What is an adequate `pilot` study leading to a stepped-wedge cluster randomized trial


I am wondering what would be an adequate pilot study that would inform about the feasibility of a large scale stepped-wedge cluster-randomized controlled trial. I have two consideration/questions:

  1. Would a simple before-after study be enough to demonstrate that the planned intervention can effectively be rolled-in in an hospital or a proper pilot with multiple clusters is needed?

  2. Would the clusters included in the pilot study could be also included in the definitive trial provided that the intervention as been rolled-out of the hospital for a pre-defined period? I realize this might depend largely on the nature of the studied intervention. I suppose that if a carry-over effect is suspected, this will affect the results in this cluster. (e.i. the attendings acquire a new prescription habit during the pilot study) Of course, this implies that the more you include clusters in the pilot trial, the less clusters is left for the main trial.

Happy to hear your thoughts.