Introduction

The COVID-19 pandemic has caused an increasing demand for research with abnormal requests for preprints, journal submissions, and COVID-related publications (Fraser, 2021; Abramo et al., 2022; Watson, 2022). However, this so called “covidization of research” has not affect all academic groups equally (Ioannidis et al., 2022). For instance, research has shown that during the first wave of the pandemic, junior women submitted proportionally fewer manuscripts than men (Squazzoni, 2021; Madsen et al., 2022), and that this penalty was even more prominent for junior women working in less prestigious academic organisations located in less gender-equal countries (Kwon et al., 2023). Studies also found that academics doing COVID-related research had the easiest and fastest publications compared to those doing non-COVID related research (Aviv-Reuven & Rosenfeld, 2021), thus determining various potentially intertwined forms of inequality (Santos, 2022).

On the one hand, the pandemic had a negative impact on the work habit and routine of many organizations and institutions due to difficult access to lab facilities, suspension of fieldwork, and competing demands from family obligations due to home-schooling and parental care (Petts et al., 2021; EU Commission, 2023; Esquivel et al., 2023). On the other hand, the public measures to contain the pandemic, including especially strong mobility restrictions, could also have potentially benefitted certain academics by providing time to complete existing work or adapting existing research to new opportunities. As for employees in other sectors (Deole et al., 2023), working from home might have caused reduced working time for busy parents, but saving time for mobility and face-to-face meetings could have benefited those who could either rely on family support or did not have children or old parents to care for (Carr, 2021; Parlangeli et al., 2022).

However, it is difficult to assess these heterogeneous effects of the pandemic at the global level without a careful comparison of pre- and post-pandemic data that considers all research areas. Here, we have tried to fill this gap by examining the impact of the COVID-19 pandemic with full and complete data on about 9,2 million submissions to 2689 Elsevier journals from 2018 to 2021. We reconstructed academic origins from the affiliation of each author and assigned a research area to each submission via journal information.

This allowed us to reconstruct the growth rate of submissions and its global temporal patterns before and during the pandemic per country of authors’ affiliation and research area. Furthermore, we considered heterogeneity effects on submissions due to country-level anti-contagion measures. This permitted us to estimate the effect of the pandemic on academic productivity more systematically than in previous research.

Methods

Data

Our dataset included about 9.2 million submissions to 2689 Elsevier journals from January 1, 2018 to May 31, 2021 in four research areas: Health & medicine sciences (HMS), Life sciences (LS), Physical sciences (PS), and Social sciences & economics (SSE) (Table 1). Data access required a confidential agreement to be signed on 12th May 2020 between Elsevier and each author of this study (Squazzoni et al., 2017). We used the e-mail (sometimes various e-mails) associated with each submission’s author in the different submission systems used by Elsevier journals (i.e., Editorial Manager, Elsevier Editorial System, and EVISE) to reconstruct the country of affiliation of each author for each submission. Each submission was also assigned a unique research area depending on the journal which it was submitted to. To distinguish COVID related and non-related manuscripts, we used an internal Boolean flag from the manuscript submission systems used by journals in the Elsevier data. This allowed us to use a taxonomy of terms related to diseases caused by the same family of viruses to track back COVID-19 related manuscripts before the start of the pandemic (Squazzoni, 2021).

Table 1 Number of journals and submissions by year and research area
Full size table

To estimate the submission growth rate over time, we performed a seasonal trend decomposition using Loess, isolated periodical events, and focused on unexplained events in our time series. This decomposition was performed by using the STL method from the StatsFootnote 1 library in R.Footnote 2 To calculate and compare changes between different periods of our time series, we used the Autocorrelation-based dissimilarity method from the package TSdist.Footnote 3

To control for mobility restrictions, we used the COVID-19 stringency index, i.e., a composite measure based on nine policy responses. These nine measures included: school closures, workplace closures, cancellation of public events, restrictions on public gatherings,; closures of public transport, stay-at-home requirements, public information campaigns, restrictions on internal movements, and international travel controls. Calculated on a daily basis, the index considered the mean score of all nine metrics and takes a value between 0 and 100. A higher score indicates the introduction of stricter policy responses to the pandemic (Hale, 2021).

Results

The COVID-19 pandemic caused an abnormal rate of manuscript submissions to journals compared to the pre-pandemic period, mostly concentrated between February and May 2020 (see Fig. 4, Table 1 in the Methods Section, and Table 3 in the Appendix). This abnormal trend showed significant country- and research area-specific differences. While before the pandemic, the number of submissions from China, India, the United States, and Western European countries was comparable, the first wave of the COVID-19 pandemic created a more prominent growth rate of submissions from authors from China and India (+ 55% of submissions in 2018–2020: 674,180 submissions in 2018 vs. 1,048,717 in 2020) compared to authors from the United States and Europe (+35% of submissions in 2018–2020: 406,336 submissions in 2018 vs. 542,706 in 2020). We also found higher growth rates from certain peripheral countries, such as Bangladesh, Vietnam, Saudi Arabia, Indonesia, and Turkey (see the left panel in Figs. 1 and 4 in the Appendix).

To provide a more robust measurement of the submission patterns that also considered seasonality and temporal trends, we first calculated an autocorrelation-based dissimilarity index between pre and post-pandemic submission time series per country. Given that this measurement considers seasonality of submission trends, it could be considered as a proxy of how academics restructured their own research agenda and manuscript submission routines. Our results suggest that the countries most positively affected by the pandemic, were Nigeria (2.22), Bangladesh (1.95), Argentina (1.88), and Italy (1.87) (see the right panel in Fig. 1).

Fig. 1
figure 1

The submission growth rate and the auto-correlation index in 2018-2020 for the top 50 countries for number of submissions

Full size image

Furthermore, besides regional effects, this abnormal trend did not have the same impact in all research areas. Indeed, in the earlier onset of the pandemic, i.e., in February–May 2020, in correspondence with the highest peaks of submission growth rate, submissions were also mostly addressed to journals in Health & Medical Sciences. However, in later periods, research directed to Social Sciences & Economics journals became more prominent, probably reflecting the persistent interest of the academic community in examining the psychological, socioeconomic, and political implications of the pandemic (see Fig. 2).

Fig. 2
figure 2

Percentage of COVID-related submissions on the total number of submissions by research area: Health and Medicine Sciences in yellow, Social Sciences & Economics in blue, Life Sciences in orange and Physical Sciences in purple

Full size image

To check the hypothesis that the anti-COVID mobility restriction measures introduced, i.e., mobility restrictions, could have affected these patterns, we estimated a mixed-effect model where we used the stringency index calculated by the Oxford COVID-19 Government Response Tracker (Hale et al., 2020). This was to predict the auto-correction distance between pre-post pandemic submission trends.

Table 2 shows a positive and significant effect of the stringency index on the abnormal rates of submissions to journals, even when controlling for other variables, including the growth rate trends, and country-specific factors. For each increase of 10 in the value of the stringency index, we found an increase of 1% in the auto-correlation distance between pre- and post-pandemic submission trends. This suggests that mobility restrictions and other anti-contagion public measures at the country level had an effect on disrupting submission trends.

Table 2 Anti-contagion measures and submission patterns. Note that the auto-correlation distance was standardized, data were grouped by country, and divided into year quarters; we included a random-effect for countries. Estimate values can be interpreted as percentages over the auto-correlation distance
Full size table

Figure 3 shows a visual representation that confirms that countries with the strongest mobility restriction policies were also those where the country-specific pattern of submissions to journals varied more compared to their own pre-pandemic trends (e.g., Italy, Argentina, and India). In Northern European countries, where mobility restrictions were either not contemplated or only softly recommended, the effect of the pandemic on changing pre and post-pandemic submission patterns was weaker (e.g., Finland, Sweden, and Denmark) (more country-based detail in Table 6 in the Appendix). Although there is heterogeneity in the effect for certain specific countries, results suggest that generally strongest mobility restriction policies had a positive effect on submission trends.

Fig. 3
figure 3

The effect of the stingency index on country-level patterns of submissions to journals in 2018–2020

Full size image

Discussion and conclusions

Previous evidence has suggested that the annual growth rate of scientific publications globally was around 4% percent (Bornmann et al., 2021). Our results showed that the COVID-19 pandemic and the associated mobility restrictions boosted this trend with huge submission growth rates for certain countries. Besides the increasing demand for research to tackle the multi-faceted challenges of the global pandemic, previous research has suggested the anti-contagion measures have contributed to shape new routines and changing practices, including lecturing from home, and reduced time from daily work commuting and in-person meetings, which could have boosted productivity in the short-term (Pellegrini et al., 2020; Commission, 2023).

Our findings showed that mobility and anti-contagion restrictions had the positive effect of boosting manuscript submissions to journals and disrupting previous submission trends and their seasonality. We also found prominent country and research area-specific effects, with the emergence of a relatively new geography of the pandemic research with certain peripheral countries and non-Western regions in a prominent position. In general, with the exception of Italy at the frontline of COVID-19 spread, and therefore introducing stringent anti-contagion policies, submissions grew especially from non-Western countries (Santos, 2022).

Furthermore, we found an interesting area-specific trend in COVID-related research. While the highest peaks of the submission growth rate were initially addressed to journals in Health & Medical Sciences, a later growth of submissions was also visible for submissions to Social Sciences & Economics journals. This showed the changing focus of research from health issues to more general societal implications of the pandemics, echoing previous studies on temporal trend dynamics of adaptation of the scientific community to the complex nature of the pandemic crisis (Santos., 2022).

This said, our study has also certain limitations. The assumption of our study is that the country of affiliation of authors and their country of residence during the pandemic in 2020/2021 were the same. It is possible that some global academics could have decided to relocate in their home country to reunite with their family, thus anticipating the possible effect of restrictions either in the country of their affiliation or in the country of their family. However, given the large scale nature of our dataset, we believe that these cases would be randomly distributed across countries in the sample without affecting our results.

Furthermore, our study cannot help to estimate the long-term effect of these changes in the submission patterns. Although previous survey research has emphasized the expected long-lasting effects of the pandemic on the academic community (Gao et al., 2021), understanding the long-term implications of these new practices, including the co-existence of on/off line meetings and lectures, the new work-life balance of academics, and the expected reduction of international conferences, in terms of potential changes in the annual growth of scientific activities and their unequal distribution between research areas, requires more research (Ashencaen et al., 2021; Jack & Glover, 2021).

Furthermore, by providing new opportunities for fast publications, the abnormal submission growth due to the pandemic could have compromised ethical standards of research integrity, and weakened the filters of peer review, thus undermining the quality of scientific publications (Bauchner et al., 2020; Horbach, 2021; Faust et al., 2023). Research that considers the interplay of these multiple, quantitative and qualitative factors is also needed to inform possible interventions and understand how to maintain rigorous standards of research and scholarly communication in periods of global shocks.