|
Activity Number:
|
539
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 10, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Statistics and the Environment
|
| Abstract - #306446 |
|
Title:
|
Prediction of State Cancer Case Counts and Rates by Hierarchical Spatio-Temporal Models
|
|
Author(s):
|
Linda Pickle*+
|
|
Companies:
|
National Cancer Institute
|
|
Address:
|
6116 Executive Blvd., Suite 504, Bethesda, MD, 20892-8317,
|
|
Keywords:
|
spatio-temporal models ; Poisson regression ; hierarchical models
|
|
Abstract:
|
Although not all U.S. states provide high quality cancer data annually, estimates of the numbers of incident cancer cases by state are useful, e.g., for measuring cancer burden and monitoring trends. We present results of a hierarchical spatio-temporal Poisson regression model that predicts the annual number of incident cancer cases expected in each U.S. state, which are then projected ahead 4 years, based on age- and county-specific incidence data available from high quality registries and the mortality/sociodemographic/lifestyle profiles of all counties. The model was assessed by usual goodness of fit statistics plus comparison of predictions to state-reported data and projections from the American Cancer Society. Methodologic issues discussed include a comparison of full Bayesian vs. approximate likelihood methods (SAS GLIMMIX) and parametric vs. robust estimates of spatial covariance.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2006 program |
