|
Activity Number:
|
251
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #307541 |
|
Title:
|
A Two-Phase Analysis of Air Pollution and Adverse Birth Outcomes
|
|
Author(s):
|
Katherine Hoggatt*+ and Sander Greenland and Beate Ritz
|
|
Companies:
|
University of Michigan and University of California, Los Angeles and University of California, Los Angeles
|
|
Address:
|
109 Observatory Street, Ann Arbor, MI, 48109,
|
|
Keywords:
|
epidemiology ; two-phase ; two-stage ; case-control
|
|
Abstract:
|
We show how a two-phase design and analysis can be used to collect detailed covariate data on a subsample and correct for possible selection bias and improve the efficiency of the estimates relative to what would be expected if only the subsample were analyzed. Our Phase 1 group was year 2003 births in select Los Angeles ZIP codes. Our Phase 2 group was births selected for the nested case-control study where the mother responded to the study questionnaire. To assess possible selection bias, we estimated the association between ambient carbon monoxide exposure and low weight birth (< 2500g) in the Phase 1 and 2 groups with only first-phase variables included in the models. We then used three two-phase estimators, which allowed us to use second-phase covariates to better control for confounding, reduce the apparent selection bias, and increase the efficiency of our estimates.
|
