NIH Pragmatic Trials Collaboratory - Choosing the Right Study Design (June 19, 2020): 19-min webinar (37-slide presentation; slides 2, 4-6, 30-62)

• Summary: Dr. Elizabeth Turner introduces individually randomized trials and cluster randomized trials and discusses reasons to randomize at the cluster-level vs. the individual-level. A flow diagram is used to depict the decision-making process for doing an individual- vs. a cluster-level randomized trial. Conceptual breakdowns of parallel vs. stepped-wedged cluster trials, as well as arguments for (and against) doing a stepped-wedged cluster trial, are also provided.

Giraudeau B, et al. Why and when should we cluster randomize? J Epidemiol Popul Health. 2024 Feb;72(1):202197. (6-page paper)

• Summary: When designing a trial, it is important to choose a robust study design and randomization scheme. It is up to investigators and their research teams to justify their choice of study design, given the scientific, practical, and logistical constraints of the project. This paper provides a detailed explanation of possible reasons for adopting cluster randomization, emphasizing that certain justifications (e.g., avoiding obtaining individual consent) are not acceptable reasons to adopt a cluster randomized design.

NIH Pragmatic Trials Collaboratory (Living Textbook): Section 6 - Choosing Between Cluster and Individual Randomization (Chapter: Design - Experimental Designs and Randomization Schemes): 1-page website.

• Summary: Given that careful consideration is needed before settling on a particular approach to randomization, this brief resource provides information to help a researcher consider whether the unit of randomization in their trial should be an individual or a cluster.

National Cancer Institute - Division of Cancer Control & Population Sciences. Cluster Randomized Designs in Cancer Care Delivery Research - Short Course (Session 2 - Justification for a Cluster Randomized Design, May 3, 2022): 19-minute webinar.

• Summary: Dr. Karla Hemming addresses the justification for choosing a cluster trial design, situating that justification within the context of risk of bias, study aims related to assessing direct and indirect effects, contamination, logistics, and unit of delivery. Examples are used to illustrate the concepts. The implications of post-randomization recruitment on risk of bias and how to mitigate such biases are also reviewed.

Easter C, et al. Cluster randomized trials of individual-level interventions were at high risk of bias. J Clin Epidemiol. 2021 Oct;138:49-59. (11-page paper)

• Summary: Describes the prevalence of risks (according to the Cochrane Risk of Bias tool) of bias in cluster-randomized trials of individual-level interventions and finds that many cluster-randomized trials evaluating individual-level interventions appear to be at risk of bias. Therefore, due to the risks of identification and recruitment bias, opting for a cluster design when individual randomization would be feasible warrants a strong justification.