A classification of incomplete stepped wedge cluster randomised trial designs

Stepped wedge designs are cluster randomised trial designs for testing new interventions through a phased rollout of the intervention across clusters. In many trial settings, a complete stepped wedge design can place a substantial burden on clusters by requiring data collection over the entire trial duration. Incomplete stepped wedge designs, for example, "staircase designs," where clusters collect data for a limited duration of time, are frequently used to overcome this issue. However, the term "incomplete stepped wedge design" is also used when transition or implementation periods are incorporated in an otherwise complete stepped wedge design. Thus, "incomplete" is an imprecise term that does not convey any specific design characteristics.

We introduce terminology to enable a classification of incomplete stepped wedge designs, with a particular focus on different variants of staircase designs.

We provide definitions of variants of staircase designs with their sub-variants. We also describe two types of mid-sequence breaks in data collection: inter-treatment breaks occurring between control and intervention durations of data collection (i.e. implementation or transition periods), and intra-treatment breaks occurring between durations of data collection under the same treatment condition. These breaks introduce different forms of incompleteness into designs.

Through the use of this terminology and associated classifications, trialists can more clearly communicate the way in which their "incomplete stepped wedge design" is incomplete, and health researchers can classify trials in systematic reviews involving complete stepped wedge designs and any incomplete variants.