Efficient cluster-randomised trials
In cluster randomised trials, clusters of participants are randomly assigned to a treatment or control group or, in a stepped wedge trial, randomly assigned to a specific time delay until the treatment is implemented. Our work focuses on how to optimise the design of these trials by minimising the number of clusters and/or measurements required, and to demonstrate new ways of understanding and reporting these trials.
What are cluster randomised trials and why are they relevant?
In a cluster randomised trial, pre-existing groups of individuals, called clusters, are randomised to treatment or control groups for the duration of the trial. These clusters could be, for example, hospitals or communities.
In contrast, in a longitudinal design, clusters are randomised to a sequence of exposures (treatment or control) that may vary over time. For example, in a stepped wedge trial, clusters being in the control exposure and the sequence determines the length of the delay until the intervention is implemented.
Cluster randomised trials are very important for evaluating interventions that are carried out at a cluster level, such as staff training, community health promotion, or hospital policies.
What have we done?
Our work focuses on optimising the design of cluster randomised trials and understanding and reporting characteristics of stepped wedge trials. More recently, we have defined estimands for cluster randomised trials.
Our work on efficient cluster-randomised trials has:
- Reviewed sample size methods for cluster randomised trials
- Identified optimal designs for cluster randomised trials with baseline data
- Identified optimal designs where the intra-cluster correlation differs between trial arms
- Described key features in the stepped wedge trial design and sample size calculation
- Identified optimal stepped wedge designs
- Developed methods for sample size calculation where clusters are open cohorts
- Clarified the implications of continuous recruitment for stepped wedge trials
- Contributed to the CONSORT extension for reporting stepped wedge trials
- Reviewed practical challenges faced in stepped wedge trials
- Defined estimands for cluster randomised trials and identified the implications of informative cluster size.
How will this make a difference?
Our work helps understanding different cluster randomised trial designs. We have also provided tools to identify optimal designs that increase efficiency and lead to quicker and cheaper identification of interventions that are effective.
Resources:
Publications:
Articles concerning the design of parallel group cluster randomised trials
- Methods for sample size determination in cluster randomized trials (2015)
- Optimal design of cluster randomized trials allowing unequal allocation of clusters and unequal cluster size between arms (2021)
Articles concerning the design of cluster randomised trials with baseline data
- Cluster randomised trials with different numbers of measurements at baseline and endline: Sample size and optimal allocation (2019)
- Optimal design of cluster randomised trials with continuous recruitment and prospective baseline period (2021)
Articles concerning the design of stepped wedge trials, reporting and challenges
- Designing a stepped wedge trial: three main designs, carry-over effects and randomisation approaches (2015)
- Reporting of stepped wedge cluster randomised trials: extension of the CONSORT 2010 statement with explanation and elaboration (2018)
- Recruitment and implementation challenges were common in stepped-wedge cluster randomized trials: Results from a methodological review (2022)
Articles defining estimands and the implications of informative cluster size