Several countries are witnessing significant increases in influenza cases and severity. Despite the availability, effectiveness and safety of influenza vaccination, vaccination coverage remains suboptimal globally. In this study, we examined the prevailing negative sentiments related to influenza vaccination via a deep learning analysis of public Twitter posts over the past five years. We extracted original tweets containing the terms 'flu jab', '#flujab', 'flu vaccine', '#fluvaccine', 'influenza vaccine', '#influenzavaccine', 'influenza jab', or '#influenzajab', and posted in English from 1 January 2017 to 1 November 2022. We then identified tweets with negative sentiment from individuals, and this was followed by topic modelling using machine learning models and qualitative thematic analysis performed independently by the study investigators. A total of 261,613 tweets were analyzed. Topic modelling and thematic analysis produced five topics grouped under two major themes: (1) criticisms of governmental policies related to influenza vaccination and (2) misinformation related to influenza vaccination. A significant majority of the tweets were centered around perceived influenza vaccine mandates or coercion to vaccinate. Our analysis of temporal trends also showed an increase in the prevalence of negative sentiments related to influenza vaccination from the year 2020 onwards, which possibly coincides with misinformation related to COVID-19 policies and vaccination. There was a typology of misperceptions and misinformation underlying the negative sentiments related to influenza vaccination. Public health communications should be mindful of these findings.
Keywords: flu; hesitancy; influenza; machine learning; negative sentiment; vaccine.