Global academic response to COVID‐19: Cross‐sectional study

Abstract This study explores the response to COVID‐19 from investigators, editors, and publishers and seeks to define challenges during the early stages of the pandemic. A cross‐sectional bibliometric review of COVID‐19 literature was undertaken between 1 November 2019 and 24 March 2020, along with a comparative review of Middle East respiratory syndrome (MERS) literature. Investigator responsiveness was assessed by measuring the volume and type of research published. Editorial responsiveness was assessed by measuring the submission‐to‐acceptance time and availability of original data. Publisher‐responsiveness was assessed by measuring the acceptance‐to‐publication time and the provision of open access. Three hundred and ninety‐eight of 2,835 COVID‐19 and 55 of 1,513 MERS search results were eligible. Most COVID‐19 studies were clinical reports (n = 242; 60.8%). The submission‐to‐acceptance [median: 5 days (IQR: 3–11) versus 71.5 days (38–106); P < .001] and acceptance‐to‐publication [median: 5 days (IQR: 2–8) versus 22.5 days (4–48·5‐; P < .001] times were strikingly shorter for COVID‐19. Almost all COVID‐19 (n = 396; 99.5%) and MERS (n = 55; 100%) studies were open‐access. Data sharing was infrequent, with original data available for 104 (26.1%) COVID‐19 and 10 (18.2%) MERS studies (P = .203). The early academic response was characterized by investigators aiming to define the disease. Studies were made rapidly and openly available. Only one‐in‐four were published alongside original data, which is a key target for improvement. Key points COVID‐19 publications show rapid response from investigators, specifically aiming to define the disease. Median time between submission and acceptance of COVID‐19 articles is 5 days demonstrating rapid decision‐making compared with the median of 71.5 days for MERS articles. Median time from acceptance to publication of COVID‐19 articles is 5 days, confirming the ability to introduce rapid increases at times of crisis, such as during the SARS outbreak. The majority of both COVID‐19 and MERS articles are available open‐access.


INTRODUCTION
Coronavirus disease 2019  caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has spread rapidly since it began in the city of Wuhan in late 2019 (Xu et al., 2020). The outbreak has been declared as a pandemic by the World Health Organization (WHO), reflecting its spread to more than 200 countries worldwide (WHO, 2020c). As of April 2020, there have been more than 100,000 deaths and the number of those affected continues to rise (WHO, 2020a).
In response to this unprecedented situation, healthcare systems around the world have taken urgent action to scale up medical staffing, equipment, and infrastructure. In the UK, this is demonstrated by the construction of new field hospitals, the return of thousands of retired healthcare workers, and the re-purposing of many acute and urgent hospital services (NHS England, 2020). The role that science and data play in the face of the COVID-19 pandemic must also be recognized. Robust and rapidly available evidence is essential for improving the detection and treatment of the virus as well as for reducing its transmission.
It is the responsibility of all members of the academic community to facilitate this transfer of knowledge so that patients around the world receive the best evidence-based care available.
There is limited information about how the academic community has responded to COVID-19 and how this can be further guided during times of global crisis. There is a need for investigators to respond urgently to new clinical priorities across all sectors of healthcare and to work in unfamiliar environments. Journal editors must facilitate rapid and robust peer review while making difficult editorial decisions based on little previous academic background. Publishers must mobilize quickly to enable timely publication of manuscripts so that research outputs can be realized and implemented into practice. An early understanding of this response is important so that the academic community can identify early challenges and respond appropriately as the COVID-19 pandemic develops.

Ethics and governance
As a review of published literature, approval by a research ethics committee was not applicable. Since only bibliometric outcomes were considered, rather than outcomes of direct relevance to research participants, this review was not eligible for registration on the PROSPERO database. There were no changes to the research design or outcomes during the course of the study. The results are reported with consideration to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRI-SMA) Checklist (Liberati et al., 2009).

Aims and objectives
The study aimed to examine the responsiveness of the academic community to COVID-19 during its early stages. The following objectives were predefined: • To explore investigator responsiveness by describing the volume and type of research accepted in peer-reviewed journals.
• To explore editorial responsiveness by describing the time taken to facilitate editorial/peer review and the availability of original data as a condition of publication.
• To explore publisher responsiveness by describing the time taken for accepted manuscripts to be available and the provision of open access publication.

Study design
This was a systematic, cross-sectional, bibliometric review of existing healthcare literature related to COVID-19. To facilitate comparisons of outcomes in a unique setting, a comparable review of Middle East respiratory syndrome (MERS) was also performed and used as control. MERS was chosen as a suitable comparator because it similarly represents a respiratory illness caused by a zoonotic coronavirus, but was not classified as a pandemic or a Public Health Emergency of International Concern by the WHO during the search period (WHO, 2020d). It therefore represents a non-COVID-19 and non-pandemic control.

Definitions
In line with the WHO, COVID-19 describes the disease caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and MERS describes the disease caused by the Middle East respiratory syndrome-related coronavirus (MERS-CoV) (WHO, 2020b). The early stages of COVID-19 was defined as the period between the first known human case and its classification as a pandemic on 11 March 2020 (WHO, 2020c). Availability of original data was considered to be an editorial responsibility since this is recommended by the International Committee of Medical Journal Editors for clinical trials and can be enforced as a condition of publication (Taichman et al., 2017).

Key points
• COVID-19 publications show rapid response from investigators, specifically aiming to define the disease.
• Median time between submission and acceptance of COVID-19 articles is 5 days demonstrating rapid decisionmaking compared with the median of 71.5 days for MERS articles.
• Median time from acceptance to publication of COVID-19 articles is 5 days, confirming the ability to introduce rapid increases at times of crisis, such as during the SARS outbreak. de-duplicated and saved offline. Two independent investigators then screened titles, abstracts, and full texts for possible inclusion, with discrepancies addressed through re-examination and discussion. A full outline of search strategies is available in Table 1.

Study eligibility criteria
For published studies, any COVID-19-or MERS-related manuscript available online or in print within the defined time periods was eligible for inclusion. Eligible studies had to report primary data. This included studies that performed analyses using open source data sets, but not studies that reported systematic reviews or meta-analyses. Studies of animals or pre-clinical models were eligible, but only if the outcomes contributed to human rather than veterinary medicine. Articles published in non-English languages and other grey literature (such as conference abstracts) were excluded since these are not practically accessible by the global academic community.

Study outcomes
Academic responsiveness was considered across three groups of the academic community. Investigator responsiveness was explored by measuring the volume and type of research (pre-clinical, clinical, and modelling) published in peer-reviewed journals.
Editorial responsiveness was explored by measuring the time taken for editorial/peer review, defined as the number of days between manuscript submission and acceptance. The availability of original data to support the results was also explored. Data were considered available if individual observations (or source code in the case of modelling studies) were provided as a supplementary file, in a controlled-access repository, or if an explicit statement of availability from authors was present. Where the status of data sharing was unclear, corresponding authors were contacted and non-responses were considered to represent an absence of data. Publisher responsiveness was explored by measuring the time taken for manuscripts to complete production, defined as the number of days between acceptance and first availability online or in print. Publisher responsiveness was also assessed by exploring the incidence of open access publication.

Data collection
Data were collected by a single investigator and checked by an independent investigator for accuracy. Data sources included fulltext manuscripts and editorial-related content. Data points of interest were country of origin (according to the corresponding institution), manuscript format (short communication/letter or full-text), journal subject category, and subject category ranking (according to Thomas Reuters Journal Citation Reports). For clinical studies, additional data of interest were population (adults, children, pregnant adults, or healthcare workers), study design (case report, case series, observational, or interventional) and clinical focus (definition of disease, diagnosis/screening, treatment, prevention, or resource use). Definition of disease related to studies describing clinical features and outcomes.

Statistical analysis
Data are presented descriptively using averages and measures of variance. Continuous data were analysed using the student's ttest or analysis of variance (ANOVA). Categorical data were analysed using χ 2 or Fisher's exact test. All statistical comparisons of COVID-19 and MERS studies were pre-planned. A single subanalysis of data sharing practices was performed with the exclusion of case reports since these may or may not provide the full

Assessment of quality
An assessment of self-declared compliance to reporting guidelines was performed for 48 full-text COVID-19 studies (n = 45 observational; n = 3 interventional) and 6 full-text MERS studies (n = 4 observational; n = 2 interventional). None of these studies declared compliance to relevant reporting checklists, as recommended by the Equator Network.

DISCUSSION
The results demonstrate a rapid academic response to COVID-19 during its early stages. Investigators responded with reports of cases and case series, many of which were published in major general medical journals. The volume of interventional studies exploring treatments was low, but this is expected to change as on-going trials reach completion. The editorial and production times for COVID-19 studies were strikingly shorter than MERS controls and almost all manuscripts were openly accessible at the point of publication. In contrast, only one-in-four COVID-19 manuscripts were published with original data available to support the results. This study raises important considerations for the dissemination of data during the COVID-19 pandemic. First, the urgency to produce evidence and the need for rigorous peer review must be balanced to ensure quality of published outputs. In this study, an absence of self-declared compliance to reporting checklists (such as STROBE and CONSORT) may indicate a relaxation of reporting standards. The average time to manuscript acceptance of 5 days may also suggest a pattern of internal editorial review, rather than review by externally invited subject-matter experts. It is not possible to explore this with certainty from the present data because details of peer review are not usually available. While there is an urgent need to report outputs quickly, other routes for provisional dissemination (such as pre-prints) may be impor-  (Piwowar et al., 2018). Consideration is needed to how this approach can be sustained as the pandemic progresses through its recovery and subsequent outbreak phases. Finally, this study provides an early insight into the challenges of data sharing during the COVID-19 crisis. As more potentially practice-changing studies emerge, data sharing will be critical for allowing scrutiny of results, reducing unnecessary duplication, and enabling rigorous pooled analyses.
Awareness of secure access data repositories as well as guidelines for data stewardship will be important (Wilkinson et al., 2016).
The academic response during times of global disease has been explored previously. In a bibliometric analysis of the 2003 severe acute respiratory syndrome (SARS) crisis, the difference in median submission-to-acceptance intervals between SARS and non-SARS articles was 106.5 days (95% CI 55.0-140.1) and the difference in median acceptance-to-publication intervals was 63.5 days (18.0-94.1) (Xing, Hejblum, Leung, & Valleron, 2010 (Hagel, Weidemann, Gauch, Edwards, & Tinnemann, 2017). More recently, two analyses of COVID-19 research found that most data were generated from China and called for increased academic output, particularly in the form of interventional trials of new treatments (Chahrour et al., 2020;Lou et al., 2020). As shown by the previous response to EVD and the current data, it is clear this must be done with global coordination and with a common research agenda (COVID-19 Clinical Research Coalition, 2020).
Strengths of this study are recognized. This review provides timely and critical information about how the academic community have responded to this global emergency during its early stages. Specifically, it identifies on-going challenges that must be considered, such as the balance between rapid and robust review, as well as the availability of original data to support published results. Weaknesses are also recognized. This review provides an incomplete representation of the entire research pathway. While investigators, editors, and publishers are essential for generating and disseminating research, other groups such as funding bodies, research ethics committees, regulatory agencies, and patient/ stakeholder groups also play important roles. Measuring the response of these groups is challenging since it relies on data that are not openly available or straight forward to measure. Another weakness is the focus on peer-reviewed publications. It is likely that a wider body of unpublished data exist following rejection or through investigators making results available on non-peerreviewed platforms such as pre-print servers. These are important and further work to confirm their rigour and stability (i.e. changes made during parallel submission and peer review) is needed.
There is no question that the academic response to COVID-19 from investigators, editors, and publishers has been prompt. Data sharing is an early challenge and a commitment to this from all members of the academic community is required. Consideration to