The Design and Analysis of N of 1 Trials

5 July, 2016, 9 AM-5 PM

Radisson Blu Edwardian Heathrow Hotel <br>140 Bath Road, Hayes, Middlesex, UB3 5AW <br><br><b>Start:</b> 09:00 <br><b>Finish:</b> 17:00

PLEASE CLICK HERE TO REGISTER!

Introduction

N of 1 trials are trials in which individual patients are repeatedly treated with experimental and control treatments in a deliberate and designed manner using principles of control, randomisation and replication. Their uses include personalising treatment and increasing efficiency by reducing the number of patients it is necessary to study.

In chronic diseases, sets of n-of-1 trials (in which a limited number of patients follow an n-of-1 protocol) have great potential as phase IV trials for understanding components of variation but may also constitute possible Phase III programmes for rare diseases. They can also be used as phase II studies for proof of concept and dose-finding. However, they are often poorly analysed and, indeed, much of the published advice is poor.

This course will present the latest thinking on n-of-1 trials and cover not only their analysis through SAS®, R® GenStat® and meta-analysis packages but also approaches to design. They will also be critically examined as to their potential use in a) establishing average effects of treatment b) studying the extent to which such effects vary from patient to patient and c) optimising treatment for individual patients.

Course leader

This course will be given by Professor Stephen Senn, of the Luxembourg Institute of Health, who is well- known for his work on the design and analysis of clinical trials and the application of statistics in drug development. 

Topics covered

• Uses of n-of-1 trials and purposes of analysis
  • Showing the treatment can work
  • Understanding variation in effect
  • Predicting effects
• Design 
  • Randomisation in cycles
  • Randomisation in patient blocks
• Graphical presentation of results
  • Trellis plots
  • Dot plots
• Causal analysis
• Analysis of variance
  • Block structure and the Wilkinson & Roger notation
  • Main effect models
  • Allowing for interaction
• Summary measures approaches
• Mixed models
  • Estimation
  • Best linear unbiased predictors (shrunk estimates)
• Meta-analysis approaches
  • Fixed effects
  • Random effects
• Binary data
• Planning issues

Registration closes 28th June 2016

PLEASE CLICK HERE TO REGISTER!