VARIAZIONE SPAZIALE DEL RISCHIO RELATIVO DI MORTE PER TUMORE DEL SISTEMA EMOLINFOPOIETICO, IN UN¿AREA PROSSIMA AD UNA RAFFINERIA PETROLIFERA

Kernel smoothing is a popular approach to estimating relative risk surface from data in the location of cases and controls in geographical epidemiology. To date, the use of a fixed smoothing parameter has dominated in literature, but it can perform poorly with highly heterogeneous population. A more intuitive approach is to utilize a spatially adaptive, variable smoothing parameter where the amount of smoothing depends inversely on the local amount of data. The interpretation of such surfaces is facilitated by plotting of tolerance contours which highlight areas where the risk is sufficiently high to reject the null hypothesis of unit relative risk. This approach has been applied to a epidemiological survey about the risk of death in the area surrounding an Italian petrochemical refinery where 172 deaths and 337 controls has been identified. Maps of the spatial variation of relative risk and tolerance contours have be produced. Only for women and those spending most of their time at home, we found a sub-region of elevated risk highlighted by plotted tolerance contours.