Measuring the impact of the COVID-19 pandemic on the activeness, efficiency, and international collaboration of clinical trials Open Access

https://www.webofscience.com/api/gateway/wos/peer-review/10.1162/qss_a_00354

The COVID-19 pandemic has disrupted global structures, creating an urgent health crisis that demands rapid and innovative research for control. During the pandemic, the international scientific community pivoted its focus towards managing and preventing the spread, treating the infected population, and accelerating vaccine development. Previous studies of COVID-19’s impact have been conducted from a variety of perspectives due to its diverse impact in almost all areas. These include its catalytic effect on scientific novelty (Liu, Bu et al., 2022), changes in international collaboration (Fry, Cai et al., 2020), evolution, disruption, and resilience in research topics (Zhang, Cai et al., 2021b), as well as some social topics such as the simultaneous rise of the COVID-19 infodemic (Zhang, Pian et al., 2021a). It is commonly known that, out of all the scientific responses, clinical trials—that is, studies conducted on humans with the goal of evaluating a medical intervention—are the primary method used by researchers to ascertain whether a new therapy, such as a medication, diet, or medical device, is safe and effective in humans (NIH National Institute on Aging, 2020). What effect the COVID-19 pandemic has had on initiating and conducting clinical trials, particularly for non-COVID-19 disorders, also merits more in-depth and fine-grained research.

Clinical trials are an indispensable component in the medical domain, serving as the critical link between basic research and clinical practice, which is particularly salient in the context of the pandemic era. They are crucial for bridging the gap between foundational research and practical healthcare applications, providing robust evidence that informs clinical decision-making, policy formulation, and healthcare system management (Weber, 2013). In the realm of bibliometrics, studies on COVID-19 have predominantly focused on scientific publications. Publication, indeed, is the most common data source in this field, which represents the documented records of research findings, typically disseminated through academic journals, books, or other media. Clinical trials are usually interventional studies that assess the impact of novel treatment modalities, pharmaceuticals, or medical devices on human health. These trials, which span multiple years and are typically structured in sequential phases, are essential for establishing safety, efficacy, and potential adverse effects. The U.S. Food and Drug Administration (FDA) requires the completion of Phases 1 through 3 of a trial to ascertain the readiness of a drug or device for broader application (National Institute on Aging, 2023).

Although clinical trials have not traditionally been a primary data type in bibliometrics, their intrinsic value and significance within the field are gaining recognition. For instance, to comprehend the current state of clinical research in a particular domain, scholars may leverage a synthesis of literature data and clinical trial registries for analysis (Dagli, Haque, & Kumar, 2024; Wang, Zhou et al., 2020; Xia, Yao et al., 2021). Additionally, the citation of clinical trials is an area of increasing interest in bibliometrics. The transition of knowledge from academic research to clinical trials is pivotal for transforming scientific insights into tangible healthcare interventions (Thelwall & Kousha, 2016; Thelwall & Maflahi, 2016). Researchers examine the citation patterns of academic papers within clinical trials to explore the extent to which academic research is translated into actionable clinical practices (Pallari, Eriksson et al., 2021; Park, Kim et al., 2019; Thelwall & Kousha, 2016; Zang & Liu, 2023). This examination is vital for assessing the impact and applicability of scientific discoveries in real-world healthcare settings.

Assessing the impact of the COVID-19 pandemic on clinical trials has drawn increasing attention from scholars in various fields with diverse approaches. In bibliometrics, Xia et al. (2021) conducted a study concentrating on COVID-19 clinical research, synthesizing literature with data from ClinicalTrials.gov. Their analysis revealed that the main emphasis of early-phase COVID-19 clinical trials was directed towards discovering efficacious treatments. Chen, Chen et al. (2021) performed semistructured qualitative interviews, which suggests that while online meetings, remote follow-up, fast medication delivery, and remote monitoring may ensure the progress of clinical trials, they cannot completely guarantee their quality as previously. As for the influence of COVID-19 on the initiation of new clinical trials, 50% of 734 researchers responded that it had considerable impacts on the initiation of new clinical trials, while almost 35% of researchers claimed the pandemic had little to no effect on the beginning of new studies, according to a survey conducted by Medidata (2020). As for other quantitative studies, several studies employed trial data from ClinicalTrials.gov and arrived at inconsistent findings (Audisio, Lia et al., 2022; Hawila & Berg, 2021; Margas, Wojciechowski, & Toumi, 2022; Nishiwaki & Ando, 2021). For instance, non-COVID-19 trials experienced a significant decrease during the first wave of the COVID-19 pandemic (Nishiwaki & Ando, 2021) while a minimal impact of COVID-19 on the number of submitted clinical trials was also observed (Hawila & Berg, 2021).

The outbreak of COVID-19 also led to a high concentration of medical and scientific research resources to address public health emergencies, which may be accompanied by a decline in the research activeness of other diseases; that is, the crowding-out effect of COVID-19 on clinical trials of other diseases. As Cohen (2020) stated early on in Forbes magazine, “research and development for treatments and vaccines targeting the novel coronavirus is crowding out R&D in other critical clinical areas, such as cancer, cardiovascular disease, Alzheimer’s, and autoimmune disorders.” Simultaneously, a comment published in Nature also stated that “The COVID-19 pandemic has placed a tremendous strain on the clinical research enterprise. With the redirection of resources and temporary halting of in-person visits, studies in other therapeutic areas have been unavoidably constrained” (Tuttle, 2020). However, such observations, noted through industry and hospital phenomena, lack systematic quantitative research to substantiate these perceptions.

In general, existing studies have exhibited varying results. Most studies are performed on the aggregate level of COVID-19 versus non-COVID-19 trials (Audisio et al., 2022). Few studies have attempted to investigate the multidimensional effect and crowding-out effect under a more fine-grained disease classification due to the difficulty of precisely identifying clinical trials for various diseases. Moreover, although ClinicalTrials.gov is a representative and international registration platform, some trial data from other national registries, such as the Chinese Clinical Trial Register (CHiCTR)¹, cannot be disregarded when performing global-level analysis. It is also worth noting that utilizing trial data collected at one time (a static perspective) cannot provide a thorough comparative analysis of trials’ efficiency, because clinical trial databases only present the latest states at the time of data collection. In addition, with the COVID-19 pandemic profoundly affecting international collaboration patterns and the need for large-scale clinical trials structured according to a master protocol in a coordinated and collaborative manner (Park, Mogg et al., 2021), few scholars have paid attention to the status quo of multi-regional clinical trials (MRCTs) for different diseases in the context of the epidemic.

The collaborative nature of clinical trials, particularly those involving multiple countries, represents a more substantial form of collaboration than the coauthorship seen in single publications. The conduct of these trials, which often span several years and are structured in sequential phases, necessitates prolonged and substantive collaboration. Hence, clinical trials are a data type that scholars studying medical research should consider more seriously due to their inherent collaborative essence. Moreover, MRCTs are a significant concept and research topic in their own right. As stated in the ICH E17 guidelines (2014–2017), MRCTs, which are not a new concept and have been in use for decades, refer to trials conducted across multiple regions under a unified protocol². Additionally, the MRCT Center, established in 2009 and affiliated with Brigham and Women’s Hospital and Harvard University, is a renowned research and policy center dedicated to addressing the complexities of multiregional clinical trials³. Given this context, employing MRCTs as a data source to reflect scientific collaboration within the medical sphere is not only representative but also reflective of the expansive, international dimensions of medical research collaboration.

Indeed, understanding the multidimensional effect of COVID-19 on medical resources of various non-COVID-19 diseases is extremely challenging because it is influenced by multiple social, economic, and political factors. Clinical trials have, however, provided one potential pathway for observing such an effect from different aspects. By constructing a disease mapping system, this study will explore the effect of COVID-19 from the aspects of trials’ activeness and efficiency, as well as the scientific collaboration from MRCTs. For activeness, we focus on the crowding-out effect of activity in research on other diseases from a quantitative perspective during the COVID-19 pandemic by using the interrupted time series (ITS) regression model. The crowding-out effect, originally an economic theory, suggests that increased public sector spending can displace private sector spending (Spencer & Yohe, 1970). In noneconomic contexts, it refers to how a dominant or externally imposed initiative can unintentionally diminish the participation, motivation, or effectiveness of other activities in the same domain. For instance, in education, Engel (2013) shows how high-stakes testing can crowd out student motivation by focusing on external rewards (test scores) rather than fostering intrinsic interest in learning. A similar dynamic can occur in health research, where the overwhelming focus on a global health crisis, like COVID-19, crowds out research on other diseases (Cohen, 2020). The ITS model is a robust tool for analyzing time-series data disrupted by an intervention, in this case, the COVID-19 pandemic. During the COVID-19 pandemic, the scientific community allocated vast resources to COVID-19 research, drawing from existing research capacity and reducing focus on other diseases. This phenomenon, referred to as the “crowding-out effect” of COVID-19, is examined by comparing research activity before and after the pandemic’s onset using the ITS regression model.

Further, through text-mining and fine-grained analysis, we have summarized and validated potential factors across three dimensions to shed light on the observed effects of COVID-19 on clinical studies. These factors encompass the digitization of clinical trials, the distinct roles played by various sponsors (those driven by wellbeing vs. profit), and the differing crisis responses of countries. In addition to global observation, we select the four most prolific countries of clinical trials as representative examples to conduct subgroup analysis, namely the United States, China, Japan, and the United Kingdom.

The literature review is structured into two targeted subsections. The initial subsection offers a comprehensive examination of the impact of the COVID-19 pandemic on clinical research, encompassing a diverse array of studies employing varied methodologies to explore the multifaceted impacts. The subsequent subsection shifts focus to the analysis of clinical trials within the bibliometrics domain, delineating the distinctions between (a) publications derived from clinical trials and (b) registered clinical trials themselves, alongside a spectrum of research applying bibliometric methods to address clinical research topics.

The COVID-19 pandemic has significantly impacted academic research, leading to a notable increase in COVID-19-related scholarly publications, as evidenced by bibliometric studies. Clinical trials, another pivotal research activity in medicine, extend over multiple years, also reflecting an enduring scientific effort. The global pandemic presents many predictable challenges in conducting clinical trials, including difficulties with patient recruiting, in-person data collecting, and sudden suspension of interventions (Audisio et al., 2022; Ledford, 2021; van Dorn, 2020). For instance, the social distancing measures and quarantines brought about by the pandemic could impede the conduct of clinical trials, potentially causing the suspension or cessation of trials and thereby negatively impacting trial efficiency. As previously highlighted, the surge of COVID-19 channeled a substantial amount of medical and scientific research resources toward addressing the pandemic, potentially leading to a relative decrease in research activity for other diseases, indicating a crowding-out effect of COVID-19 on clinical trials of non-COVID-19 conditions. Additionally, while the importance of coordination and collaboration in response to the pandemic is widely recognized, challenges arise in large-scale clinical trials across multiple regions (Park et al., 2021). Unlike publications, which can often be discussed through virtual meetings, clinical trials require in-person interactions involving patients, clinicians, and researchers. This requirement for physical interaction might make clinical trials, particularly those with international collaborations, more vulnerable to disruptions due to pandemic-induced social distancing and quarantine protocols. These challenges include increased pressure on coordinating centers to maintain oversight, lack of standardized workflows across research sites, and differences in laws and regulations among various countries (Hashem, Abufaraj et al., 2020).

Digital factors can potentially shape the overall crisis response patterns of clinical trials. Clinical trials are not an exception to the expected increase in the usage of digital technology brought on by the COVID-19 pandemic (Chen et al., 2021; Smith, Thomas et al., 2020; Upadhaya, Yu et al., 2020). As telemedicine has grown in acceptance and popularity among patients, physicians, and health authorities (Holtz, 2021; Sathian, Asim et al., 2020), it is plausible to suppose that the transition from a traditional to a remote digital mode for clinical trials may be beneficial for their continued operation during the pandemic. A 2020 Nature publication, leveraging 18 years of data from ClinicalTrials.gov, highlighted the significant rise in the utilization of connected digital products (CDPs) throughout all stages of clinical trials and by a diverse array of sponsors (Marra, Chen et al., 2020). Studies examining the nature of clinical trial sponsors have revealed notable disparities between industry-sponsored and nonindustry-sponsored trials (Atal, Trinquart et al., 2015; Cooper, Lee, & Waldron Lechner, 2021). The extent to which these disparities are magnified during a pandemic could further impact the allocation of resources for medical research on non-COVID-19 diseases.

The COVID-19 pandemic has had a multifaceted impact on clinical research, affecting the activity levels of research on various diseases, the implementation efficiency of clinical trials, and international collaborations. However, certain factors, such as the integration of digital technologies and the driving forces of different types of sponsors, may have mitigated some of these effects. Existing studies have shown mixed results regarding these impacts. During the pandemic, several scholars have qualitatively discussed the impact of COVID-19 on clinical research (Sathian et al., 2020; Tuttle, 2020). Medical researchers have also explored the impact of COVID-19 on clinical trials and clinical research by manually reviewing the relevant literature, with six full-text records ultimately included in their analysis. The study found that globally, most sites conducting clinical trials unrelated to COVID-19 experienced delays and a complete halt of operations due to the pandemic. This result suggests that there may be a crowding-out effect of clinical trials for other diseases during the COVID-19 period. However, this study was based on a small sample and lacked large-scale analysis methods, which bibliometrics can provide (Sathian et al., 2020).

Therefore, this study attempts to use clinical trial data, combined with bibliometric methods, for a more multidimensional analysis. The analysis of activity in this study is not only focused on the trend and growth rate of COVID-19-related research, because a high level of attention to COVID-19 during the pandemic is to be expected. More importantly, this study attempts to combine bibliometrics with text analysis, by constructing a disease mapping system to systematically focus on the crowding-out effect of activity in research on other diseases from a quantitative perspective during the COVID-19 pandemic. This approach aims to provide a comprehensive understanding of the broader impacts of the pandemic on the landscape of clinical research and to identify potential shifts in research focus across different disease areas.

Clinical trials can indeed be understood as two distinct types of data in the context of research and publication: (a) publications resulting from clinical trials and (b) registered clinical trials. Publications resulting from clinical trials are documents that report the outcomes of clinical studies where participants receive one or more interventions to evaluate their effects on biomedical or health-related outcomes. In PubMed, this article type is categorized separately as “clinical trial” and represents a significant focus within bibliometrics (National Library of Medicine, 2023). Research in the bibliometric field often centers on analyzing the “clinical trial” literature to understand trends, impact, and the dissemination of clinical trial results (Dagli et al., 2024; Rosas, Kagan et al., 2011; Tao, Zhao et al., 2012; Tsay & Yang, 2005). For instance, Rosas et al. (2011) used clinical trials publications from the National Institute of Allergy and Infectious Diseases (NIAID) HIV/AIDS extramural clinical trials networks to explore the presence, performance, and impact of papers published in 2006–2008. Tsay and Yang (2005) utilized bibliometric techniques to investigate the characteristics of randomized controlled trials (RCTs, one specific type of clinical trials) literature, including publication types, languages, countries of publication, and research topics. Similarly, Tao et al. (2012) aimed to characterize the most highly cited clinical research articles on sepsis, further illustrating the application of bibliometric analysis in clinical research assessment.

Clinical citation is indeed a significant focus within the bibliometrics community. Bibliometric methods are essential for evaluating the impact of biomedical research in translational science, offering quantitative and objective metrics of research influence (Jones, Cambrosio, & Mogoutov, 2011; Ke, 2019; Kim, Levine et al., 2020). By analyzing citation patterns of academic papers within the context of clinical trials, researchers can gauge how effectively academic research translates into practical clinical applications (Pallari et al., 2021; Park et al., 2019; Thelwall & Kousha, 2016; Zang & Liu, 2023). Recent studies have also aimed to understand the factors influencing the clinical translation of research and to predict clinical citation counts (Chen & Liu, 2022; Li, Tang, & Cheng, 2022; Liu, Wang, & Wang, 2024). For example, Li et al. (2022) developed a four-layer multilayer perceptron neural network (MPNN) model to forecast the future clinical citation count of biomedical papers. Such approaches underscore the ongoing efforts to leverage bibliometric techniques for predicting research impact in clinical settings.

Differing from publications resulting from clinical trials, registered clinical trials are not publications but rather the trials themselves, which can be either ongoing or completed. Ongoing trials may not yet have outcomes and, consequently, may not have resulting publications. This category of data represents the current activity and dynamism within a specific field of research. Although it reflects the vitality of research in real time, it has been less explored in the bibliometric domain. Registered clinical trials offer a valuable perspective on the research landscape. Integrating this data into bibliometric analyses offers a broader view of the research ecosystem, encompassing both published findings and the range of ongoing experimental work.

Several studies have employed bibliometric analysis of clinical trial data to explore the landscape of clinical research within particular fields. In the context of COVID-19, studies such as those by Xia et al. (2021) have utilized a combination of literature and ClinicalTrials.gov data to identify hotspots and trends in clinical research. Similarly, Gianola, Jesus et al. (2020) compared the volume and reporting characteristics of COVID-19-related academic articles, preprints, and the number of ongoing clinical trials and systematic reviews by integrating data from PubMed, preprint servers, and 18 clinical trial registries. In other disease areas, Nye, D’Souza et al. (2021), for instance, combined trial data with publication records to explore the research productivity, influence, and collaborative efforts within the HIV Vaccine Trials Network (HVTN) over the past two decades.

However, current utilization of clinical trial data often focuses solely on quantitative statistics to examine the landscape of specific fields, neglecting in-depth exploration and discussion of various dimensions of clinical trials. This approach fails to yield robust quantitative evidence based on large-scale data. A study published in BMJ Open investigated the characteristics of registered clinical trials evaluating treatments for COVID-19, concluding that global coordination and increased funding for high-quality research could optimize scientific progress in swiftly identifying safe and effective treatments during the pandemic (Mehta, Ehrhardt et al., 2020). Yet, such conclusions are typically the result of qualitative reviews and lack empirical support. The significance and irreplaceability of clinical trial data in the medical field, combined with the systematic nature of bibliometric methods, provide a natural intersection that the bibliometrics community should prioritize examining. Consequently, this study aims to deeply integrate clinical trial data with bibliometric methods, not only to address trends in the number of COVID-19 trials but, more importantly, to employ text mining techniques and the ITS model to comprehend the potential crowding-out effect of the COVID-19 outbreak on trials for non-COVID-19 diseases.

To thoroughly explore the effect of COVID-19 on trials of other diseases from the perspectives of activeness, efficiency, and MRCT, various fields of trial data, such as the start year, country/region, trial phase and status, trial outcomes, and number of enrollments, are essential. More importantly, defining and classifying non-COVID-19 diseases in our analysis required gathering associated disease information from each trial. As in previous studies, ClinicalTrials.gov⁴ is used as the primary data source in this study, owing to its comprehensiveness and the standardization of various fields, as well as its global representativeness of clinical trial data. ClinicalTrials.gov is an international clinical trial registry database maintained by the National Institutes of Health (NIH) and the Food and Drug Administration (FDA) in the United States. It serves as one of the largest global repositories for clinical trial information, offering a platform for researchers to register their trials and disclose trial designs and outcomes to the public. Although ClinicalTrials.gov is administered by US entities, it accepts clinical trial registrations from around the world, encompassing a diverse array of trials conducted globally. ClinicalTrials.gov is a widely used Web-based resource at the NIH since 2000, which enables quick access to worldwide data on clinical trials on a variety of diseases. Table 1 presents a representative sample of clinical trial data, exemplified by trial NCT04129216, which is dedicated to investigating breast cancer. Registered in 2019, the trial is in the recruiting phase (data acquisition time), sponsored by Johns Hopkins University, and is designed as an interventional study. For comprehensive clarification on each data field, readers can refer to the “Data Element Definitions” provided by ClinicalTrials.gov⁵.

Trial data fields are designed to reflect the most recent status at the time of data collection. For example, the field of recruitment status indicates the current stage of a trial: whether it is in the planning phase, ongoing, or completed. When assessing the impact of the COVID-19 pandemic on the progress or efficiency of trials related to different diseases, it would be inappropriate to solely compare the recruitment status of trials initiated before and after 2020 using data collected at a single point in time. This is because trials that start in different years but are collected at a single time point have varying durations. For instance, consider a trial initiated in 2018 (prepandemic) and another in 2020 (postpandemic), with data extracted uniformly in 2022. Assuming a trial initiated in 2018 has reached a completed status and one that began in 2020 is still recruiting by 2022, it would be a misinterpretation to conclude that the efficiency of the 2018 trial is inferior to the 2020 trial. This assessment would not account for the fact that the 2018 trial had a 4-year duration by the time data was extracted in 2022, while the 2020 trial had only been active for 2 years. Obtaining clinical trials initiated before and after the pandemic with the same duration is of great significance for a comparable analysis of trials’ efficiency. We acquired trial data from ClinicalTrials.gov in the format of independent XML files in June 2021 and July 2022 separately, referred to as the 2021 set and 2022 set hereinafter. As a result, trials started in 2019 in the 2021 set, and those started in 2020 in the 2022 set, can be regarded as two comparable subsets of trial data in this study, as shown in Figure 1.

In addition, although ClinicalTrials.gov is an international registration platform and its data is representative, some trial data from national registries cannot be disregarded. Further trial data from Dimensions⁶ was supplemented to present a study that is reasonably thorough and capable of generating reliable results for national comparison. Digital Science launched Dimensions.ai in January 2018, which is a partly free scholarly database containing multiple types of scientific data, including clinical trials registered with 15 major national registries worldwide. Note that supplemental data from Dimensions was only used for activeness analysis, as the database did not contain the required fields for subsequent analysis. Trials listed on ClinicalTrials.gov are assigned unique NCT numbers as their identifiers. These NCT numbers are also recorded in Dimensions, facilitating the deduplication of trials extracted from both platforms.

In sum, we gathered our trial data from ClinicalTrials.gov and Dimensions. Regarding the COVID-19 pandemic, clinical trials that began in 2015–2019 and in 2020–2021 were viewed as before-and-after comparisons. Note that we included only interventional studies due to the distinct experimental designs in comparison to observational trials.

Identifying and categorizing diseases of trial data is the essential procedure for this research. Figure 2 illustrates several important procedures used in disease mapping.

In this study, we utilized Medical Subject Headings (MeSH), a biomedical indexing vocabulary maintained by the US National Library of Medicine (NLM), as the primary tool for disease mapping. To obtain relevant MeSH terms for the trials, we primarily relied on the “condition-MeSH” field within trial data from ClinicalTrials.gov. For trials with empty “condition-MeSH” fields and those supplemented from Dimensions, we employed the NLM Medical Text Indexer (MTI)⁷ to extract MeSH terms from the titles and abstracts. The MTI, a key component of the NLM’s Indexing Initiative project, has been providing indexing recommendations based on the MeSH vocabulary since 2002 (Mork, Aronson, & Demner-Fushman, 2017). It operates by analyzing the titles and abstracts of MEDLINE citations to extract MeSH terms, and its design and workflow have been described in a previous study (Zhang, Zhao et al., 2020a).

Regarding the disease classification, we adopted the Global Health Estimates (GHE) cause category, which is a hierarchical structure with four levels based on the International Classification of Diseases, 10th edition (ICD-10) and established by the World Health Organization (WHO, 2017). The GHE cause category comprises four hierarchical levels. At level 1, there are three broad cause groups: Group I—communicable, maternal, perinatal, and nutritional conditions (CMNNs); Group II—noncommunicable diseases (NCDs); Group III—injuries. We also created a separate category for COVID-19 to facilitate appropriate comparisons with CMNNs and NCDs. GHE level 2 provides further stratification of the three major disease categories from level 1, offering a more detailed classification to facilitate a granular analysis of health conditions. We did not include Group III (the injuries category) in our analysis, as it primarily addresses transportation and intentional/unintentional injuries rather than diseases. A detailed explanation of the GHE cause category can be found from a previous study (Zhao, Wang, & Zhang, 2022).

The mapping between MeSH and GHE cause category is established by taking advantage of the concordance table between ICD-10 and MeSH terms built by Yegros-Yegros, van de Klippe et al. (2020). The concordance table used in Yegros-Yegros et al. (2020) was primarily designed for retrieving publication data from PubMed, which has a built-in feature for automatically searching both the MeSH headings and the subordinate terms beneath those headings in the MeSH hierarchy, known as “automatic explosion.” In our effort to map diseases to trial data, we enhanced the concordance table by including all subordinate terms for each MeSH term, based on the MeSH tree hierarchy⁸. Additionally, we manually examined each MeSH term extracted from trial data that wasn’t initially recognized as a disease in the extended mapping relationship. A manual double check was also applied to prevent the overlooking of disease-relevant MeSH terms and to ensure that subordinate terms were not categorized into multiple disease categories simultaneously. A full-count assignment scheme was applied in calculating the number of trials for each disease; that is, the trials count of each disease expressed how many trials the disease was labeled in.

MTI can also process arbitrary biomedical texts to provide an ordered list of MeSH terms. Such a process can enable disease mapping for multiple types of scientific data (publications, grants, patents, etc.) when used with the extended mapping relationship between MeSH terms and the GHE cause category established in this study. Overall, the number of clinical trials obtained and utilized for this study is shown in Table 2. After the de-duplication of trials, Table 3 further presents the distribution of clinical trials by disease type in GHE level 2.

We classified the trial’s sponsorship type using the Global Research Identifier Database (GRID), employing the “Sponsors/Collaborators” field in ClinicalTrials.gov and “GRID IDs” in Dimensions⁹. Notably, “Sponsors/Collaborators” encompasses organizations that initiate and have authority and control over the trial, along with other entities involved in the clinical trial. GRID is an open repository of unique identifiers for research-related organizations, designed for public use within the research community¹⁰. We utilized GRID data released on September 16, 2021¹¹, which includes organizational details like GRID ID, organization name, address information, and relationships with other organizations. Furthermore, GRID classifies organizations into eight distinct types: healthcare, education, company, facilities, government, nonprofit, archive, and others. The official website provides a detailed interpretation of the above types¹².

We matched GRID ID and classification information for ClinicalTrials.gov trials with the organization names provided by Sponsors/Collaborators, corresponding to the organization names in the GRID data set. For Dimensions data, we directly obtained classification information based on the GRID ID. Figure 3 illustrates the overall distribution of clinical trials by sponsorship type. It’s important to clarify that “archive” designates organizations responsible for preserving research and cultural heritage materials, encompassing entities like libraries, museums, and zoos. We consolidated “archive” and “other” into a single category labeled as “others” due to their relatively low number of trials.

During the pandemic, digital technology telemedicine played a role in sustaining ongoing clinical trials and partially alleviating the disruptions caused by COVID-19. Connected digital products (CDPs) are software and sensor-based technologies that are portable and intended for patient use. CDPs gather health-related data from patients, facilitating remote monitoring and virtual consultations within clinical research. Identifying clinical trials that utilize CDPs offers a tangible approach for extracting trials utilized telemedicine technology.

CDPs encompass various tools, such as wearable activity trackers, heart rate monitors, mobile apps for data management, ingestible sensors, and mobile platform-based health assessments (Marra et al., 2020). Marra, Gordon, and Stern (2021) also developed a comprehensive list of 140 search terms to identify telehealth and remote monitoring use in clinical trials. By applying their search terms and adding several additional relevant terms, we employed an automated algorithm to search for each term in various relevant trial fields, such as “title,” “abstract,” “interventions,” and “eligibility criteria,” which allowed us to identify trials utilizing CDPs. This approach enables us to investigate the shift from traditional to remote digital modes in clinical trials during the pandemic. In general, 15,854 interventional trials have been identified using CDPs. The search term list is provided in the Supplementary material, Table S1.

This study employs a variety of indicators and analytical models to assess the impact of COVID-19 on the activeness, efficiency, and international collaboration of clinical trials.

In this study, trial efficiency is operationalized through two dimensions: speed and enrollment scale. The “recruitment status” field for each clinical trial is utilized to evaluate the impact of the pandemic on trial progression speed. As previously stated, we have ensured comparability by analyzing trials initiated both before and after the pandemic over equivalent time frames. Regarding “enrollment scale,” this pertains to the number of participants enrolled in a trial. The analysis of scale focused on whether the pandemic and the imposition of quarantine restrictions have altered the trial’s enrollment scale.

Regarding collaboration, our study primarily focuses on international collaboration, specifically on multi-regional clinical trials (MRCTs), a pivotal concept within the clinical trials domain. Table 4 presents the descriptive statistics on the number of collaborators and countries involved in clinical trials from 2015 to 2021. The data show that the average number of countries involved in a single clinical trial remains relatively stable, with a value consistently below 2. Similar to studies that utilize publication data to explore international cooperation, this research primarily employs the quantity and proportion of MRCTs as foundational analytical metrics. Additionally, we utilize the Strength of Collaboration index (Salton’s index), which is defined as the number of joint publications divided by the square root of the product of the total publications of the two countries (i.e., the geometric mean), to discuss the intensity of collaboration among different countries in the realm of clinical trials research.

In this study, we characterize the disease-specific trials’ activeness using the total number of clinical trials started before and after the pandemic for each disease. Figure 5 demonstrates the number of clinical trials by disease category from 2015 to 2021 worldwide and for four countries. The United States, China, and Japan are the three countries with the largest number of trials. The United States ranks first in all categories of disease. Within each disease, the distribution patterns across disease categories are quite similar (i.e., a higher proportion of publications on noncommunicable diseases (NCDs) than for communicable diseases, and maternal, neonatal, and nutrition conditions (CMNNs)).

In terms of dynamic trends, there has been a consistent increase in the overall number of trials globally over the past 7 years. Interestingly, the global total trial volume has not been significantly affected by the pandemic. This suggests that clinical trial activity, irrespective of the specific disease, has not experienced substantial disruptions due to the epidemic. This finding aligns with the observations made by Agarwal and Gaule (2022), which indicated that the overall volume of clinical trials increased in 2020, with a limited impact of COVID-19 on overall trial volume. In all countries, the quantity of COVID-19-related trials declined in 2021 to varying degrees. The NCD trials’ activeness in the United States, the United Kingdom, and Japan showed a recovery pattern relative to 2020, whereas it declined sharply in China.

The increase in CMNN and COVID-19 trials during the first year of the pandemic is entirely expected, as COVID-19 can be categorized as a respiratory infection. However, the number of NCD trials did not rise globally in 2020, indicating the potential crowding-out effect on non-COVID-19 diseases. In particular, the number of NCD trials in the United States, the United Kingdom, and Japan all drastically decreased in 2020, by 13.80%, 20.32%, and 33.84%, respectively. From 2018, there have been fewer trials in Japan; this drop peaked in 2020. In contrast to the other three countries, China exhibits a quite distinct pattern. The pandemic has led to a decrease in the number of NCD trials in other countries, with China being the only nation where the number of NCD and CMNN trials increased during the epidemic. The growth in NCD trials in China was also a major factor in the comparatively low declining rate of NCD trials worldwide (0.14%).

While the overall activity of clinical trials was not affected by COVID-19, we found this insignificant effect was due to offsetting among different diseases, indicative of a crowding-out effect of COVID-19 on clinical trials for other diseases. This observation was substantiated by the ITS regression analysis, as shown by Figure 6 and Table 5. In this context, NCDs and CMNNs pertain to disease categories that exclude trials related to COVID-19 and NCD/CMNN comorbidities. For all diseases, there was a consistent upward trend prior to the pandemic, with the overall number of trials showing little impact from the pandemic, aligning with the aforementioned observations. Regarding NCDs, a stable growth pattern was observed before the pandemic; however, there was a notable decline in the number of trials in 2020 (p = 0.003 < 0.005), followed by a recovery trend post-2020. For CMNNs, a relatively stable, slight downward trend was present before the pandemic. However, the intense focus on COVID-19 during the pandemic led to a significant decrease in trials unrelated to COVID-19 among CMNNs (p = 0.003 < 0.005), with a continued stable decline observed subsequently. That is, COVID-19 has had a significant crowding-out effect on trials for non-COVID-19 diseases, affecting both NCDs and CMNNs.

Figure 7’s growth rate offers a more detailed picture of COVID-19’s impact on non-COVID-19 trial activity, with a clearer illustration of the crowding-out effect on trials for other diseases under the more detailed GHE level 2 classification, excluding COVID-19 and related comorbidities. That is, the rising trend of the number of trials on most NCD subtypes was “interrupted” by the outbreak of COVID-19 in 2020. Malignant neoplasms represented more than 25% of all disorders in terms of the number of trials, and interestingly, their growth trend continued in 2020. An in-depth analysis revealed that China’s substantial increase in neoplasm trials in 2020, with a rise of 47.44%, was the primary factor behind the global increase in trials related to this disease, unlike the decline in cancer trials in other countries.

Regarding CMNN, the correlation between COVID-19 and respiratory infections led to a natural increase in the total number of respiratory infections-related trials, with a growth of over 120%. However, the exclusion of trials related to COVID-19 and respiratory infections comorbidities has led to the shift in the growth rate of respiratory infection trials from positive to negative, as observed in Figure 7. This change indicates that the increase in respiratory infection trials in 2020 was predominantly related to COVID-19. Likewise, the number of trials exclusively focused on infectious and parasitic diseases decreased more significantly when excluding COVID-19-related comorbidities.

Overall, the COVID-19 pandemic has had a minimal disruptive effect on the overall activeness of registering clinical trials. However, given the massive attention and effort directed toward COVID-19 and its potential comorbidities, our results reveal varying degrees of crowding-out effects on non-COVID diseases, encompassing both NCDs and CMNNs. Additionally, the activeness of NCD trials during the pandemic displays significant variation among nations, with China exhibiting a distinct pattern compared to the other three nations.

This study defines the efficiency of a trial from two aspects: speed and enrollment scale.

Figure 8 demonstrates the distribution of trials’ status by disease and by country, using two comparable subsets of trial data (Figure 1) before and after the pandemic. There are nine types of statuses established by ClinicalTrials.gov, which we further organized into four categories for clarity: ongoing (“recruiting,” “active, not recruiting,” “not yet recruiting,” “enrolling by invitation”), stopped early (“withdraw,” “terminated,” “suspended”), completed (“completed”), and unknown (“unknown status”).

Except for COVID-19, most clinical trials related to NCDs and CMNNs are in the “ongoing” status, with the majority of these trials falling under the “recruiting” category. More than 43% of COVID-19 trials have been completed in 2 years, which is a significant indication of the intensive and effective efforts contributed by academia to address the pandemic. Moreover, COVID-19 trials had the highest proportion of “stopped early” trials, indicating a rapid response and decision-making approach, with the trials being halted swiftly if risks or issues arose. Initially, “unknown” status refers to the status not last verified within the past 2 years. Interestingly, the relatively high proportion of COVID-19 trials with an “unknown” status, when compared to NCD and CMNN trials, suggests that researchers may register COVID-19 trials rapidly to address urgent concerns, follow emerging hotspots, or for other reasons without providing subsequent follow-up or status updates.

Despite the rapid response and decision-making pattern observed in COVID-19 trials, the distribution of trial statuses remained consistent from a national and a disease perspective. This finding indicates that COVID-19 has not significantly affected the pace of trials for other diseases. That is, when researchers worked to combat the outbreak, the progress of studies on other diseases was not significantly disrupted by the COVID-19 pandemic.

The enrollment scale in clinical trials varies considerably for a variety of factors. In addition to disparities among nations and diseases, the trial phase is a nonnegligible factor that significantly determines the enrollment scale. For instance, the US Food & Drug Administration (FDA) offers suggestions for the enrollment scale for various trial phases (US Food & Drug Administration, 2018). This study divided the enrolment number into six segments by referring to the recommendation from the FDA (0; 1–20; 21–100; 101–300; 301–3,000; >3,000).

It is important to note that the enrollment information offered by ClinicalTrials.gov for the recruiting status of “completed” and other types of trials differs. Enrollment for “completed” studies refers to the actual participants in the trial, whereas for other types of trials it refers to the target number of participants. The completed trials are applied here to reflect the enrollment scale accurately. Figure 9 presents the distribution of the actual enrollment scale of completed trials by phase and country. The average number of enrollments by disease category is shown in Table 6.

The distinct patterns across the five phases are depicted in Figure 9. Phase 2 trials exhibit the highest proportion of trials with enrollments exceeding 100 and 300 participants, followed by phases 1 and 3 trials. From the viewpoint of before-and-after pandemic comparisons, there was no structural change in the fraction of trials with varied scales for the five phases. Nevertheless, in all five phases, the scale fraction with 1∼20 participants was reduced, which in part suggested a little widening of the enrollment scale. This trend is also evident in the distribution across the four nations; the proportion of trials with more than 100 participants has increased following the pandemic. Upon closer examination of the data, we find that studies involving over 3,000 volunteers have been conducted in all four nations (the United States, the United Kingdom, China, and Japan) since the pandemic. Nearly 50% of the 73 studies with over 3,000 participants were COVID-19 trials.

The number of enrollments by disease category, as shown in Table 6, makes this tendency easier to see. Compared to NCDs and CMNNs, COVID-19 has a significantly greater average participant count. Furthermore, enrollment in NCDs and CMNNs also improved even when trials on COVID-19 and NCDs/CMNNs comorbidities were excluded from the calculation of the average number of participants for these two categories.

Effective emergency response in clinical research necessitates collaboration, particularly among clinical researchers exchanging clinical insights with colleagues across various specialties (Eke, Morone et al., 2021). Similar to coauthored academic papers, MRCTs are also a vital form of medical research collaboration. A clinical trial defined as an MRCT is conducted under a single protocol in more than one region (ICH, 2018). A significant advantage of MRCTs is the acceleration of global clinical development and the facilitation of registration in every part of the world, which can enable the worldwide distribution of novel medications to patients as quickly as is scientifically possible. According to Fry et al. (2020), the geographic loci of coronavirus research generally, as well as the structure of academic teams, were shifted by the onset of COVID-19. Yet, although most studies on COVID-19 involve collaborative analysis with publication data (Pathak, 2020; Wu, Yang et al., 2021; Zhang, Zhao et al., 2020b), little attention has been given to examining the collaboration patterns exhibited by clinical trial data.

The identification and calculation of MRCTs can be performed by utilizing the field “Locations” and extracting country information in trial data. Figure 10 presents the proportional changes in MRCTs by country/region. With an average percentage of MRCTs of 8.69% between 2015 and 2019, most trials are carried out independently globally. Only China has a lower rate of MRCTs than the global average. Notably, half of the clinical trials conducted in the United Kingdom and Japan during the COVID-19 pandemic were MRCTs. The finding sheds light on the fluctuating trends observed in the number of trials conducted in these two countries, as illustrated in Figure 5. In contrast to the United States, both the United Kingdom and Japan have a comparatively lower number of trials, with a considerable proportion of them being international collaborations. Due to the smaller scale and preponderance of collaborative trials, multiple factors are likely to influence the magnitude of changes, making it easier to demonstrate significant variations.

From 2015 to 2019, the percentage of MRCTs continuously decreased globally. Yet again, the COVID-19 pandemic “interrupted” this downward trend. That is, a significant increase in the MRCT percentage was seen in 2020, and this tendency was observed in all four nations, indicating strengthened collaboration among countries with academic capabilities. Further, most MRCTs are conducted with involvement from two to four countries. Our results also demonstrate that COVID-19-related MRCTs involve, on average, fewer countries than NCDs and CMNNs, consistent with Fry’s finding of narrowing team membership in coronavirus research during the COVID-19 pandemic. The preference for strong alliances among countries with solid research capacities when responding to emergent outbreaks may contribute to the “shrinking scale” of collaboration (Fry et al., 2020; Zhao, Zhang et al., 2024).

A further observation is provided from the perspective of national comparative analysis, as shown in Figure 10b. Compared to the United States, which participated in 65% of the MRCTs worldwide, China participated in less than 8%. China has progressively boosted its involvement in the worldwide MRCTs since the China Food and Drug Administration (CFDA, now known as the National Medical Products Administration) joined the ICH¹³ in 2017 (WCG FDAnews, 2017). In addition to independent trials, bilateral collaboration is the most common type of MRCT for China and the United States, whereas the United Kingdom and Japan have engaged more in MRCTs with a greater number of countries.

We further constructed a co-occurrence matrix for pairs of collaborated countries and calculated collaboration intensity using Salton’s measure. VOSviewer (https://www.vosviewer.com/) was applied to establish the global collaboration network of MRCTs before and after the pandemic, as depicted in Figure 11. When comparing the network before and after the pandemic, we observe that the post-pandemic network has a higher link density, indicating increased global collaboration on clinical trials. By calculating the collaboration strength between the four countries we focused on, we observed an increase in collaboration strength for each pair of these countries following the pandemic, as shown in Figure 11c. These findings align with the results presented in Figure 10.

Another notable observation is that countries conducting the most trials, such as the United States and China, do not necessarily have the most extensive or highest collaboration strength. This contrasts with general findings from previous literature (Zhang, Zhao et al., 2020b; Zhao et al., 2024), where countries with the largest literature volume tend to exhibit extensive and close collaboration networks. However, our results align with the unique characteristics of clinical trials, which often necessitate the enrollment of a large number of participants for MRCTs. As a result, it can be challenging for countries with smaller populations to conduct large trials independently, leading them to engage more in MRCTs, as exemplified by the United Kingdom and Japan.

In general, our results provide evidence that collaboration among countries with academic capabilities in research and innovation intensified during the pandemic (Guimon & Narula, 2020), in addition to the results from previous studies using publications data.

Previous findings reveal the multidimensional impact of COVID-19 on clinical trials. While the total number of trial registrations remained largely unaffected, a pronounced crowding-out effect was observed for non-COVID-19 diseases, particularly NCDs. Nevertheless, the academic community demonstrated a rapid response to COVID-19 research, and research on other diseases maintained its original efficiency in speed and scale. Additionally, COVID-19 prompted increased global collaboration in clinical research. Investigating the reasons underlying our observational results could be an intriguing avenue for a more comprehensive understanding of the COVID-19 impact on clinical studies. By reviewing existing related literature and reports (Cohen, 2020; Inan, Tenaerts et al., 2020; Marra et al., 2021) and conducting the above analysis, we have summarized and further validated the potential factors within three dimensions.

Digitizing clinical trials presents new opportunities for improving trial efficiency, such as enabling remote patient monitoring and enhancing data collection quality, quantity, and frequency (Garg, Williams et al., 2018). Using the approach outlined in Section 3.4, we identified trials utilizing connected digital products (CDPs) and quantified the trend from traditional to remote digital modes in clinical trials. Figure 12 illustrates a significant increase in the total number and growth rate of trials utilizing CDPs after the pandemic, with the most notable surge occurring in 2021. These findings align with a previous study published in NPJ Digital Medicine by Marra et al. (2020), highlighting a significant increase in the use of connected digital products in clinical research.

Additionally, Figure 12 presents the distribution of trials using CDPs by disease, highlighting that mental and substance disorders, malignant neoplasms, and diabetes are the three disease categories with the highest number of digital trials. The COVID-19 outbreak in 2020 led to the immediate adoption of digital trial technologies for combating this infectious disease. In 2021, while the actual number of COVID-19 digital trials remained stable, their proportion decreased. This suggests that the pandemic accelerated the broader use of CDPs for trials related to diseases beyond COVID-19. In essence, the pandemic promoted the digitization of clinical trials for various diseases, including COVID-19 and others.

The increasing use of CDPs during the pandemic and the ongoing digitization of clinical trials have contributed to trial efficiency. This context provides additional support for the interpretation of our findings. This trend was also anticipated to persist during the pandemic as efforts were made to mitigate COVID-19-related challenges, such as the “virtual” clinical trial guidance issued by the FDA early in the pandemic (FDA, 2020). Moreover, the digitization of clinical trials may lower the cost of collaborations, which could be an additional incentive for intensified collaboration.

To some extent, this observation explains that COVID-19 had a more pronounced crowding-out effect on trials sponsored by nonindustry section (government and nonprofit organizations) than those sponsored by companies. This difference can be attributed to their distinct goals (wellbeing-driven vs. profit-driven). The ITS regression analysis substantiates these observations for various sponsorship categories. We refined the seven sponsorship types depicted in Figure 13 into two broad categories—industry and nonindustry—to assess the effect size by sponsorship type, revealing distinct crowding-out effects, as detailed in Table 7. We find that the crowding-out effect of COVID-19 on clinical research resources is predominantly seen in nonindustry-sponsored clinical trials, with both NCDs (p = 0.007 < 0.01) and CMNNs trials being significantly crowded out (p = 0.006 < 0.01). In contrast, no significant crowding-out effects were observed for industry-sponsored clinical trials. This suggests that the allocation of resources for nonindustry-sponsored trials to COVID-related research has served to meet immediate public health emergencies. However, this reallocation has also had a direct impact on the activeness of clinical trials focused on long-term, high-burden diseases, such as neoplasms and cardiovascular diseases, particularly those initiated by nonindustry institutions. The relatively stable performance of industry-sponsored trials may have played a role in maintaining the overall efficiency of clinical studies during the pandemic. Therefore, we argue that different sponsors might play different roles in responding to the COVID-19 pandemic, indirectly affecting the number of trials for various diseases.

As evident in Section 4.1, China is the only nation where the number of NCDs and CMNNs trials increased during the pandemic, exhibiting a distinct pattern compared to the other three countries. To further illustrate the impact of China’s number of trials on the global total, we calculated the growth rate of clinical trials by disease, excluding China’s trials and trials related to COVID-19 and NCD/CMNN comorbidities. As depicted in the Supplementary material, Figure S1, nearly all NCDs and CMNNs trials exhibit a noticeable downward trend. This reveals that China’s distinct crisis response pattern significantly contributed to maintaining the overall growth trend in the total number of trials on a global scale. Specifically, the negligible variation in the number of NCD trials observed globally (Figure 5) could be attributed to China’s increasing number of trials and other countries’ decreasing number of trials, effectively neutralizing each other.

China’s unique crisis response pattern may be attributed, in part, to the differential progression and phase of the pandemic within the country compared to the rest of the world. For example, based on the confirmed and death cases (Center for Systems Science and Engineering (CSSE) at Johns Hopkins University (JHU), 2020; Zhao et al., 2024), the initial outbreak in China occurred in late 2019 to early 2020, followed by containment and mitigation from early 2020 to mid-2020. China’s efficient centralized government allowed for the quick implementation of strict measures during the pandemic’s initial phase, leading to rapid virus containment in affected regions. Meanwhile, many other countries implemented lockdowns, travel restrictions, mask mandates, and social distancing measures during the early to mid-2020 period, as the pandemic was still in its early stages in those regions. Overall, China had relatively more resources at its disposal, including time, a larger population, and a greater pool of available researchers and trial participants for conducting clinical trials in 2020. These factors may explain why Chinese academia exhibited high activity in registering trials in 2020.

In this study, we investigated the impact of COVID-19 on medical research resources, particularly its crowding-out effect on non-COVID-19 diseases, using clinical trial data. We examined this effect through trial activeness, efficiency, and scientific collaboration in MRCTs. Further, by identifying trials using connected digital products (CDPs) and classifying sponsorship types, we summarized and validated potential factors across three dimensions to shed light on the observed effects of COVID-19 on clinical studies. These factors encompass the digitization of clinical trials, the distinct roles played by various sponsors (those driven by wellbeing vs. profit), and the differing responses of countries, such as China, in contrast to other nations. In addition to global observation, the United States, China, Japan, and the United Kingdom are chosen as representative examples to conduct a more fine-grained comparative study.

Our findings highlight the multifaceted impact of COVID-19 on clinical trials. Notably, the overall clinical trial activeness, regardless of the disease categories, has not been significantly disrupted by the pandemic, as evidenced by the minimal change in the total number of trial registrations. However, a pronounced crowding-out effect was observed in both the number and growth rate of trials related to non-COVID-19 diseases, including both NCDs and existing CMNNs. We further employed the interrupted time series (ITS) model to verify those findings. Contrary to the assumption that the COVID-19 pandemic might substantially affect the efficiency of clinical trials, our analysis, based on clinical trial data, did not reveal significant effects of the pandemic on the speed and scale of enrollment in trials. Even though researchers have been working intensively on COVID-19 to contain the outbreak, the progression and completion of other disease trials have not been significantly impacted by the pandemic. Instead, the academic community responded rapidly to COVID-19 research, with 43% of COVID-19 trials being completed within 2 years and an increase in the average number of participants. Moreover, enrollment in trials related to NCDs and CMNNs also improved during the pandemic. Additionally, COVID-19 prompted increased global collaboration in clinical research, as demonstrated by a significant increase in MRCTs in 2020 and intensified global collaboration.

Three potential factors on the observed effects of COVID-19 on clinical studies were further summarized and validated through text-mining and statistical analysis. By examining the development of digital clinical trials, our study evidenced the significant increase in the total number and growth rate of trials utilizing CDPs after the pandemic and their expanded use for diseases beyond COVID-19. Such a trend offers new opportunities for enhancing trial efficiency and serves as an additional incentive for intensified collaboration. Thus, temporary policies to support digital health innovation during the pandemic should be encouraged when facing unknown diseases. Moreover, given the distinct goals of different sponsors (well-being driven vs. profit driven), we found that government and nonprofit organizations play significant roles in responding to COVID-19 and CMNNs. Therefore, we contend that the establishment of a rapid-response mechanism, emphasizing government and nonprofit organization-sponsored research in addressing future unknown diseases is crucial, as such unknown diseases demand substantial research resources. Considering the large number of potential pathogens and limited resources for disease R&D, our study also highlights the increased collaboration among countries with academic capabilities, again, emphasizing the significance of collaboration in the time of COVID-19.

In sum, this paper makes the following contributions. First, it provides an updated study and comparison of clinical trials initiated before and after the COVID-19 pandemic with worldwide trial data using an ITS model, leading to some discoveries regarding COVID-19’s crowding-out effect on medical research resources for non-COVID-19 diseases. Second, by employing the MTI and establishing the mapping relationship between the MeSH term and GHE cause category, this study identifies disease-specific trials and attempts to investigate the crowding-out effect under a more fine-grained disease classification. It is worth noting that the methods and tools used in this study can be further applied to analyze the allocation of research resources to different diseases in a broader range with different types of data. Third, the analysis of MRCTs performed in this study provides additional evidence on intensified scientific collaboration among countries with academic capabilities in addition to previous studies using publications data.

Two limitations need to be addressed in this study. The first one is the data-related limitation. To contribute a comprehensive and systematic analysis of the clinical trial before and after the pandemic, we drew our trial corpus from ClinicalTrials.gov and Dimensions. Although trial data from the major national registration platforms on clinical trials are included, certain trials may have been missed due to registration, location, and other reasons. Additionally, we used the concordance table between ICD-10 and MeSH terms from Yegros-Yegros et al. (2020), approved by a medical doctor, to guarantee the accuracy of mapping relationships between MeSH terms and GHE cause categories. Although we included additional subordinate terms based on the MeSH hierarchy and manual examination of our mapping relationship, certain trials with MeSH terms related to no particular disease will have been neglected.

The present study is an extended version of a conference paper presented at the 19th International Conference on Scientometrics and Informetrics: Zhao, W., & Du, J. (2023). Has COVID-19 crowded out medical research resources from non-COVID-19 diseases? Observations from clinical trials. Proceedings of ISSI 2023 – the 19th International Conference of the International Society for Scientometrics and Informetrics, 1, 773–786. https://doi.org/10.5281/zenodo.8280486

Wenjing Zhao: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Resources, Software, Validation, Visualization, Writing—original draft, Writing—review & editing. Chi Yuan: Data curation, Formal analysis, Investigation, Methodology, Software, Validation, Writing—review & editing. Zixuan He: Formal analysis, Investigation, Methodology, Validation, Visualization, Writing—review & editing. Jian Du: Conceptualization, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Writing—review & editing.

The authors have no competing interests.

This work was supported by the National Natural Science Foundation of China (Project numbers 72074006 to Jian Du); China Postdoctoral Science Foundation (Certificate Number: 2023M740154; 2024T170039 to Wenjing Zhao).

For reproducibility, the mapping relationship between GHE cause categories and MeSH terms generated during this study is made available to researchers worldwide in Figshare: https://doi.org/10.6084/m9.figshare.24720087.v1.