Cluster RCT book
The textbook was published in June 2014.
Cluster randomised trials are now one of the key tools in Medicine and Health Service Research, and they are being increasingly used. There have been a number of books published on cluster randomised trials but this new book has book has a number of features that distinguish it from other text books.
• We cover a wide range of trials, for many of which we have first-hand experience, and we use real data from these trials to illustrate our methods. The emphasis in the book is on practical issues and things we have learnt from experience.
• We show how to do all the analyses in two commercial packages SPSS and Stata and also the free software R. Many investigators use one of these commercial packages for analysing data and so will find the extensions to cluster randomised trials useful. The advantage of R is that any reader can reproduce the analyses and figures in the book without having to invest financially in a package. Since many cluster randomised trials take place in developing countries, where cost is an issue, the use of free software is also an advantage. R also lends itself to small bespoke programs for doing specific tasks such as randomisation or estimating sample sizes and these are given in the book and available of the accompanying website. We make no claim to programming expertise and these programs are designed to illustrate methods, but are not necessarily the most efficient or convenient method for analysis.
• The book is also intended as a textbook, and contains exercises and answers. We believe that one cannot learn anything without doing it. These exercises have been used in courses we have given on cluster randomised trials. The use of R also facilitates teaching since all students can obtain easily and so follow the examples in the book.
Many of our intended readers may not design and run cluster trials, but will certainly need to interpret them. We have special sections and exercises on reading and interpreting cluster trials.
The sequence of the book follows the sequence an investigator is likely to follow: after design issues in the first three chapters, analysis follows in the next three. To plan a study an investigator needs to follow a protocol, which is in Chapter 7. When the trial is completed the investigator needs to report the trial and this is described in Chapter 8. This chapter will also be useful for people who need to read and interpret cluster randomised trials. Chapter 9 covers more practical issues and Chapter 10 describes how to implement the methods on a computer. Finally, the emphasis in this book is primarily frequentist. This does not mean we do not think Bayesian methods are useful, but simply reflects the prevailing literature. "