統計学輪講（第２２回）

日時　　  2001年10月 2日（火）　15時00分〜16時40分
場所　　  経済学部5階視聴覚室
講演者　　丹後俊郎（国立公衆衛生院）
演題　    A spatio-temporal model for excess environmental risk
          around putative sources

概要：
　　Since the 1980s there has been growing interest in the analysis of 
small area data to investigate the relation between the risk of a disease 
and proximity of residence to a prespecified putative source of hazard. 
It is well known that the apparent excess of cases of childhood leukaemia 
near the nuclear reprocessing plant such as that in the village of 
the Seascale at Sellafield has been extensively investigated. More recently, 
there is great public concern on the health effects of so called dioxin, 
organic compounds such as polychlorinated dibenzodioxins (PCDDs) and 
dibenzofurans (PCDFs), emitted from municipal solid waste incinerators.　
　　 Many statistical models to detect such an excess environmental risk 
or a cluster of cases around a putative source of hazard have been proposed. 
However, most of them are spatial models which cannot account for temporal 
trends. In this talk, a spacio-temporal model is presented for detecting 
such a subtle excess risk based on small area data with many zero frequencies.
The proposed model is illustrated with small area data regarding an excess 
risk of perinatal undesirable outcomes associated with maternal residential 
proximity to nuclear power plants in Japan.