Multi-City Time Series Analyses of Air Pollution and Mortality Data Using Generalized Geoadditive Mixed Models

• Activity Number: 600
• Type: Contributed
• Date/Time: Thursday, August 6, 2009 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Lung-Chang Chien*+ and Shrikant Bangdiwala and Jiu-Chiuan Chen and Mark Weaver and Todd Schwartz and John S. Preisser
• Companies: The University of North Carolina at Chapel Hill and The University of North Carolina at Chapel Hill and The University of North Carolina at Chapel Hill and Family Health International and The University of North Carolina at Chapel Hill and The University of North Carolina at Chapel Hill
• Keywords: air pollution ; mortality ; spatio-temporal data ; generalized geoadditive mixed model

Abstract

In air pollution research, traditional modeling approaches don't have efficient procedures to fit air pollutant effects on national level and city level, and are lack of a convenient way to handle geographical variation. Here we introduce the generalized geoadditive mixed model (GGAMM) by combining GAM, GLMM and spatial function within an unified model structure. We illustrate its application using air pollutant and mortality data with 15 US cities in 1991--1995. We found that, at the national level a 10-µg/m^3 increase in PM10 was associated with 0.11% (95% CI: -0.46, 0.67) increases in the relative risk of respiratory mortality in elders aged =65 years, after adjusting for nonlinear time trend and temperature. The spatial function showed Northeast US cities had higher geographical effect than the others. As a result, we recommend the GGAMM to analyze spatio-temporal data.