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

Chapter 46

Multiple Endpoints

Frank Bretz and Michael Branson

46.1 Introduction

A common problem in pharmaceutical research is the comparison of two treatments for more than one outcome measure, which in the clinical context are often referred to as endpoints. A single observation or measurement is often not sufficient to describe a clinically relevant treatment benefit. In respiratory studies, for example, several endpoints (such as FEV₁, respiratory symptoms, and health-related quality of life) are frequently considered to determine a treatment-related benefit. In cardiovascular trials, possible endpoints include time to myocardial infarction, congestive heart failure, stroke, and so on. In such instances, the experimenter is interested in assessing a potential treatment effect while accounting for all multiple outcome measures.

The aspects of a highly multidimensional and complex syndrome are usually assessed by means of various symptoms or ordinal items in scales. In order to map these observations on one (or a few) ordinal scale(s) that best represents the severity (or improvement) of the disease, one can try to reduce the dimensionality by combining the univariate projections of this syndrome to a single (or a few) measure(s) of efficacy. This approach is also in agreement with the ICH E9 guideline [1], in which it is recommended to use a single (primary) endpoint, where possible, thus reflecting the need to efficiently reduce the dimensionality. If, nevertheless, certain ...