Development and Application of Spatially Disaggregated Exposure Series in Time-series Analyses of Air Pollution-related Health Effects in Chennai, India

O-30A5-2 Background/Aims: A recent meta-analysis of the health effects of outdoor air pollution conducted in Asia emphasizes the need to include countries and cities where few studies have previously been conducted. We report here the results of such a time-series analysis of short-term effect on mortality due to measured PM10 and modeled PM10, PM2.5, and O3 in Chennai, India. The primary focus in the study was on developing a representative exposure series of PM10 for Chennai population in order to estimate the associated relative risk (RR) for all-cause mortality within the limitations imposed by routinely collected retrospective data and to validate the estimates of RR by modeled PM10, PM2.5, and O3 data. Methods: We used data on PM10 collected from AQMs throughout the city and all-cause mortality collected from all city zones for the period 2004–2007. Modeled pollutant data were generated for 1 year (2007), using the WRF-CHEM model. We developed a zonally disaggregated exposure series and subsequently used quasi-Poisson regression with smooth functions of time, weather parameters to estimate the RR for mortality, using zonally disaggregated exposure series and the modeled data exposure series. Results: The model estimated an RR of 1.0037 (1.0008, 1.0067) per 10 μg/m3 increase in daily PM10 concentrations while using empirical data and estimated an RR of 1.0014 (0.9994–1.0034), 1.0033 (0.9878–1.0189), and 1.0103 (0.9787–1.0429) for modeled PM10, PM2.5, and O3, respectively. Conclusion: Our estimates obtained using empirical data are robust and comparable to similar estimates from other countries. The development of methodological refinements to address specific data limitations (missing measurements and small footprints of AQM) could be useful for routine analysis in the developing country where similar issues prevail. However, the modeled data estimates indicate the relative inability of the emission inventory based models to capture daily difference in emission. But, the refinements being made are expected to improve their application in India.