2013 Environment and Health - Basel
Design and implementation of household air pollution intervention studies in relation to birth outcomes: Results from Phase I activities in Ghana and India
Background: Randomized controlled trials (RCTs) offer the best opportunity to quantify the health impacts of household air pollution (HAP), as exposures correlate with income and other intractable confounders. Yet RCTs pose substantial challenges, including cost and potentially limited external validity.
Methods: We present and critically assess two RCT pilots that focus on birth outcomes and child pneumonia. Experiences in Ghana (with the Kintampo Health Research Centre) and in India (with UC Berkeley, SRU, and INCLEN) at the SOMAARTH site are discussed. Results: In India, an initial user preference assessment of two candidate fan stoves revealed a clear winner. A subsequent pilot demonstrated the feasibility of recruiting pregnant women, ultrasound for gestational age, automated tracking of stove use, and assessing personal CO exposure. In Ghana, the search for viable interventions has been particularly arduous. Pilot work suggested that exposure reductions from passive rocket stoves are small in magnitude, but that LPG reduces exposures substantially. Securing LPG supplies presented major difficulties. Other pilot work focused on refining exposure assessment protocols to incorporate personal PM2.5 monitors.
Conclusions: First, HAP research should offer a clear path from research to large-scale public health interventions. In the cases under consideration, research interventions, if successful, could be grafted on to existing antenatal care programs. Second, we underscore the importance and difficulty of identifying interventions that are both acceptable to village cooks and truly clean. In many settings this will require clean fuels, given current biomass stove technology. Third, while RCTs provide robust evidence of causality, exposure response analysis remains key, as intention-to-treat results may not extend to other interventions. RCTs should collect the detailed exposure and covariate data necessary for ER analysis.