Publication and Impact of Preprints Included in the First 100 Editions of the CDC COVID-19 Science Update: Content Analysis

Introduction

Preprints are publicly available, non-peer reviewed articles posted to various servers such as medRXiv and bioRxiv. Preprints first appeared as information exchange groups in 1961. However, due to resistance from journals, information exchange groups closed in 1967 [1]. In 1991, an automated email server that later became arXiv, a popular modern preprint server, was established [1]. In 2019, Cold Spring Harbor Laboratory collaborated with Yale University to launch medRxiv [1]. Preprints have gained popularity and credibility during the COVID-19 pandemic due to the need for rapid, relevant information to respond to the pandemic. To help inform the public health response to COVID-19, the Centers for Disease Control and Prevention (CDC) created the COVID-19 Science Update [2].

The COVID-19 Science Update provides brief summaries of new COVID-19–related articles on topics such as health equity, vaccines, variants, natural history, and testing, among others [2]. To provide the most relevant and timely information, the COVID-19 Science Update includes both published (peer-reviewed) articles and manuscript preprints. In collaboration with the World Health Organization (WHO), the Stephen B. Thacker CDC Library does a daily systematic (exhaustive, reproducible, and defensible) search for all COVID-19–related articles, which are then cleaned and deduplicated to be included in the WHO COVID-19 Database. The full search strategy for creating this database can be found on the WHO COVID-19 Database website [3]. The resulting articles are sent to the Science Update team within the CDC COVID-19 Response, who then select peer-reviewed articles and preprints on public health priority topics in the CDC Science Agenda for COVID-19 [4] and CDC COVID-19 Response Health Equity Strategy [5].

The COVID-19 Science Update was piloted biweekly beginning in April 2020 and has been publicly available since September 2020 (previous editions were retroactively posted on the internet) [2]. In November 2020, public release became weekly, and an email subscription became available in February 2021. The objective of this analysis is to evaluate the publication rate (percent of preprints published in peer-reviewed journals) and impact (eg, Altmetric Attention Score [6] and citation count) of preprints included in the COVID-19 Science Update and to assess the performance of the COVID-19 Science Update team in selecting impactful preprints.

Methods

All preprints in the first 100 editions (April 1, 2020, to July 30, 2021) of the COVID-19 Science Update were included in this analysis. Some preprints were eventually published (categorized as “peer-reviewed preprint”) whereas others were not published (categorized as “unpublished preprint”). Time to publish was the time interval between the date on which a preprint was first posted and the date on which it was first available as a peer-reviewed article. Impact was quantified (median and interquartile range) using both Altmetric Attention Score (a weighted measure of attention received from academic, news, and social media sources) and citation count (also from Altmetric) for all available articles (peer reviewed or not) on August 6, 2021. For peer-reviewed preprints, impact metrics only measure impact accumulated following publication. To allow time for preprints to be published and to accumulate impact, we restricted comparisons to items included in the 2020 editions (through Edition 70) of the COVID-19 Science Update. Statistical comparisons (Mood median test) were performed in Minitab with P<.05 for statistical significance. Preprints were analyzed by publication status, publication rate, preprint server, and time to publication.

Results

Among the 1,971 articles in the COVID-19 Science Update among the analysis period, 275 (14%) were preprints (Figure 1), most of which came from 1 of the following 3 servers: medRxiv (n=201, 73.1%), bioRxiv (n=41, 14.9%), or Lancet preprints with SSRN (n=25, 9.1%), with 8 (2.9%) coming from other sources. More than half (152/275, 55.3%) were published within the analysis period (April 1, 2020, to August 6, 2021). The median time to publish was 2.3 months (IQR 1.4-3.7). When preprints posted in the last 2.3 months were excluded (to account for the time to publish), the publication rate was 67.8% (143/211). Preprints included in the COVID-19 Science Update were published in 76 different journals, and 18 journals published at least three. Moreover, 2 journals (New England Journal of Medicine and Clinical Infectious Diseases) published 11 preprints each.

Peer-reviewed articles included in the COVID-19 Science Update (n=1696) had a median Altmetric Attention Score of 365 (IQR 64-1316) and median citation count of 33 (IQR 9-111) (Table 1). For peer-reviewed preprints (n=152), median Altmetric Attention Score was 265 (IQR 29-1896) and median citation count was 19 (IQR 3-101). For unpublished preprints (n=123), Median Altmetric Attention Score was 146 (IQR 22-552) and median citation count was 2 (IQR 0-8).

To account for the differences in time that articles have been publicly available, the analytic sample was restricted to only published articles (n=1,140) and preprints (n=73, 50 peer reviewed and 23 unpublished) that were included in the 2020 editions of the COVID-19 Science Update (through edition 70; Table 1). Among the 73 preprints, 50 (68.4%) were published. The median Altmetric Attention Score for published articles (328, IQR 57-1224) was higher than that of unpublished preprints (83, IQR 9-231; P=.002). The difference in Altmetric Attention Score between peer-reviewed (260, IQR 43-1817) and unpublished preprints was not statistically significant, likely due to small sample sizes (P=.09). The median citation counts for peer-reviewed preprints (55, IQR 9-148) and published articles (43, IQR 15-149) were significantly higher (P<.001) than that for unpublished preprints (4, IQR 1-9). There was no significant difference in citation count between peer-reviewed preprints and published articles (P=.23).

‎View this figure

Discussion

Prior analyses of COVID-19 preprints have found publication rates between 5.7% and 21.1% [7,8]. Preprints included in the COVID-19 Science Update were published at a higher rate than reported elsewhere [7,8], and those that were ultimately published received higher attention scores than unpublished preprints. A high Altmetric Attention Score is indicative of only the total attention a publication receives. It does not differentiate between positive and negative attention, so a high Altmetric Attention Score could equally reflect a publication of high scientific value or one that is widely refuted [6]. Preprints facilitate rapid access to information; however, they are ultimately limited by the absence of peer review and are thus subject to change and may never be published. The ability to discern high-quality impactful preprints is therefore an important tool for public health decision-making. Despite these limitations, the findings of our analysis indicate that the COVID-19 Science Update process for selecting preprints is robust, with high fidelity in terms of the likelihood of preprints to be published and to be impactful. The incorporation of high-quality preprints into the COVID-19 Science Update improves this activity’s capacity to provide timely information, which could have an impact on meaningful public health decision-making.