Vaccination behavior as a critical phenomenon

The 2014–15 measles outbreak at Disneyland in California brought heightened attention to the decision by some people to refuse vaccination for themselves and their children. Such personal decisions play into the complex relationship between vaccination behavior and disease dynamics, which influence each other in a nonlinear feedback loop that also incorporates societal norms and perceived risks. Chris Bauch and his colleagues at the University of Waterloo, Dartmouth College, and École Polytechnique Fédérale de Lausanne have now analyzed Twitter and Google search data related to measles and the measles-mumps-rubella (MMR) vaccine before and after the outbreak. Treating trends in the social-media data as a proxy for trends in the evolution of people’s attitudes toward vaccination, they find signs that the coupled vaccination–disease system was approaching a critical transition, a tipping point leading to a new dynamical regime—in this case an epidemic. The numbers of tweets and Google searches containing measles-related terms peaked sharply in early 2015, just after the outbreak began. For several years leading up to the outbreak, a period in which the vaccination rate was decreasing, the variations in the data from both platforms revealed so-called critical slowing down—a declining rate of recovery from small perturbations that is a signature of critical phenomena. In the wake of the outbreak, the vaccination rate increased and the system moved away from the tipping point, even before new requirements went into effect in California. The team’s mathematical model qualitatively reproduces both trends. Since statistical indications of an approaching tipping point appeared well before it was evident in the raw time series, the researchers suggest that online social data can provide early warning signals of approaching outbreaks in vaccination and disease dynamics. (A. D. Pananos et al., Proc. Natl. Acad. Sci. USA 114, 13762, 2017, doi:10.1073/pnas.1704093114 .)