The current study aimed at assessing the effects of exposure to Particulate Matter (PM) on the incidence of respiratory diseases in a sub-sample of participants in the longitudinal analytical epidemiological study in Pisa, Italy. Three hundred and five subjects living at the same address from 1991 to 2011 were included. Individual risk factors recorded during the 1991 survey were considered, and new cases of respiratory diseases were ascertained until 2011. Average PM10 and PM2.5 exposures (µg/m3, year 2011) were estimated at the residential address (1-km2 resolution) through a random forest machine learning approach, using a combination of satellite data and land use variables. Multivariable logistic regression with Firth's correction was applied. The median (25th-75th percentile) exposure levels were 30.1 µg/m3 (29.9-30.7 µg/m3) for PM10 and 19.3 µg/m3 (18.9-19.4 µg/m3) for PM2.5. Incidences of rhinitis and chronic phlegm were associated with increasing PM2.5: OR = 2.25 (95% CI: 1.07, 4.98) per unit increase (p.u.i.) and OR = 4.17 (1.12, 18.71) p.u.i., respectively. Incidence of chronic obstructive pulmonary disease was associated with PM10: OR = 2.96 (1.50, 7.15) p.u.i. These results provide new insights into the long-term respiratory health effects of PM air pollution.
Keywords: air pollution; incidence; long-term exposure; particulate matter; questionnaire; random forest; respiratory symptoms/diseases.