Article Text

Download PDFPDF

OP VII – 5 Long-term pm10 exposure and cause-specific mortality in latium region: a difference-in-differences approach
  1. Matteo Renzi,
  2. Francesco Forastiere,
  3. Marina Davoli,
  4. Paola Michelozzi,
  5. Massimo Stafoggia
  1. Department of Epidemiology, Lazio Regional Health Service, ASL Rome 1, Rome, Italy


Background/aim The link between particulate matter (PM) exposure and health has been widely explored in epidemiological literature. However, the availability of environmental data is limited to urban settings and little is known about the health effects of PM in rural areas. Our aim is to evaluate a causal association between PM10 and cause-specific mortality in Latium Region (centre Italy) during 2006–2012.

Methods We used satellite data combined with spatial predictors in 3-stage mixed models to obtain daily PM10 values on a grid map of 1 × 1 km cells in the Region. For each municipality, we averaged daily PM10 values of each cell in a single yearly value. We used a difference-in-differences approach to estimate a causal relationship between annual PM10 exposure and non-accidental, cardiovascular, and respiratory mortality. We applied Poisson regression models adjusted for municipality and calendar year in order to focus on municipality-levels annual fluctuations of exposure and mortality. In addition, we also added terms for mean and standard deviation (SD) of winter and summer temperatures to account for potential meteorological confounders displaying different temporal trends across municipalities.

Results During 2006–2012 we observed 3 47 699 deaths for non-accidental, 92 787 for cardiovascular and 16 509 for respiratory causes in 378 municipalities of Latium Region. The average PM10 concentration during the period was 21.9 mg/m3 (SD 4.9) in Latium. The mean winter and summer temperatures were 11.9, and 17.8°C, while the respective SDs were 2.7°C and 3.9°C. For each IQR (6.8 µg/m3) increase in annual PM10, we estimated a 5.9% (95% CIs: 3.3, 8.5%) increase in non-accidental, 6.2% (1.3, 11.4) cardiovascular, and 17.0% (4.1, 31.5%) in respiratory mortality, respectively. As sensitivity approach, we excluded Rome from the analysis and we found percent increases of risk of 4.2% (1.4, 7.1%) for non-accidental, 4.9% (-0.5, 10.5%) for cardiovascular, and 18.1% (3.8, 34.3%) for respiratory mortality.

Conclusion Our analysis suggests a causal effect of PM10 on cause-specific mortality in Latium Region, with a strongest effect on mortality for respiratory causes. We can conclude that PM10 represents a strong risk factor for human health not only in urban settings but also in suburban and rural areas.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.