Using GIS to Estimate Personal Exposure in an Epidemiologic Study
A key issue in exposure assessment is how well an exposure metric estimates exposure to the individual. Exposure has been defined as “the contact of a chemical, physical, or biological agent with the outer boundary of an organism” (Berglund et al. 2002). Exposure is a function of concentration and time: “An event that occurs when there is contact at a boundary between a human and the environment with a contaminant of a specific concentration for an interval of time” (NRC 1991). Thus, in the context of exposure assessment for an epidemiologic study, it is important to distinguish between environmental concentration, exposure concentration, and dose. The environmental concentration of an agent refers to its presence in a particular carrier medium [for example, polycyclic aromatic hydrocarbons (PAH) in ambient air], expressed in quantitative terms (for example, micrograms per cubic meter). Similarly, the exposure concentration of an agent refers to its presence in its carrier medium at the point of contact (for example, PAH in breathing zone air) expressed in quantitative terms (for example, micrograms per cubic meter). Finally, the dose refers to the amount of a pollutant that actually enters the human body, i.e., is taken up through absorption barriers. A number of variables can influence the exposure and dose. These include physiologic factors such as age, sex, physical condition, disease, and genetics, as well as exposure factors related to human behavior and activities (e.g., the amount of time spent commuting to work each day), and contact rates (e.g., the amount of drinking water ingested per day). In epidemiologic studies, environmental concentration will often be used as a surrogate for both exposure concentration and dose.
We could not find an example of the use of GIS to estimate personal exposure for an epidemiologic study. In our review of the literature, questionnaire data were generally used as a surrogate for deriving personal exposure. Only recently have researchers started using GIS to study activity patterns in a study population, which conceivably could be linked to environmental data for exposure assessment. Phillips et al. (2001) reported on a test of GPS data recorders as a means of validating time-location data recorded in study diaries of a subset of participants enrolled in the Oklahoma Urban Air Toxics Study. Elgethun et al. (2003) describe the development and testing of a data-logging GPS unit designed to be integrating into clothing. Both studies concluded that GPS units could be useful in developing time–location information for use in exposure assessment. GPS is a satellite-based technology composed of a system of satellites encircling earth and emitting a radio frequency detectable by GPS receivers. GPS receivers are designed to use this information and calculate coordinates of the receiver location. Precision of these coordinates can vary based on receiver design and signal quality. Phillips et al. (2001) reported precision of about 10 m for most readings, whereas Elgethun et al. (2003) reported mean root mean square error of 3.2 m outdoors and 5.8 m indoor in positional accuracy for two GPS units tested. This level of precision should be sufficient for most studies attempting to link location of a participant with a particular environmental setting where contaminant monitoring or modeling data are available for linkage using a GIS. A major advantage of the technology, as reported by Phillips et al. (2001), was not only that its use confirmed all reported trips over a 12- to 23-hr monitoring period, but it provided time–location data on travel events not recorded in the participant diary.
Both Phillips et al. (2001) and Elgethun et al. (2003) reported limitations of the technology as a sole source of space–time data for an exposure assessment study. Both studies found the reception of the satellite signals to be adversely impacted by shielding from buildings of certain materials (concrete, steel), electrical power stations, and to some extent vehicle body panels. Signal blockage continues to be an issue with GPS today. Phillips et al. (2001) also reported extensive failure including battery failure, data-logging failure, and data storage limitations, which resulted in capturing only about 30% of the total monitoring time attempted in 25 trials. Elgethun et al. (2003) reported reception efficiencies of 79% outdoors, 20% in homes, 12% in vehicles, and 6–9% in schools and businesses. These findings indicate that although GIS using GPS technology hold promise in terms of integrating study population activity data with measured or predicted levels of environmental contaminants in the exposure assessment process, their use is still very much in the developmental research stage for use in epidemiology studies.