General Structure of the SAS Macros Available for the Analysis of Surrogate Endpoints

The SAS macros, developed for the analysis presented in the book, have the same general form. Depending on the surrogacy setting, the macros require the user to prepare a dataset with a specific structure. Note that different macros require different dataset structures. A generic call of a surrogacy SAS macro has the following form:

% macroname (data=,true=,surrog=,trial=,treatment=,patid=, optional-arguments)

The following arguments are used in all macros:

  • • data: a dataset containing one record per patient with measurements for both the true and the surrogate endpoints.
  • • true: a measurement of the true endpoint.
  • • surrog: a measurement of the surrogate endpoint.
  • • treatment: treatment indicator variable (1= active, -1=control).
  • • trial: the unit of the study for which trial-level surrogacy will be estimated.
  • • patid: the patient’s identification number.
  • • optional-arguments: optional arguments that should be provided to conduct a specific analysis (for example, the leave-one-out trial analysis or the exploratory plots); all of this will be discussed further in Section 12.3.1.

Graphical and numerical outputs are customized in such a way that users only see relevant information, depending on the surrogacy setting. Note that the standard SAS output tables are not produced and that it is recommended to check the SAS log window for possible problems.

For the remainder of the chapter we present different surrogacy settings and their corresponding macros, case studies, and output. Note that, for some surrogacy settings, the models are only briefly presented and we refer to the relevant section elsewhere in the book, where the necessary technical details and model formulations are presented and explained.

< Prev   CONTENTS   Source   Next >