Randomised Trials
Altman DG.
© 2016 John Wiley & Sons, Ltd. All rights reserved. This chapter concentrates on randomised trials of healthcare interventions, better known as clinical trials. The simplest and most frequently used design for a clinical trial is the parallel group design, in which two groups of patients receive different treatments concurrently. The chapter illustrates how the design and analysis of trials have to be tailored to individual circumstances. It also provides a brief overview of a complex topic. Randomisation deals with possible bias at the treatment allocation, but bias can also creep in while the study is being run. For parallel group studies, the calculation of sample size is based on comparing either means. A detailed protocol must accompany an application for a grant for a trial, and the local research ethics committee will also require most of the above information. Subgroup analyses pose serious problems of interpretation similar to those resulting from multiple outcome measures.