Short-term effects of PM10 on cause-specific mortality in municipalities across industrial sites in Italy.

BACKGROUND: the association between air pollution and human health has been extensivelyinvestigated in the epidemiological literature. However, evidence in non-urban areasaffected by industrial activities is still limited. OBJECTIVES: to assess the association between daily exposure to PM10 and cause-specific mortality in municipalities affected by emissions from industrial plants during the period 2006-2015. DESIGN: two-stage time-series design was applied to assess the relationship between lagged PM10 and the outcomes. In the first stage, the association at the municipal level was analyzed with Poisson regression models adjusted for space-time confounders (trend and temperature). In the second stage, a mixed effects metanalysis of municipal estimates was applied to obtain a pooled estimate. Effect modification for individual variables (sex, age) and type of industrial site (chemical, steel, energy, or mining plant) was assessed. SETTING AND PARTICIPANTS: using the European Pollutant Release and Transfer Register, industrial plants entailing a combustion process were selected. 4x4 km² buffer around the plant was constructed to select the municipalities to be included in the analysis. Daily PM10 was estimated using machine-learning models based on satellite data. MAIN OUTCOMES MEASURES: daily counts of natural, cardiovascular, and respiratory deaths. RESULTS: the average exposure to PM10 was 28.4 μg/m³ (±SD 16.9). During the period under study, 568,804 deaths from natural causes were observed in the 100 municipalities near the 61 industrial sites identified. Percentage risk variations for 10-units increments of lagged 0-1 PM10 were 1.04% (95%CI 0.67;1.41), 1.04% (95%CI -1.21;3.34), and 7.89% (95%CI 0.16;16.23) for natural, cardiovascular, and respiratory mortality, respectively. Higher risk estimates were observed in municipalities near steel plants, especially for respiratory mortality (8.13%; 95%CI -2.85;20.35). No differences were observed in estimates between different age and sex classes. CONCLUSIONS: although not fully able to capture the industrial component of pollution, the results indicate excess risk of mortality in residents of the municipalities under study, especially when considering the presence of steel plants.