Desktop version

Home arrow Economics

  • Increase font
  • Decrease font


<<   CONTENTS   >>

The binary-binary setting

Here, it is examined whether the dichotomized BPRS is an appropriate surrogate for the dichotomized PANSS. The Rhind equaled 0.5392, with 95% confidence interval [0.5089; 0.5688]. The results thus indicate that the uncertainty about T = clinically relevant change on the PANSS that is reduced when S = clinically relevant change on the BPRS becomes known (individual-level surrogacy) is moderate.

FIGURE 10.2

Schizophrenia Data. Binary-binary setting. Plot of the estimated R)ind (and their 95% confidence intervals) per cluster (treating physician) for S = clinically relevant change on the BPRS and T = clinically relevant change on the PANSS.

Figure 10.2 shows the estimated Rhind and their 95% confidence intervals for each of the clusters (treating psychiatrists) separately. As can be seen, there was substantial variability in the point estimates for Rhind across clusters. Nonetheless, the 95% confidence intervals for most clusters overlapped, indicating that the individual-level surrogacy estimates were of similar magnitude for the different clusters.

FIGURE 10.3

Schizophrenia Data. Continuous-binary setting. Plot of the estimated Rhind (and their 95% confidence intervals) per cluster (treating physician) for S = clinically relevant change on the CGI and T = PANSS.

The continuous-binary setting

Here, it is examined whether the dichotomized CGI is an appropriate surrogate for the change in PANSS scores (measurement on the original scale). The R2hind equaled 0.4545, with 95% confidence interval [0.4229; 0.4859]. It can thus be concluded that the uncertainty about T = change on the PANSS that is reduced when S = clinically relevant change on the CGI becomes known (individual-level surrogacy) is low to moderate.

Figure 10.3 shows the estimated Rhind (and 95% confidence intervals) in each of the clusters (treating psychiatrists). As can be seen, there was quite some variability in the point estimates for Rhind across clusters, though their

95% confidence intervals largely overlapped. This indicates that the individual- level surrogacy estimates across clusters were of the same magnitude.

 
<<   CONTENTS   >>

Related topics