Spatial analysis of dengue fever and exploration of its environmental and socio-economic risk factors using ordinary least squares: A case study in five districts of Guangzhou City, China, 2014

Abstract

ObjectiveSpatial patterns and environmental and socio-economic risk factors of dengue fever have been studied widely on a coarse scale; however, there are few such quantitative studies on a fine scale. There is a need to investigate these factors on a fine scale for dengue fever. MethodsIn this study, a dataset of dengue fever cases and environmental and socio-economic factors was constructed at 1-km spatial resolution, in particular ‘land types’ (LT), obtained from the first high resolution remote sensing satellite launched from China (GF-1 satellite), and ‘land surface temperature’, obtained from moderate resolution imaging spectroradiometer (MODIS) images. Spatial analysis methods, including point density, average nearest neighbor, spatial autocorrelation, and hot spot analysis, were used to analyze spatial patterns of dengue fever. Spearman rank correlation and ordinary least squares (OLS) were used to explore associated environmental and socio-economic risk factors of dengue fever in five districts of Guangzhou City, China in 2014. ResultsA total of 30553 dengue fever cases were reported in the districts of Baiyun, Haizhu, Yuexiu, Liwan, and Tianhe of Guangzhou, China in 2014. Dengue fever cases showed strong seasonal variation. The cases from August to October accounted for 96.3% of the total cases in 2014. The top three districts for dengue fever morbidity were Baiyun (1.32%), Liwan (0.62%), and Haizhu (0.60%). Strong spatial clusters of dengue fever cases were observed. Areas of high density for dengue fever were located at the district junctions. The dengue fever outbreak was significantly correlated with LT, normalized difference water index (NDWI), land surface temperature of daytime (LSTD), land surface temperature of nighttime (LSTN), population density (PD), and gross domestic product (GDP) (correlation coefficients of 0.483, 0.456, 0.612, 0.699, 0.705, and 0.205, respectively). The OLS equation was built with dengue fever cases as the dependent variable and LT, LSTN, and PD as explanatory variables. The residuals were not spatially autocorrelated. The adjusted R-squared was 0.320. ConclusionsThe findings of spatio-temporal patterns and risk factors of dengue fever can provide scientific information for public health practitioners to formulate targeted, strategic plans and implement effective public health prevention and control measures.