Skip to main content
         

Vol. 08 Issue 1, Early Spring 2003

Mapping Disease: Deciphering Geographic Patterns from Cholera to Breast Cancer
The Ribbon 

Zev Ross
Zev Ross is a graduate student in the Department of Natural Resources at Cornell University. He spent last summer assisting California's Environmental Health Investigation Branch with a spatial analysis of breast and testicular cancer data.

The more than 200-year history of disease mapping is filled with examples of maps that helped provide etiological clues to diseases from cholera to lung cancer. Geographic patterns in disease, such as those discussed in this issue of The Ribbon, can help provide insight into disease incidence and mortality or help identify environ-mental sources of risk. While geographic patterns can be an important research tool, these patterns can also pose statistical challenges. Patterned data violates the fundamental statistical assumption of independence and ignoring this violation can lead to distorted statistical results. In 2002, as part of a larger project to look at geographic variations in breast cancer rates, we initiated a project to look at geographic patterns in breast cancer and, more specifically, to evaluate their affect on statisti-cal models. We expect to have results from our analysis in May 2003.

Maps to Solve Medical Mysteries

It was the terrifying epidemic diseases of the 18th and 19th centuries that first prompted researchers to develop maps of disease. In contrast to often dreary statistical tables, maps can (and did)sharpen otherwise obscured relationships between disease and possible environmental influences. With maps, researchers could look at the location of cholera or yellow fever deaths in relation to water wells, garbage dumps,or outhouses.

Dr.John Snow 's map of cholera deaths in relation to London 's water pumps, for example, was one of the first,and perhaps the most celebrated, disease maps. With the help of his famous map, Snow was not only able to track the source of what he called "the most terrible outbreak of cholera which ever occurred in this kingdom,"but he was able to convince authorities to take action against the disease (2). His map (next page) demonstrated for future epidemiologists the value of maps as both a research and a communication tool.

But clearly disease mapping is not limited to epidemic diseases in the 1800s. Modern researchers have also successfully used geographic patterns to identify etiological clues. Examples include:

  • Mapping of cancer mortality rates in the United States 1950-69 revealed "exceptionally "high rates of lung cancer along the eastern seaboard,particularly in parts of Georgia, Florida and Louisiana (3,4). These patterns led to further investigation. Several studies found that exposure to asbestos among employees at shipyards accounted for a significant part of the excess mortality from lung cancer (3,5,6).
  • Researchers in the 1950s used the geographic patterns in malignant melanoma to identify causes.In his influential 1956 paper,H.O.Lancaster noted that rates of malignant melanoma were higher in the people of Southern Africa and Australia than in the countries from which they originated. He also noted that incidence increased with proximity to the equator in the United States, Australia and New Zealand and correctly proposed that sunlight may be a cause of the increased incidence.Due to the fact that tumors tend to be more common on non sun-exposed sites (the trunk in men and legs on women) sun had not been considered a risk factor prior to analyzing the geographic patterns (7,8).

    Signif icant Innovations Since the Sixties

    In the time since Dr.Snow 's cholera map,and more specifically in the past forty years,we 've seen enormous innovations in mapping and techniques to analyze spatial patterns. Geographic Information Systems (GIS), for example,the topic of this issue of The Ribbon were developed in the 1960s and have become widespread only in the past 10 years. Similarly,while traditional statistical techniques (such as regression)are hundreds of years old,statistics specific to spatial data evolved out of research in the 1960s. These innovations come alongside an exponential increase in the volume and quality of data on disease and environmental pollutants.

    These innovations are particularly significant in light of the challenges presented by modern diseases. In contrast to diseases such as cholera or yellow fever, whose sources can be traced to a single bacterium or virus,diseases such as breast cancer appear to have more complex causes.

    Breast Cancer's Enigmatic Geographic Patterns and Statistical Considerations

    Breast cancer has posed one of the greatest challenges to researchers investigating geographic patterns. Age-adjusted breast cancer incidence and mortality exhibits a strong geographic pattern. For decades,researchers in the United States have noted substantial regional variation in breast cancer mortality rates,most notably a "regional excess "of breast cancer in the Northeast United States (9-13). Nationally,even after controlling for variables that,themselves,have a geographic pattern (e.g.,race, socioeconomics),breast cancer mortality still exhibits a (slight)regional pattern. The source of the regional variation is hotly debated and has led to significant research on the issue.

    In conjunction with Dr.Peggy Reynolds and collaborators at the California Department of Health Services and Dr.Patrick Sullivan, a biostatistician and Professor at Cornell University, I am looking at the statistical implications of geographic patterns in breast cancer incidence in California. Specifically,we are in the process of evaluating the affect of autocorrelation -the tendency for near things to be more similar than distant things -on parameters, predictions and confidence intervals.

    Our analysis of geographic patterns relies on geostatistics, a statistical method that describes how data (in this case rates)are related with distance and direction. Unlike the more traditional nearest-neighbor analyses that rely on an arbitrarily defined neighborhood (e.g.,adjacent neighbors), geostatistics can analyze the strength of relationships at the full range of distances.

    Our preliminary results suggest that even after accounting for race, age,socioeconomics and urbanization, breast cancer incidence rates in California still exhibit autocorrelation. The known risk factors, however, account for a significant amount of the autocorrelation (and the variation) seen in unadjusted rates and the "left-over "autocorrelation appears to be minimal. The statistical and etiologic implications of the autocorrelation are still under investigation.

    In the tradition of Dr.John Snow, the process of analyzing the geographic patterns will hopefully lead us to insights on breast cancer 's origins and influences.This issue of The Ribbon describes several ongoing studies pursuing these questions in California, Massachusetts and New York State.

    "The cartography of disease owes its genesis to the abrupt, terrifying challenge which epidemic outbreaks presented, whereas endemic disease, more or less constantly active, offered no comparable stimulus to cartographic creativity. Plague, yellow fever, and cholera - all exotic - accomplished what tuberculosis could not." Jarcho (1969)(1). (Center quote for page 2)

    "Whether they are stalking bioterrorists or tracing the origin of a hepatitis outbreak, today's epidemiologists stand on the shoulders of John Snow (1813- 1858), the English anesthesiologist whose methodical investigations showed that cholera spreads through polluted water." www.ph.ucla.edu/epi/snow.html (Center insert on page 1)

    References

    1.Jarcho S. The contribution of Heinrich and Hermann Berghaus to medical cartography. Journal of the History of Medicine and Allied Sciences 25:131-142(1969).

    2.Snow J. On the Mode of Communication of Cholera. London:John Churchill,1855.

    3.Blot WJ, Harrington JM, Toledo A, Hoover R, Heath CW, Jr., Fraumeni JF, Jr. Lung cancer after employment in shipyards during World War II. N Engl J Med 299:620-4(1978).

    4.Devesa SS,Grauman DJ, Blot WJ, Pennello GA, Hoover RN, Fraumeni JF. Atlas of Cancer Mortality in the United States: 1950-1994 NIH Publication No.99-4564. Washington: National Cancer Institute,1999.

    5.Blot WJ, Davies JE, Brown LM, Nordwall CW, Buiatti E, Ng A, Fraumeni JF, Jr. Occupation and the high risk of lung cancer in Northeast Florida. Cancer 50:364-71(1982).

    6.Blot WJ, Morris LE, Stroube R, Tagnon I, Fraumeni JF, Jr. Lung and laryngeal cancers in relation to shipyard employment in coastal Virginia. J Natl Cancer Inst 65:571-5(1980).

    7.Hutt M, Burkitt D. The Geography of Non-Infectious Disease. Oxford:Oxford University Press, 1986.

    8.Lancaser H. Some geographical aspects of the mortality from melanoma in Europeans. Medical Journal of Australia 1:1082-7(1956).

    9.Blot WJ, Fraumeni JF, Jr., Stone BJ. Geographic patterns of breast cancer in the United States. J Natl Cancer Inst 59:1407- 11(1977).

    10.Blot WJ, McLaughlin JK. Geographic patterns of breast cancer among American women. J Natl Cancer Inst 87:1819-20(1995).

    11.Sturgeon SR, Schairer C, Gail M, McAdams M, Brinton LA, Hoover RN. Geographic variation in mortality from breast cancer among white women in the United States. J Natl Cancer Inst 87:1846-53(1995).

    12.Canto MT, Anderson WF, Brawley O. Geographic variation in breast cancer mortality for white and black women:1986-1995. CA Cancer J Clin 51:367-70(2001).

    13.Devesa SS, Grauman DJ, Blot WJ, Pennello GA, Hoover RN, Fraumeni JF. Atlas of Cancer Mortality in the United States: 1950-94: National Cancer Institute,1999.

    Last Update 04.04.03