An investigation of the performance for ordered subset analysis to a gene-environment interaction model.

摘要

Ordered subset analysis (OSA) is a recent (∼1999) addition to statistical methodology and its application has been limited to genetic linkage analysis. This study extends the OSA to a test of association in a gene-environment interaction model, where the explanatory variable is continuous. The application of the OSA to this model, entails: (a) Data Simulation, (b) Regression Analysis, and (c) Empirical p-value determination. The power of the OSA is determined, based on the Empirical p-values obtained via repetition of steps (a) - (c) above, and is used as a performance marker for comparison to "traditional" linear regression analysis.; The goal of the OSA is to determine the level of the continuous environmental factor, which provides maximum evidence of a difference in the mean levels of the outcome between the two gene levels for a given dichotomous gene factor.