|
Activity Number:
|
166
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 9, 2004 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #301554 |
|
Title:
|
When Does Studying Only the Higher Relative Risk Subgroup Increase Efficiency for Evaluating an Effect?
|
|
Author(s):
|
Abhijit Dasgupta*+ and Sholom Wacholder
|
|
Companies:
|
National Institutes of Health and National Institutes of Health
|
|
Address:
|
Biostatistics Branch, DCEG, NCI, DHHS, Bethesda, MD, 20852-7244,
|
|
Keywords:
|
subgroup analysis ; study design ; sample size ; relative efficiency ; gene-environment interaction
|
|
Abstract:
|
Many studies are designed to assess the joint effects of genetic and environmental factors in the development of disease. We consider the efficiency of different designs for testing whether a dichotomous environmental factor X is related to disease. The alternative hypothesis is that X is related to disease in at least one subgroup defined by dichotomous genetic factor G. When the relative effect of X is the same in both subgroups, clearly there is no efficiency advantage in determining G. On the other hand, if the effect of X is present only when G=1, a strategy that excludes all cases and controls with G=0 will be more efficient. We investigate the relative efficiency of these two strategies over a wide range of realistic scenarios. We find that often there is no benefit in efficiency from using G to choose subjects for measurement of X, even when the relative risk due to X is much greater in G=1 than in G=0. We provide a spreadsheet to assess the relative efficiencies of the strategies for arbitrary joint distributions of X and G and arbitrary joint effects models. We also identify situations where a particular strategy is at least 10% more efficient.
|
