Evolution of scientific collaboration on COVID-19: A bibliometric analysis.

This paper considers the pattens of international collaboration by analysing publications on COVID-19 published in the first 6 months of the pandemic. The data set comprised articles on COVID-19 indexed in the Web of Science Core Collection (WoS CC) downloaded four times between 1 April 2020 and 1 June 2020. The analysis of 5,827 documents revealed that 128 countries, 23,127 authors, and 6,349 institutes published on the pandemic. The data reveal that the three main publishing countries were the USA, China, and England with Italy closely following. Although publication was widely spread, most of the institutions with the highest volume of output were in China. Network analysis showed growth in international cooperation with an average degree of country/region cooperation rising to 23.06 by 1 June. There was also a clear core-periphery structure to international collaboration. Institutional collaboration was shown to be highly regionalized. The data reveal a high and growing incidence of international collaboration on the pandemic.

The US, China, England, and Italy published the most articles on COVID‐19 in the first 6 months, with the US overtaking China by June 2020.

International collaboration on articles about COVID‐19 grew rapidly between April and June 2020.

Institutional collaborations on COVID‐19 articles tend to be localized indicating close research networks.

Network analysis reveals a clear core‐periphery structure of international collaboration on COVID‐19 articles with growing participation of different countries.

INTRODUCTION

Practice has long proved that international cooperation is not only the leading force in the global exploration of cutting‐edge science but also the best way for the world to respond to issues such as resource and environment, climate change, health, and public safety (Adams, 2013; Adams & Loach, 2015; Choi et al., 2015; Freeman, 2010; Narin et al., 1991; Wagner et al., 2019). It took only 6 months from the discovery of the Novel Coronavirus (COVID‐19) to more than 6 million confirmed cases and 300,000 deaths, which not only proves that the COVID‐19 is too contagious to be overcome but also demonstrates the common destiny of all countries and regions in the era of globalization (Nature Editorial, 2020c; Washington, 2020). In fact, when this outbreak was declared as a Public Health Emergency of International Concern (PHEIC) on 30 January 2020 by the WHO, it was already indicated that international cooperation is the key to combating this pandemic (Berkley, 2020; Duan et al., 2020; Nature Editorial, 2020a, 2020b; Nature Medicine Editorial, 2020).

International scientific collaboration, an important part of international cooperation, has been given growing attention in innovation economics (Andersen, 2019; Bauder et al., 2018; Cassi et al., 2012, 2015; Gui et al., 2018a, 2018b; Wuestman et al., 2019), S&T policy (Chen et al., 2019; Fung & Wong, 2017; Gazni et al., 2012; Hou et al., 2008; Sun & Cao, 2020), and knowledge production and technology transfer (Aldridge & Audretsch, 2011; Ankrah & Al‐Tabbaa, 2015; Bekkers & Freitas, 2008). Increasingly common and frequent knowledge flows crossing borders not only speed up the process of scientific globalization but also constantly re‐shape the global scientific landscape (Adams, 2013; Adams & Loach, 2015; Royal Society, 2011). International scientific collaboration is the key support of national competitiveness (Bathelt & Henn, 2014; Freeman, 2010). In the era of pandemic, cooperation in virus research is and important win‐win for participating countries/regions. While improving the scientific research capacity, international cooperation also strengthens the capacity in pandemic prevention and control for each country and region (Nature Editorial, 2020c). In the past 5 months, researchers around the world have conducted a large number of in‐depth studies on the structural morphology, gene sequence, pathogenic mechanism, diffusion mode, etc. of the COVID‐19 virus, giving us a gradually clearer understanding of the virus and how to prevent and control the epidemic (Corey et al., 2020; Guan et al., 2020; Tian et al., 2020; Wu et al., 2020; Zhu et al., 2020). Within this are influential achievements jointly completed by researchers from multiple countries and institutions (Drew et al., 2020; Tian et al., 2020).

By exploring scientific collaboration among countries/regions and among institutes on COVID‐19, this paper aims to answer the following two questions: (1) what is the structure of the international scientific collaboration network and the inter‐institution collaboration network on COVID‐19 research? (2) Who are the major contributing countries/regions and institutions participating in the scientific collaboration? The main contributions of this paper are twofold. Firstly, this paper seeks to enrich the literature on scientific collaboration through sorting out the relevant research about COVID‐19. Specifically, it intends to test whether international scientific collaboration on COVID‐19 is consistent with the existing findings on the structure of global scientific cooperation. It also tries to deepen our understanding of international collaboration in virus research.

METHODOLOGY AND DATA

Data

Although widely being criticized for its limitations (Cantner & Rake, 2014; Royal Society, 2011), co‐publication is still one of the best ways to characterize scientific collaboration between authors, between countries/regions or between organizations (Basu & Kumar, 2000; Gui, Liu, & Du, 2019; Gui, Liu, Du, et al., 2019; He, 2009; Lemarchand, 2012; Liu & Gui, 2016; Sun & Cao, 2020; Sun & Grimes, 2016). The publications data analysed here was retrieved from Web of Science Core Collection (WoS CC), by adopting the full counting method (full credit to a country/institutes when at least one of the authors is from this country/institutes) to count the scientific collaborations among countries/regions or among institutes (Gauffriau & Larsen, 2005).

To clearly describe the development of scientific cooperation in the research of COVID‐19, we counted all related publications (articles, reviews, letters and so on) collected on April 1, and collected new publications every half month thereafter. As of June 1, we had collected publications about COVID‐19 at five points in time, which are April 1, April 15, May 1, May 15, and June 1. In addition, due to the difference in the initial naming of the new coronavirus, the publications search was sequentially retrieved through four topic words: novel coronavirus, SARS‐CoV‐2, 2019‐nCoV, and COVID‐19. All publications were published in 2020, and each search was conducted cumulatively, not discretely. The detailed description is as follows. On the Web of Science literature search page, we first selected WoS CC as the search database. Secondly, we selected the advanced search strategy, and use field identifiers and Boolean operators to create the search query, specifically, TS (topic) = novel coronavirus or TS = SARS‐CoV‐2 or TS = 2019‐nCoV or TS=COVID‐19. Thirdly, we selected the literature data published in 2020 in the search results. We repeated the above three‐step search method at five points in time to obtain the accumulated data at each point of time. To understand the changes between every two points in time, by deleting the duplicated part of the data collected at the later point of time, we obtained the newly added data during every time period.

Bibliometric tools

In this article, the bibliometric method is used to analyse the scientific cooperation on COVID‐19. In the process, two kinds of software were used: VOSviewer and ArcGIS. VOSviewer is a software tool for constructing and visualizing bibliometric networks which can be constructed based on citation, bibliographic coupling, co‐citation, or co‐authorship relations (Perianes‐Rodriguez et al., 2016; Van Eck et al., 2010; Van Eck & Waltman, 2010). ESRI's ArcGIS is a geographic information system for processing maps and geographic information. Its ArcMap product can be used to display and analyse the geographic structure of the cooperative network among authors, institutions, cities, and countries (Gui, Liu, & Du, 2019; Gui, Liu, Du, et al., 2019; Liu & Gui, 2016).

By integrating these two kinds of software, we analysed scientific cooperation around COVID‐19 research both at national level and institute level. Specifically, we first used the VOSviewer to analyse the bibliographic data downloaded from WOS CC, drawing the scientific cooperation network among institutes or among countries/regions, obtaining the list of participating institutes or countries/regions, and the cooperation matrix between institutes or between countries/regions. Second, we used GPS Visualizer's Address Locator (www.gpsvisualizer.com/geocoder/) to geocode all participating institutes or countries/regions. Third, we imported the cooperation matrix with geographic information into ArcMap to analyse the geographical structure of scientific cooperation among institutes or among countries/regions.

Network analysis

Network analysis is a powerful tool to reveal the structural characteristics of a scientific cooperation network (Gui, Liu, & Du, 2019; Gui, Liu, Du, et al., 2019). In this article, network analysis was applied to measure the structural characteristics of the scientific cooperation network on COVID‐19. Specifically, the number of nodes and edges indicates the size of the network, that is, the number of countries/regions, institutes, or authors participating in cooperation. Density and average degree measure the cohesion of the network. The average clustering coefficient and the average path length are measures of the small world network (Watts & Strogatz, 1998). In addition, we also applied block modelling in network analysis to study the core‐peripheral structure of the international cooperation network on COVID‐19. The significant core‐peripheral characteristics of the world economic system have been widely proven (Nemeth & Smith, 1985; Smith & White, 1992), and the core‐peripheral structure of the global scientific cooperation network have also been discussed many times (Gui, Liu, & Du, 2019; Gui, Liu, Du, et al., 2019). We used the PAJEK program for block modelling (Waltman et al., 2010), which is a program for network analysis and visualization.

RESULTS

Descriptive analysis

We are interested in the distribution of publications by countries/regions, institutes and authors, and the leading contributing economies and institutes participating in scientific cooperation on COVID‐19. Table 1 shows the descriptive statistics of the main indicators. During the 2‐month observation from April 1 to June 1, the number of articles about COVID‐19 published worldwide grew rapidly, from 808 as of April 1 to 5,827 as of June 1. The number of countries/regions and institutes participating in the research (sourced from author affiliations) also increased from 62 and 851 as of April 1 to 128 and 6,349 as of June 1, respectively. Cooperation is particularly evident in COVID‐19 research. Most of the countries/regions, institutes and authors involved in the research have cooperated with others to some degree.

TABLE 1

Descriptive statistic of publications and collaborations about COVID‐19.

	As of April 1	April 2–15	April 16–May 1	May 2–15	May 16–June 1	As of June 1
In terms of publication
Number of documents	808	457	878	1,493	2,191	5,827
Number of countries/regions	62	68	66	93	103	128
Number of institutes	851	1,044	2,160	2,378	3,241	6,349
Number of Authors	3,029	2,787	4,021	6,433	9,736	23,127
In terms of collaboration
Number of countries/regions participating in scientific collaboration	60	62	57	83	96	122
collaborations among countries/regions	537	642	947	1,614	2,143	5,886
Number of institutes participating in scientific collaboration	801	950	1,760	2,190	2,960	5,879
collaborations among institutes	2,995	4,199	6,420	11,145	15,602	40,384
Number of authors participating in scientific collaboration	2,976	2,547	3,614	6,142	11,245	21,014
collaborations among authors	21,176	27,786	30,561	36,166	81,739	197,428

Note: The data in the table are de‐duplicated. Institution data are matched by country and institution name, and author data is matched by institution and author name.

The growth of COVID‐19 studies

Despite the increasing number of countries/regions participating in the research, publications on COVID‐19 were highly concentrated in a few countries/regions. China, the US, and England have consistently ranked among the top three in terms of cumulative publications. China was originally leading in terms of publication volume, indicating that China's leading research work laid a solid knowledge base for the world's knowledge of COVID‐19. With the development of the pandemic, the US became prominent as a global scientific centre. As of June 1, the US had surpassed China in the number of publications, reaching 1,389. China ranks second with 1,295 publications, and England ranks third with 616 publications. In addition, Italy, Canada, India, Germany, Australia, and France also have published a large amount of literature on COVID‐19 (Table 2).

TABLE 2

Number of documents published by main countries/regions at five points in time.

Country/region	As of April 1	As of April 15	As of May 1	As of May 15	As of June 1
US	118	248	442	810	1,389
China	246	460	650	934	1,295
England	41	108	196	358	616
Italy	23	68	164	345	599
Canada	29	53	78	150	262
India	12	35	66	165	252
Germany	31	55	86	151	245
Australia	21	45	77	136	242
France	18	29	40	106	202
Iran	6	25	92	125	177
Switzerland	16	38	65	108	151
Spain	6	14	24	59	141
Singapore	12	39	47	83	139
Brazil	8	16	36	68	118
Netherlands	14	26	33	61	102
Japan	20	33	35	58	99
South Korea	27	42	61	75	97
Turkey	0	4	14	55	96
Saudi Arabia	26	32	41	45	68
Chinese Taiwan	11	14	21	43	51

Similarly, the publication pattern of COVID‐19 at the institute‐level also showed a high uneven degree of concentration (Table 3), that is, most institutes only published one document, and the number of institutes publishing more than 20 documents is only 86 as of June 1. Institutes from China have the highest volume of scholarly output on COVID‐19 research. According to the literature statistics as of April 1, 17 of the top 20 institutes in terms of publications were from China. The CAS, HKU, and HUST ranked among the top three with 27, 21 and 18 publications, respectively. As of June 1, although the number of Chinese institutes in the top 20 decreased to 10, 4 of the top 5 came from China. HUST, WU, and HKU ranked first, second, and third with 143, 102, and 81 documents, respectively. Moreover, institutes from the US, England, Canada, Italy, Iran, Australia also played an important role in COVID‐19 research.

TABLE 3

Top 20 institutes with the most publications on COVID‐19 at five points in time.

As of April 1		As of April 15		As of May 1		As of May 15		As of June 1
Ins.	Articles	Ins.	Articles	Ins.	Articles	Ins.	Articles	Ins.	Articles
CAS	27	HUST	38	HUST	63	HUST	101	HUST	143
HKU	21	HKU	38	HKU	44	WU	75	WU	102
HUST	18	CAS	35	WU	44	HKU	59	HKU	81
FU	15	FU	33	CAS	37	ZJU	54	ZJU	76
CMU	14	WU	28	ZJU	37	FU	50	HMS	71
ZJU	14	ZJU	25	FU	35	CUHK	47	FU	66
WU	13	CMU	21	CMU	34	CMU	44	UT	65
CUHK	11	SYSU	21	UTMS	32	HMS	44	OU	63
GMU	11	CUHK	20	CAMS	29	OU	44	CUHK	62
SCAU	11	CAMS	19	SYSU	27	UTMS	41	UoM	62
SYSU	11	SJTU	18	SJTU	26	UT	41	CMU	58
UoS	11	SCU	18	OU	26	CAS	40	UCL	58
CAAS	10	LSHTM	17	UCL	25	CAMS	39	UTMS	54
HU	10	GMU	16	CUHK	24	SJTU	39	NUS	53
SJTU	10	PU	16	HMS	24	UCL	36	CAMS	52
SCU	10	TSU	15	PU	22	PU	33	CAS	52
CUMB	9	UoS	15	SCU	22	UoM	32	SJTU	51
HZAU	9	UCL	14	SBUMS	21	SCU	31	UMG	51
U. CAS	9	OU	14	LSHTM	20	SYSU	31	CU	50
CAMS	8	CQMU	13	ICU	19	CU	30	UMB	47

Abbreviations: CAAS, Chinese Academy of Agricultural Science; CAMS, Chinese Academy of Medical Sciences; CAS, Chinese Academy of Sciences; CMU, Capital Medical University; CQMU, Chongqing Medical University; CU, Columbia University; CUHK, Chinese University of Hong Kong; CUMB, Charité‐University Medicine Berlin; FU, Fudan University; GMU, Guangzhou Medical University; HKU, University of Hong Kong; HMS, Harvard Medical School; HU, Hokkaido University; HUST, Huazhong University of Science and Technology; HZAU, Huazhong Agricultural University; ICU, Imperial College London; LSHTM, London School of Hygiene & Tropical Medicine; NUS, National University of Singapore; OU, Oxford University; PU, Peking University; SBUMS, Shahid Beheshti University of Medical Sciences; SCAU, South China Agricultural University; SCU, Sichuan University; SJTU, Shanghai Jiao Tong University; SYSU, Sun Yat‐Sen University; TSU, Tsinghua University; U. CAS, University of CAS; UCL, University College London; UMB, University of Melbourne; UMG, University of Michigan; UoM, University of Milan; UoS, University of Sydney; UT, University of Toronto; UTMS, Tehran University of Medical Sciences; WU, Wuhan University; ZJU, Zhejiang University.

More and more researchers also participated in COVID‐19 research. The literature statistics as of April 1 showed that 3,029 researchers published studies of COVID‐19 and related fields, and this increased to 23,127 by June 1. In addition, China's noticeable performance at the national and institutional level has not been confirmed at the individual level. In the literature statistics on April 1, only 6 of the top 20 authors were from China (and two authors also received partial support from Chinese institutions), while eight authors were from the HU in Japan. As of April 1, Shi Zhengli, a researcher from CAS published the largest number of articles in the world on COVID‐19 research, reaching 8. As of June 1, 8 of the top 20 authors were from China, with 4 of them from Chinese Hong Kong. As of June 1, Wiwanitkit Viroj, a researcher from DDYPU and HMU had published the largest number of research articles in the world, reaching 26 publications (Table 4).

TABLE 4

Top 20 authors with the most publications and their related information.

As of April 1			As of June 1
Author	Institute	Publications	Author	Institute	Publications
Shi, Z. L.	CAS	8	Wiwanitkit V.	DDYPU and HMU	26
Holmes E. C.	FU and UoS	7	Lippi G.	VU	17
Drosten C.	CUMB	7	Joob B.	SMA	15
Akhmetzhanov A. R.	HU	7	Memis Z. A.	EMU and AU	14
Linton N. M.	HU	7	Drosten C.	CUMB	12
Nishiura H.	HU	7	Nishiura H.	HU	12
Memish Z. A.	EMU and AU	7	Cowling B. J.	HKU	11
Yuen K. Y.	HKU	6	Leung G. M.	HKU	11
Zhang W.	CAS	6	Rodriguez‐Morales A. J.	ACI, UTP and FUAA	11
Hayashi K.	HU	6	Yang L.	HKPU	11
Jung S. M	HU	6	Yang Y.	ISMMS	11
Kinoshita R.	HU	6	Zhang W	CAS	11
Kobayashi T.	HU	6	He D. H.	HKPU	10
Xiao S.	HAU	6	Jiang S. B.	NYBC and FU	10
Yang Y.	HU	6	Li H.	CJFH and CAMSPUMC	10
Zumla A.	UCL	6	Zumla A.	UCL	10
Baric R. S.	UNC	5	Akhmetzhanov A R.	HU	9
Fang L.	HAU	5	Cao B.	CJFH, CAMSPUMC, TSU and CMU	9
Feng L.	CAAS	5	Li T. S.	CAMSPUMC	9
Jiang S. B.	NYBC and FU	5	Linton N. M.	HU	9

Abbreviations: ACI, Asociación Colombiana de Infectología; AU, Alfaisal University; CAMSPUMC, Chinese Academy of Medical Sciences & Peking Union Medical College; CJFH, China‐Japan Friendship Hospital; DDYPU, Dr. DY Patil University; EMU, Emory University; FUAA, National Autonomous University of Mexico; HKPU, Hong Kong Polytechnic University; HMU, Hainan Medical University; ISMMS, Icahn School of Medicine at Mount Sinai; NYBC, New York Blood Center; SMA, Sanitation 1 Medical Academy Centre; UTP, Technological University of Pereira; VU, Verona University.

Contributions to scientific cooperation

This section traces network evolution on scientific cooperation around COVID‐19 articles and analyses the countries/regions, and institutions contributing to the promotion of COVID‐19 scientific cooperation.

Cooperation network evolution

According to Table 5, the international cooperation network on COVID‐19 is moving towards intensiveness, with the network density increasing from 0.163 as of April 1 to 0.191 as of June 1. The average degree also increases continuously from 9.633 to 23.06, which means that a country/region has cooperated with 23.06 other countries/regions on average. As of June 1, the density of international cooperation network was only 0.191, indicated that in the first few months of the outbreak, the international cooperation network was relatively sparse. This shows that although the number of countries/regions participating in the COVID‐19 research is increasing, international cooperation is mainly found in a few countries/regions.

TABLE 5

Topological characteristics of scientific cooperation network on COVID‐19.

Indicators	As of April 1	As of April 15	As of May 1	As of May 15	As of June 1
International cooperation network
Nodes	60	77	96	112	122
Edges	289	487	777	1,055	1,407
Density	0.163	0.186	0.170	0.170	0.191
Average degree	9.633	12.649	16.188	18.839	23.06
Average clustering coefficient	0.752	0.749	0.775	0.769	0.766
Average path length	2.095	2.065	2.041	2.028	1.955
Inter‐institute cooperation network
Nodes	801	1,495	2,454	3,980	5,879
Edges	2,725	6,530	12,329	22,572	36,180
Density	0.009	0.006	0.004	0.003	0.002
Average degree	6.804	8.736	10.048	11.343	12.308
Average clustering coefficient	0.857	0.857	0.860	0.851	0.846
Average path length	4.094	3.816	3.849	3.761	3.694

The density of inter‐institute cooperation networks is generally lower than 0.009 with a continuous downward trend. While the average degree shows an upward trend, increasing from 6.804 as of April 1 to 12.308 as of June 1 (Table 5). Although it is said that the cooperation among countries/regions is undertaken by institutes, when the research scale is placed at the institute level, the global cooperation network on COVID‐19 appears abnormal coefficient and cooperation becomes extremely precious. Besides, based on Watts and Strogatz's work (Watts & Strogatz, 1998) about small‐world network's features, we also found that the scientific cooperation network on COVID‐19 both at national‐level and institute‐level is a typically small‐world network with higher clustering coefficients and shorter average path length compared with a random graph.

Meanwhile, the international cooperation network on COVID‐19 has an obvious core‐periphery structure (Fig. 1), which can be divided into four categories: core, strong semi‐periphery, semi‐periphery, and periphery (Nemeth & Smith, 1985; Smith & White, 1992; Wallerstein, 1974). The international cooperation network on COVID‐19 as of April 1 was a remarkable double‐core pyramid structure, only the US and China located in the core position. As of June 1, China moved down to the strong semi‐periphery group, a single‐core structure of the international cooperation network on COVID‐19 led by the US has been taking shape. In the strong semi‐periphery layer, from April 1 to June 1, except for the change in China, India rose from the semi‐periphery to this level at May 1 but returned at June 1, Saudi Arabia fell to the semi‐periphery at May 1 and remained its status at June 1. However, the number of countries or regions located in the strong semi‐periphery is relatively stable. In the semi‐periphery, the number of countries or regions increased significantly from 9 at April 1 to 40 at June 1. Surprisingly, countries with large numbers of publications were also located in this layer, such as Iran, Switzerland, Spain, Singapore, etc.

FIGURE 1

The core‐periphery structure of international cooperation network on COVID‐19 at three points in time.

The contributing countries/regions

Using the ArcMap platform, the international scientific cooperation on COVID‐19 at three points in time, as shown in Fig. 2, is visualized geographically. The Changing geography of international cooperation on COVID‐19 confirmed that COVID‐19 research gradually developed from individual countries leading to global participation. The tri‐polar landscape of global science dominated by North America, Asia‐Pacific, and Europe has also been proven in COVID‐19 research. Cooperation between countries generally occurs within or between these three regions, and the US, China, and England are the three key nodes (Tables 6 and 7).

FIGURE 2

Geographic pattern of international cooperation on COVID‐19 research.

TABLE 6

International cooperation on COVID‐19 of main countries (regions).

Country/region	As of April 1		As of May 1		As of June 1
	Partners	Collaborations	Partners	Collaborations	Partners	Collaborations
US	35	112	70	476	95	1,304
China	31	132	52	353	72	776
England	28	77	60	351	84	972
Italy	21	40	47	245	67	710
India	7	16	41	120	63	289
Germany	30	69	51	215	72	575
Canada	22	57	45	171	68	514
Australia	20	39	36	153	60	472
Iran	2	2	26	52	46	145
Switzerland	19	28	42	121	61	360
France	17	38	37	120	52	374
Singapore	8	13	22	56	44	164
South Korea	13	14	28	62	37	115
Brazil	6	8	38	88	54	223
Netherlands	16	24	34	108	53	309
Spain	18	20	31	108	58	352
Japan	12	17	24	48	55	199
Turkey	0	0	2	2	39	87
Saudi Arabia	24	59	41	115	51	172
Chinese Taiwan	9	13	16	29	34	83

Note: “Partners” = number of countries (regions) they cooperated with, “Collaborations” = number of international collaborations.

TABLE 7

Top 20 partnerships (country‐level) with the most frequent cooperation on COVID‐19.

As of April 1		As of May 1		As of June 1
Cooperation pairs	Times	Cooperation pairs	Times	Cooperation pairs	Times
China and the US	29	China and the US	86	China and the US	189
The US and Saudi Arabia	13	The US and England	45	The US and England	129
China and England	12	China and England	38	The US and Italy	102
China and Canada	11	The US and Italy	36	The US and Canada	89
China and Australia	10	England and Germany	26	China and England	88
The US and Canada	9	China and Australia	23	England and Italy	77
The US and England	9	The US and Canada	23	The US and Australia	70
England and Germany	8	The US and Germany	22	The US and Germany	59
China and Germany	7	England and Italy	21	England and Germany	58
Canada and Saudi Arabia	6	The US and Australia	20	England and Australia	52
Germany and France	6	China and Canada	20	England and Canada	52
The US and France	6	China and India	18	China and Australia	43
China and Saudi Arabia	6	China and Germany	17	Italy and Germany	43
China and Thailand	6	India and Thailand	17	The US and France	40
Germany and Saudi Arabia	5	The US and Saudi Arabia	17	Italy and Spain	40
China and India	5	England and Canada	16	The US and Switzerland	40
India and Thailand	5	China and Thailand	16	China and Canada	38
China and Italy	5	Germany and Italy	15	China and India	38
Canada and Australia	4	The US and Switzerland	15	China and Italy	35

In the early stage of the outbreak, China played a vital role in promoting international scientific cooperation. Literature statistics as of April 1 showed that China cooperated with 31 countries/regions 132 times. And among the top 20 partnerships, there are 9 pairs with China's participation, 4 of which are in the top 5. Meanwhile, the US and England also performed well in the international scientific cooperation of COVID‐19, conducting 112 and 77 collaborations with 35 and 28 countries/regions respectively. In addition, the US also participated in 5 of the top 20 partnerships. As of May 1, the US cooperated with 70 countries/regions 476 times, surpassing China both in the number of partners and collaborations. While China conducted 353 collaborations with 52 countries/regions and England carried out 351 collaborations with 60 countries/regions. Of the top 20 partnerships, 8 pairs have US's participation, and China and England participated in 7 and 5 pairs respectively. By June 1, as the hub of COVID‐19 global scientific cooperation, the United States was further consolidated. It has cooperated with 95 countries/regions 1,304 times, far more than other countries/regions both in the number of partners and collaborations. Among the top 20 partnerships, there were 8 pairs with US participation, 4 of which are in the top 5. England also surpassed China by conducting 972 collaborations with 84 countries/regions, while China cooperated with 72 countries/regions 776 times. And in the top 20 partnerships, both China and England participated in 6 of them.

Canada in North America, India, Australia, Iran, Singapore, etc. in the Asia‐Pacific, and Italy, Germany, France, Switzerland, etc. in Europe also greatly participate in scientific cooperation on COVID‐19. However, as of now, China and the US are the two most important countries for COVID‐19 research and scientific cooperation. At the five points in time, the closest cooperation relationship always existed between China and the US, increasing from 29 as of April 1 to 189 as of June 1.

Contributing institutes

Chinese institutes also played an important role in promoting cooperation on COVID‐19 among institutes. But over time, the role of institutes in the US (e.g. Harvard Medical School, HMS), Canada (e.g. University of Toronto, UT), England (e.g. University College London, UCL), Germany (e.g. Charité‐University Medicine Berlin, CUMB), and Australia (e.g. University of Sydney, UoS) in scientific cooperation on COVID‐19 also grew rapidly, even more than most institutes in China. Literature statistics as of April 1 showed that CAS and Capital Medical University (CMU) cooperated with 61 and 64 institutes 87 and 83 times, respectively, becoming the double‐core of the inter‐institute cooperation network on COVID‐19. In addition, HUST, CUMB, UoS, and Fudan University (FU), carrying out 59, 58, 56, and 56 collaborations with 40, 38, 47, and 43 institutes, respectively, also played an important role in the scientific cooperation on COVID‐19 (Table 8). Among the top 20 institutional partnerships, Chinese institutes participated in 9 of them. Cooperation between CAS and University of CAS (U. CAS) was the greatest with nine collaborations. By May 1, HKU, CMU, and HUST ranked among the top three with 187, 186, and 178 collaborations, respectively. Regarding the number of partners, HKU, UCL, and HUST ranked among the top three with 155, 126, and 119 partners, respectively. In addition, FU, WU, CAMS, and Oxford University (OU) also played an important role in promoting cooperation on COVID‐19 between institutes. Among the top 20 institutional partnerships, there were 10 pairs with Chinese institute participation. The collaborations between CAS and U. CAS also ranked highest with 15 collaborations.

TABLE 8

The top 20 institutional cooperation on COVID‐19.

Institution	As of April 1		Institution	As of June 1
	Partners	Collaborations		Partners	Collaborations
CAS	61	87	HUST	235	418
CMU	64	83	HMS	309	409
HUST	40	59	UT	291	398
CUMB	38	58	UCL	254	362
UoS	47	56	UMB	244	343
FU	43	56	CUHK	223	338
CUHK	39	56	CU	244	324
Ins. Pa	50	55	HKU	224	315
UT	49	51	WU	186	306
UCL	33	49	CMU	174	295
PU	41	47	UoS	222	278
ZJU	41	47	CUMB	209	277
AU	34	47	UoM	215	271
CAMS	33	42	OU	189	268
WU	31	42	UW	194	266
GMU	39	41	UP	221	265
HKU	35	39	CAMS	159	260
KAU	30	38	PU	183	258
EMU	32	36	FU	140	236
UW	28	33	UMG	190	235

Abbreviations: AU, Alfaisal University; Ins. Pa, Institut Pasteur; KAU, King AbdulAziz University; UP, University of Pennsylvania; UW, University of Washington.

As of June 1, HUST and HMS had conducted 418 and 409 institutional collaborations, respectively. There are also 7 institutes that conducted more than 300 institutional collaborations, namely, UT, UCL, University of Melbourne (UMB), CUHK, Columbia University (CU), HKU, and WU. In terms of the number of partners, HMS, the only institute with more than 300 partners, has cooperated with 309 institutes. There are also 11 institutes with more than 200 partners, of which UT and UCL have more than 250 partners. Among the top 20 partnership institutes, there were 6 pairs of Chinese institutes' and 5 pairs from Germany. The collaborations between HUST and WU reached 22, ranking highest among institutional cooperation. An interesting phenomenon is that, contrary to international cooperation, cooperation on COVID‐19 among institutes exhibits significant geographic proximity, that is, inter‐institute cooperation on COVID‐19 mostly occurred within the country or even within the city. Among the top 20 institutional partnerships as of June 1, there was only one transnational partnership (Table 9).

TABLE 9

Top 20 partnerships (institute‐level) with the most frequent cooperation on COVID‐19.

As of April 1		As of June 1
Cooperation pairs	Collaborations	Cooperation pairs	Collaborations
CAS and UCAS	9	HUST and WU	22
HU and JSTA	7	DDYPU and HMU	20
CUHK and UCL	5	CAS and UCAS	16
CICSPP and HZAU	5	SBUMS and UTMS	16
HUST and WU	5	CUHK and HKU	14
BIH and CUMB	4	BIH and GUMB	13
BIH and FUB	4	BIH and FUB	13
BIH and HBU	4	CMU and CAMS	13
CMU and CAMS	4	CUMB and HBU	13
CMU and HUST	4	FU and SJTU	13
CUMB and FUB	4	NUHS and NUS	13
CUMB and HBU	4	RMH and UMB	13
CAS and CCDCP	4	BIH and HBU	12
DDYPU and HMU	4	CUMB and FUB	12
FUB and HBU	4	FICGOMP and UoM	12
FU and NYBC	4	FUB and HBU	12
HZAU and UGA	4	IUMS and SBUMS	12
AU and CUHK	3	IUMS and UTMS	12
AU and EMU	3	CMU and HUST	11

Abbreviations: BIH, Berlin Institute of Health; CCDCP, Chinese Center for Disease Control and Prevention; CICSPP, Cooperative Innovation Center for Sustainable Pig Production; FUB, Free University of Berlin; HBU, Humboldt—Universitat zu Berlin; IUMS, Iran University of Medical Sciences; JSTA, Japan Science and Technology Agency; UGA, University of Georgia.

DISCUSSIONS AND CONCLUSIONS

At the time of writing, the COVID‐19 pandemic is still ravaging the world. Tens of thousands of confirmed cases and thousands of deaths are confirmed and announced every day. More extensive and in‐depth cooperation should be carried out on a global scale (Nature Editorial, 2020a, 2020b). This paper attempts to provide a comprehensive picture of scientific collaboration on COVID‐19 research among countries/regions and among institutes within the first few months of the pandemic. The study included 5,827 papers about COVID‐19 published by 6,349 institutions from 128 countries/regions.

We admit that there are some shortcomings in this study. Firstly, we limited our data to the publications retrieved from the Web of Science. Although it is known for its huge amount of data (Cassi et al., 2012; Gui et al., 2018b; Leydesdorff & Wagner, 2008), it is still limited in its inclusion. Secondly, although co‐publications are widely accepted as proxies of scientific collaboration, as mentioned before, scientific cooperation does not necessarily lead to the publication of papers (Cantner & Rake, 2014; Royal Society, 2011). Moreover, cooperation in publishing papers may only be a small aspect of scientific cooperation on COVID‐19. Thirdly, this paper mainly focused on the cooperation, other bibliometric features are not involved, such as citation analysis, hotspot analysis, and community analysis.

Through this bibliometric study, we found some interesting phenomena. First of all, scientific cooperation on COVID‐19 has become more frequent. As of June 1, an increasing number of countries/regions, institutions, and researchers participated in scientific cooperation on COVID‐19. The international scientific community generally recognizes that collaboration is the right way to work to overcome the epidemic and build a community of human health. Secondly, we discovered that the tri‐polar pattern of international scientific cooperation controlled by North America, Asia‐Pacific, and Europe (Gui, Liu, & Du, 2019; Gui, Liu, Du, et al., 2019) is clearly portrayed in COVID‐19 research. In these three regions, the US, China, England, Canada, Germany, India, and Australia are the core hubs of the international cooperation network for COVID‐19 research. Particularly, the US is playing an increasingly important role in research and international cooperation on COVID‐19, reflecting its status as a global scientific centre. Most countries/regions regard the US as the strongest scientific partner. Thirdly, China has played a vital role in the scientific research and cooperation on COVID‐19, which is not only reflected in the number of published papers (Duan et al., 2020) but also in its extensive international cooperation (Mo & Zhou, 2020; Wu et al., 2020; Zhou et al., 2020). Fourth, China and the US were the closest partners in the current international scientific cooperation of COVID‐19. Regardless of the current tense international relations between China and the US, in the face of the epidemic the institutions and researchers of the two countries still carried out close scientific cooperation.