[Evaluation of the use of the socioeconomic deprivation index at area level in ecological studies on environment and health].

BACKGROUND: in SENTIERI surveillance system, the health risk assessment in the contaminated sites (CSs) was carried out considering the socioeconomic deprivation index (ID-SENTIERI) at the municipal level as a confounder. Pasetto et al. discussed the use of IDs in ecological studies and proposed to examine the role of ID-SENTIERI as a confounder.

OBJECTIVES: to evaluate the initial conditions necessary for the role of confounding, i.e., the verification of an association of the ID-SENTIERI with the risk of some of the diseases for which evidence of a relationship with the socioeconomic state (SES) is documented in the literature.

DESIGN: the methodological approach is to exploit ecological correlations in order to assess the relationship between ID and the occurrence of mortality or congenital malformations (CMs).

SETTING AND PARTICIPANTS: mortality and CM data from municipalities in the regions of Emilia-Romagna (Northern Italy), Tuscany (Central Italy), and Campania (Southern Italy) were used. Mortality for all causes, all tumours, and lung cancers were analysed for the period 1995-2008. For the CMs, data from the regional registries of Emilia-Romagna, Tuscany, and Campania in the periods 1995-2011, 1992-2011, 2004-2010, respectively, were used. Analyses of the relationship between mortality/CMs and ID-SENTIERI were carried out according to cause, region, gender, and demographic amplitude classes using a GAM-Poisson regression model.

RESULTS: the mortality analyses show that, for the male population of the three regions under study, the relationship between ID and mortality is increasing for each outcome for municipalities with less than 50,000 inhabitants; while results for female population differ from region to region. The results of the CMs show growing associations, considering only municipalities under 10,000 inhabitants.

CONCLUSIONS: the results of the study suggest that scientific evidence of an association between SES and health should be evaluated taking into account the gender, the study area, and the demographic size of the municipalities. For ID to be used as a confounding variable, it must be associated with the outcomes for which the influence of socioeconomic factors in the reference areas has been ascertained. If there is no such association, the ID cannot be used as a confounder, as it could lead to distortions of estimates and such distortions are difficult to predict. In these cases, ad hoc IDs should be produced.