Biases Arising from the Use of Ambient Measurements to Represent Personal Exposure in Evaluating Inflammatory Responses to Fine Particulate Matter: Evidence from a Panel Study in Beijing, China

Abstract

Ambient particulate matter (PM) is often used as a proxy of personal exposure in epidemiological studies of PM-induced health effects, yet whether this proxy biases the estimates of health effects is still unknown. On the basis of a panel study in Beijing, China, we investigated the dependence of 24 h personal exposure concentration to fine particles (PM2.5) and its carbonaceous components on the corresponding 24 h ambient concentration. The associated changes in inflammatory biomarkers with personal and ambient exposure were further examined using linear mixed-effect models. At ambient PM2.5 levels of <25 μg m–3, personal exposure to PM2.5 was often several times higher, with a median personal/ambient ratio of ∼3. The ratio declined with an increase in ambient concentration, approaching ∼1 at ambient PM2.5 levels of >75 μg m–3. Similar trends were also observed for organic carbon and elemental carbon. Personal exposures were significantly associated with both respiratory and systemic inflammatory biomarkers, such as fractional exhaled nitric oxide and white blood cell count. When ambient data were used, the association with systemic inflammation weakened. Our findings imply that the use of the ambient pollutant concentration as a proxy for personal exposure may be inaccurate and could bias the estimates of PM-induced health effects.