Small clinical trials: are they all bad?

Abstract

Statisticians have long argued that randomized controlled trials should be sufficiently large to achieve their purpose, and for common diseases with major public health implications this has brought many benefits. However, there are many instances where it is unrealistic to expect clinicians to provide the information on which sample sizes are calculated and undue emphasis on trial size can be counterproductive. Examples of such trials are given and some general issues discussed, including the unhelpful contribution of hypothesis testing, the need for replication and the role of the statistician.