Abstract

Air pollution epidemiologic studies are intrinsically difficult because the expected effect size at general environmental levels is small, exposure and misclassification of exposure are common, and exposure is not selective to a specific pollutant. In this review paper, epidemiologic study designs and analytic methods are described, and two nationwide projects on air pollution epidemiology are introduced. This paper also demonstrates that possible confounding issues in time-series analysis can be resolved and the impact on the use of data from ambient monitoring stations may not be critical. In this paper we provide a basic understanding of the types of air pollution epidemiologic study designs that be subdivided by the mode of air pollution effects on human health (acute or chronic). With the improvements in the area of air pollution epidemiologic studies, we should emphasize that elaborate models and statistical techniques cannot compensate for inadequate study design or poor data collection.

• Keywords: Air pollution epidemiologyEpidemiologic methodsStudy designsGeneralized additive modelCase-crossover study