Cross-over Trials in Clinical Research: An Introduction

Course leader: Stephen Senn, Consultant Statistician, Edinburgh


This course is based on the book with the same name and is designed to provide the attendee with a basic grounding in the techniques used for analysing cross-over trials although, in fact, some of the course will cover design also. The course will be firmly practical with the emphasis on using cross-over trials to say something about the effects of treatments rather than as some sort of mathematical guinea pig. Controversies and issues will be covered as well as techniques. The analysis of examples using Genstat, SAS and R will also be illustrated.


a. Nature and uses of cross-over trials
b. Philosophy and approaches to analysis
c. Some examples

2.The AB/BA design

a. The basic estimator approach
b. Adjusting for period effects
c. The problem of carry-over
d. Problems with the two-stage analysis
e. Analysis using proc glm and proc mixed
f. Use of baselines

3. Other outcomes for the AB/BA design

a. Nonparametric approaches
b. Binary data
c. Count data
d. Time to event data

4. Normal data from designs with three or more treatments

a. The degrees of freedom problem
b. Basic estimator approach
c. Analysis using proc glm and proc mixed

5. Various complications

a. Incomplete block designs
b. Basic estimator approach
c. Ordinary least squares
d. Mixed models
e. Trials in infertility

6. Various design issues

a. Sample size determination
b. Choosing sequences
c. Incomplete block designs

7. Mathematical approaches to carry-over

a. Reason why the simple carry-over model is not useful
b. How to adjust if you must
c. Analysis using proc glm and proc mixed

Return to Statwise courses