ABSTRACT
This chapter is intended principally for clinicians who need to use the results of clinical grails in their routine practice, and who want to know why some types of evidence are much more reliable than others. It is concerned with treatments that might improve survival (or some other really major aspect of long-term disease outcome), and its chief point is that as long as doctors start with a healthy scepticism about the many apparently striking claims that appear in the medical literature, large-scale trials do make sense. The main enemy of common sense is overoptimism: there are a few striking exceptions of treatments for serious disease that really do turn out to work extremely well, but in general most of the claims of vast improvements from new therapies turn out to be evanescent. Hence, clinical trials need to be able to detect or to refute really reliably the more moderate differences in long-term outcome that it is medically realistic to expect. Once this common-sense idea is explicitly recognized, the rest follows naturally, and it becomes fairly obvious what types of evidence can and cannot be trusted. Although this chapter may also be of some interest or encouragement to doctors who are considering participating in (or even planning)
some large trial, its main intended audience is the practising clinician, for even the most definite results from largescale randomized evidence cannot save lives unless practising clinicians accept and apply them. The chapter does not go into a lot of statistical details: instead, it tries to communicate the spirit that underlies the increasing emphasis over the past decade or so on large-scale randomized evidence.
