Cluster randomised trials with repeated cross sections: alternatives to parallel group designs | Synapse