RESUMO
Web-based methodologies may provide a new and unique insight into public response to an infectious disease outbreak. This naturalistic study investigates the effectiveness of new web-based methodologies in assessing anxiety and information seeking in response to the 2009 H1N1 outbreak by examining language use in weblogs ("blogs"), newspaper articles, and web-based information seeking. Language use in blogs and newspaper articles was assessed using Linguistic Inquiry and Word Count, and information seeking was examined using the number of daily visits to H1N1-relevant Wikipedia articles. The results show that blogs mentioning "swine flu" used significantly higher levels of anxiety, health, and death words and lower levels of positive emotion words than control blogs. Change in language use on blogs was strongly related to change in language use in newspaper coverage for the same day. Both the measure of anxiety in blogs mentioning "swine flu" and the number of Wikipedia visits followed similar trajectories, peaking shortly after the announcement of H1N1 and then declining rapidly. Anxiety measured in blogs preceded information seeking on Wikipedia. These results show that the public reaction to H1N1 was rapid and short-lived. This research suggests that analysis of web behavior can provide a source of naturalistic data on the level and changing pattern of public anxiety and information seeking following the outbreak of a public health emergency.