|
Activity Number:
|
284
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, July 31, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #309454 |
|
Title:
|
Inverse Probability-Weighted Estimators in the Stratified Nested Case-Control Sampling Methods
|
|
Author(s):
|
Sachiko Tanaka*+ and Yutaka Matsuyama and Shiro Tanaka and Seiichiro Yamamoto
|
|
Companies:
|
Tokyo University of Science and University of Tokyo and University of Tokyo and National Cancer Center
|
|
Address:
|
Faculty of Engineering Course, Tokyo, 102-0073, Japan
|
|
Keywords:
|
absolute risk ; counter-matching ; matching ; nested case-control study ; weighted estimators
|
|
Abstract:
|
Stratified nested case-control designs are widely used in the expectation that these methods may increase precision relative to simple nested case-control design. Although conditional logistic regression is used for analysis of data with matching, it will be inefficient because it does not use all subjects at risk. Furthermore, usual method does not give the estimate of absolute risk. In this talk, we introduce the inverse probability weighted estimator of Cox regression parameter under stratified sampling designs. We estimate selection probability and conditional expectation of missing covariate using nonparametric method that incorporates information of matching factors. Estimates of absolute risk are obtained based on Nelson-Aalen estimators. We also extend our method to counter-matching design. The performance of the proposed method is investigated through simulation studies.
|
