Multiple Rating Scales
A second situation in which surrogate evaluation methods are attractive is when the true endpoint is not clearly defined, such as is frequently the case in psychiatric disorders. The case of schizophrenia described in Section 2.2.2
is typical of such situations, with several scales being used to assess the patient’s condition, in this case the Clinical Global Impression (CGI), the Brief Psychiatric Rating Scale (BPRS), and the Positive and Negative Syndrome Scale (PANSS). In fact any of these scales could be considered the true endpoint of interest, with the other two being surrogates for it (Molenberghs et al., 2010). Mathematically, the models are symmetric in the surrogate and the true endpoint, so either endpoint can be considered a surrogate for the other. The evaluation methods described in this book can provide valuable insight in the predictive value of one scale for another, both at the individual and trial level, and as such they can inform the choice of a preferred endpoint for future clinical trials when multiple rating scales are being used.