Variance and Dissent - Stepped Wedge Design
- We thank Kotz et al.  for the interesting commentary on our article . We agree with the authors in a number of areas. First, there is no question that in terms of the hierarchy of evidence, a classic cluster randomized controlled trial (CRCT) is preferred to the stepped wedge CRCT for many of the reasons the authors outline. Second, in an ideal world, an unevaluated intervention should not be given to all potential recipients because we do not know whether it is effective. However, there are some circumstances where the stepped wedge CRCT is preferable to the alternative: no randomized trial at all.
- In a conventional randomized controlled trial (RCT), participants are randomly allocated to either an intervention or a control/comparison condition. However, for certain types of interventions, it is not valid to randomize participants at an individual level because contamination between the intervention and control groups can occur. In those situations, an alternative approach is to randomize entire groups of participants (clusters) to the two conditions. Studies using this type of design are typically called cluster RCTs.
- We thank Mdege et al.  for their response to our critical appraisal  of the stepped wedge design (SWD). They agree with us that the SWD has many disadvantages compared with the cluster randomized controlled trial (cluster RCT) and that the cluster RCT is preferable in most circumstances. However, they also maintain that the SWD may be useful when the alternative is to conduct no randomized trial at all and that some of the cluster RCT variants that we suggest in our article are in fact variants of the SWD.