United Kingdom’s contribution to European research output in biomedical sciences: 2008–2017

Background: On 31 January 2020, the United Kingdom (UK) formally left the European Union (EU). Only a short transition period, until 31 December 2020, is available to negotiate collaborations for research in biomedical sciences and health care. Within the European scientific community, two opinions are common: 1) Brexit is an opportunity to obtain more funding at the expense of the departing British; and 2) UK colleagues should continue to collaborate in EU scientific efforts, including Horizon Europe and Erasmus+. To provide evidence for more informed negotiations, we sought to determine the contribution of the UK to EU’s research in biomedical sciences. Methods: We performed a macro level scientometric analysis to estimate the contribution of the UK and EU member states, including those associated with EU-funding (EU+) namely Albania, Armenia, Bosnia-Herzegovina, Faroe Islands, Georgia, Iceland, Israel, Macedonia, Moldova, Montenegro, Norway, Serbia, Switzerland, Tunisia, Turkey, and Ukraine, to preclinical, clinical and health sciences. We searched the Web of Science database to count the total number of scientific publications and the top 1% most cited publications in the world between 2008 and 2017, calculated the performance efficiency by dividing the top 1% by the total number, and calculated the odds ratios to create a ranking of performance efficiency. We then compared the contribution of the UK to all the EU+ -based publications and the top 1% to the contributions of the ten EU member states with the largest biomedical research output and also compared the respective contributions to EU+ publications that resulted from collaborations with other regions in the world. Results: We found 2,991,016 biomedical publications from EU+ during 2008–2017, of which 19,019 (0.64%) were in the world’s top 1% of the most cited publications. The UK produced 665,467 (22.3%) of these publications and had over two and a half times more top 1% most cited publications than the EU+ (odds ratio 2.79, 95% CI 2.71–2.88, p< 0.001). The UK’s share in the EU+ co-publications with regions outside Europe ranged between 23.0% for the Arab League and 50.6% for Australia and New Zealand and its share of the top 1% ranged between 48.6% for the USA and Canada and 70.7% for the African Union. Conclusions: The UK contributed far more highly cited publications than the rest of the EU+ states and strongly contributed to European collaborations with the rest of the world during 2008–2017. This suggests that if the UK ceases to participate in EU scientific collaborations as a result of Brexit, the quantity and quality of EU’s research in biomedical sciences will be adversely affected.


Introduction
On 31 January 2020, the United Kingdom (UK) left the European Union (EU). However, a transition phase will last until 31 December 2020 during which the future relationship between the UK and the EU will be negotiated. The borders of Northern Ireland and trading relationships are the main topics of the ongoing negotiations; the future of collaborations in biomedical sciences and health care has received little attention. UK Prime Minister Johnson promised a substantial investment (over £18 billion, or approximately €22.6 billion within 5 years) in health and life sciences and, if the assumptions about a boosted post-Brexit economy quickly become a reality, even more. 1 The UK government appears to have allocated £800 million (about €920 million) to a future new research agency, but the size of the investments in biomedical research is not known. 2 The implications of the UK leaving the EU will be significant for both parties in many ways, including scientific research and collaboration. The UK has been the most successful country in obtaining Horizon 2020 grants, receiving €1.3 billion (about £1.1 billion) annually on average, 2

and one in five European Research
Council winners is working in the UK. 4,5 For the call for proposals on Health, demographic change, and well-being under the Horizon 2020 theme Societal Challenges, this translates to 15% of the theme's budget for UK institutions. 6 Closing access to EU funding might result in potentially inadequate funding of UK research. An example in which a number of these things converge is the withdrawal of UK colleagues from European Reference Networks (ERNs). The 24 thematic ERNs are the infrastructure for pan-European collaboration on about 6000 complex or rare diseases severely affecting 30 million people in the EU. The networks connect centres of expertise in specific diseases, health care providers, patients, and laboratories in EU member states to share knowledge and infrastructure and to improve the care and access to care for these patients. The UK's contribution to this infrastructure is substantial: in the UK, 35 hospitals and 129 health care providers are responsible for approximately

United Kingdom's contribution to European research output in biomedical sciences: 2008-2017
23% of the total patients in the 24 ERNs. However, after the coordinators' meeting on 25 June 2018, all six coordinatorships of the ERNs by UK colleagues were transferred to their partners on the Continent, although UK health care providers continue to be involved in 23 of the 24 networks. 7 Continuation of the participation depends on the outcome of Brexit negotiations during the transition period: UK participation might end after a hard or otherwise poorly negotiated Brexit. 8 Support from the Continent to keep UK colleagues in the ERNs is mixed, because some groups may see Brexit as an opportunity for increased funding for groups from countries remaining in the EU.
Taken together, Brexit looks like a political accident that will mainly affect the UK, but it will also have implications for European research. However, the extent of contribution to EU research by UK colleagues is unknown. We therefore sought to determine the contribution of UK scientists to EU research in quantitative terms to support the negotiations during the transition period with hard evidence.

Database
We extracted data from the Web of Science the same way as described by Asubiaro and Badmus 9 for African countries for a single research area; however, our work encompassed not one research area but several, namely preclinical, clinical, and health sciences within the larger domain of biomedical sciences, and we looked for the total number of publications and the world's top 1% most cited publications each year for the decade 2008-2017.
The details of how we searched the Web of Science are given in online Supplement 1.

Indicators
We calculated the indicators for EU member states including those associated with EU funding (EU+), because many important publications during the decade under study were based on FP6, FP7, and Horizon2020 funding, with and without the UK and for the UK separately. Specifically, we counted the total number of publications, the top 1% of the most cited publications, and the number of publications resulting from collaboration with other regions of the world. We calculated the performance efficiency by dividing the top 1% by the total number of publications. Next, we calculated odds ratios and confidence intervals (and p values) to create top-ten rankings: position of the country in total output, in the top 1% most cited publications, position in terms of collaboration with other regions of the world, and position in terms of publications based on EU-funded research. In each ranking, separate columns were assigned to the contributions of the UK (Tables 1 to 3). The share of a country in the publications from EU+ as a whole was calculated as a percentage (both total or top 1% most cited publications of the country divided by the total or top 1% most cited publications of the EU+ without this specific country, multiplied by 100).
Analysis for Tables 1 and 2 was performed on 4 February 2020; analysis for supplemental Table 1 was performed on 11 February 2020.

Results
From 1 January 2008 to 31 December 2017, we found 2,991,016 biomedical publications of institutes in the UK and EU+ in the Web of Science, of which 19,019 belonged to the world's top 1% most cited publications (in the field of research in the year of publication). This corresponds to a performance efficiency of 0.6%. The UK institutions contributed the most, with 665,467 publications, of which 8412 were in the top 1% most cited publications. The UK occupied the 4th position within the EU, with more than two and a half times the publications from EU+ (odds ratio 2.79, 95% confidence interval (CI) 2.71-2.88, p <0.0001; Table 1). Overall, the EU+ without the UK contributed 10,607 of the top 1% most cited publications, with a performance efficiency of 0.5%. The share of the UK in the output from EU+ countries was 22.2% and that in the top 1% of the most cited publications was 44.2% (Table 1).
The UK also led in collaborating with other more productive EU member states: 9.4%-16.3% of all publications from the ten most productive members were from research based on collaboration with the UK, and 43.0%-59.2% of the top publications from the individual countries were in collaboration with scientists affiliated to British institutions ( Table 1). These ten countries and the UK together contributed 79.7% of the biomedical scientific publications from EU+.
Germany, second only to the UK in the total number of publications, contributed 16.8% of the total publications and 27.3% of the top 1% most cited publications, and was followed, in that order, by Italy and France, accounting for, respectively 12.6% and 10.6% of the total publications and 20.0% and 20.9% of the top 1% most cited publications.
Publications from Belgium, Denmark, and the Netherlands featured significantly more often than those from the UK in the top 1% most cited publications (relative to the EU+, the odds ratio was 3.07 for Belgium (95% CI 2.93-3.22), 2.90 for Denmark (95% CI 2.76-3.06); and 2.84 for the Netherlands (95% CI 2.74-2.94). The p value in each case was <0.001 (including those directly comparing the four highest-ranked EU member states).
Since the EU+ region collaborates actively with other regions in the world, we investigated the contribution of the UK and EU+ to the total and top 1% most cited publications resulting from collaborations with institutions outside Europe ( Table 2). The UK's share in the total EU+ co-publications with regions outside Europe ranged from 23.0% with the Arab League to 50.6% with Australia-New Zealand and that in the top 1% from 48.6% with USA-Canada to 70.7% with the African Union ( Table 2).
The above pattern remained more or less unchanged in the analyses restricted to publications arising from EU-funded research. The UK's share was exceptionally high, being 25.1% in the total output and 44.2% in the top 1% and ranged from 31.5% to 55.3% in total publications and from 53.4%-76.2% in the top 1% in terms of output from collaborative research (Table 3).

Discussion
Biomedical research institutions from the UK were key contributors to the EU+ research output, the UK being among the top four countries collaboration with which increases the chances of research papers from such collaborations being highly cited. The UK plays a substantial part in EU+ research collaborations and is less of an island than its geography may suggest when it comes to collaborating with the Continent and other regions of the world.
The UK's share in EU collaborations with other regions in the world ( Table 2) is even larger than its share in EU publications as a whole. These collaborations with other regions in the world reflect the extensive international network of UK universities. The UK government's Science and Innovation Network is based in 31 countries, supporting international collaboration of UK companies and researchers. We believe the possibility to use funding from the UK's research councils-Research Councils UK (RCUK) is the strategic partnership of the UK's seven research councils-for overseas collaborations, especially when setting them up, is a significant advantage, even when the budget is restricted to the UK partner, because the funding sources (EC-funded or national) available to universities from the Continent are too limited to set up such collaborations.
Although the presence of co-authors affiliated to UK institutes does not mean that the research would not have been realised without the UK contribution nor that the quality would have been lower, our results emphasize a fair role for the UK and indicate that science on the Continent might, at least temporarily, suffer from reduced access to high-quality After the Brexit transition period, European science without UK participation might result in top continental scientists receiving more EU funding. However, a more probable outcome is that the money will go to those who failed to obtain funding before Brexit, which may lower the quality of EU-funded research. It is difficult to predict what will happen after the transition period. The UK might experience difficulties in retaining top scientists, especially those not from the UK, and the EU might, for instance, open its research programmes to participation from non-EU countries.
Recently, higher education and research organisations from the UK and the EU have asked their national governments and the European Commission to agree on continued collaboration by keeping the UK associated with Horizon Europe and Erasmus+. 10 The biomedical sciences account for roughly half of all science, and this study provides concrete evidence that collaboration between the UK and EU+ in the biomedical disciplines is essential.
We argue that Brexit should not be a reason for excluding UK researchers from such European consortia as the ERNs. Most of the UK scientists voted against Brexit 11 and protested against it in March 2019. 12 Moreover, our results suggest that science on the continent of Europe is stronger for the participation by UK researchers. From a clinical viewpoint, continued collaboration as before will ensure early access to innovations for patients throughout Europe and the UK. If we cannot keep biomedical research out of the political consequences of Brexit, the weakest people, those with diseases that offer limited options for treatment, will suffer the most, probably on both sides of the new border. Author contributions EJGS initiated the study with initial data and performed the final statistics; TLRT analysed the Web of Science data summarised in the tables; both together discussed the data and wrote the manuscript.   As search strings we used for data shown in Table 1:

Table 3. Publications from EU+ and UK based on EU-funded research and in collaboration with other regions of the world in preclinical, clinical, and health sciences: 2008-2017.
For the row named "EU+" we used WC = (as As search strings we used for data shown in Table 2: As search strings for data shown in supplemental Table 1 we used: