|
Activity Number:
|
46
|
|
Type:
|
Invited
|
|
Date/Time:
|
Sunday, July 29, 2007 : 4:00 PM to 5:50 PM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #307797 |
|
Title:
|
Meta-Analysis at the Individual Level: Synthesis of Inferences Across Multiple Gerontological Databases
|
|
Author(s):
|
Karen Bandeen-Roche*+
|
|
Companies:
|
Johns Hopkins Bloomberg School of Public Health
|
|
Address:
|
Department of Biostatistics, Baltimore, MD, 21205,
|
|
Keywords:
|
aging ; selection ; differential measurement ; bias
|
|
Abstract:
|
Multiple cohort studies have studied health declines in older adults. Health aspects assessed overlap, hence combining individual-level information across the studies for addressing research questions on elders' health has promise. We propose methodology to heighten accuracy and precision of inferences in analyses that combine so. We address proper accounting for stochastic heterogeneity across studies, as well as two challenges that may be primarily biasing rather than variability heightening: the least or most healthy are often screened out, thus target populations may differ considerably across studies; and key outcomes may be subject to appreciably differential measurement. The project exemplifies multiplicity of challenges in aging statistics and aims to improve accuracy and precision of findings obtainable through the combined use of multiple data sets on aging.
|
