Summary measures and statistics in the analysis of quality of life data: an example from an EORTC-NCIC-SAKK locally advanced breast cancer study

Quality of Life (QL) is now included as an endpoint in many phase III cance r clinical trials. Numerous statistical techniques have been presented in t he literature to analyse QL data but there is still no agreement as to what is the optimal approach of analysis. In this paper we, therefore, present and compare various techniques which have all appeared in the literature an d which may be globally described as summary measures and summary statistic s. These techniques are illustrated using data from an EORTC clinical trial in locally advanced breast cancer (EORTC trial 10921). It is also explaine d in this paper how and when these techniques may be used in other cancer s ettings. For EORTC trial 10921, it is shown that by choosing different tech niques different conclusions may be drawn concerning the QL outcome. This h ighlights the importance of choosing an appropriate primary statistical met hod and for describing it a priori in the protocol and analysis plan. In th is paper, we show the importance of performing sensitivity or supportive an alysis to support conclusions drawn from the primary analysis. (C) 2000 Els evier Science Ltd. All rights reserved.