Exposure indicators
We used the geo-coded addresses to define residential exposures. Each subject in the cohort was assigned a value of heavy metals and NO2 corresponding to the estimated map values at the residence from the ADMS dispersion model. Quartiles of the distribution of the total population were calculated for heavy metals and NO2. The final categorization, however, was finely adjusted to consider the natural cut-off points.
Based on census data from 2001 for the area of Forlì, an area-based indicator (census block) of socio-economic status (SES) was defined. A total of 1,116 census areas with at least 20 people for the town of Forlì were considered. Four census variables were chosen to represent different levels of social advantage: education, employment, housing conditions, and family composition. A factorial analysis was conducted, to define a composite indicator of socioeconomic position, combining algebraic indicators and using the weight factor scores. The final categories of the SES indicator (from low to high social class) were based on the population distribution (quintiles) of the composite index.