Prospective Associations of Regional Social Media Messages with Attitudes and Actual Vaccination: A Big Data and Survey Study of the Influenza Vaccine in the United States

"Combating an 'infodemic'...on social media is important in preventing the spread of misinformation in the community."

Health messages, such as those found on social media, can influence attitudes, which in turn guide people's decisions about what to do, including whether or not to vaccinate against a particular disease. Prior studies have obtained cross-sectional evidence of relations between the use of or exposure to social media and vaccination status. However, social media do not exist in a vacuum: Tweets posted in a region may be relevant to the local context, and their impact may also depend on discussions of vaccines with family and friends in real life. Thus, this study uses longitudinal data and a combination of individual- and regional-level data to understand the relations between social media messages posted within a region, interpersonal conversations, and individual attitudes and behaviours in that region related to actual vaccination against influenza in the United States (US).

Believing that Tweets, including retweets, can be informative about popular topics and conversations within a community, the study analyses 115,330 geolocated tweets about the flu and vaccination along with data from a survey of 3,005 US adults conducted from September 2018 to May 2019. The tweets were analysed against individual responses gathered in 5 waves of US survey data from the 2018-19 flu season. The respondents (ranging from 1,591 to 3,005 per wave) answered questions about vaccine attitudes, vaccination, and real-life discussions about vaccination.

Ten topics were found to be common across US counties during the 2018–2019 influenza season. In the overall analyses, two of these topics - Big Pharma and Vaccine Science Matters - were associated with attitudes and behaviours. The conspiracy theory "Big Pharma" was associated with negative vaccination attitudes concurrent with the tweets (using terms such as "autism" and "murder"). On the other hand, the topic concerning vaccine science in November-February (using terms such as "vaxwithme" and "cancer") in November-February were positively correlated with vaccination attitudes in February-March. These findings imply that social media posts about scientific evidence of vaccines could promote the spread of factual information of vaccines, thereby contributing to more favourable attitudes towards the flu vaccine.

Moreover, among respondents who did not discuss the influenza vaccine with family and friends, the topic Vaccine Fraud and Children in November-February was negatively correlated with attitudes in February-March and with vaccination in February-March, and April-May. (The language of this topic, which included the terms "child" and "worldwide", also included tweets describing kidney pathology and references to what are now known to be falsified claims of vaccine fraud made in 2014.) However, this correlation was absent when participants had discussions about the influenza vaccine with family and friends. The finding that discussing vaccines with family and friends appeared to eliminate negative effects from social media could encourage public health officials to promote real-world conversations about the benefits of vaccination.

One of the researchers explained the findings in an interview, as follows: "What we find is that some online discussions appear to have a negative influence on people's attitudes and vaccine behavior - which makes the people exposed to them less likely to get a flu shot. That's the case if they do not have real-world discussions about vaccination with family and friends. But if they discuss it with others, that effect goes away." Looking ahead, she mused, "What's going to happen when we have a COVID-19 vaccine? If public health officials don't offer clear, consistent messaging on vaccination, whatever circulates on Twitter - however crazy it is - may have an impact."

The researchers stress that, while the study found "strong to very strong associations" between the social media topics and vaccine attitudes and behaviour, the associations do not necessarily imply causation. Future studies could assess, for example, whether the presence of an unobserved or third variable, such as the change of regional patients' minimum age for convenient immunisation at pharmacies, increases discussions on Twitter while also increasing vaccine attitudes and actual vaccination.