Assessing air pollution exposure and health effects on children's respiratory health in four Chinese cities

The objective of this dissertation is to construct appropriate variables to represent exposure to ambient air pollutant mixture, exposure to heating coal smoke, exposure to cooking coal smoke, and other potential risk factors. These variables were used in unconditional logistic regression models to study exposure - health effect relationships in children residing in eight districts of four Chinese cities.; The eight districts were classified into four district clusters based on the integrated ambient levels of total suspended particles (TSP), size fractionated particulate matter (PM_2.5, PM_2.5–10, and PM₁₀), sulfur dioxide (SO₂), and oxides of nitrogen (NO_x). These district clusters were used as ‘ecological’ variables in logistic regression models. The results indicate that the integrated ambient pollution levels were associated with wheeze and with cough and phlegm together. PM_2.5 was associated with cough, with phlegm, with bronchitis, and with asthma. PM_2.5–10 and TSP were both associated with wheeze and with cough and phlegm together.; Based on personal and household information collected through a questionnaire survey, the use of a factor analysis technique generated five potential risk factors including heating coal smoke, cooking coal smoke, socio-economic status, ventilation, and children's passive smoking. The use of these uncorrelated factors, derived from a large set of household/lifestyle variables, in logistic regression models minimized potential problems of multicollinearity. The results indicate that four of the five factors (except cooking coal smoke ) were significantly associated with most of the concerned health outcomes. In addition, this dissertation developed semi-quantitative variables for heating coal smoke exposure and cooking coal smoke exposure by using a Scenario Evaluation Approach incorporating parameters of residential ventilation conditions and exposure duration estimates. Logistic regression analyses of these semi-quantitative variables show that heating coal smoke exposure was significantly associated with most of the health outcomes. In contrast, no consistent associations were observed between cooking coal smoke exposure and any of the health outcomes.