A Comparison Study Of Spatial Clustering Analysis Method Based On Spatial Point Data Of TB/HIV Co-infection Cases In Urumqi Area

Objective:The spatial point data of TB/HIV co-infection in Urumqi based on the scale of street was used to(1): explore the clustering of TB/HIV co-infection and the spatio-temporal variation of TB/HIV epidemic;(2)to study and discuss the feasibility of spatial point data in application of spatial clustering detection in TB/HIV co-infection and the applicability of each spatial clustering analysis method.Methods : TB/HIV co-infection data in Urumqi during the period of 2008 to 2015 were collected from the internet-based National Infectious Diseases Repotting System(NIDRS).The geocoded address for 81 streets was transformed into a dot map used to as the basis of TB/HIV spatial datasets.Besag-Newell method,CEPP method,and spatial scan statistics was utilized to detecting the spatial clusters of TB/HIV co-infection epidemic based on the scale of street.Composite Reference Standards(CRS)method was also used to calculate the sensitivity and specificity of each method to provide the evidence for evaluating the detectability of each method.Results:(1)498 TB/HIV cases was screened during the period of 2008 to2015.There are spatial variations in TB/HIV co-infection epidemic.The incidence of TB/HIV co-infection in urban area was higher than that in rural area.From the temporal and spatial pattern changes of TB/HIV epidemic,there was a tendency to spread from center to periphery after 2012 year;(2)The spatial clusters existed in TB/HIV co-infection epidemic in Urumqi.Besag-Newell method and CEPP method with different parameter,and spatial scan statistics have detected the spatial clusters of TB/HIV epidemic in Urumqi.Despite the number of clusters of TB/HIV epidemic through three spatial clustering methods was different,however,the main clustered area was similar through the detection of three methods.And the clusters mainly located in the area with higher economic level and larger population density,which.in accordance with the distribution of high risk area.Whereas the clusters of low risk area of TB/HIV epidemic were located in that area with a lower economic level and less population.(3)The comparing results show that Besag-Newell method and CEPP method were highly dependent on their parameter setting,and the clusters varied in accordance with the parameter setting,however,the global spatial-temporal changes of clusters detected by three methods was similar.There was a trend that the clusters of TB/HIV epidemic that varied from peripheral to central area during a period of 2008 to 2012,however,there is an opposite trend that varied from central to peripheral area during a period of 2012 to 2015.The global distribution of the clusters of TB/HIV epidemic in Urumqi was located in central area accompanied by the peripheral area.(4)The sensitivity and specificity evaluated by CRS method suggest that the highest sensitivity in three methods was spatial scan statistic method,and the lowest one was CEPP method.Additionally,the better setting for parameter k of Besag-Newell method was the mean of the incidence,and the sensitivity of CEPP method was higher when parameter N* was set as the number of population corresponding to the mean of incidence.Additionally,spatial scan statistics has a good power in detecting the clusters in rural area or suburbs.Conclusions: There were spatial variations and spatial clusters in the distribution of TB/HIV cases in Urumqi.The global distributed pattern of high risk area and spatial clusters were central aggregation accompanied by peripheral aggregation.This study,based on the spatial point-event data,would provide more clues for the risk factors of TB/HIV epidemic in Urumqi.Moreover,a spatial clustering analysis of disease combined different kinds of spatial analysis methods could provide the reference for the in the practical application of disease prevention and control.