|
Activity Number:
|
440
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 5, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #305448 |
|
Title:
|
Confidence Interval for the Product of a Relative Risk and a Probability
|
|
Author(s):
|
William M. Reichmann*+ and Charles R. Horsburgh, Jr. and David Gagnon and Elena Losina
|
|
Companies:
|
Brigham and Women's Hospital and Boston University School of Public Health and Boston University School of Public Health and Boston University
|
|
Address:
|
75 Francis Street, Boston, MA, 02115,
|
|
Keywords:
|
product of RR and a probability ; bias ; confidence interval ; coverage
|
|
Abstract:
|
When the prevalence of an exposure and/or diseases is low, exposure-specific risks have been calculated from two different samples by multiplying a relative risk (RR) by the overall probability of disease. We performed a simulation study that evaluated the bias of the product estimator for a rare exposure and disease. We also, propose two CIs and evaluate their coverage probabilities. The first CI was calculated by multiplying the bounds of the CIs of the RR and the overall probability together to get the bounds for the product. The second CI was derived using the delta method. Both the bias of the estimator and the coverage probability of the CIs exhibited worse characteristics when the relative risk was larger. Correcting the bias of the estimator will go a long way in ensuring accurate confidence intervals in this setting.
|
