|
Activity Number:
|
26
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, August 3, 2003 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #300383 |
|
Title:
|
Detection of Spatial Clustering Using Case-Control Data in the Presence of Covariates
|
|
Author(s):
|
Ronald Gangnon*+ and Murray K. Clayton
|
|
Companies:
|
University of Wisconsin and University of Wisconsin
|
|
Address:
|
207 WARF Office Building, Madison, WI, 53726-2336,
|
|
Keywords:
|
case-control study ; cluster detection ; spatial analysis ; point pattern ; hypothesis testing
|
|
Abstract:
|
A basic problem in spatial epidemiology is the detection of spatial clusters. Many hypothesis testing procedures for this purpose have been proposed for use with case-control data. The significance of the test statistic is assessed using the randomization distribution of the cases and controls. A major limitation of this approach is that it does not account for known risk factors for the disease in question. If the spatial distribution of subjects with these known risk factors is not random, incorrect inferences about the presence and/or locations of clusters may be drawn. We extend model-based spatial clustering tests for case-count data such as the spatial scan statistic and the weighted average likelihood ratio test to the case-control setting. We consider the computational issues that arise in the change from a Poisson model suitable for case-count data to a binomial model suitable for case-control data. We also discuss a simple sampling scheme for generating the null distribution of the test statistics in the presence of covariates. The methods are illustrated using data from an ongoing, statewide population-based case-control study of breast cancer in Wisconsin.
|
