Files
Abstract
Background: The Household Air Pollution Intervention Network (HAPIN) is a randomized controlled trial that measures pregnant women and children’s exposure to carbon monoxide (CO), fine particulates (PM2.5), and black carbon (BC) with a liquified petroleum gas (LPG) intervention in Guatemala, India, Peru, and Rwanda. Objectives: 1) To investigate the short-term CO exposures by comparing rolling averages to World Health Organization (WHO) Air Quality Guidelines (AQGs) stratified by study arm and location, 2) build multivariable models using household questionnaire variables to try to predict CO concentrations, and 3) characterize relative exposures between pairs of pollutants and to explain observed variations in the CO:PM2.5 and BC:PM2.5 correlations and mass ratios. Methods: Implementing the Enhanced Children MicroPEM (ECM) and Lascar CO Datalogger, personal exposures from 3,195 women were investigated within the HAPIN trial. Evaluating minute-by-minute personal CO exposure we compared maximum rolling averages with established WHO AQGs to determine frequency of exceedances. Additional analyzation of 24-hour CO concentrations included over 60 time-variant and -invariant characteristics to build site-specific and HAPIN-wide multivariable models using backwards stepwise regressions. To evaluate the strength of relationships among pollutants and HAP composition, correlations and mass ratios were evaluated between BC and PM2.5 and between CO and PM2.5. Results: Post-randomization HAPIN-wide, we observed significant reductions in the interventions arm for all mean maximum short-term personal CO exposures. In the control arm post-randomization, 4% of exposure visits exceeded a short-term CO AQG but not the 24-hour AQG. We found that the largest associations with covariates for determining CO concentrations were stove type, fuel type, and study site, while other time-invariant measures were more strongly associated than time-variant. The variation in the BC:PM2.5 and CO:PM2.5 ratios were almost all attributed to study arm and study site. When averaging together pollutant concentrations from the two repeated exposure visits, the Spearman Rho’s improved majority of the time for HAPIN-wide comparisons. Conclusions: Applying three unique exposure assessment methodologies, a fuller understanding of LPG intervention stove success was evaluated in low- middle- income households that predominantly use biomass for cooking.