|
Activity Number:
|
393
|
|
Type:
|
Invited
|
|
Date/Time:
|
Wednesday, August 1, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
ENAR
|
| Abstract - #307832 |
|
Title:
|
Methods for Correcting Regression Dilution and Reverse Causation Biases in Regression-Model Survival Analyses
|
|
Author(s):
|
James Greenberg*+
|
|
Companies:
|
Brooklyn College
|
|
Address:
|
2900 Bedford Avenue, Brooklyn, NY, 11210,
|
|
Keywords:
|
Within-person variation ; measurement error ; regression dilution ; reverse causation ; body mass index ; blood pressure
|
|
Abstract:
|
Three methods were tested: 1) correcting for regression dilution by using the average of several replicates as predictor; 2) correcting for reverse causation by excluding ever smokers and also excluding participants with age>60 yrs and evidence of secular decrease in predictor; and 3) correcting for both biases by using the average of several replicates as predictor and the baseline observation as a covariate. These methods were tested using data from two prospective cohorts. Two predictors were tested, blood pressure and body mass index. All three methods were found to cause substantial increases in hazard ratios at elevated levels of the risk factor. These methods therefore appear to be effective in correcting for some of the effects of regression dilution and reverse causation. Further research is need to uncover their modus operandi, and to define when their use is appropriate.
|
- 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 2007 program |
