Abstract:
In trials of group administered treatments clustering may occur due to the nesting of patients in therapy groups. Heterogeneity of the clustering effect may arise particularly where a group administered therapy is compared with a placebo or pharmacological control. Where individual randomisation of patients is used, this leads to a partially nested design with clustering in one arm but not the comparator. Where two group administered treatments are compared this leads to a fully nested design. Alternatively, patients can be randomised as groups to either intervention resulting in a cluster randomised trial design. This talk will compare the robustness of the analysis methods, and consider some of the practical implications for trialists.
Registration:
No need to register, just turn up.