Workshop type: Training.

Objective: To introduce review authors to the design and statistical issues associated with cluster-randomised trials that may impact the conduct of systematic reviews.

Summary:
Cluster-randomised trials (CRTs) randomise intact social units (for example, communities, health care organisations) and collect multiple observations per unit (for example on citizens within communities or patients cared for by health care organisations).  Cluster randomisation has considerable implications for the design, power and analysis of studies which frequently have been ignored in both primary studies and systematic reviews.  In particular, sample sizes need to be inflated and methods of analysis need to account for the non-independence of the observations within units.

At the end of this workshop, participants will:
• Be able to identify cluster-randomised trials;
• Be able i dentify inappropriately analyzed cluster-randomised trials
• Be aware of methods of reanalysing the data appropriately
• Be familiar with methods of incorporating inappropriately analysed cluster-randomised trials into reviews
• Be aware of methods of reporting the results of CRTs into systematic reviews

Intended audience: (i) Review authors; (ii) statisticians.

Technical level expected of participants: Intermediate.