Spatiotemporal Analysis Of Lung Cancer Incidence In Hangzhou,China:2009-2012

BackgroundLung cancer is the most common tumor in the world and the global public health problem. There were about1.8million new cases and1.6million people dead in2012. Lung cancer had become the number one cause of death among all the malignant tumors in China since2008. The aim of this study was to characterize the spatiotemporal pattern of lung cancer in Hangzhou. Recognition of spatial distribution characteristic of lung cancer incidence would be greatly helpful to study of multi-factorial etiology of lung cancer and to allocate the public health resources for cancer control.MethodsThe data of lung cancer incidence case at township level between2009and2012were obtained from Hangzhou Center for Disease Control and Prevention. Firstly, descriptive analysis for lung cancer crude incidence was conducted with statistical methods. Then spatial empirical Bayes smoothing was employed to smooth the lung cancer crude incidence rate. Spatial autocorrelation analysis was implemented to identify the spatial pattern of lung cancer incidence. Two spatial cluster analysis methods, local indications of spatial association (LISA) and spatial scan statistic were employed to evaluate the geographic distribution of lung cancer incidence for both genders in Hangzhou during2009-2012. Space-time scan statistics was used for detecting spatial-temporal distribution of lung cancer incidence.Results(1).A total of15575cases of lung cancer had been reported in Hangzhou in2009-2012. The ration of male and female in the four years was2.26:1, including10825male cases and4750female cases. The four-year average crude incidence rate of lung cancer was57.4per100,000,79.6per100,000for male and35.2per100,000for female, respectively.(2). After spatial empirical Bayes smoothing, the data of every year for male and female are all approximated to normal distribution. (3). The Moran’s I values (an indicator for Global Spatial Autocorrelation) of lung cancer incidence for gender-specified were all larger than0, implying the spatial autocorrelation at township level. Moreover, the Moran’s I values of female were larger than the male’s, indicating that the distribution of female incidence was more centralized.(4). Two high-high clusters, four low-low clusters and one high-low cluster were identified for four-year average incidence of male lung cancer by the Local Moran’s test. The two high-high clusters were located in the main urban area of Hangzhou and Tonglu County, respectively. For female lung cancer incidence, one single high-high cluster, three low-low clusters and one high-low cluster were identified. The only one high-high cluster was located in the main urban area of Hangzhou.(5). Two significant high risk clusters and four significant low risk clusters were identified for male lung cancer incidence. One significant high risk cluster and three significant low risk clusters were detected for female lung cancer incidence.(6). Two significant high risk clusters and five significant low risk clusters were identified for male lung cancer incidence and one high risk clusters and three low risk clusters were identified for female. The year of gathering was both in2010-2012for male and female high risk clusters.Conclusions(1). There are significant spatial-temporal distribution for lung cancer incidence in Hangzhou and different spatial distribution characteristic between male and female.(2). Two or more methods applied together could reflect characterize the spatiotemporal of cancer more accurate and comprehensive.(3). The identification of lung cancer incidence clusters lays a foundation to explore the environmental factors responsible for increased cancer risk and has vital practical value for health services and policies implementation.