Summer School 2023: Design and analysis of longitudinal cluster randomised trials
Cluster randomised trials are often conducted when randomising individuals to treatments isn’t possible. This can occur when the treatment is delivered at the cluster level, but we still measure outcomes on individuals. For example, if we want to assess the impact of a new hospital-wide cleaning program on the number of patients treated at that hospital who acquire infections, we can’t randomise the individual patients to cleaning programs, and instead must randomise the entire hospital. Longitudinal cluster randomised trials extend cluster randomised trials over time, and clusters may switch between the control and intervention conditions. The stepped wedge design and cluster randomised crossover design are key examples of these trials.
In this workshop, you will learn how to design and analyse longitudinal cluster randomised trials, including the stepped wedge and cluster randomised crossover design. You will learn about the latest developments in these trials, including complex within-cluster correlation structures, “incomplete” stepped wedge designs such as the staircase design, and what can go wrong when these trials are not designed and analysed appropriately. Computer practicals will give you hands-on experience in designing and analysing longitudinal cluster randomised trials, using Stata or R.
The target audience is statisticians and researchers with some statistical background. Knowledge of mixed effects models will be assumed.