Methods

There is no agreement on a single optimal method to validate a surrogate marker or surrogate endpoint. Multiple statistical frameworks have been proposed for evaluating surrogate markers including the meta-analytic framework, the principal stratification framework, and the proportion of treatment effect explained (PTE) framework. In the content and videos below, we describe these frameworks and then focus primarily on implementation of the PTE framework. This includes methods that can handle censored data, multiple surrogates, surrogates measured with error, longitudinal surrogates, high-dimensional surrogates and more. Each method has publicly available software in R to implement estimation and inference. These are described below with more detailed information about software available here.