czech Republic).
This table reveals clearly that the large countries in scientific output received by far the bulk of EU support (the top seven obtained almost two thirds of the total), but some of the smaller countries, notably Latvia, Slovakia and the Czech Republic, depended highly on the EU for their research on RESPI.
Our survey of leading researchers elicited some 30 responses, of which 22 gave useable data (some suggested unrealistically high or low costs for a paper). As expected, the responses from high-income countries (Norway and Switzerland) gave a figure (€412 k) higher than that from the researchers in middle-income countries (Belgium and Finland; €260 k), and much higher than that from respondents in low-income countries (Bulgaria, Czech Republic, Romania, Slovakia and Slovenia; €142 k). These mean costs parallel the differences in income per caput between the groups, but are less divergent. Since the large majority (about 85%) of all RESPI papers came from middle-income countries (mainly France, Germany, Italy and the UK), the overall average cost per paper worked out at close to their figure, at €255k. (This was also similar to results found in earlier surveys, adjusted for inflation [34,35].) This means that the average annual research expenditure on respiratory diseases in Europe during these years was of the order of €407 million, barely one fifteenth of that on cancer research (€5977 million p.a.). Since the burden from respiratory diseases in Europe is just over one quarter of that from cancer, it is apparent that respiratory disease research is relatively under-funded. If it were to  receive as much research money as cancer in relation to its disease burden, it would need an additional €1225 million per year, or three times its existing research budget.

Discussion
This study has mapped the main features of research into non-communicable respiratory diseases in Europe in 2002-13, the outputs of the 31 countries and their citation scores, and the funding sources for papers in the last five years . The study has some limitations. We were not able to check how comprehensive the filter was at capturing relevant papers, and indeed we expect that it will have omitted some basic research papers where the title or the journal did not directly indicate the name of the disease. However, this limitation affects other disease-related research topics. The restriction of the database to articles and reviews is standard bibliometric practice, although it is arguable that other document types such as editorials, letters and meeting abstracts sometimes contain interesting results, and are indeed occasionally cited among the references on clinical guidelines. The large number of funding bodies acknowledged on the Scandinavian papers is attributable to the many small (and some large) endowed foundations in these countries. In Finland, private-non-profit sources accounted for over half the total funding (54%), and most of these were endowed foundations. (So far, we have encountered over 50 different named Finnish foundations, although not all of them support respiratory disease research.) In the former socialist countries of central and eastern Europe, although there are some private-non-profit funders, notably professional associations, there does not appear to be a tradition of public charitable funding of medical research, or the creation of foundations by wealthy families and individuals. The fiscal regime may also not be so favourable to such activities as it is in western Europe (and north America).
Of the main respiratory diseases, asthma is better established as a research subject than COPD; its integer count output from the EUR31 countries was 8489 papers in 2002-13, or 0.35% of all biomedical research output. This is still much lower than its disease burden of 1.07%, but the ratio is more favourable than for COPD which had 5286 papers or 0.22% compared with a burden of 2.95%, an adverse ratio of 13.6 compared with 3.1 for asthma.
Why should this be? "COPD mainly affects people over the age of 40 and becomes more common with increasing age. The average age when it is formally diagnosed is around 67 years. It is more common in men than women. COPD accounts for more time off work than any other illness and a flare-up (exacerbation) of COPD is one of the most common reasons people seek emergency hospital assistance" [38]. On the other hand, asthma affects people of all ages, and in England one in five sufferers are children, over 1.1 million, or one in 11 children. It is not surprising, therefore, that asthma has a far higher public profile, and there are medical research charities to support asthma research whereas COPD has largely to rely on public funding, plus support from pharma companies. Moreover, because it is so often caused by smoking, it does not attract public sympathy in the same way as asthma.
There are also charities in at least eight European countries (Belgium, France, Germany, Ireland, Italy, Netherlands, Sweden and the UK) that support research in cystic fibrosis, sometimes called mucoviscidosis. Inevitably this disease mainly affects young people as life expectancy for patients is quite low, though it has increased to about 40 years [39] as better treatments have become available.
It appears, therefore, that COPD has an image problem, and does not attract research funding from the European public, who are understandably more concerned with the health of children than with that of old men, many of whom have been smokers. However, COPD is an unpleasant disease, and as the quotation above suggests, it is an expensive one to treat because of the numbers of patients and the difficulty of doing more than provide palliative care. It should therefore have a greater claim on public sources of research funding. There should not be a problem in finding applications of high quality, as COPD research is actually better cited than papers in asthma or CF, as Table 6 shows.
We have not analysed the outputs by the type of research, such as epidemiology, genetics, pharmaceutical and other treatments, and palliative care. This would provide information on which aspects of research may be neglected relative to others and might suggest changes in approach in individual countries. In particular, this would show the types of research that are most frequently cited on clinical guidelines, and so are of medical utility.

Conclusions
The main conclusion is that respiratory disease research is seriously under-funded. The research outputs of the leading European countries vary between 0.4% of all their biomedical research (Austria) and just over 1.0% (Netherlands, Sweden and Denmark), but these values are all very small compared with the European disease burden of 4.7% of all DALYs. If respiratory disease research were to be funded on the same scale as cancer, relative to its disease burden, funding would need to be quadrupled. And within RESPI research, the main need is for more work on COPD, which is even more under-resourced than asthma research but causes more distress and expense.