RESUMO
OBJECTIVE: Sporadic cases occurring in mall geographic unit could lead to extreme value of incidence due to the small population bases, which would influence the analysis of actual incidence. METHODS: This study introduced a method of hierarchy clustering and partitioning regionalization, which integrates areas with small population into larger areas with enough population by using Geographic Information System (GIS) based on the principles of spatial continuity and geographical similarity (homogeneity test). This method was applied in spatial epidemiology by using a data set of thyroid cancer incidence in Yiwu, Zhejiang province, between 2010 and 2013. RESULTS: Thyroid cancer incidence data were more reliable and stable in the new regionalized areas. Hotspot analysis (Getis-Ord) on the incidence in new areas indicated that there was obvious case clustering in the central area of Yiwu. CONCLUSION: This method can effectively solve the problem of small population base in small geographic units in spatial epidemiological analysis of thyroid cancer incidence and can be used for other diseases and in other areas.