An evaluation of the edge effects in disease map modelling

The aim of this work is to find out how the estimation of the mortality relative risk from a disease near boundaries can be affected by the edge position. Also, the behaviour of a small selection of models for disease mapping is evaluated when different edge conditions exist in the data. One data set is examined and to make the results more rigorous a simulation study was set up where various relative risk patterns (from simple risk gradients to more complex risk structures, including spatial correlation) were simulated.