The Virtual Subject Response (VSR) tab allows the user to explicitly define virtual subject response profiles that dictate the distribution that patients’ endpoints should be simulated from. VSRs can be specified explicitly by specifying the control distribution and treatment effects, or by importing externally simulated patient responses.
If specifying, the VSR in the “Explicitly Defined” tab you will always have the dose response VSR to specify, and if subjects have multiple visits, then a Longitudinal VSR must be specified as well. If using a continuous endpoint and simulating a baseline, then a baseline simulation VSR must be specified as well.
A dose response VSR specifies how all arms in the study should have their final endpoint value simulated. For continuous endpoints this is the mean and standard deviation of the normal distribution. For dichotomous endpoints it is the probability of response, and for time-to-event endpoints it is the hazard rate for each arm.
The longitudinal VSR dictates how subject visits are correlated with the final endpoint value. Each endpoint type has different methods of simulating longitudinal correlation, which are explained in detail in the following sections.
It is common, and advisable, to create a variety of VSR scenarios. Each scenario is a combination of a dose response VSR, a longitudinal VSR (if it exists), and a baseline VSR (if it exists). Generally, at least one null VSR, and a set of alternative scenarios are created. In a null scenario the treatment arms have the same efficacy For a superiority study. In a non-inferiority study the treatment arms will have an efficacy profile that lies on the control minus the non-inferiority margin. profile as the control arm. Alternative This term (and the term ‘null scenario’) is borrowed from hypothesis testing. It indicates that the assumed scenario belongs to the alternative space of a traditional hypothesis test. VSRs have treatments with a variety of treatment effects, usually with treatment arms simulated to be better than control.
If an external file is used to specify the subject responses to be simulated, it replaces the dose response, longitudinal, and baseline profiles specified in an explicitly defined VSR tab. An entire sequence of visit responses for a subjects is drawn from the uploaded patient responses file. This is elaborated on more in each endpoint’s VSR description.
Each endpoint type (Continuous, Dichotomous, and Time-to-event) has its own method for specifying the dose response VSR and the visit to visit correlation.