Dear all,
Here are details of a joint meeting of the RSS medical section and primary health care study group in February. Everyone welcome.
Best wishes
Sandra
JOINT MEDICAL SECTION AND PRIMARY HEALTH CARE STUDY GROUP MEETING
24th February 2004
Royal Statistical Society, Errol St, London
2-5pm
Cluster randomised trials: Should we avoid them, improve them, or use new approaches?
MIKE CAMPBELL (Sheffield University)
Cluster trials: can we avoid them?
A common argument for the use of cluster randomised trials is that they are necessary to avoid contamination, but at a cost of increased sample size. Some researchers have argued that this cost is too great and that individually randomised trials are more efficient. The debate comes down to bias versus efficiency and the talk will discuss both sides of the coin.
SANDRA ELDRIDGE (Queen Mary, University of London)
What's wrong with recent cluster randomised trials and how can we make them better?
Based on a review of 200 cluster randomised trials in primary care, this talk will consider the quality of design, conduct, analysis and reporting of these trials, and problems faced by investigators. I will discuss recruitment, generalisability, intervention design, non-adherence to protocol, sample size and analysis, and make suggestions for improvement.
REBECCA TURNER (MRC Biostatistics Unit, Cambridge)
Bayesian approaches to design and analysis of cluster randomised trials, and the
role of the intracluster correlation coefficient.
This talk will discuss construction of prior distributions to represent relevant external information on the intracluster correlation coefficient (ICC). In design of cluster trials, we make use of multiple available ICC estimates, while acknowledging their imprecision. A Bayesian approach to analysis allows incorporation of prior information and offers flexibility in modelling.
Sandra Eldridge
Department of General Practice and Primary Care
Barts and The London
Queen Mary University of London
Medical Sciences Building
Mile End Road
E1 4NS
telephone 020 7882 7920
fax 020 7882 6396
e-mail [log in to unmask]
|