Biostatistics Group Seminar
School of Primary, Community and Social Care
Keele University
Title: Treatment effect and time effect heterogeneity in stepped-wedge
cluster randomised trials
Authors: James Martin and Karla Hemming
Speaker: James Martin, Institute of Applied Health Research, University
of Birmingham, UK
Date: Thursday 5th December 12:00-13:30
Venue: Dinwoodie Lecture Theatre, David Weatherall Building
https://www.keele.ac.uk/connect/howtofindus/maps/keele-campus-guide-colour.pdf
Abstract: In a standard parallel cluster randomised trial, the effect of
time is balanced across intervention and control arms. In a
stepped-wedge cluster randomised trial (SW-CRT), calendar time is
associated with the intervention exposure and potentially the outcome.
Standard methods to estimate the power in a SW-CRT assume that the
underlying time trend is consistent across all clusters, and that the
observed treatment effect is uniform across all clusters. Increasingly,
SW-CRTs are being conducted across multiple countries or across other
settings where these assumptions are expected to be violated.
We begin by outlining the different cluster randomised trial designs,
and discuss why time trends are important in a stepped-wedge trial
design. Through the use of a recently conducted study, we illustrate
potential approaches for evaluating a SW-CRT with treatment effect and
time effect heterogeneity across clusters, and discuss the implications
of using standard methods.
We discuss approaches that could be used to help design future trials
that allow for strata or cluster specific secular trends and strata or
cluster specific intervention effects.
Forthcoming seminar: Gabriela Czanner, School of Applied Mathematics,
Liverpool John Moores University, Tuesday 14th January 12:00-13:00,
Title TBC
--
Dr Ivonne Solis-Trapala
Deputy Director, NIHR Research Design Service, West Midlands (Keele Hub)
Senior Lecturer in Medical Statistics
Biostatistics Group and Keele Clinical Trials Unit
School of Primary, Community and Social Care
Room 1.104, David Weatherall Building
Keele University, Staffordshire, ST5 5BG
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|