Mapping the Environment and Breast Cancer on Cape Cod, MA
The Ribbon

Theresa C. Kennedy and Julia G. Brody, Silent Spring Institute

Ever Since John Snow's famous work, thematic maps have emerged as valuable tools in understanding the relationship between our environment and the occurrence of disease.

Among breast cancer activists and researchers, geographic patterns have prompted concern in high incidence communities and spawned hypotheses about environmental factors. One of the most widely known environmental epidemiology studies,the Long Island breast cancer study,came about in part because women, concerned at the number of their neighbors with breast cancer, began pinpointing their neighbors 'homes on maps laid out on their kitchen tables.

Today, Geographic Information Systems, or GIS, replace paper maps and transparent overlays with a sophisticated mix of computer hardware, software and expertise, all focused on the visualization and analysis of spatial data. Beyond simple mapping, GIS allows the researcher to integrate layer upon layer of different kinds of spatial information, and describe or quantify the spatial and temporal relationships between them. Gaining insight into these relationships can help form causal hypotheses, which in turn moves us closer to the goal of disease prevention.

The power and versatility of GIS is apparent in the many applications seen thus far in health studies. GIS is used in disease surveillance to monitor incidence by geographic units, such as zip code or county, and in environmental surveillance, for example, to track toxic releases or map air and water quality. By putting disease and environmental data together,GIS is also used in ecological analysis, to analyze 'clusters' of disease with respect to the sociological or environmental aspects of the affected population and place. Lastly, GIS is beginning to be used to assess environmental exposures to individuals in health studies.

Researchers approach this last use of GIS - individual exposure assessment for health studies - with caution because limitations in the quality of spatial data can be formidable. Differences in scale, resolution, accuracy, and completeness of the datasets are a constant hindrance, particularly for studies of diseases involving long latencies, like breast cancer, where exposures of interest may have occurred decades in the past. Collecting new data can be time intensive,expensive, and,for historical exposures, sometimes impossible.

Despite these challenges, the Cape Cod Breast Cancer and environment study (Cape Cod Study), set out to confront the limitations and develop new techniques to reconstruct historical exposures to various contaminants on Cape Cod, MA, in an effort to see if certain exposures are related to breast cancer risk in the region. The study is conducted by Silent Spring Institute, a non-profit scientific research organization, in collaboration with researchers at Applied Geographics Inc.,Boston University, Harvard, and Tufts.

The use of GIS in the Cape Cod Study was motivated by the gaps in other exposure assessment methods that limit our ability to identify environmental factors that may one day lead to prevention. Most of what we know about breast cancer risk relates to factors women can report in interviews: age at menarche or a first pregnancy, use of oral contraceptives and hormone replacement therapy, history of exercise and alcohol use, and so on. Biological and environmental sampling hold promise for the future, but tell us little about the past, and sampling is expensive at the scale needed in breast cancer epidemiology to uncover risks that are probably moderate in scale. Geographic data can add another dimension -answering questions about their environment that women cannot answer for themselves, and allowing estimation of environmental influences on a large scale. While the data developed in the study are specifically relevant to Cape Cod, the methods are widely applicable for environmental health studies and planning.

In the Cape Cod Breast Cancer and Environment Study, Silent Spring Institute unites GIS, health, and environmental data to assess the exposures of 2,100 individual women to multiple environmental pollutants throughout the past 40 years. (Center quote for page 1)

Breast Cancer on Cape Cod, MA

The study was instigated in 1993 when the Massachusetts Cancer Registry (MCR) reported elevated incidence in a majority of Cape Cod towns compared with the rest of the state (1). Using the GIS to better estimate how many breast cancer cases would be expected in each census block group on the Cape over the 14-year history of the MCR, detailed disease surveillance revealed approximately 20%higher age-adjusted breast cancer incidence on Cape Cod for the period 1982 through 1994. (See Figure 1.) Using data from the Collaborative Breast Cancer Study, we learned that incidence was elevated in comparison with other areas of Massachusetts, even after statistically controlling for a long list of established and hypothesized risk factors for breast cancer, including family and reproductive history, physical exercise, alcohol, tobacco, and certain aspects of diet (2).

Cape Cod's history of pesticide use in support of tourism, cranberry cultivation and other agriculture is also quite distinctive(3). Forests were repeatedly sprayed for gypsy moth and other tree pests, and wetlands were sprayed for mosquito control. Other wide area uses include applications to manage golf courses and rights of way. The Cape 's sandy soils allow pollutants to travel quickly to groundwater, which is the sole drinking water source and Cape Cod residences have been developed in or adjacent to pesticide use areas or on land where pesticides were previously applied. Persistent organochlorine chemicals including DDT and dieldrin were widely used on the Cape from the late 1940s to the mid 1970s, and less persistent compounds including carbaryl, malathion, and carbamates have been applied in more recent years (3).

In assessing which environmental exposures may be important to measure on Cape Cod, the study team started with what they knew about the already well-established risk factors. The most promising candidates were substances that mimic estrogen -a known breast cancer risk factor -or that have been shown to cause mammary tumors in animals. These substances include pesticides (insecticides,herbicides and fungicides)and other chemicals found in detergents, plastics and personal products which have been widely used on Cape Cod in the past, and may lead to exposure from their use and disposal in conjunction with certain characteristics of the Cape Cod environment.

The Cape Cod Breast Cancer and Environment Study

The Cape Cod study involves 2100 women who lived on Cape Cod between 1988 and 1995. Women diagnosed with breast cancer during those years were compared with women of similar age who had not been diagnosed. Interviewers collected information on where the women had lived on Cape Cod and on established risk factors for breast cancer, including family history of breast cancer, menstrual and reproductive history, height and weight, and education, an indicator of socioeconomic status. Interviews also included topics of interest in recent studies as possible breast cancer risk factors:alcohol and tobacco use, physical activity, and pharmaceutical hormone use. Other questions assessed use of home pesticides, tap water, and certain consumer products.

To measure historical exposures from pesticides and other chemicals that could not be ascertained by interview we designed and developed special GIS tools to model the historical Cape environment. One of these tools, the Spatial Proximity Tool, is described in more detail later.

Thematic data layers which served as input to the models (Figure 2)were gathered from federal and state sources, such as the United States Geological Survey (USGS), state agencies, and the Cape Cod Commission and were integrated into the GIS.

Land use maps from four periods beginning in 1951, supplemented with local information and other federal Figure 1 - Breast Cancer Incidence 1982 - 1994, compared to state average datasets were used to assess the historical locations of forested areas, cranberry bogs, and other land use types known to have been regularly sprayed with pesticides. Information on local spraying activities was exhaustively researched, compiled, and mapped by Institute staff.Data on public water supply systems and private wells was gathered, and most importantly, the information collected at interview identifying our women's Cape Cod addresses was used to 'geocode 'each woman 's residential history during the study period, 1948 -1990, when the target chemicals were used.

A new GIS Tool to Measure Environmental Exposures: The Spatial Proximity Tool

The Spatial Proximity Tool was developed to relate the women's residential address history with the historical environmental data like pesticide spraying areas (Figure3). The tool is based on modeling of spray drift and deposition,and takes into account the distance of a residence from the pesticide use area, the size of the area, and the wind direction typical during the early morning hours in spring when pesticides were applied. The goal was to reconstruct spatial, temporal, and intensity or "dose "information. Given the expected limitations of historical records, the focus was on assessing relative intensity -i.e., on correctly ranking higher and lower exposures and differentiating exposed from unexposed residences.

While these measures do not approach the ideal of historical biological exposure assessment, which, indeed, cannot be attained retrospectively, they offer insight into historical exposure patterns.

Mapping Residential History over a 40 year period

One of the significant challenges in the study was the 'geocoding'of residential history data gathered at interview. Geocoding is the process of taking a street address, such as '1 Main St.Hyannis', and translating it into a latitude and longitude so that it can be placed on a map. Early automated methods of geocoding women in our study involved using town parcel maps to place them in the center of the land parcel corresponding to their street address. The parcel maps however, like many publicly available datasets, had been created independently by each town on the Cape, using different source data, and at varying resolution and scale. The net result is that while the parcel maps are useful basemaps for each individual town, they do not represent a uniform basemap across the Cape, and are subject to local variation in accuracy and completeness. The environmental datasets, originating from uniform state-or nationwide basemap data, did not therefore 'fit 'with the parcel maps in some areas, introducing the possibility of underestimating (or overestimating) a woman's exposure to a particular source. These problems of exposure misclassification are common in environmental epidemiology and remain serious barriers to accurately identifying health impacts of pollutants.

In order to overcome the limitations of the parcel maps, we used high resolution aerial photographs to move the women directly onto their house rooftops, using a process of on-screen editing and digitization. The aerialphotographs, available from MassGIS (4), constitute the state basemap for Massachusetts, and represent a consistent cape-wide reference on which to map other information. This process, undertaken by Institute researchers and Applied Geographics Inc. (AGI), maximizes the quality of the residential history data, and has the additional benefit of reducing error associated with the significant number of large parcels found on the Cape, in which the center of a parcel may be some distance away from the actual residence.

Building an Historical Exposure Profile for each woman

The Spatial Proximity Tool calculates relative exposure intensities for each source (e.g.an actively used Cranberry Bog) at each residential address during a particular time period in the study. A woman's total exposure score for each source, or group of like-sources (e.g.all aerially sprayed pesticides) was calculated by adding together all appropriate exposures for each year at each address over a woman's residential history (Figure 4). Cumulative exposure to residual pesticides, stemming from the previous application of persistent chemicals to residential areas or adjacent land, was also calculated. Results of the analysis comparing aggregated exposure scores with breast cancer risk are in preparation.

Conclusions and Recommendations

Given that limitations due to disparate datasets and missing exposure information (such as we found in this study) are typical in retrospective environmental health studies, GIS may be seen as a tool for creatively, but judiciously,constructing proxies and developing methods for estimation and interpolation (3). The promise of GIS for this purpose is held back by the lack of statewide data,both present day and historical, and also by gaps in our understanding of the dynamics of chemicals in our environment. Standardized data collection, consistency across geographical boundaries, and long-term monitoring, are critical to the success of initiatives such as Health Track (5), in which the integration of many kinds of health and environmental data offer hope for studying the long term effects of the environment on health. Despite the challenges, the data and tools developed during the Cape Cod study represent a rich resource for studying a wide range of health and environmental questions. The Spatial Proximity Tool in particular, while demonstrated here for historical analysis, can also be used to model future hypothetical events. With applications limited only by imagination and available data, GIS represents a promising tool for environmental stewardship, and protecting our health.

More information about Silent Spring Institute and the Cape Cod Breast Cancer and Environment Study, including an environmental data atlas (Figure 5), can be found at http://www.silentspring.org

1. Brody, J.G.; Rudel, R.A.; Maxwell, N.I.; Swedis, S.R.1996. Mapping out a search for environmental causes of breast cancer.Public Health Reports, 6(3):494-507.

2. Silent Spring Institute,1997. Cape Cod Breast Cancer and Environment Study: Final report, December 8, 1997, Newton, MA.

3. Brody, J.G.; Vorhees, D.J.; Melly, S.J.; Swedis, S.R.; Drivas, P.J.; Rudel, R.A. 2002. Using GIS and historical records to reconstruct residential exposure to large-scale pesticide application. Journal of Exposure Analysis and Environmental Epidemiology, 12:64-80.

4. MassGIS. Massachusetts Geographic Information System. Boston, MA: MassGIS, Executive Office of Environmental Affairs. accessed on January 31, 2003 [accessed at: http://www.state.ma.us/mgis/massgis.htm ].

5. Trust for America's Health. Trust for America's Health Reports. Washington, DC. Available at: http://healthyamericans.org/resources/reports/; accessed February 2003.

Back to the top

Mapping the Environment and Breast Cancer on Cape Cod, MA The Ribbon

Mapping the Environment and Breast Cancer on Cape Cod, MA
The Ribbon