Sample size determination for stepped wedge cluster randomized trials in pragmatic settings

Research output: Contribution to journalArticlepeer-review

Abstract

The stepped-wedge cluster randomized design has been increasingly employed by pragmatic trials in health services research. In this study, based on the GEE approach, we present closed-form sample size calculation that is applicable to both closed-cohort and cross-sectional stepped wedge trials. Importantly, the proposed method is flexible to accommodate design issues routinely encountered in pragmatic trials, such as different within- and between-subject correlation structures, irregular crossover schedules for the switch to intervention, and missing data due to repeated measurements over prolonged follow-up. The closed-form formulas allow researchers to analytically assess the impact of different design factors on sample size requirement. We also recognize the potential issue of limited numbers of clusters in pragmatic stepped wedge trials and present an adjustment approach for underestimated variance of the treatment effect. We conduct extensive simulation to assess the performance of the proposed sample size method. An application example to a real clinical trial is presented.

Original languageEnglish (US)
Pages (from-to)1609-1623
Number of pages15
JournalStatistical Methods in Medical Research
Volume30
Issue number7
DOIs
StatePublished - Jul 2021

Keywords

  • Clinical trials
  • GEE
  • pragmatic
  • sample size
  • stepped wedge

ASJC Scopus subject areas

  • Epidemiology
  • Statistics and Probability
  • Health Information Management

Fingerprint

Dive into the research topics of 'Sample size determination for stepped wedge cluster randomized trials in pragmatic settings'. Together they form a unique fingerprint.

Cite this