BIOM&S Seminar: Analysis of the stepped wedge cluster randomized trial with small number of clusters: An empirical assessment of available methods for a binary outcome

SPEAKER:   Stephanie Dixon, ICES Scientist, Institute for Clinical Evaluative Sciences (ICES) Western, London Ontario

ABSTRACT:

Stepped wedge cluster randomized trials (SW-CRTs) commonly include a small number of clusters: a recent systematic review found that nearly half of published SW-CRTs have fewer than 10 clusters. The feasibility and implications for inferences of various individual- and cluster-level methods is explored using data from the Register-Now SW-CRT, recently completed in Ontario, Canada. This trial evaluated an intervention to promote registration for organ donation in six family physician offices. The primary outcome was the proportion of patients who are registered organ donors seven days following their office visit. Empirical estimates are presented from 3 cluster-level methods and 4 individual-level methods. The talk will highlight advantages, disadvantages and analytical challenges when producing intervention estimates on different scales and applying small sample corrections.