Abstract:
|
In randomized clinical trials, the use of validated surrogate markers can be highly desirable in trial design, monitoring and analysis, as they may reduce sample size and follow-up to save cost and effort, and facilitate scientific discoveries. However, challenges exist to identify a reliable marker. One particular statistical challenge arises on how to measure the surrogacy of potential markers, and rank the markers. Two measures, namely, the proportion of treatment effect (PTE) explained by a marker (Freedman et al., 1992) and the F-measure (Wang and Taylor, 2002), have been proposed to quantify a marker's surrogacy. In this article, we examine these two measures, and propose a new surrogacy measure, the so-called "treatment surrogacy fraction" in the setting of clinical trials. We show that both the PTE and the F -measure can be expressed by the new measure, but it is more reliable for estimation and inference, and carries an appealing interpretation. We illustrate and compare these three measures by the HIV Prevention Trial Network 052 Study, a landmark HIV/AIDS prevention trial.
|