Compared application of local and national deprivation indices to study cancer incidence in Italian regions

Aims The goal of the study was to compare the different results in analysing cancer incidence by sites through socio-economic deprivation indices, computed at local (LDI) and national (NDI) levels, stressing the different ability in identifying the associations with the socio-economic status (SES) of the populations. Materials and methods The variables building both, the LDI and NDI, came from the National Italian Census 2001. The NDI was computed for all the Italian population but re-parameterised for the considered geographical areas (Liguria, Umbria, Campania), while the LDIs were computed specifically for the same areas. The variables in the LDIs were chosen according to their correlation with the general mortality and the indices were computed by the Principal Component Analysis technique. Cancer data came from the registries operating in the considered areas. Results LDIs performances seemed to be more efficient with respect to the NDI ones in discriminating the cancer incidence across SES, because they were able to take into account the peculiarities of the local population, particularly where the population is less urbanised. In fact, the number of statistical significant associations is higher than the ones found by the NDI and they were generally homogeneous with the international literature. The NDI and LDI produced similar results where the populations were more urbanised, stressing the main feature of the NDI as descriptor of the national population as a whole (67% of urbanised people). Conclusions We propose the use of the two types of index for different public health purposes. LDIs are able to provide specific information for an improvement of the health and social support policies and actions, identifying specific health needs in targeted groups of population. NDI can give a more general description of the relationships between deprivation and health status, allowing a comparison among different regions and providing a validation benchmark for the LDIs.