This is an archived version of the Handbook. For the current version, please go to or search for this chapter here.

16.3.7  Issues in the incorporation of cluster-randomized trials

Cluster-randomized trials may, in principle, be combined with individually randomized trials in the same meta-analysis. Consideration should be given to the possibility of important differences in the effects being evaluated between the different types of trial. There are often good reasons for performing cluster-randomized trials and these should be examined. For example, in the treatment of infectious diseases an intervention applied to all individuals in a community may be more effective than treatment applied to select (randomized) individuals within the community since it may reduce the possibility of re-infection.


Authors should always identify any cluster-randomized trials in a review and explicitly state how they have dealt with the data. They should conduct sensitivity analyses to investigate the robustness of their conclusions, especially when ICCs have been borrowed from external sources (see Chapter 9, Section 9.7). Statistical support is recommended.