Exploring cancer patterns in Norway through cluster analysis

Abstract

Cluster analysis may be applied to group different diseases based on similarities in their geographical distributions as well as to group different areal units based on their disease ‘profiles’. In this study both approaches are being demonstrated using mortality data for different types of cancer for the period 1966–1987. Results from the use of different options in the SPSS/PC program are compared. Two alternative regionalizations are used, counties and rural-urban dichotomies of counties. In southern Norway rural and urban areas respectively tend to cluster with similar areas in neighbouring counties, whereas in northern Norway disease regions are formed through clustering of urban and rural areas within the same county.