COVID-19 Public Health Communication on X (Formerly Twitter): Cross-Sectional Study of Message Type, Sentiment, and Source.

Abstract

Background: Social media can be used to quickly disseminate focused public health messages, increasing message reach and interaction with the public. Social media can also be an indicator of people's emotions and concerns. Social media data text mining can be used for disease forecasting and understanding public awareness of health-related concerns. Limited studies explore the impact of type, sentiment and source of tweets on engagement. Thus, it is crucial to research how the general public reacts to various kinds of messages from different sources.

Objective: The objective of this paper was to determine the association between message type, user (source) and sentiment of tweets and public engagement during the COVID-19 pandemic.

Methods: For this study, 867,485 tweets were extracted from January 1, 2020 to March 31, 2022 from Ireland and the United Kingdom. A 4-step analytical process was undertaken, encompassing sentiment analysis, bio-classification (user), message classification and statistical analysis. A combination of manual content analysis with abductive coding and machine learning models were used to categorize sentiment, user category and message type for every tweet. A zero-inflated negative binomial model was applied to explore the most engaging content mix.

Results: Our analysis resulted in 12 user categories, 6 message categories, and 3 sentiment classes. Personal stories and positive messages have the most engagement, even though not for every user group; known persons and influencers have the most engagement with humorous tweets. Health professionals receive more engagement with advocacy, personal stories/statements and humor-based tweets. Health institutes observe higher engagement with advocacy, personal stories/statements, and tweets with a positive sentiment. Personal stories/statements are not the most often tweeted category (22%) but have the highest engagement (27%). Messages centered on shock/disgust/fear-based (32%) have a 21% engagement. The frequency of informative/educational communications is high (33%) and their engagement is 16%. Advocacy message (8%) receive 9% engagement. Humor and opportunistic messages have engagements of 4% and 0.5% and low frequenciesof 5% and 1%, respectively. This study suggests the optimum mix of message type and sentiment that each user category should use to get more engagement.

Conclusions: This study provides comprehensive insight into Twitter (rebranded as X in 2023) users' responses toward various message type and sources. Our study shows that audience engages with personal stories and positive messages the most. Our findings provide valuable guidance for social media-based public health campaigns in developing messages for maximum engagement.