clusterrandomized

Cluster randomized, or cluster-randomized trials (CRTs), are experiments in which intact groups or clusters are assigned to intervention or control conditions rather than individual participants. Clusters may be defined by geographic units such as villages or districts, institutions such as schools or clinics, or other natural groupings. Interventions are often delivered at the cluster level, while outcomes are typically measured at the individual level. CRTs are frequently used when individual randomization is impractical or when there is a risk of contamination among participants within the same cluster.

Key design considerations include the distinction between the unit of randomization and the unit of analysis. The number of clusters and the average cluster size strongly influence power; outcomes within a cluster tend to be correlated, quantified by the intra-cluster correlation coefficient (ICC). This correlation reduces the effective sample size, described by the design effect, and necessitates more clusters rather than merely more individuals. Unequal cluster sizes complicate sample size calculations and analyses.

Common designs encompass parallel CRTs, where clusters are randomized to arms and followed over time, and stepped-wedge CRTs, in which all clusters eventually receive the intervention but at different time points. Other designs include factorial CRTs and, less frequently, cluster-based cross-over trials with attention to carryover effects.

Analysis methods must account for clustering. Approaches include mixed-effects (multilevel) models with random cluster effects, generalized estimating equations with robust standard errors, or cluster-level summaries. Intention-to-treat principles are typically applied, and reporting should include ICC estimates, design effects, number of clusters, and handling of baseline imbalances.

Ethical and practical considerations involve consent at both cluster and individual levels, potential equitable access to interventions, and adherence to reporting guidelines such as CONSORT extensions for cluster trials.