Methods and Applications of Statistics in Clinical Trials, Volume 1: Concepts, Principles, Trials, and Designs

Chapter 34

Global Assessment Variables

Barbara C. Tilley, Peng Huang, and Peter C. O’Brien

34.1 Introduction

Although most clinical trials are designed using a single primary outcome, in some trials it is difficult to find the single most appropriate outcome for the main objective of the study. In the NINDS Stroke Trial, a consistent and persuasive post-stroke improvement on multiple outcomes was required to define treatment efficacy [1]. In Parkinson’s disease clinical trials, a treatment would not be considered to slow progression if the treatment improved only motor score and other measures of outcome deteriorated. In studying quality of life, we may be interested in treatment effects on all subscales being measured. In all cases, no single measure is sufficient to describe the treatment effect of interest, and no validated summary score exists. In a quality-of-life example, we could measure an overall quality-of-life score by summing across subscales, but we may under or overestimate the effect of treatment depending on how the treatment affects the individual subscales.

34.2 Scientific Questions for Multiple Outcomes

The choice of statistical method used for assessing treatment effect on multiple outcomes depends on the scientific question under investigation. Two types of questions lead to two different classes of tests. The first type is a directionless question: “Is there any difference, helpful or harmful between the two treatments in these outcomes?” A treatment difference ...