Articles with original data published in scholarly journals
Among the first 2118 articles on COVID-19 published in scholarly journals, 533 (25%) contained original data. We have excluded 1585 articles for the following reasons: not original research (N=1386), duplicate articles (N=118), unrelated to the topic (N=56), correction (N=18), preprint server publication (N=4), study protocol (N=2), and retraction (N=1). The list of analyzed and the list of excluded studies is available on OSF (https://osf.io/dzvxc/). The first article was published on January 21, 2020. The majority of articles were published in English (N=405; 75%); a quarter was published in Chinese (N=131; 24%), and one article was published in Persian.
The median number of authors was 7 (range: 1 to 63). Articles were published in 207 different journals. The highest number of articles was published in the Journal of Virology (N=33; 6.1%) (Table 1). For 377 articles published in journals with a JIF, the median JIF was 5.099 (range: 0.364 to 70.670).
The median number of countries in the authors’ affiliations was 1 (range: 1 to 9). Authors from 48 countries authored the articles, the majority of affiliations were from China (N=402; 75%), followed by the USA (N=62; 12%) (Table 1).
In 312 (58%) journal articles, authors self-reported study design. The most common self-reported study designs were retrospective study (N=88; 28%) and case report (N=86; 28%) (Table 1). Our classification of articles in three major groups showed that there were 503 (94%) observational studies, 19 (4%) evidence syntheses of various types, and 11 (2%) experimental studies.
Among the 533 articles, 456 were in the EPPI-Centre living map of evidence; the majority were classified as case reports (N=173; 38%) (Table 1). In 381(71%) articles unit of analyses were humans; in the majority (N=236; 62%) only adults were included. Declaration about study funding was reported in 324 (60%) of the journal articles; among those, there were 268 (83%) articles that reported that the study received funding. Sponsors were most commonly from China (N=202; 75%) (Table 1).
Table 1. Characteristics of analyzed journal articles with original data
Variable (N of denominator)
|
N (%)*
|
Journals (N=533)
Journal of Medical Virology
Journal of Infection
International Journal of Infectious Diseases
Clinical Infectious Diseases
Radiology
Other
|
33 (6.1)
18 (3.3)
16 (3.0)
15 (2.8)
14 (2.6)
437 (82)
|
Country in the author affiliation (N=533)
China
USA
UK
Japan
Italy
Other
|
402 (75)
62 (12)
21 (3.9)
20 (3.7)
19 (3.5)
9 (1.7)
|
Self-reported study design (N=312)
Retrospective study
Case report
Case series
Modelling
Systematic review with or without meta-analysis
Other
|
88 (28)
86 (28)
46 (15)
18 (5.7)
16 (5.1)
58 (18.6)
|
Thematic classification (N=456)
Case reports – patients
Transmission/risk/prevalence
Genetics/biology
Health impacts
Diagnosis
Other
|
173 (38)
104 (23)
57 (13)
54 (12)
41 (9)
27 (5.9)
|
Country of study funders (N=268)
China
USA
Japan
Korea
Canada
Other
|
202 (75)
13 (4.9)
11 (4.1)
5 (1.9)
4 (1.5)
33 (12.3)
|
*Denominator is provided in the first column, as “N of trials with reported variable”; for some variables due to rounding the sums may not be exact 100%, for variables we presented only five most frequent categories.
Preprint articles
From the exported 1102 preprint articles we excluded 4 that were withdrawn and 10 that were about SARS and MERS; we included the remaining 1088 preprint articles in the analysis. The list of analyzed preprint articles is available on OSF (https://osf.io/dzvxc/). The majority was posted on medRxiv (Table 2). The first preprint article on COVID-19 was posted on bioRxiv on January 19, 2020; it reported a mathematical model of transmission of the novel virus [8], the first article was posted on medRxiv on January 24, 2020; it reported early estimation of epidemiological parameters and epidemic predictions regarding the novel virus [9].
The median number of authors was 7 (range: 1 to 178). The most common country in the authors’ affiliations was China (51%) (Table 2). In 494 (45%) preprint articles, authors self-reported study design. The most common self-reported study design was a modeling study (Table 2).
The most frequent thematic classification of the preprint articles was transmission/risk/prevalence (43%; Table 2). Study funding was reported in 681 (63%) of the preprint articles. The majority of funders were from China and the USA (Table 2).
Table 2. Characteristics of analyzed preprint articles
Variable (N of denominator)
|
N (%)*
|
Preprint server (N=1088)
medRxiv
bioRxiv
|
842 (77)
246 (23)
|
Country in the author affiliation (N=1088)
China
USA
UK
Italy
Hong Kong
Other
|
563 (52)
298 (27)
92 (8.4)
51 (5.3)
43 (3.9)
41 (3.7)
|
Self-reported study design (N=494)
Modelling
Retrospective study
Cross-sectional study
Cohort study
Systematic review with or without meta-analysis
Other
|
306 (62)
59 (12)
35 (7.1)
22 (4.4)
21 (4.3)
51 (10)
|
Thematic classification (N=1088)
Transmission/risk/prevalence
Health impacts of COVID-19
Genetics/biology
Diagnosis
Treatment development
Other
|
470 (43)
163 (15)
127 (12)
101 (9.2)
84 (7.7)
137 (13)
|
Country of study funders (N=681)
China
USA
UK
Japan
China and USA
Other
|
312 (46)
107 (16)
14 (2)
13 (1.9)
11 (1.6)
224 (33)
|
*Denominator is provided in the first column, as “N of trials with reported variable”; for some variables due to rounding the sums may not be exact 100%, for variables we presented only five most frequent categories.
Registered clinical trials
By April 7, 2020, there were 927 clinical trials indexed on WHO ICTRP. The list of analyzed registered trials is available on OSF (https://osf.io/dzvxc/). The first trial was indexed on January 27, 2020. The majority (N=581; 63%) of trials were primarily registered on the Chinese Clinical Trials Registry (ChiCTR), followed by ClinicalTrials.gov (N=286; 30%). Few trials were primarily registered with other platforms (Table 3).
Recruitment status was available for 915 (99%) of registered protocols, and among them about half were either “not recruiting” or “recruiting” (Table 3). None of the trials retrieved from WHO ICTRP were labeled as “withdrawn” in the recruitment status. However, 38 (4%) of protocols were labeled as “Cancelled” in the name of the study; all these protocols were indexed primarily in ChiCTR.
In 744 trials, the minimal age of participants was specified. In the majority, the minimal age of participants was 18 years (N=532; 72%) (Table 3). In 663 trials, information about the maximum age of participants was provided. In about a third of them (N=197; 30%), it was specified that there was no upper age limit (Table 3). In 921 protocols there was information about the inclusion of participants based on sex; the majority (N=892; 97%) reported they will include both men and women (Table 3).
The majority of registered trials were described as interventional (N=535; 58%), followed by descriptor “observational” (N=322; 35%) (Table 3). Among registered “trials”, there were even 7 that were described as “basic science” (Table 3).
The median number of planned study participants was 140 (range above zero: 1 to 15,000,000). For eight protocols, the planned number of participants in the WHO ICTRP data was zero; we checked web sites of all those protocols and found that five of them were from ClinicalTrials.gov where they were labeled as withdrawn, the remaining three were from ChiCTR, whereas one had information about the number of patients in the wrong field, but the remaining two did not have any explanation for zero number of patients.
Five protocols did not have any information about the number of participants; two were canceled protocols from ChiCTR, two were protocols labeled as “Expanded access status” in ClinicalTrials.gov, and we were unable to verify the fifth because the web link was not functional. In interventional studies, the median number of planned participants was 108 (range from 1 to 55,000), while in the observational median was 200 (range from 8 to 15,000,000). Three protocols reported that the planned number of participants was higher than one million.
In 825 registrations, the location, where the trial will be conducted, was reported. Only 20 (2.4%) reported that the trial will be conducted in more than one country. Most of the trials conducted in a single location were located in China (N=522; 63%), followed by United States (N=33; 4%) (Table 3).
In 535 trial protocols described as interventional, 532 (99%) provided information about the primary outcome. Most of the protocols (N=260; 49%) had multiple primary outcomes that were not described as composite. In studies with a single or composite primary outcome (N=272), highly heterogeneous primary outcomes were used (Supplementary file 1). Few outcomes were used more commonly. The most commonly used outcome was time to recovery, used in 40 (15%) protocols, and phrased differently such as “time to clinical recovery”, “time to clinical improvement”, “time to disease recovery”, “time to remission”, “clinical recovery time”, etc. The second most common outcome was mortality, found in 23 (8.4%) protocols with a single or composite primary outcome, described variously as mortality, all-cause mortality, in-hospital mortality, or mortality at certain time points (28 days, 30 days, 60 days).
In registered trials of interventions, various heterogeneous interventions were tested; the most frequently studied interventions were hydroxychloroquine (N=39; 7.2%) and chloroquine (N=16; 3%) (Table 3).
Table 3. Characteristics of analysed clinical trial registrations
Variable (N of trials with reported variable)
|
N (%)*
|
Clinical trial registry (N=927)
Chinese Clinical Trials Registry (ChiCTR)
ClinicalTrials.gov
EU Clinical Trials Register
Australian New Zealand Clinical Trials Registry (ANZCTR)
ISRCTN
IRCT
Other
|
581 (63)
286 (30)
21 (2.2)
9 (1)
8 (0.9)
8 (0.9)
14 (1.5)
|
Recruitment status (N=915)
Not recruiting
Recruiting
Authorized
|
453 (50)
441 (48)
21 (2.2)
|
Minimal age of participants (N=744)
18 years
0 years
14 years
16 years
1 year
Other
|
532 (72)
26 (3.5)
18 (2.4)
15 (2)
13 (1.7)
140 (15)
|
Maximal age of participants (N=663)
Not applicable/no upper limit
80 years
75 years
90 years
65 years
Other
|
197 (30)
59 (9)
55 (8.2)
42 (6.3)
32 (4.8)
278 (42)
|
Eligibility of participants based on sex (N=921)
Both men and women
Only men
Only women
|
892 (97)
18 (1.9)
11 (1.2)
|
Self-reported study type (N=927)
Interventional
Observational
Diagnostic test
Observational (patient registry)
Epidemiological research
Other
|
535 (58)
303 (33)
35 (3.8)
19 (2)
10 (1)
25 (2.7)
|
Location of trials located in single countries (N=825)
China
United States
France
Italy
United Kingdom
Other
|
622 (63)
33 (4.0)
21 (2.5)
17 (2.1)
10 (1.2)
122 (15)
|
Tested interventions (N=535)
Hydroxychloroquine
Chloroquine
Tocilizumab
Lopinavir/ritonavir combination
Convalescent plasma
Other
|
39 (7.2)
16 (3.0)
10 (1.9)
10 (1.9)
9 (1.7)
451 (84)
|
*Denominator is provided in the first column, as “N of trials with reported variable”; for some variables due to rounding the sums may not be exact 100%, for variables we presented only five most frequent categories.