Designing stepped wedge trials to evaluate physical activity interventions in schools: methodological considerations

Abstract Background In a stepped wedge design, schools are randomised to a sequence of measurements, with each sequence transitioning to intervention status at a different time. There are several advantages to such designs, including increased statistical power, logistical benefits and the ability to explore change over time. However, stepped wedge designs have not previously been used to evaluate school-based physical activity interventions in children. This paper aimed to explore the feasibility of this design, by identifying school constraints, balancing these with statistical considerations and exploring the power of this chosen design under different scenarios. Methods We conducted three interlinked studies, with the results from one informing the next. Study 1 was a qualitative study to identify school constraints that inform the choice of stepped wedge configuration. Study 2 used simulation to choose a configuration that balanced these school constraints and statistical properties. Study 3 explored the statistical power for the chosen design for different school and pupil sample sizes, using an open cohort design (a mixture of new and repeated pupils). Results School staff considered the proposed data collection feasible, and supported a maximum of 3–4 measurements per year and an implementation period of one school term. Study 2 therefore considered incomplete stepped wedge designs with five steps. Statistically, the best designs had a mix of control and intervention measurements in terms 2–4 and a spread of measurements across the whole study duration. Power depended on a combination of the overall recruitment rate and the retention rate. For 20 schools with an eligible class size of 30 pupils, we would be able to detect a 6 min difference in average weekday moderate-to-vigorous physical activity with 80% power, provided there were > 50% of pupils measured per school at each time. A similarly powered cluster randomised controlled trial would require 42 schools. Conclusion Stepped wedge trials are a viable design for evaluating school-based physical activity interventions. Incomplete designs, where not all schools are measured at each point, offer the flexibility to work around practical constraints.