Metodi geografici per la sorveglianza epidemiologica

The improved availability of georeferenced health information, population metrics, satellite imagery portraying environmental influences on disease rates, and the progress in Geographic Information Systems (GIS) and geocoding software have significantly eased the investigation of geographical and spatiotemporal fluctuations of disease occurrences in public health surveillance framework. Access to data is crucial for efficiently detecting and addressing public health challenges. These data are essential for preventing and controlling various health issues, such as infectious diseases, non-communicable diseases, injuries, and health-related behaviours. However, the increasing availability and granularity of geospatial data have raised awareness about privacy protection. In the first contribution, an ecological study is conducted to assess the presence and extent of the risk of ovarian cancer associated with asbestos exposure in a large population (10 million people) by applying a series of Bayesian hierarchical shared models to the bivariate spatial distribution of ovarian and pleural cancer mortality by municipality in the Lombardy Region (Italy). The data come from Italian National Statistical Institute (ISTAT) death certificates for the period 2000-2018. Evidence of a shared risk factor between ovarian and pleural cancer was found at the small geographical level. The second contribution aims to estimate attributable cases and fractions of ovarian cancer resulting from asbestos exposure by performing a trivariate Bayesian joint disease model to extract information on attributable cases and attributable fractions from the geographic distribution of ovarian, pleural and breast cancer mortality at municipality level in the Lombardy Region (Italy). We found attributable fractions in the 34-47% range, consistent with known heavy asbestos pollution in some municipalities. The impact of asbestos can be relevant but may go unnoticed when the background risk of ovarian cancer is low. Bayesian modelling provides helpful information in such context, and it is useful to tailor epidemiological surveillance. The third work deals with geoprivacy concerns. This study aims to use a geomasking procedure identifying the amount of displacement that protects confidentiality and avoids the deterioration of geostatistical inference assessing the spatial distribution of contamination risk of Per- and poly-fluoroalkyl substances (PFAS) using data from a food monitoring campaign from 2016 to 2017 by the National Health Institute sponsored by Veneto Region. A simple procedure is proposed through a simulation study to calculate the displacement required for georeferenced data to ensure citizens' right to access information (data transparency) compared to the institutions responsible for public health activities. We used a uniform displacement to facilitate sharing data; we found that there seems to be a threshold delta (up to 2 km) at which the results of analyses of masked data become substantially different from the original results. When carefully implemented case-to-case basis, geographic masking appears to be a workable solution for safeguarding geoprivacy while providing researchers access to georeferenced, individual-level data. By showing the different results of these three contributions, this dissertation aims to illustrate complex spatial methods for areal and geostatistical data to furnish helpful information to plan and implement public health interventions.