|
Activity Number:
|
17
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Sunday, August 2, 2009 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
| Abstract - #303889 |
|
Title:
|
Comparing Covariates Selection in Propensity Score Modeling: A Simulation Study
|
|
Author(s):
|
Jian Huang*+ and Alan Yu and Lan Pan
|
|
Companies:
|
Boston Scientific Corporation and Boston Scientific Corporation and Boston Scientific Corporation
|
|
Address:
|
4100 Hamline Avenue North, St. Paul, MN, 55112,
|
|
Keywords:
|
Propensity Score ; Covariates ; Bias ; Imbalance ; Simulation
|
|
Abstract:
|
Propensity score methods are increasingly popular in observational studies to reduce bias of subject selection in the estimation of treatment effects. In order to estimate the propensity score, one must model the distribution of the treatment indicator variables given the observed covariates. There is no consensus on which covariates should be included in the propensity score model. In this presentation we compare different covariates selection approaches (i.e. include all potential covariates; or include covariates significantly related to treatment assignment, or include only covariates that are predictors of outcomes, etc.). Stratification on the quintiles of the propensity score method is employed. The bias of treatment effects and statistical power are evaluated via simulation. Comparability of two treatment groups through estimated propensity score distributions is also discussed.
|
