Household Cooking with Solid Fuels Contributes to Ambient PM2.5 Air Pollution and the Burden of Disease

Background: Approximately 2.8 billion people cook with solid fuels. Research has focused on the health impacts of indoor exposure to fine particulate pollution. Here, for the 2010 Global Burden of Disease project (GBD 2010), we evaluated the impact of household cooking with solid fuels on regional population-weighted ambient PM2.5 (particulate matter ≤ 2.5 μm) pollution (APM2.5). Objectives: We estimated the proportion and concentrations of APM2.5 attributable to household cooking with solid fuels (PM2.5-cook) for the years 1990, 2005, and 2010 in 170 countries, and associated ill health. Methods: We used an energy supply–driven emissions model (GAINS; Greenhouse Gas and Air Pollution Interactions and Synergies) and source-receptor model (TM5-FASST) to estimate the proportion of APM2.5 produced by households and the proportion of household PM2.5 emissions from cooking with solid fuels. We estimated health effects using GBD 2010 data on ill health from APM2.5 exposure. Results: In 2010, household cooking with solid fuels accounted for 12% of APM2.5 globally, varying from 0% of APM2.5 in five higher-income regions to 37% (2.8 μg/m3 of 6.9 μg/m3 total) in southern sub-Saharan Africa. PM2.5-cook constituted > 10% of APM2.5 in seven regions housing 4.4 billion people. South Asia showed the highest regional concentration of APM2.5 from household cooking (8.6 μg/m3). On the basis of GBD 2010, we estimate that exposure to APM2.5 from cooking with solid fuels caused the loss of 370,000 lives and 9.9 million disability-adjusted life years globally in 2010. Conclusions: PM2.5 emissions from household cooking constitute an important portion of APM2.5 concentrations in many places, including India and China. Efforts to improve ambient air quality will be hindered if household cooking conditions are not addressed. Citation: Chafe ZA, Brauer M, Klimont Z, Van Dingenen R, Mehta S, Rao S, Riahi K, Dentener F, Smith KR. 2014. Household cooking with solid fuels contributes to ambient PM2.5 air pollution and the burden of disease. Environ Health Perspect 122:1314–1320; http://dx.doi.org/10.1289/ehp.1206340


Model Methodologies
The GAINS data used in Equation 1 represent the PPM 2.5 attributable to cooking at the household level. They do not include, in either the numerator or denominator, secondary particle formation. The TM5-FASST data, used in Equation 2, include PPM 2.5 and associated secondary particles; they represent do not include dust or sea salt. To estimate the dust/salt increment, country-or regional-level estimates of combustion-derived PM 2.5 obtained from TM5-FASST were compared with country-level estimates of total APM 2.5 (including dust and sea salt) developed in Brauer et al.(Brauer et al. 2012) for the Global Burden of Disease project, using inputs from TM5-FASST as well as ground and satellite observations. Current air quality legislations are included for both 2005 and 2010 PM 2.5 estimates. While ambient air quality legislations do not necessarily have a large impact on cooking-related emissions, the total APM 2.5 in 2010 would have been higher if no legislation was assumed in the 2000-2010 period.

MESSAGE covers all greenhouse gas (GHG)-emitting sectors, including power plants, industry
(combustion and process), road transport, households, international shipping and aviation, agricultural waste burning, and biomass burning (deforestation, savannah burning, and vegetation fires) for a full basket of greenhouse gases and other radiatively active gases.
To estimate the impacts of these spatially explicit emissions, atmospheric concentrations of average ambient population-exposure weighted anthropogenic PM 2.5 and also specifically the household-related fraction are further derived using the TM5 -FASST source-receptor model.
Modeled PM 2.5 includes contributions from (i) primary PM 2.5 released from anthropogenic sources and forest fires, and (ii) secondary inorganic aerosols formed from anthropogenic emissions of SO 2 , NO x and NH 3 (including water vapor). The data are reported on a spatial level and are then aggregated by country and GBD region. (See Supplemental Material, Table 1.) Population-weighted annual average ambient PM 2.5 concentration estimates (APM 2.5 ) were produced by overlaying APM 2.5 concentrations (from TM5-FASST, 1°x1° resolution) with highresolution population maps (0.042°x0.042°). This allowed us to adjust the primary components of PM 2.5 concentration gradients within the 1°x1° gridcell using the underlying population gradients as a proxy. This method parametrizes the so-called urban increment and allows for a populationweighted APM 2.5 concentration which is higher than or equal to the 1°x1° area-averaged PM 2.5 concentration (Brauer et al. 2012;Rao et al. 2012).
Table S1. Regional Groupings Used in Global Burden of Disease 2010 (With Regional

Population in 2010 of Countries Included in this Analysis and Total Population in Region). Note:
countries that were not included in this analysis, because of data gaps, are shown in italics.  a Population x 10 6 . b Percent of primary PM 2.5 household emissions attributable to household cooking (GAINS). c Percent of combustion-derived emissions attributable to household cooking and heating (TM5-FASST).
Figure S1. Emissions and particle coverage in the major databases and models used in this analysis. Note that sea salt, dust, and some secondary particle precursors are not included in the models used here; however, they are represented in the total ambient PM 2.5 concentrations calculate for GBD 2010, published in Brauer et al. (2012) and used in the final stages of the analysis presented in this analysis.