Ranking spatial areas by risk of cancer: modelling in epidemiological surveillance