Detecting Association for Low-Frequency Exposures in Case-Control Studies

• Activity Number: 623
• Type: Contributed
• Date/Time: Thursday, August 2, 2012 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Guan Xing*+ and Chang-Yun Lin and Chao Xing
• Companies: and The University of Texas Southwestern Medical Center and The University of Texas Southwestern Medical Center at Dallas
• Keywords: Association ; low-frequency ; case control

Abstract

In testing association between two dichotomous variables, there exist popular metrics such as odds ratio (OR) and Phi coefficient to measure the strength of association, and there exist popular tests such as Pearson's chi-squared test, Fisher's exact test, and the likelihood ratio test to test the strength of association. However, these measures and tests are not sensitive to detect association for low-frequency exposures. In this study, we propose a new statistic for the purpose of detecting low-frequency events associated with an outcome. The association test based on the new statistic is more sensitive than existing methods in detecting association for low-frequency exposures. We are able to numerically prove that the new statistic offers higher power while maintains the same type I error rate as Pearson's chi-squared test in a balanced study design. Application of the new method in genetic studies of detecting association of rare genetic variations with diseases is shown.