Spatial analysis of prostate cancer incidence and race in Virginia, 1990-1999

Am J Prev Med. 2006 Feb;30(2 Suppl):S67-76. doi: 10.1016/j.amepre.2005.09.008.


Background: Racial disparities exist in prostate cancer incidence. An important contributor to these disparities may be socioeconomic status.

Methods: Virginia Cancer Registry data, 1990-1999 (37,373 cases) were geocoded to the Census tract and county level. The annualized, age-adjusted incidence rates for African Americans and whites were calculated, and crude and smoothed maps of these rates were produced. Statistical tests for clustering of cases were conducted. Prostate cancer incidence was statistically modeled as a function of area-based measures of poverty, median household income, education, rural status, ratio of physicians to population in each county, percentage of men in each county obtaining prostate cancer screening with the prostate-specific antigen (PSA) test, and percent of households headed by females.

Results: Prostate cancer incidence was elevated in the eastern and central portions of the state. Statistical tests for clustering were highly significant (Tango's test, p<0.008; spatial scan statistic, p<0.001). Poverty and lower education were associated with a decreased incidence among whites but not African Americans. Median household income and urban status were positively associated with incidence for both populations. Among whites, increased percent of female heads of households and ratio of physicians per population were associated with increased incidence. Associations between predictor variables and prostate cancer incidence were seen only in the census tract level analyses.

Conclusions: Overall, the findings support the argument that area measures of poverty and education do not explain the increased incidence of prostate cancer among African Americans. Other factors, such as dietary practices, may help explain racial disparities in prostate cancer incidence. Because of the large differences between tract and county level results, the time and expense of obtaining data geocoded to the tract level seems worthwhile.

Publication types

  • Research Support, N.I.H., Extramural
  • Research Support, Non-U.S. Gov't

MeSH terms

  • Adult
  • Aged
  • Aged, 80 and over
  • Cluster Analysis
  • Humans
  • Male
  • Middle Aged
  • Prostatic Neoplasms / epidemiology*
  • Racial Groups
  • Registries
  • Topography, Medical*
  • Virginia / epidemiology