AAAR 36th Annual Conference October 16 - October 20, 2017 Raleigh Convention Center Raleigh, North Carolina, USA
Abstract View
Quantifying the Sensitivity of Mortality Calculations to Household-level, Ambient-level, and Combined Exposure to PM2.5 from Solid-fuel Combustion
JACK KODROS, John Volckens, Michael Brauer, Ellison Carter, Kelsey Bilsback, Christian L'Orange, Michael Johnson, Jeffrey R. Pierce, Colorado State University
Abstract Number: 245 Working Group: Aerosol Exposure
Abstract Many studies have estimated the burden of human disease due to PM2.5 exposure from household combustion of solid fuel. Central estimates of mortality from this exposure source range from 2.9 to 4.3 million deaths for 2012, globally. While it is clear that residential solid-fuel combustion poses a substantial health burden, the sensitivity of estimated mortality to uncertainties in individual input parameters (such as PM2.5 concentration, baseline mortality rate, or concentration-response functions) has not been quantified. We perform a variance-based sensitivity analysis on mortality due to exposure to PM2.5 from solid-fuel combustion at the ambient level and household level. We also consider the mortalities attributable to solid-fuel use from the combined effect of exposure to PM2.5 from solid-fuel combustion and PM2.5 from ambient sources. When considering the combined effect of ambient- and household-level exposure, we estimate 300,000 fewer deaths per year attributable to solid-fuel use than when only household-level exposure is considered. This is likely attributable to the log-linear health-response to PM2.5 exposure. Exposure to PM2.5 from solid-fuel combustion in regions with already high (>40 µg m-3) concentrations of PM2.5 from other anthropogenic sources does not proportionately increase disease burden. We find that concentration-response functions contribute the most to the variance in mortality (50-60% across Asia and 40-50% in South America and Africa) with the fraction of the population using solid fuel for energy the next largest contributor (15-30%). Conversely, the ambient and household PM2.5 exposure contributes very little to the estimated variance in mortality estimates (1-7%). This suggests that in order to reduce uncertainty in mortality calculations at the ambient- and household-level exposures, future studies should focus on more precise quantification of exposure-response relationships to PM2.5.