Background: Annual influenza epidemics occur worldwide, resulting in considerable morbidity, mortality, and economic burden. Despite this regular occurrence, few studies have systematically examined the regional spatiotemporal patterns of influenza outbreaks.
Methods: Weekly incidence data for influenza cases in Fukuoka Prefecture (Japan) from 29 August 2005 to 3 June 2007 (n = 113,503) were geocoded at sentinel medical institutions (mainly pediatric hospitals and clinics). Space-time permutation scan statistics were then used to identify weekly temporal and spatial clusters of reported cases of influenza during the 2006-2007 influenza season.
Results: In the early phase of the influenza season, clusters were detected in the Fukuoka and Kitakyushu urban areas. The clusters then diffused gradually and spread into the rural areas of Fukuoka Prefecture. After a period of middle- to large-scale epidemic, the outbreak gradually waned.
Conclusions: Based on the permutation model, space-time permutation scan statistics can play an important role in detecting influenza outbreaks at an early stage and identifying the spatiotemporal patterns of geographic spread.