DESCRIPTION
The stepped wedge cluster randomized design has received increasing attention in pragmatic trials and implementation science research. The key feature of the design is the unidirectional crossover of clusters from the control to intervention conditions on a staggered schedule, which induces confounding of the intervention effect by time. While such designs first appeared in the 1980s, the associated statistical methods were not formally introduced until 2007. Since then, a variety of novel methods have been introduced for improving the design, analysis and conduct of these trials. In this talk, we explore these new methods under a unified perspective. We describe essential ingredients in analytical models for stepped wedge designs and discuss their implications for study planning, data analyses and reporting. Recent trial examples, challenges and opportunities are also discussed throughout this talk.
LEARNING OBJECTIVES
- To describe common variants of stepped wedge design and when such designs are appropriate choices
- To describe essential ingredients in analytical models for stepped wedge designs, and to understand key concepts and steps for determining the sample sizes
- To recognize challenges and opportunities in using stepped wedge designs for pragmatic trials
PRESENTER(S)
Fan Li, PhD, is an Assistant Professor in the Department of Biostatistics at the Yale School of Public Health. I am also a faculty member in the Center for Methods in Implementation and Prevention Science (CMIPS) and the Yale Center for Analytical Sciences (YCAS). My research focuses on statistical methodology to evaluate comparative effectiveness with real-world data arising from pragmatic clinical trials, observational studies or a combination of both. I am an expert in the design, monitoring, analysis and interpretation of parallel-arm, crossover and stepped-wedge cluster randomized studies, which are increasingly common in pragmatic trials embedded in the health care delivery systems. I have also been developing new propensity score methods to enable causal inference with real-world observational data, with a focus on improving overlap and internal validity.