Risk factors and autocorrelation characteristics on severe acute respiratory syndrome in Guangzhou

Most of the traditional epidemiological studies are based on the classic statistical analysis instead of spatial information. Spatial analysis of risk factor and autocorrelation characteristics of epidemic can guide scientific prevention and control measures. Spatio-temporal data of 1277 cases of infected persons in 2003 in Guangzhou are studied. Map of incidence rate based on 1 km × 1 km grids is gained by kriging and kernel methods. Nine spatial risk factors, such as population density, traffic net, hospital, shopping mall, school, etc., are explored, results show that these risk factors are significantly correlated to incidence rate of SARS. Strict control measures to these risk factors can effectively prevent and control SARS epidemic. Global and local spatial autocorrelation characteristics are quantitatively measured with Moran's I and LISA statistics. Spatial cluster of incidence rate has experienced a weak-strong-weak process. High-high cluster areas are mainly in the center of Guangzhou city, where have high population density, economically active, and well-developed traffic net. The focus of high-high cluster areas did not transfer in the whole SARS epidemic process. The Government has taken successfully the prevention and control measures to prevent the further spread of SARS; however, the strategy of taking infectors to the nearest hospital contributed to the result that the spread risk has been high in the city centre. SARS incidence emerged in Guangzhou provides a sample for studying SARS and other unexpected new epidemics emerged in urban areas. Spatial autocorrelation analysis of SARS in Guangzhou provides a scientific basis for the emergency plan of the outbreak of SARS or other unexpected new epidemics in urban areas.