[Cluster randomised trials: an important method in primary care research].

In primary care research interventional studies often address organisational changes or educational interventions, for example, in the context of guideline implementation. For pragmatic reasons randomisation is often conducted at practice level instead of at the individual patient level. Patients from one practice form a cluster, thus violating the basic assumption of independent patient samples. Hence an increased number of participants and a more complex analysis are required. Using the example of two cluster randomised trials the present article provides insights into the advantages and disadvantages of cluster randomisation as well as its practical significance for the planning and analysis of cluster randomised trials.