Tracking vaccine misconceptions–by mining online data

This 1802 British cartoon skewers the cowpox vacccine, newly introduced against smallpox. Read more at http://en.wikipedia.org/wiki/File:The_cow_pock.jpg#file
This 1802 British cartoon skewers the cowpox vaccine, newly introduced against smallpox. Read more at http://en.wikipedia.org/wiki/File:The_cow_pock.jpg#file
Fifty years after Boston Children’s Hospital faculty developed a vaccine against measles, the United Kingdom is seeing a surge of cases. Last year, it tracked a record 2,000 measles diagnoses—unusual for a country that used to average only a dozen cases every year. With 1,200 cases reported this year so far, that record could be broken.

The cases are the legacy of parents who decided to forgo vaccinating at least 1 million children against measles, based on a 1998 study in The Lancet linking the measles vaccine to autism. That now-retracted study became the origin of its own epidemic, carrying misinformation through a network of parents and media outlets that believed the author had discovered the cause of autism.

Until recently, tracking the spread of vaccine-related rumors was even more difficult than tracking the outbreaks such misinformation engenders. A study in The Lancet Infectious Diseases, involving Boston Children’s Hospital’s HealthMap data collection system and funded by the Bill & Melinda Gates Foundation, has taken a huge step toward turning that around.

Launched in 2006 by Boston Children’s John Brownstein, PhD, and Clark Freifeld, HealthMap uses scraping technology—software that uses keywords to identify and collect specific content from across the Internet—to mine for information about public health events around the world in 15 different languages. In 2010, it was used to track cholera outbreaks in post-earthquake Haiti.

For Brownstein and researchers at the London School of Hygiene and Tropical Medicine, as well as the International Society for Infectious Diseases, using the system to track anti-vaccine rumors was a logical next step.

“What we’ve frequently learned is that the same tools we’re developing to track disease outbreaks can also uncover people’s attitudes towards health and vaccines,” Brownstein explains.

So Brownstein and his colleagues adapted HealthMap to search the web for online articles, blogs and government reports about vaccines, vaccine programs and vaccine-preventable diseases in English, French and Spanish. (In areas with little or no access to the web, Brownstein and his colleagues have started and continue to build networks of local disease experts who can provide single channels of information.)

The accumulated data—10,380 reports from 144 countries—were analyzed and sorted by vaccine, the concern/complaint, disease, location and source of the report, allowing the team to monitor how the information spread geographically over time. Of the 10,380 reports, 7,171 (69 percent) were positive or neutral attitudes towards vaccines; the remaining 3,209 (31 percent) represented negative attitudes such as safety concerns or suspicions about the motive behind vaccination.

Anti-vaccine rhetoric is dangerous not only because it encourages some people to forgo vaccination, but also because it can breed distrust between public health agencies and the populations they serve. Matters only get worse when that distrust coincides with politics, as health workers in Somalia recently discovered when extremists mounted a campaign against the polio vaccine.

Now that the study has proven the feasibility of real-time monitoring of vaccine concerns over time and locations (soon to be expanded by mining social media like Twitter), immunization programmers can identify categories of rumors and develop communications strategies to try to counter them. “It’s difficult to change minds given the conspiracy ideas people have around government, sterilization, autism and pharmaceutical companies,” Brownstein admits.

Citation:

Larson HJ, Smith DM, Paterson P, Cumming M, Eckersberger E, Freifeld CC, Ghinai I, Jarrett C, Paushter L, Brownstein JS, & Madoff LC (2013). Measuring vaccine confidence: analysis of data obtained by a media surveillance system used to analyse public concerns about vaccines. The Lancet Infectious Diseases PMID: 23676442

ResearchBlogging.org