The aims are to: 1) conduct the first randomized intervention trial in air pollution history, thus increasing confidence in air pollution risk estimates, which are currently based on observational studies; 2) estimate personal particulate (PM) exposures for children across a wide range of exposures associated with an improved stove (intervention) and the traditional open fire (control) thus potentially assisting efforts to understand the shape of the exposure-response curve of PM impact in young children. After a 7-year search and 14 pilot studies, an excellent site has been characterized in highland Guatemala where PM exposures are dominated by open wood-burning cookstoves producing daily 24-hours PM 2.5 exposure levels of 1000 ug/m3 and higher, i.e., some 60 times more than current United States standards. Such conditions are common in less-developed countries (LDCs), where some two-thirds of households rely on biomass fuels (wood, dung, crop residues). Pilot work has identified a socially and economically acceptable intervention, a chimney stove that is capable of reducing mean exposures by 6- 10x. The primary health outcome to be measured, acute lower respiratory infection (ALRI), is the chief cause of morbidity in children under 5 worldwide and the chief cause of death among LDC children. It thus accounts for nearly 10% of the entire burden of global disease, making it the single largest category of ill-health. The need to examine this relationship in more detail is highlighted by some dozen case-control or cohort studies in the United States and LDCs that have found significant odds ratios for ALRI in young children living in households using wood or other biomass fuels, suggesting that reducing this exposure to pollution may be a powerful preventive intervention. 500 children, allocated randomly to control and intervention groups, will be visited weekly from birth to 18 months to detect a 25% difference in ALRI incidence (power=0.8; alpha=5%). Diagnosis will be done using international protocols and physician verification. Child personal exposures and household microenvironments will be monitored twice each season (four times per year), and more intensely in a subsample. Because pilot studies consistently show passive CO diffusion tubes to be reliable indicators of PM arising from use of wood fuel, they will be used as the primary exposure monitors for the children. Total individual PM exposure will be modeled for each child using the information from the personal CO monitors, from microenvironmental PM and CO measurements in a subsample of households, and activity pattern information for each child. Previous pilot studies at the site have shown that exposures in the intervention and control groups will extend from PM levels similar to those found outdoors in United States cities, where most previous epidemiology has focused, to levels more than an order of magnitude higher. Active smoking risks provide evidence that the curve must become less steep at exposures much higher than ambient air pollution, but it is not known how the curve is shaped in the wide gap between, within which lie the exposures experienced by this study population. An important secondary objective therefore is to describe the relationship between exposure and ALRI incidence across this range. Neither the primary nor secondary aim is likely to be achieved in the United States or other developed country today, because the needed conditions have ceased to exist, i.e., exposures are not dominated by household sources suited to randomized intervention and also lie in a relatively limited range because they are heavily influenced by widespread outdoor sources. Thus, in addition to being directed toward a serious health problem in a large vulnerable population worldwide, this research can assist the worldwide inquiry into PM health effects by moving air pollution epidemiology closer to the strongest stage of the (Bradford) Hill criteria for establishing causality, the """"""""gold standard"""""""" of randomized intervention.
Showing the most recent 10 out of 23 publications
