Ensuring Statistical Power in Step-Wedge Designs with Unequal Cluster Sizes

Jan 22, 2018 - 12:00pm to 1:00pm
700-828 W. 10 Ave., VGH Research Pavilion
Hubert Wong

Phase II of the TEC4Home project is a randomized trial of the effectiveness of Home Health Monitoring in patients with COPD. Patients will be recruited from 22 hospitals across BC with hospitals randomized using a step-wedge design.  If the randomization is unconstrained, large variability across hospitals in the number of patients enrolled increases the risk that the trial will be underpowered. I will discuss the heuristics underlying this problem and potential solutions.
Learning Objectives:

  1. Complex study designs involve complex power calculations.
  2. The risk that the realized power (achieved in the actual study) will be lower than the (pre-study) target power should be considered when designing a randomization plan.