Global Estimation of Exposure to PM2.5 from Household Air Pollution

Abstract

Background: Household air pollution (HAP) exposure from cooking with dirty fuels is a major global health risk factor. Epidemiological studies have demonstrated significant variation in particulate matter concentrations of diameter ≤ 2.5 micrometers (PM2.5), an important metric for using integrated exposure–response functions to assess risks. To characterize global HAP-PM2.5 exposures, novel estimation methods are needed, as financial/resource constraints render it difficult to monitor exposures in all relevant areas.Methods: A Bayesian, hierarchical HAP-PM2.5 global exposure model was developed using kitchen and female HAP-PM2.5 exposure data available in published, peer-reviewed studies. Cooking environment characteristics and quantitative HAP-PM2.5 measurements from 47 studies were used to model urban and rural, fuel- and country-specific (traditional wood, improved biomass, coal, dung and gas/electric stoves) 24-hour HAP-PM2.5 kitchen concentrations and male, female and child exposures for 106 countries in Asia, Africa and Latin America.Results: A model incorporating fuel/stove type, urban/rural location and the socio-demographic index resulted in a Bayesian R2 of 0.57. Estimated global average 24-hour HAP-PM2.5 concentrations in rural kitchens using traditional, improved biomass, animal dung, and coal stoves were 320 μg/m3, 180 μg/m3, 1,760 μg/m3 and 400 μg/m3, respectively, higher than in rural kitchens using gas/electricity. Modeled female exposures from traditional wood stoves varied from 90–260 across countries, on average, with urban area exposures 40 μg/m3 less than those in rural areas. Male and child rural area exposures from traditional wood stoves ranged from 60-190 and 80-230 μg/m3, respectively; urban area exposures were 10 μg/m3 less than rural area exposures, among both sub-groups.Conclusions: A global exposure model incorporating type of fuel-stove combinations adds specificity and reduces exposure misclassification for estimation of HAP risk.