Anti-Vaccine Discourse on Social Media: An Exploratory Audit of Negative Tweets about Vaccines and Their Posters

"As public health authorities, health communicators and the media hasten to work with technology firms to fight vaccine mis/disinformation..., it is urgent for research to classify the characteristics of anti‐vaccine messages into very specific categories as well as investigate how regularly users of social media are exposed to each category."

Declining public confidence in vaccines has manifested in the rise of vaccine hesitancy. The anti-vaccination movement is spreading around the world. This paper addresses the need for health professionals, communicators, and policymakers to grasp the nature of vaccine mis/disinformation on social media. It provides quantitative and qualitative insights into the nature of anti‐vaccine discourse on Twitter based on data collected prior to the COVID‐19 health crisis in order to capture a "normal" scenario and minimise "noise" caused by the global pandemic.

The paper begins with a brief literature review of anti‐vaccination movements on social media. The movement itself ignited in the modern area following the 1998 publication in The Lancet of a study by Andrew Wakefield that falsely established a causal link between the measles, mumps, and rubella (MMR) vaccine and autism. Although the paper was later retracted, the massive media coverage that it received sparked a public distrust in vaccines that has persisted. Since that time, social media has enabled the exponential speed in the growth and communication of mis/disinformed anti‐vaccine communities online - and, by extension, in the offline world - despite efforts to disseminate pro‐vaccine content on social media. Soon after the first COVID‐19 cases were reported, for instance, conspiracy theories - such as that the coronavirus was created by the pharma industry to sell a vaccine or that Bill Gates created a microchip that would be implanted with the COVID‐19 vaccine - began to spread on various social media platforms, especially Facebook, across the globe.

Such growth has often been attributed to social media's ability to foster like‐minded communities that are not always formed on the basis of rational evidence‐based reasoning. That is, people are emotional beings and, as such, are not always rational consumers of science information. Many posts are therefore promoted due less to their factual value than to their social appeal. Researchers have observed that anti‐vaccination posts on social media tend to stimulate more content interactions than pro‐vaccine ones. Over time, it is argued, this phenomenon results in each user developing an exclusive network of contents and connections with people having very similar beliefs and attitudes - so-called echo chambers. Studies show that this sort of closed-group communication on social media only reinforces exposure to biased, misleading, or deceptive vaccine information, which can fuel vaccine hesitancy.

For the present study, a one-by-one coding of 4,511 vaccine-related tweets posted from the United Kingdom (UK) over a 3-year period in 2019 resulted in 334 anti-vaccine tweets generated by 196 users. In terms of identities, 48.5% of these unique users were classified as anonymous (did not provide enough data for others to identify them), which could include bots and trolls that have been found to be significantly more productive and more likely to spread anti‐vaccination content or to sow discord. Of non‐anonymous users, most were ordinary lay people posting as parents and activists (39% of the sample), followed by so‐called "alternative health" practitioners (e.g., natural therapists, holistic therapists, homeopaths), celebrities, health professionals and experts, the media, and so on.

The researchers' analysis shows that: (i) anti-vaccine tweeters are quite active and widely networked users on their own; (ii) anti-vaccine messages tend to focus on the "harmful" nature of vaccination, based mostly on personal experience, values, and beliefs rather than hard facts - as for the origin of tweeted information, the majority (57%) of the sample did not include a clear source; and (iii) anonymity does not make a difference to the types of posted anti-vaccine content, but it does so in terms of the volume of such content: About 60% of the tweets had an anonymous author.

In line with previous studies, the study found that anti‐vaccine posts constitute a very small part of vaccine discourse on Twitter, accounting for just over 7% of all tweets about vaccination during the study period. Also, the data show that the top 10% of anti‐vaccine Twitter accounts produce most of the sampled tweets and have more following and followed users than the rest of UK tweeters. However, most anti‐vaccine tweet authors do not seem to interact much with each other. In particular, nearly a half of sampled tweets mentioned no other user, and another 28% did one; 82% used no hashtags; and about 81% were not retweeted at all. This somewhat contradicts previous social network analyses and needs further investigation with larger samples over longer time periods to confirm and explain.

Based on the findings, the researchers suggest that communication initiatives to tackle anti‐vaccine discourse on social media should:

• Work closely with technological platforms to tackle anonymous anti-vaccine tweets. They need to be closely monitored, flagged to users and, when necessary, removed from the system.
• Focus efforts on mis/disinformation in 3 major arears (in order of importance): (i) the medical nature of vaccines - e.g., supposedly harmful side effects (framed in 45% of the sampled tweets); (ii) the belief that vaccination is a tool of manipulation and control for money and power, and (iii) the "freedom of health choice" discourse against mandatory vaccination.
• Start by reaching out to the tiny proportion of hyperactive anti‐vaccine users who are not only productive but also well connected. Of the 3 most followed anti‐vaccine Twitter accounts during the study period, 2 are sister news outlets, one with nearly 1.5 million followers. In addition, 5 of the 10 most followed anti‐vaccine accounts in the sample were owned by famous people.
• Go beyond common factual measures - such as detecting, labelling, or removing fake news - to address emotions induced by personal memories, values, and beliefs. "[E]each group will require a different set of tools and techniques to suit, inter alia, its emotional intelligence level, trust in healthcare systems, health and science literacy, news/media literacy, and openness to other sides of the debate."

In conclusion, the researchers hope this study can help guide those seeking to know what to prioritise when designing a campaign to combat anti‐vaccine on social media. The findings might also provide an empirical basis for the development of content‐based methods to tackle anti‐vaccine mis/disinformation.