Skip Navigation LinksHome > November 2009 - Volume 20 - Issue 6 > Nitrogen Dioxide Spatial Variability in Rome (Italy): An App...
Epidemiology:
doi: 10.1097/01.ede.0000362420.36474.59
Abstracts: ISEE 21st Annual Conference, Dublin, Ireland, August 25-29, 2009: Symposium Abstracts

Nitrogen Dioxide Spatial Variability in Rome (Italy): An Application of the LUR Model Over a Decade

Porta, Daniela*; Cesaroni, Giulia*; Badaloni, Chiara*; Stafoggia, Massimo*; Meliefste, Kees†; Forastiere, Francesco*; Perucci, Carlo Alberto*

Free Access
Article Outline
Collapse Box

Author Information

*Department of Epidemiology Local Health Authority RME, Rome, Italy; and †Institute for Risk Assessment Sciences, University Utrecht, Utrecht, Netherlands.

ISEE-0184

Back to Top | Article Outline

Background and Objective:

Land Use Regression (LUR) models have been increasingly used to assess exposure to air pollution within urban areas. There are no available studies comparing the performance of two or more LUR models over long time periods. To estimate exposure to air pollution among subjects enrolled in a large cohort study in Rome, we compared two LUR models from NO2 measurements collected in two periods.

Back to Top | Article Outline

Methods:

We measured NO2 at 67 and 78 locations in 1995/96 and in 2007, respectively, over three three-weeklong periods (winter, spring and fall). The sites stayed the same, but 11 were added in 2007. Several land-use and traffic variables were available. The association between each land-use variable and NO2 concentrations was assessed by univariate and multiple linear regressions. The final model was constructed through a backward elimination procedure (P>0.20).

Back to Top | Article Outline

Results:

Mean NO2 concentration was 45.4 μg/m3 (SD = 6.9) in 1995/96 and 44.6 μg/m3 (SD = 11.0) in 2007. There was a high correlation between the surveys (r = 0.79). The most important predicting variables in 1995/96 were circular traffic zones, altitude, geographic coordinates, inverse population density, distance from the nearest high traffic roads and traffic density in a 150 meter buffer zone, A multiple regression model including these variables resulted in an adjusted R2 of 0.724. The model for 2007 included circular traffic zones, altitude, geographic coordinates, size of census block, and meters of high traffic road in a 150 meter buffer zone (R2 = 0.659).

Back to Top | Article Outline

Conclusion:

The models developed for the two periods were sufficiently comparable, although some of the variable performed differently. The approach is useful for studying a large cohort over a long period of time.

© 2009 Lippincott Williams & Wilkins, Inc.

Twitter  Facebook

Login

Article Tools

Share