The "Infodemic" of Journal Publication Associated with the Novel Coronavirus Disease.

Abstract

The novel coronavirus disease 2019 (COVID-19) pandemic has now surpassed 1 million documented cases globally. The pandemic has forced social distancing measures and cancellation of nonessential services at an unprecedented scale. COVID-19, caused by the viral pathogen severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has permeated communities at unparalleled rates. Within just a few months after the initial documentation of the disease in December 2019, the World Health Organization (WHO) declared a pandemic. Prior to this, however, the WHO had already warned the public of an “infodemic,” described as “an overabundance of information—some accurate and some not.”1 The WHO warned primarily of an infodemic related to news and social media; however, much of the source of opinions and commentaries online is predicated, to some extent, on what has been published in scientific journals2. Although communicating rapid advancements in a crisis is critical to academic and social discourse, too much information can overwhelm even experienced clinicians, making it difficult to discern opinion from evidence and data-driven facts from expert conjecture. Most concerning is the potential for “misinformation” or “disinformation” when publications across scientific journals are used to justify false or misleading claims3. The extent to which opinions and commentaries published in scientific journals dominate the evolving base of evidence during a crisis may, in part, contribute to information overload. We quantified and characterized the scientific literature pertaining to SARS-CoV-2 and COVID-19 in order to understand the type of information that is produced and the speed with which it is made available to the scientific community. Materials and Methods A systematic PubMed literature search was performed covering the period from December 1, 2019, to March 31, 2020. The search used the following 8 terms combined with the “OR” operator: coronavirus, SARS-CoV-2, COVID-19, severe acute respiratory syndrome coronavirus 2, coronavirus disease-2019, coronavirus disease 2019, novel coronavirus 2019, and 2019-nCoV. Articles that did not have a primary focus on SARS-CoV-2 were excluded. No language restrictions were placed. Six reviewers screened and extracted pertinent data in a collaborative online data-extraction form that was piloted prior to formal data extraction. Interobserver agreement for the assessments of study eligibility was calculated with use of the Cohen kappa (κ) statistic. The following data were extracted for each record: name of publishing journal, date of submission, date of publication, country of origin, study type, surgical relevance, and publication theme. Study type was categorized as randomized controlled trials; prospective cohort studies; retrospective cohort or case control studies; case series or reports; basic science; systematic and/or scoping reviews; opinion-based commentaries, editorials, or narrative reviews; guidelines; protocols; and surveys. Following data collection, all clinical studies using primary patient data were grouped with basic science as “primary research,” and all other review and commentary studies were grouped together. Types of articles published in the top journals (as determined by impact factor) were compared with those in the rest of the journals. Descriptive statistics were reported as the mean and standard deviation. Chi-square tests were performed to compare proportions across groups, with post-hoc testing when necessary with a Tukey correction. The Spearman rho (ρ) was used to assess for correlation between mean time to publication and publications in each week. Results The search returned 3,212 records, of which 1,741 articles were eligible for inclusion. Reviewers achieved perfect agreement regarding inclusion (κ = 1.0). The studies originated from 59 countries and were published across 447 unique journals (Table I). Countries with the largest number of publications included China (748; 43.0%), the United States (275; 15.8%), and the United Kingdom (178; 10.2%). The BMJ (117; 6.7%), The Lancet (80; 4.6%), and the Journal of Medical Virology (72; 4.1%) published the greatest number of articles. A total of 24 (5.4%) surgery-focused journals published articles on the topic. The mean time from submission to publication was 13 ± 12 days (range, 0 to 113 days). The most common publication types were commentaries and narrative reviews, accounting for 1,103 publications (63.4%). TABLE I - Summary Data No. of publications* 1,741 Surgical articles 89 (5.1%) Type of article Commentary 1,103 (63.4%) Case report/series 293 (16.8%) Basic science 188 (10.8%) Guidelines 52 (3.0%) Systematic or scoping review/meta-analysis 46 (2.6%) Retrospective cohort/case control 43 (2.5%) Prospective cohort 7 (0.4%) Survey 4 (0.2%) Protocol 3 (0.2%) Randomized controlled trial 1 (0.1%) Top-publishing journals The BMJ 117 (6.7%) The Lancet 80 (4.6%) Journal of Medical Virology 72 (4.1%) Journal of the American Medical Association 44 (2.5%) New England Journal of Medicine 43 (2.5%) Top-publishing countries China 748 (43.0%) United States 275 (15.8%) United Kingdom 178 (10.2%) Italy 82 (4.7%) Canada 44 (2.5%) No. of journals* 447 Surgical journals 24 (5.4%) No. of countries 59 Time to publication (days) Median 10 Mean ± standard deviation 13 ± 12 Range 0-113 *The values are given as the count, with the percentage in parentheses. When categorized by “primary research” versus all other article types, clinical and basic science research, which included actual patient data, bench work, or mathematical modeling, accounted for 532 publications (30.6%). Commentaries and reviews and other study designs accounted for 1,209 publications (69.4%). The number of publications produced weekly over the 13-week query period increased over time, mirroring the global spread of COVID-19 (Fig. 1). There was a significant, strong positive correlation between the mean time to publication and number of publications in a given week (ρ = 0.74; p = 0.01). Articles relevant to surgery and/or perioperative care accounted for 89 publications (5.1%).Fig. 1: Trends in types and volume of publications over time, compared with the pandemic curve.Journals with high impact factors (≥40) published a significantly lower proportion of primary research compared with journals with impact factors <40 (22.1% compared with 31.9%; chi-square = 9.02; p = 0.003). The proportion of primary research published varied significantly across journals with a high impact factor (chi-square = 20.7; p = 0.002), such that our post-hoc analyses revealed that the New England Journal of Medicine and Nature published a significantly higher (17; 39.5%) and lower (4; 9.3%) proportion of clinical and basic science research, respectively (Fig. 2). There were no other significant differences among the remaining high-impact-factor journals.Fig. 2: Study types among the top 5 journals with impact factor of ≥40 and >1 publication. N Engl J Med = New England Journal of Medicine, JAMA = Journal of the American Medical Association, and * = significant difference compared with other journals in the relative proportion of primary versus review/commentary literature (p = 0.002).Discussion The present review identified 1,741 articles pertaining to COVID-19 and/or SARS-CoV-2 published in scientific journals during the short 13-week period since the initial documentation of the disease. We identified 3 key findings: (1) an exponential increase in publications over the 13-week period since initial documentation of COVID-19, (2) a rapid time from submission to publication, and (3) a higher proportion of commentaries and opinion papers, especially within high-impact-factor journals. This review has a unique strength in the comprehensive review of the entirety of the scientific literature, which allowed for an understanding of literature-production trends over time. The review had limitations, including the surface-level nature in which data were collected and grouped. Furthermore, in the interest of efficiency given the rapidly evolving nature of the literature, only 1 database (PubMed) was utilized. Also, the assumption was made that all publications within each category were of the same value and validity. The trend that reviews and commentaries have far outpaced clinical and basic science research to date is not particularly surprising. This is a novel virus, and data-driven, high-quality studies require considerable planning to design, execute, analyze, and publish. As of March 24, 2020, there were 555 relevant registered clinical trial records, 188 of which were underway at the time of this writing4. The volume of publications over the 13-week query period demonstrated the global focus on this virus and a concerted effort of scientific journals to prioritize and rapidly disseminate knowledge. There are clear advantages to articles being published at an average of 13 days after submission, and certainly some drawbacks. In this sample, 260 articles were published within 7 days of submission. Researchers have cautioned about interpretation of research findings during a crisis5. Although the warning against an infodemic was largely based on misinformation propagated by mainstream news and social media, much of the information originates within the scientific community—nearly two-thirds of all publications were commentaries or narrative reviews on the topic. For instance, a preprint article suggesting similarities between the viral characteristics of SARs-CoV-2 and the human immunodeficiency virus (HIV) fueled conspiracy theories about bioengineering6. Although the paper was highly criticized and later withdrawn, it was among the most widely discussed scientific publications in the last 8 years despite its short lifespan, with the highest Altmetric score of any publication during that time frame. The disproportionate number of commentaries and opinion pieces in high-impact-factor journals was particularly troubling. These journals are among the most widely distributed and read journals globally and may be where clinicians turn first for high-quality evidence in a time of crisis. A commentary in The Lancet Global Health highlighting experiences of front-line health-care workers was later found not to be a first-hand account and was retracted7. Similarly, the first documented case of asymptomatic transmission was published by the New England Journal of Medicine8. The article documenting that case was later revised after it was discovered that the patient had been symptomatic and the researchers had not actually spoken to her firsthand. Given the global reach of these top medical journals, it is more important than ever to focus on the publication of high-quality research that has been rigorously peer-reviewed. The findings of this review highlight the need for more primary research to overcome this pandemic. A deliberate attempt to focus on basic research, clinical studies, systematic reviews, and guidelines from governing bodies may not only help to “flatten” the infodemic curve but also provide timely and relevant data to guide medical management and policy-making.