Desktop version

Home arrow Economics

  • Increase font
  • Decrease font


<<   CONTENTS   >>

Information-Theoretic Approach for Two Binary Endpoints

For the Schizophrenia study, presented in Section 2.2.2 (see also Chapters 12 and 13 for a similar analysis in R and SAS, respectively), the two binary endpoints are defined as:

In R, using the Surrogate package, for a multi-trial setting with two binary endpoints, the function FixedBinBintIT can be used to estimate both

FIGURE 14.6

Schizophrenia Study. Analysis using the information-theoretic approach for two binary endpoints.

individual-level and trial-level surrogacy measures. For the schizophrenia study, the function is called in the following way:

Sur<-FixedBinBinIT(Dataset=Schizo, Surr=Panss_Bin,

True=CGI_Bin, Treat=Treat,

Trial.ID=InvestId, Weighted=TRUE,

Pat.ID=Id, Model="Reduced",

Number.Bootstraps=500,Seed=1)

In the Surrogate Shiny App, the following specifications should be used in the data loading screen in Figure 14.1: the true (CGI bin) and the surrogate endpoints (PANSS bin), the treatment variable (Treat), the unit for which will be calculated (InvestId), and the patient’s identification number (Id). In the same screen, the tab Fixed effects information theory (BinaryBinary ) is selected in order to perform the analysis. The number of bootstrap samples (Number.Bootstraps=500) and the seed (Seed=1) are specified in the left panel in Figure 14.6. For the schizophrenia study, trial and individual- level surrogacy measures are equal to Rht = 0.8213 (0.7469, 0.87864) and Rh = 0.3305 (0.2992, 0.3623), respectively.

 
<<   CONTENTS   >>

Related topics