Spatial Parameterization of Infant Mortality in Anhui Province, China
First page number:
Last page number:
The study is motivated by an unexplained and relatively high infant mortality in Anhui Province, where a suspected unknown source may contribute to it. However, spatially focused disease clustering tests tend to base on a known location, and in many cases, use the point source as the referent. When elevated risks in some areas are caused by unknown sources, it becomes necessary to infer these sources spatially. The paper extends spatial cluster parameterization models from known sources to unknown sources by a profile likelihood method. In both simulation and extended case study, we show that the spatial parameterization method is able to effectively identify the cluster influence center and measure cluster strength. We found that the center is located where the borders of several counties met in a relatively well-to-do part of the province. In addition to its ability to identify cluster influence away from the traditional centroid point, the parameterized method tends to perform better than the common spatial cluster detection method in terms of goodness of fit statistic and location specificity.
Spatial parameterization; Cluster point source; Profile likelihood; Bootstrap
Kan, G. L.
Spatial Parameterization of Infant Mortality in Anhui Province, China.
Spatial Statistics, 14(C),