Abstract: PO0864
The COVID-19 Infodemic
Session Information
- COVID-19: Clinical and Basic Science Characteristics
October 22, 2020 | Location: On-Demand
Abstract Time: 10:00 AM - 12:00 PM
Category: Coronavirus (COVID-19)
- 000 Coronavirus (COVID-19)
Authors
- Desai, Tejas, NOD Analytics, Harrisburg, North Carolina, United States
- Conjeevaram, Arvind, The Bangalore Hospital, Bangalore, India
Background
In Situation Report #13 by the World Health Organization and 39 days before declaring COVID-19 a pandemic, the WHO declared a “COVID-19 infodemic”. The volume of coronavirus tweets was far too great for one to find accurate or reliable information. Healthcare workers were flooded with “noise” which drowned the “signal” of valuable COVID-19 information. To combat the infodemic, physicians created healthcare-specific micro-communities to share scientific information with other providers.
Methods
We analyzed the content of six physician-created communities and categorized each message in one of five domains (Symptoms, Diagnostics, Therapeutics, Prevention, Pathophysiology). We programmed 1) an application programming interface to download tweets and their metadata in JavaScript Object Notation beginning 11 March and 2) a reading algorithm using visual basic application in Excel to categorize the content. We superimposed the publication date of each tweet into a timeline of pandemic events. Finally, TD created a free repository of the dataset in the #NephTwitter Archives (https://bit.ly/2M6HJQ2) to help healthcare workers find quality information when treating patients.
Results
From 11 March to 27 April, 45% of the 19270 tweets in the dataset were categorized (signal). Tweets about Therapeutics (34%) and Prevention (32%) were the most prevalent. Tweets about Therapeutics spiked six times; the first coming 4 days after the WHO declared COVID-19 a pandemic. The largest spike came on day 8: 5 days after the US President suggested hydroxychloroquine as a potential treatment. Tweets about antimalarial therapy comprised 15% of tweets in this category. Tweets about Prevention spiked five times; the largest coming 21 days after the pandemic declaration when 1 million global cases were reported. Protective equipment comprised 13% of tweets in this category. There were 2210 searches performed of the signal tweets in the #NephTwitter Archives.
Evidence-based tweets comprised 1 in every 8 tweets in the categorized corpus. That ratio was better for tweets about antimalarials (1 in 3) and vaccines (2 in 3), the same for protective equipment, and worse for mechanical ventilation (1 in 31).
Conclusion
Algorithmic coding can 1) mitigate the COVID-19 infodemic and 2) identify & elevate illuminating evidence-based tweets. Both outcomes help healthcare workers find higher-quality information to combat the pandemic.