Author: 
Adam G. Dunn
Didi Surian
Julie Leask
Aditi Dey
Kenneth D. Mandl
Enrico Coiera
Publication Date
April 29, 2017
Affiliation: 

Centre for Health Informatics, Australian Institute of Health Innovation, Macquarie University (Dunn, Surian, Coiera); School of Public Health and Sydney Nursing School, The University of Sydney (Leask); National Centre for Immunisation Research and Surveillance, The Children's Hospital at Westmead (Dey); Computational Health Informatics Program, Boston Children's Hospital (Mandl); Department of Biomedical Informatics, Harvard Medical School (Mandl)

Editor's note: This summary was written by The CI.

"[N]ews media may reflect, amplify, or influence vaccine acceptance, and...measures of information exposure derived from Twitter may be a surrogate indicators for localized differences in acceptance."

The research reported in this paper explored a healthcare decision that appears susceptible to influence of news and social media: the choice of administering the human papillomavirus (HPV) vaccine to an adolescent. The aim was to determine whether measures of information exposure derived from Twitter could be used to explain differences in coverage in the United States (US). In short: Measures of information exposure derived from Twitter explained differences in coverage that were not explained by socioeconomic factors. Vaccine coverage was lower in states where safety concerns, misinformation, and conspiracies made up higher proportions of exposures, suggesting that negative representations of vaccines in the media may reflect or influence vaccine acceptance.

Trawling through 258,418 tweets about the HPV vaccination globally between October 1 2013 and October 30 2015, the researchers localised over 34 million US Twitter users to their state or county using information in their Twitter profiles. They monitored users' exposures to positive and negative information about the HPV vaccine on Twitter (up to 291.8 million exposures in total), including misinformation, conspiracies, and scare tactics, to gauge the broader media diets. Tweets were classified by topic using machine learning methods. Proportional exposure to each topic was used to construct multivariable models for predicting state-level HPV vaccine coverage, and compared to multivariable models constructed using socioeconomic factors: poverty, education, and insurance. Outcome measures included correlations between coverage and the individual topics and socioeconomic factors, as well as differences in the predictive performance of the multivariable models.

The topics varied between: generally positive topics describing new evidence or advocating for the use of HPV vaccines; mixed topics debating mainstream news media stories; and negative topics describing safety concerns, conspiracies, and politics (see Figure 2). A mainstream news media story related to a television programme had the strongest overall negative correlations with coverage in females and males, and its highest proportional exposures were found in states with lower coverage (see Figure 3). Generally positive topics (describing evidence or advocating for the use of the vaccine) tended to reach much larger audiences than topics related to safety concerns or conspiracies (see Figure 4). A newspaper story that was retracted and replaced with evidence-based responses had the strongest overall positive correlations with coverage in males, and its highest proportional exposures were found in states with higher coverage (see Figure 5).

Topics corresponding to media controversies were most closely correlated with coverage (both positively and negatively); education and insurance were highest among socioeconomic indicators. Measures of information exposure explained 68% of the variance in one dose 2015 HPV vaccine coverage in females (males: 63%). In comparison, models based on socioeconomic factors explained 42% of the variance in females (males: 40%). "While socioeconomic and health factors are expected to capture differences in access to healthcare, proxies for information diets derived from social media data account for more of the variance in coverage, and thus appear to capture acceptance in a way that socioeconomic factors do not. The results also showed that HPV vaccine coverage is most closely correlated with topics that were covered in mainstream news media, suggesting a relationship between the quality of information in broader public discourse and vaccine acceptance."

Vaccine coverage was lower in states and counties where users were exposed to higher proportions of tweets dealing with safety concerns, misinformation, and conspiracies, the study concluded. "Routine systems that monitor exposure to relevant topics could be constructed to identify locations where misinformation or low-quality evidence is over-represented in the news media, and guide public health interventions to amplify high-quality evidence and guide social media interventions....More precise identification of the safety concerns, misinformation, and conspiracies that are important to a segment of the public could provide a basis from which to construct targeted and cost-effective news and social media based interventions."

It is unclear whether decisions people made about vaccination were influenced by their exposure to certain tweets and media, or whether they simply chose to inhabit an online community that reinforced their beliefs, a tension known as contagion versus homophily. They feed into each other, said Julie Leask, a co-author of the study. She believes that social media generally reinforced existing sentiments in the community.

The social media platform could prove a faster, cheaper tool for governments using costly community surveys to identify areas with high rates of vaccine hesitancy or refusal and the possible reasons at play, the researchers suggested.

Health professionals and advocates of vaccination appear to be engaging a lot more on social media. "Generally speaking this is a really good thing because we don't want parents to raise unfounded concerns about vaccination in a vacuum," Professor Leask said. "As professionals we need to be engaged in those spaces, but how we engage is really important," she stressed. There were two competing schools of thought that informed how they interact. Some vaccine advocates expose, even vilify the anti-vaccination movement in an effort to call them to account on social media. But this tactic can backfire, drawing more attention to anti-vaccination messages, Professor Leask said. The alternative involves ignoring the hardline activists and vaccine refusers, and instead engaging with the fence sitters. "Talk to parents who are hesitant about vaccination but not entrenched in their decision and certainly not an anti-vaccination advocate," Professor Leask said. "We need to listen and validate their concern for their children as parents and reasonably and calmly address the issues they raise. Anger and vitriol and attempts to shout louder across the divide may disengage these important audiences."

Source: 

"How Twitter more accurately maps vaccination coverage than census data", The Sunday Morning Herald, by Kate Aubusson, May 12 2017 - accessed on May 18 2017. Image caption/credit: "Tweets form users opposing vaccination in response to an article shared by The Washington Post." Photo: Twitter