Use of GIS and exposure modeling as tools in a study of cancer incidence in a population exposed to airborne dioxin

Environmental Health Perspectives, June 15, 2004 by Poulstrup. A., H.L. Hansen

In environmental health research there is a recognized need to develop improved epidemiologic and statistical methods for rapid assessment of relationships between environment and health. Exposure assessment is identified as a major challenge needing attention. In this study an exposure simulation model was used to delimit almost exactly in space and time an urban population exposed to airborne dioxin. A geographic information system (GIS) was used as the electronic environment in which to link the exposure model with the demographic, migration, and cancer data of the exposed population. This information is available in Denmark on an individual basis. Standardized incidence ratios (SIRs) for both men and women in 10-year age bands were calculated for three different exposure areas. Migration patterns were outlined. SIRs showed no excess of cancer incidences during the time span chosen (13 years; 1986-1998) in the whole exposed area or in the medium or higher polluted areas. The exposure model appeared very useful in selection of the appropriate exposure areas. The integration of the model in a GIS together with individual data on addresses, sex, age, migration, and information from routine health statistics (Danish Cancer Registry) proved its usefulness in demarking the exposed population and identifying the cancers related to that population. Less than one-third of the study population lived at the same address after 13 years of observation, and only half were still residing in the area, indicating migration of people as a major misclassification. Key words: air pollution, cancer, dioxin, environmental epidemiology, exposure model, GIS, health registers, migration. Environ Health Perspect 112:1032-1036 (2004). doi:10.1289/ehp.6739 available via http://dx.doi.org/[Online 15 April 2004]

**********

A major challenge for investigators in environmental epidemiology is to correctly identify populations at risk from exposure to environmental contaminants. To date, three methods have been used to identify the populations at risk from point sources of air pollution: physical monitoring, environmental monitoring, or mathematical modeling (Williams and Ogston 2002). This article is a discussion of the utilization of a computerized air pollution model, normally used by the environmental protection authorities for assessing pollution values (immissions), and the putative offense of legally set thresholds of emission. To test the model for its appropriateness as an improved tool for assessment of exposure, an actual case was used of known dioxin air pollution in an urban area.

In the town of Kolding in the southern part of the Jutland peninsula, Denmark, three outlets of dioxin were identified. All three emitted dioxin into the air through their chimneys. The dioxin consisted mainly of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD, or dioxin). One outlet in particular, an aluminum recycling plant, was found on two occasions--early November and early December 2000--to have emitted large quantities of dioxins, up to 180 ng/[m.sup.3]/hr. All three plants had been operating for years. The main culprit was the aluminum recycling plant, which had been in continuous operation since 1970 with an almost unchanged method of production and output.

Our plan was to layer the computer-simulated exposure model in a geographic information system (GIS) and use the simulated immission concentrations to more accurately demarcate the exposed population. The information on addresses, vital statistics, migration, and cancer of the population of Denmark or any subset was available on individuals and on the delineated population in this study. This information was layered into the same GIS environment, enabling a more exact identification of the exposed population in both space and time.

All malignant cancers were used as the health indicators of the exposed population to assess eventual negative health outcomes caused by the dioxin pollution. TCDD is a major environmental carcinogen causing various types of cancers (IARC 1997).

Materials and Methods

The air pollution simulation model used in Denmark to assess hourly immissions of airborne pollutants is a Gaussian air dispersion model based on emission data of the actual pollutant(s) and time series of meteorological data such as wind speed, wind direction, wind temperature, rain, snow, number of stacks, their heights, surrounding buildings, and surrounding terrain. The model [OML, Operationel Meteorologiske Luftkvalitets-modeller (Danish)] has been widely validated both in Europe and North America and is reliable in predicting hourly immissions of one or more airborne pollutants (DMU 2001). Only two measurements of dioxin were available as input source, and both were obtained in November and December 2000. The GIS used was the software package ArcGis version 8.12 (ESRI, Atlanta, GA, USA).

All current and past addresses in Denmark since 1999 were geocoded with Universal Transverse Mercator coordinates with a precision of a few meters and subsequently layered into the GIS. The Address Project is described elsewhere (Briggs et al. 2002). By linkage of all individuals to these addresses using the unique Central Population Register (CPR) number (10-digit number in the Civil Registration System), the GIS eventually contained all the following information in addition to the addresses of each individual: date of birth, sex, migration (into, out of, and around the study area), and date of death (Figure 1). Each green spot represents an address and a table of the mentioned attributes.