Household Air Pollution in Ghana: Stove Use, Health Impacts, and Policy Options

Background: Three billion individuals worldwide rely on biomass fuel (crops, dung, wood) for cooking and heating, mostly in the developing world. Incomplete combustion of these biomass fuels in inefficient cookstoves leads to high levels of household air pollution (HAP). Health conditions resulting from HAP are responsible for approximately 1.6 million premature deaths each year. Of the diseases associated with HAP exposure, lower respiratory infections (LRIs) are the leading cause of death for children under five worldwide. There is a great need to understand the etiology of HAP-associated LRIs to inform health interventions and to improve treatments. Ultimately, however, the only way to prevent the disease burden from HAP is to stop exposure. Policies and programs to promote the use of clean fuels for cooking are a pivotal prevention strategy. Methods: All three studies draw from an established cohort in Ghana. The Ghana Randomized Air Pollution and Health Study (GRAPHS), was a cookstove intervention trial in Kintampo, Ghana. Participants were randomized to a more efficient biomass cookstove arm, a liquefied petroleum gas (LPG) stove arm, or the traditional cookstove arm (baseline). The principal outcome of GRAPHS was childhood pneumonia. The first chapter utilizes banked nasal swabs from GRAPHS to assess the relationship between HAP exposures and a panel of known respiratory pathogens. In the second chapter we leverage data on stove use during GRAPHS, and then follow a sub cohort 6 months prior to and 6 months after the GRAPHS termination date. We employ a novel construct, suspended use, to understand the factors associated with people stopping LPG use. The third chapter tests a new randomized intervention on a subset of the GRAPHS participants. We provide free cookstoves, and allocate participants to one of four arms: a behavior change intervention, an intervention where LPG fuel is directly delivered to their home, a dual intervention of behavior change and fuel delivery, or a control arm. We track their stove use to identify the most effective intervention on sustained use. Results: In Chapter 1, we find that the traditional cookstove users had a higher mean number of microbial species than the LPG (LPG: 2.71, 3-stone: 3.34, p<0.0001, n = 260). This difference was driven by increased bacterial (p<0.0001) rather than viral species presence (non-significant). Adjusted exposure-response analyses, however, produced null results. Chapter 2 identifies several factors associated with reduced or suspended LPG use of intervention cookstoves, including: experience of burns, types of food made, and access to biomass fuels. Finally, in Chapter 3 results show increased use for all three intervention arms, the largest for the direct delivery arm with an increased weekly use of 4.7 minutes per week (p