VALUTAZIONE DI IMPATTO SANITARIO: METODOLOGIA E APPLICAZIONI

The aim of our Ph.D. Research Program was to improve the knowledge and test the implementation in the real world of that combination of procedures, methods and tools defined as Health Impact Assessment (HIA), whose aim is to assess the impact that the construction of industrial plants, infrastructures, housing, transport, energy plants and, broadly, any policy, plan and project in diverse economic sectors may cause on the health of the general population. An HIA procedure was applied to the world exposition “EXPO 2015” (Milan, Italy), focusing on three different phases. The analysis on the PRE-CONSTRUCTION period (ante-operam) allowed us to estimate the effects of exposure to particulate matter with aerodynamic diameter ≤ 10µm (PM10) and nitrogen dioxide (NO2) on the health of the study population, before the opening of the construction site (2007-2011). The effect estimates were expressed as percent variations in the frequency of expected health events in the population (deaths and hospitalizations) per 10µg/m3 increase in the pollutants concentration. Those estimates (i.e. concentration-response functions, CRF) were obtained by applying Poisson regression models to the daily counts of health events, and were then used to estimate the impact that the exposure to the same pollutants would have entailed during the subsequent CONSTRUCTION phase (January 2013 – April 2015). The impact was thus quantified in terms of deaths and hospital admissions attributable to the difference between the mean PM10 and NO2 annual concentrations estimated for the study period and the mean PM10 and NO2 annual concentrations corresponding to selected counterfactual scenarios. Our reference scenario was represented by the mean PM10 and NO2 annual concentrations measured by the network of air pollution monitoring stations of the regional environmental protection agency (ARPA Lombardia) in the last year of the PRE-CONTRUCTION phase (2011). In the CONSTRUCTION phase, we estimated 0.54 natural deaths and 0.70 hospitalizations (due to cardiac, cerebrovascular, and respiratory causes) attributable to PM10 levels exceeding the mean concentrations of the PRE-CONSTRUCTION period, in one year. As regards NO2, we estimated 0.36 and 0.56 additional deaths and hospitalizations, respectively. During the second year of the Ph.D. Program the HIA procedure was applied to the EVENT phase (May-October 2015). We obtained estimates of the traffic flow expected during the opening of the exhibition site. Such estimates were then inputted into dispersion models to obtain maps of the ground fallout of PM10 and NO2 emitted by the traffic due to the exhibition. We observed a mild contribution of the EXPO-related traffic flow to the air pollutants concentration averages in the investigated area. The ground fallout levels (annual means) ranged between 0.019 and 0.067µg/m3 for PM10 and between 0.270 and 0.684µg/m3 for NO2. Deaths and hospital admissions data were updated with information from the local health authorities. We were thus able to estimate, on a historical basis, the baseline number of health events we could have expected in our population during the study period. Using the CRFs obtained for the PRE-CONSTRUCTION phase and up-to-date exposure and health data, we estimated, for the EVENT phase, 0.11 natural deaths and 0.15 hospitalizations (due to cardiac, cerebrovascular, and respiratory causes) attributable to PM10 levels exceeding the mean concentrations of the PRE-CONSTRUCTION period, in one year. As regards NO2, we estimated 1.45 and 2.33 additional deaths and hospitalizations, respectively. The HIA methodology was subsequently applied to another study, conducted within the ESSIA project (Effetti Sulla Salute degli Inquinanti Aerodispersi in regione Lombardia: Health Effects of Air Pollutants in Lombardy). The project has been ongoing for several years and quantified the association between air pollutants and population health (effect and impact), based on Lombardy specific characteristics (see Baccini et al., 2011. Am J Epidemiol. 174, 1396-405 and Baccini et al., 2015. Environ Health Perspect. 123, 27-33). The published investigations focused on particulate matter exposure and all-cause mortality, taking into account between-city commuting as well. During the second year of our Ph.D. Program, we expanded the investigated exposures and health events and verified whether the reduction in air pollution levels observed in Lombardy in the last ten years paralleled a decrease in the number of health events. We considered exposure to PM10 and NO2, mortality and hospitalization data for a non-opportunistic sample of the most polluted and densely populated areas of the region (years 2003-2006). We obtained area-specific effect estimates for PM10 and NO2 applying Poisson regression models to the daily count of all-cause deaths and cause-specific hospitalizations (cardiac, cerebrovascular, and respiratory causes). Area-specific estimates were then combined in a random-effect Bayesian meta-analysis. For cardiovascular and respiratory mortality, we applied a case-crossover analysis. Effect estimates were expressed as percent variation of deaths and hospital admissions per 10µg/m3 increase in PM10 or NO2 concentrations. Natural mortality was positively associated with both pollutants (0.30%, 90% Credibility Interval [CrI]: -0.21; 0.70 for PM10; 0.70%, 90%CrI: 0.20; 1.18 for NO2). Cardiovascular deaths were more strongly associated to NO2 (1.12%, 90% Confidence Interval [CI]: 0.30; 1.95), while respiratory mortality was highest in association with PM10 (1.64%, 90%CI: 0.56; 2.72). The effect of both pollutants was more evident in the summer season and a trend in mortality with increasing age classes was apparent for PM10 only. Air pollution was also associated to hospitalizations, the highest variations being 0.77% (90%CrI: 0.31; 1.32) for respiratory diseases and PM10, and 1.70% (90%CrI: 0.60; 2.66) for cerebrovascular diseases and NO2. The effect of PM10 on respiratory hospital admissions increased with age. For both pollutants, effects on cerebrovascular hospitalizations were more evident in subjects aged less than 75 years. In a sub-analysis on all-cause mortality, we evaluated how the mortality burden due to PM10 exposure estimated for the period 2003-2006 (see Baccini et al., 2011. Am J Epidemiol. 174, 1396-405) varied when considering its concentrations in 2014. Assuming our study population and its mortality rates remained constant over time and applying the previously estimated CRFs, we quantified the number of deaths attributable to exposure levels exceeding the threshold of 20µg/m3 for PM10 annual average (WHO Air Quality Guidelines, 2005). The difference between attributable deaths estimated in 2003-2006 and 2014 represents the variation of the mortality burden between the two periods or, in other words, the number of deaths “avoided” thanks to the reduction in PM10 concentrations measured in the last decade. In 2014 we estimated 162 natural deaths less than in the period 2003-2006 (CrI80%: 24.2; 311.6) attributable to PM10 levels exceeding the WHO threshold. As expected, the bigger impact was observed in the capital city of Milan, with a difference between the two periods of 116 deaths due to natural causes (CrI80%: 65; 176).