Delivering actionable infodemic insights and recommendations for the COVID-19 pandemic response/Fournir des donnees d'observation de l'infodemie et des recommandations exploitables pour la riposte a la pandemie de COVID-19.

Citation metadata

From: Weekly Epidemiological Record(Vol. 97, Issue 27)
Publisher: World Health Organization
Document Type: Article
Length: 7,727 words
Lexile Measure: 2120L

Document controls

Main content

Article Preview :


Since the beginning of the COVID-19 pandemic, digital communications and social networks have been used to support rapid real-time information-sharing about the virus and the disease in the public domain. The breadth of the exchanges, the diversity of sources and the polarity of opinions have sometimes resulted in indiscriminate amplification of both verified and unverified information. This infodemic (1) (too much information, including false or misleading information, in digital and physical environments during an acute public health event, which leads to confusion, risk-taking and behaviour that can harm health and lead to mistrust in health authorities and public health response) can prolong or amplify outbreaks and reduce the effectiveness of responses and interventions.

To address this challenge, WHO's COVID-19 incident management support team's (IMST) Pillar 2 (risk communication, community engagement and infodemic management, Figure 1), in collaboration with research partners, developed an integrated method for public health infodemic analysis and insights generation for weekly analysis of social media, traditional media and other data sources, such as user search trends, epidemiological data and socio-behavioural data, to identify, categorize and understand the concerns and narratives expressed. (2,3) When good-quality information about topics of concern to individuals and populations is lacking, the topics can fast become subjects of conjecture, speculation, poor-quality information and viral misleading content, (4,5,6) potentially harming communities. The approach therefore focuses on characterization and detection of narratives by identifying or anticipating areas of concern, questions, confusion, misinformation and information voids in narratives circulating in the information eco-system and providing immediately actionable insights for use in decision-making and risk communication, (7) to complement rumour-tracking and provide the right health information at the right time in the right format to the people who need it.


Characterization, integrated analysis and generation of insights about the infodemic was conducted in 3 steps in a mixed-methods approach (2,3,5,6) (Figure 2). First, data were collected from publicly available social and news media sources and categorized into conversations according to a COVID-19-public health taxonomy to identify potential topics of concern and information voids quantitatively. Secondly, the dataset was analysed qualitatively and compared weekly to identify narratives and their changes and to characterize changes in sentiment in the conversation, supplemented by user searches in Google Trends, and a digital infodemic intelligence report was written. Thirdly, the digital infodemic intelligence was reviewed by a group of subject matter experts and triangulated with other data sources to derive insights on the infodemic and recommend action for the week.

Step 1: Quantitative identification of potential topics of concern and information voids

The first step was based on a weekly aggregation of a sampled dataset comprising approximately 20 million publicly available social and news media and online search data in English, French and Spanish over 7 days and identification of questions, concerns, information seeking and engagement with themes in a public health COVID-19 conversation taxonomy. A natural language processing method that has been developed, validated, and published (2,3) was used to aggregate and categorize the data into 35 conversations with...

Source Citation

Source Citation   

Gale Document Number: GALE|A710235709