The 131st Annual Meeting (November 15-19, 2003) of APHA

The 131st Annual Meeting (November 15-19, 2003) of APHA

4198.0: Tuesday, November 18, 2003 - 3:05 PM

Abstract #73118

Use of hierarchical linear modeling and GIS in analyzing prostate cancer incidence data

Frances J. Mather, PhD1, Vivien W. Chen, PhD2, Leslie Morgan, MS3, Catherine A. Correa, PhD2, Sudesh Srivastav, PhD3, Janet Rice, PhD4, Xiao Cheng Wu, MD2, Jeffrey Shaffer, MS3, George Blount, BS2, Christopher Swalm, MS4, and Richard Scribner, MD, MPH5. (1) Department of Biostatistics, Tulane University School of Public Health and Tropical Medicine, 1430 Tulane Avenue, New Orleans, LA 7 0112, 504-587-7329,, (2) Department of Public Health and Preventive Medicine, Louisian State University Health Sciences Center/Louisiana Tumor Registry, 1600 Canal Street , Suite 900A, New Orleans, LA 70112, (3) Department of Biostatistics, Tulane University School of Public Health & Tropical Medicine, 1440 Canal St., 20th Floor, New Orleans, LA 70112, (4) Tulane University SPHTM, 1440 Canal Street, New Orleans, LA 70112, (5) Public Health and Preventive Medicine, Louisiana State University Health Sciences Center, 1600 Canal Street, Suite 800, New Orleans, LA 70112

Background: Recent trends in the incidence of prostate cancer have been difficult to interpret. The etiology of prostate cancer is not well understood and the established risk factors are few (or limited to age, race and gene-susceptibility). Besides risk factors, many feel that screening tests, such as the PSA test, may have contributed to trends in incidence. Attempts to evaluate the effect these factors might have on incidence is complicated by the absence of data, estimating exposure to screening tests at the individual level. The use of hierarchical linear modeling can be used to model risk factors estimated at other than the individual level, for example, the proportion of persons having PSA tests in the last year per geographic area, or the number of urologists in the geographic area, as well as measures of deprivation.

Methods: These issues were addressed in a study of risk factors associated with geographic variations of prostate cancer incidence in Louisiana from 1988 to 1999. Prostate cancer cases (31,580) from the Louisiana Tumor Registry, were geocoded to parish and urban census tracts. Hierarchical linear modeling, by means of SAS, were used to explain the variability among parishes/selected census tract groups and to provide estimates of adjusted incidence with which to illustrate the changes in incidence over time and geographical region on maps. The results of this study will be presented.

Key Words: Prostate cancer, GIS, hierarchical linear modeling

Learning Objectives:

Presenting author's disclosure statement:
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.

Epidemiologic Research on Social Determinants of Prostate Cancer: The Role of GIS Methods and Spatial Statistics

The 131st Annual Meeting (November 15-19, 2003) of APHA