Exploring Content of Misinformation about HPV Vaccine on Twitter

"Twitter is considered an informal mechanism for spreading rumors..., such that misinformation with weak sources can easily circulate and impact and amplify beliefs..."

Research indicates that a strong provider recommendation addressing parents' concerns about human papillomavirus (HPV) vaccine is an effective way of combatting vaccine hesitancy and increasing uptake among adolescents. However, provider messages are not the only HPV vaccine messaging to which parents are exposed. Although social media can be a source of guidance about HPV vaccination for parents, the information may not always be complete or accurate. This retrospective content analysis sought to identify content and frequencies of occurrence of disinformation and misinformation about HPV vaccine posted on Twitter.

Analysis was conducted among 3,876 unique, English-language #HPV tweets, excluding retweets, posted between December 15 2019 and March 31 2020. The researchers found that 24% of tweets contained disinformation or misinformation, and the remaining 76% contained supportive/educational content. The most prevalent categories of disinformation/misinformation were: adverse health effects (59%), mandatory vaccination (19%), and inefficacy of the vaccine (14%). Among the adverse health effects tweets, non-specific harm/injury (51%) and death (23%) were most frequent. Misinformation posts that are vague and nonspecific to a gender, age, or specific health condition may increase their saliency to a broader audience: "keeping the threat vague and general is more psychologically threatening to the audience....

Misinformation posts had higher rates of audience engagement, including likes, retweets, and replies, representing greater audience reach. For instance, disinformation/misinformation tweets vs. supportive tweets had 5.44 (95% confidence interval (CI) 5.33-5.56) times the incidence rate of retweet. Also, Tweets that included personal narratives had higher levels of audience engagement as compared to those that did not. Tweets that included a personal narrative vs. those that did not have an incidence rate ratio (IRR) of 5.05 (95% CI 4.90-5.17) for retweets and 7.92 (7.79-8.04) for audience engagement. Furthermore, compared to tweets without a personal narrative, tweets that included a personal narrative about children had a retweet IRR of 6.93 (95% CI 6.56-7.33) and an audience engagement IRR of 11.94 (95% CI 11.50-12.40). One conclusion from this and others' research cited in the paper is that "statistical, fact-based information may not garner as much engagement in these social media spaces and with already vaccine hesitant parents, and interventions which seek to counter this misinformation may also need to employ narrative tactics to address vaccine hesitancy."

In multivariable models, tweets containing concerns about vaccine mandates had the highest rates of retweets and engagement. Certain types of misinformation were less likely to be retweeted than if the post did not contain this type of misinformation, including: tweets containing misinformation about Pharma (tweets that expressed deception or money-making incentives on the part of pharmaceutical companies) [IRR 0.68 (95% CI 0.63-0.74)], common concerns from the literature (e.g., that the child is too young for vaccination against a sexually transmitted infection/or is not sexually active) [IRR 0.42 (95% CI 0.31-0.56)], and government (tweets that expressed fear of government or group conspiracy, along with money-making incentives of these groups) [IRR 0.40 (95% 0.37-0.43)].

Reflecting on the findings and ways forward, the researchers discuss inoculation therapy, which operates on the premise that warning an individual of an impending persuasive attack and providing them with an argument to counteract it will inoculate the individual to potential attitude changes when faced with the attack. For example, an intervention could confer resistance to the effect of non-specific health misinformation by exposing parents to the dangers of ambiguous misinformation and providing examples of what that information looks like and why it is bad.

Per the researchers: "Misinformation evolves rapidly and requires a nimble counter-messaging approach, whether inoculative (prebunking) or post-hoc. For inoculation messages to be successfully implemented, they must be used on emerging misinformation before the beliefs have been widely adopted. Social listening using machine learning and natural language processing computational methods would allow for large-scale data surveillance from multiple social media platforms to understand any changing and emerging concerns and attitudes about vaccines and identify nascent misinformation."

In conclusion, this study found that almost one-quarter of #HPV tweets contained disinformation or misinformation about the HPV vaccine and these tweets received higher audience engagement, including likes and retweets. "Given that many parents rely on social media for health-related information, future interventions, such as those that prebunk misinformation to prevent the harmful effects on vaccination acceptance are urgently needed."