Results of a similar analysis using New York State data
Francis Boscoe, New York State Department of Health
11 April 2005
Research staff the New York State Cancer Registry recently completed a similar analysis as the one presented by Klassen, Kulldorff and Curriero. These results were originally presented in a poster session at the 2003 CDC Cancer Conference [1] and we briefly summarize them here.
The goal of the study was to test whether socioeconomic factors might explain patterns of elevated incidence of prostate cancer in New York State as identified by the spatial scan statistic. Records were obtained for 60,556 cases of prostate cancer diagnosed among New York State men between 1995 and 1999. Adjusting for age and race alone, 24 areas or clusters of elevated incidence were identified, widely distributed throughout the state. These included some “nested” areas with varying levels of relative risk: for instance, an area with a relative risk above 2 might be part of a larger area with a relative risk above 1.5 [2]. After additional adjustment for socioeconomic variables using 1990 census tract-level data (high income, education, crowding; other variables were considered and found to be either redundant or insignificant), the relative risks in 16 of the areas declined substantially, remained about the same in 6 of the areas, and increased in 2 areas (two nested areas within predominantly black neighborhoods in Brooklyn). Overall, 16 of the 24 areas remained statistically significantly elevated.
These findings provide evidence for a moderate association between prostate cancer incidence and socioeconomic variables that is consistent with other published studies [3]. This association has been attributed in part to utilization of PSA testing, which is generally higher in areas with higher levels of income and education. Since PSA testing detects some early-stage prostate cancers that might otherwise go undetected during a man’s entire life, the incidence rates in these locations are likely to be higher. The Brooklyn neighborhoods cited earlier are a major exception, in that they have high incidence rates despite lower income and education levels. Based on this information and related evidence, these neighborhoods have been targeted for awareness and outreach activities by the New York State Department of Health. Much of the geographic variation in prostate cancer remains unexplained, as evidenced by the 16 areas that remained statistically significantly elevated after adjustment for socioeconomic variables. These could be areas with unusually active screening programs, areas with better data reporting by physicians or outpatient clinics that traditionally have not diagnosed cancer cases, or the result of local environmental factors.
While greater clarity as to the precise contribution of each socioeconomic risk factor could have been obtained using traditional epidemiological methods, our approach had the advantage of highlighting specific geographic locations, which is a useful framework for population-based surveillance programs that must frequently respond to questions about rates and exposures in specific locales. Further details of our study may be obtained by contacting fpb01@health.state.ny.us.
[1] Schmit KJ, McLaughlin CC, Johnson G, Schymura MJ, Negoita S. Incorporating risk factor data into geographic analyses of prostate cancer. Poster exhibited at the CDC Cancer Conference, Atlanta, September 15-18, 2003.
[2] Boscoe FP, McLaughlin CC, Schymura MJ, Kielb CL. Visualization of the spatial scan statistic using nested circles. Health and Place 2003; 9: 273-277.
[3] Liu L, Cozen W, Bernstein L, Ross RK, Deapen D. Changing relationship between socioeconomic status and prostate cancer incidence. Journal of the National Cancer Institute 2001; 93: 705-709.
International Journal of Health Geographics
Results of a similar analysis using New York State data
11 April 2005
Research staff the New York State Cancer Registry recently completed a similar analysis as the one presented by Klassen, Kulldorff and Curriero. These results were originally presented in a poster session at the 2003 CDC Cancer Conference [1] and we briefly summarize them here.
The goal of the study was to test whether socioeconomic factors might explain patterns of elevated incidence of prostate cancer in New York State as identified by the spatial scan statistic. Records were obtained for 60,556 cases of prostate cancer diagnosed among New York State men between 1995 and 1999. Adjusting for age and race alone, 24 areas or clusters of elevated incidence were identified, widely distributed throughout the state. These included some “nested” areas with varying levels of relative risk: for instance, an area with a relative risk above 2 might be part of a larger area with a relative risk above 1.5 [2]. After additional adjustment for socioeconomic variables using 1990 census tract-level data (high income, education, crowding; other variables were considered and found to be either redundant or insignificant), the relative risks in 16 of the areas declined substantially, remained about the same in 6 of the areas, and increased in 2 areas (two nested areas within predominantly black neighborhoods in Brooklyn). Overall, 16 of the 24 areas remained statistically significantly elevated.
These findings provide evidence for a moderate association between prostate cancer incidence and socioeconomic variables that is consistent with other published studies [3]. This association has been attributed in part to utilization of PSA testing, which is generally higher in areas with higher levels of income and education. Since PSA testing detects some early-stage prostate cancers that might otherwise go undetected during a man’s entire life, the incidence rates in these locations are likely to be higher. The Brooklyn neighborhoods cited earlier are a major exception, in that they have high incidence rates despite lower income and education levels. Based on this information and related evidence, these neighborhoods have been targeted for awareness and outreach activities by the New York State Department of Health. Much of the geographic variation in prostate cancer remains unexplained, as evidenced by the 16 areas that remained statistically significantly elevated after adjustment for socioeconomic variables. These could be areas with unusually active screening programs, areas with better data reporting by physicians or outpatient clinics that traditionally have not diagnosed cancer cases, or the result of local environmental factors.
While greater clarity as to the precise contribution of each socioeconomic risk factor could have been obtained using traditional epidemiological methods, our approach had the advantage of highlighting specific geographic locations, which is a useful framework for population-based surveillance programs that must frequently respond to questions about rates and exposures in specific locales. Further details of our study may be obtained by contacting fpb01@health.state.ny.us.
[1] Schmit KJ, McLaughlin CC, Johnson G, Schymura MJ, Negoita S. Incorporating risk factor data into geographic analyses of prostate cancer. Poster exhibited at the CDC Cancer Conference, Atlanta, September 15-18, 2003.
[2] Boscoe FP, McLaughlin CC, Schymura MJ, Kielb CL. Visualization of the spatial scan statistic using nested circles. Health and Place 2003; 9: 273-277.
[3] Liu L, Cozen W, Bernstein L, Ross RK, Deapen D. Changing relationship between socioeconomic status and prostate cancer incidence. Journal of the National Cancer Institute 2001; 93: 705-709.
Competing interests
None