|
Activity Number:
|
344
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 11, 2004 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #301605 |
|
Title:
|
A Multiscale Method for Disease Mapping in Spatial Epidemiology
|
|
Author(s):
|
Mary Louie*+ and Eric Kolaczyk
|
|
Companies:
|
Channing Laboratory and Boston University
|
|
Address:
|
181 Longwood Ave., Boston, MA, 02115,
|
|
Keywords:
|
disease mapping ; multiscale spatial process ; standardized mortality ratio ; statistical inference ; confidence map
|
|
Abstract:
|
Maps by their very nature display spatial information that varies with resolution (or "scale"). The effects of scale in disease mapping are well recognized. Here we provide a statistical inference framework for describing the distribution of relative risk simultaneously across a hierarchy of multiple scales. This is accomplished via an extension of the Poisson multiscale spatial process model of Kolaczyk and Huang (2001) to measures of relative risk for one of two canonical data structures in disease mapping, that of tract count data. We develop the basic framework for the standardized mortality ratio (SMR), derive (Bayesian) posterior-based estimation strategies, and provide methods for characterizing uncertainty. We illustrate the method through an application to data on gastric cancer in Tuscany. The overall framework provides a tool from which informative disease and confidence maps can be produced across a hierarchy of multiple scales, without the need to first try to identify a single appropriate scale of analysis.
|
