USING GIS FOR EXPOSURE ASSESSMENT

Abstract

Exposure misclassification has long been characterized as the Achilles heel of environmental epidemiological studies. A plausible explanation for inadequate exposure assessment in environmental epidemiology is the lack of cross-disciplinary training and communication between health, earth, and engineering scientists. As health scientists, epidemiologists have a keen understanding of the biological factors that need to be studied to determine if a population is at risk to disease because of exposure to particular contaminants in their environment. However, they generally are not trained in the sciences necessary to determine the level at which that contaminant was present in the environment of their study population. Conversely, earth scientists and environmental engineers are trained in methods for quantifying source, fate and transport of contaminants in an environment, but have little knowledge of how to apply this information in a human health study. By this presentation, I hope to demonstrate how geographic information system (GIS) and related computer-based technologies can serve as a means for improving exposure assessment for environmental epidemiological studies, and in that context, serve as a learning and communications bridge between health, earth, and engineering sciences. The term ‘environment’ implies a spatial context. Thus, the study of interactions between humans and their environment requires spatial information. We have identified at least 15 elements in the exposure assessment process for an environmental epidemiological study. In this presentation I will review the key elements of this process. Particular emphasis will be given to those steps in which GIS has the potential or has been demonstrated to enhance the state of the science of exposure assessment. For example, one step in exposure assessment is source identification for the chemical compound of interest. A typical method for conducting exposure assessment in epidemiology is to interview study participants in order to obtain information that can relate or infer exposure to such sources. However, in many studies, sources of these compounds cannot be ascertained through questionnaires. Regions exist in the USA where the general population resides in intensive agriculture landscapes, but are not engaged in agricultural production. It is highly unlikely that such individuals would know what chemicals are used on crops near their homes. I will present and discuss a GIS-based method for identifying crops and linking crop-type to pesticide use for use in epidemiological studies concerning non-occupational exposure to agricultural pesticides that holds promise for exposure classification. The use of GIS in exposure assessment requires attention to fundamental elements of geospatial science, including resolution and scale. I will present a comparative analysis of two GIS-based exposure metrics using different resolution and operational scale data. The two methods resulted in significantly different exposure classification and resultant risk estimates when compared in the context of an epidemiology study concerning non-occupational exposure to agricultural pesticides.