Keywords: Trial design, time trend, power
Stepped wedge designed trials are a popular alternative to cluster randomized trials because roll-out of the intervention to sites is staggered over time and all sites receive the intervention by the final roll-out period. Also, multiple roll-out periods allow the intervention to be refined to become more effective over time. In a traditional stepped wedge design, it is necessary to adjust for the effect of time on outcomes; however, a temporal increase in treatment effect can be partially confounded with time trend since all sites will eventually receive the intervention. This limitation may be overcome by modifying the design to include sites that act strictly as controls and whose data are often easily obtained through the electronic health record. Through simulation, we explore the reduction in bias in the estimate of a treatment effect that increases over time by including these strictly control sites. We further assess how power in the study increases through both inclusion of additional sites and more accurate estimation of treatment effect. We illustrate use of such designs in a trial aimed at monitoring and improving medication adherence in heart disease patients.