Cancer expected relative survival in Italian regions and provinces: comparison between regional and provincial models based on socio-economic variables and health resources

INTRODUCTION -The ISES Working Group built models for calculating regional and provincial expected relative survival in Italy using demographic, socio-economic and health resources variables (SEH). OBJECTIVES - The aim is to compare the contents of provincial and regional SERTS models in terms of variables which influence the expected relative survival for the main considered cancer sites and to validate the results. MATERIALS AND METHODS - The 5-years relative survival (RS-5%) of 1995-1999 patients’ cohorts was considered for all cancers (but non melanoma skin cancers) and the most levant sites (e.g. colon-rectum, prostate, breast). The SEH are collected at the regional and provincial levels for 1995-2005 period. The principal components factor analysis on SEH extracted two groups of factors, used in multiple linear regression models to compute expected RS-5% at the regional and provincial levels. RESULTS - The extraction of two groups of factors (composed by different SEH) developed the 86% of variance at the egional level and the 73% at the provincial one. The two groups have produced two sets of regression factorial models for each of the considered cancer sites at both geographical levels. The models present good correlations, particularly for all cancers and colon-rectum, stomach, lung, kidney and NHL in both sexes, prostate in men and breast, cervix and melanoma in women. The models allow the calculation of expected RS-5% very similar to the observed RS-5% of the same area. The homogeneity between provincial expected RS-5% and the one of the same region seems to validate the results. The comparison between the observed and expected provincial survival in the regions completely covered by CRs (Veneto, Friuli Venezia-Giulia, Romagna, Umbria) will be the next validation step. DISCUSSION AND CONCLUSIONS - The differences in the groups of factors express the different role in deciding of the regional and provincial administrative levels in allocating and using resources for diagnosis and treatment. The calculation of expected RS-5% seems to produce encouraging results and can explain the differences in outcome and the characteristics of resources management, allowing the calculation even in those areas where no CR works.