From: Implementation of a stepped wedge cluster randomized trial to evaluate a hospital mobility program
Design decision | Function QUERI SW-CRT | Advantages | Disadvantages |
---|---|---|---|
Time period length | 3 months | Shorter length and larger number of time periods increased power | Longer time period length would have given sites longer window to launch program |
Number of sites per sequence | 2 | Allowed study team to provide significant implementation support to each site; study conducted faster than if only 1 site/time period | Power implications; maximum power with one cluster per sequence |
Recruitment and randomization strategy | Recruited and randomized clusters to sequences in 2 separate blocks | Reduced maximum possible time from enrollment to program launch time; ability to start study before all sites recruited; potentially fewer dropouts; ability for earlier sites to serve as mentors for later sites | Not able to collect patient-reported data from block 2 sites in time periods before enrollment Potential confounding of treatment with block Full statistical implications of blocked randomization for SW-CRT have not been studied. Calendar time versus time since randomization may be an issue for blocks |
Complete vs. incomplete design and type | Incomplete design—outcome data during implementation interval not used for evaluation | Avoided including data when program is partially implemented | Power implications; loss of data from time periods not included in evaluation |