Bayes at Red Door Analytics
Alessandro Gasparini is a statistician at Red Door Analytics.
He will talk about his ongoing projects incorporating Bayesian joint modelling and advanced Bayesian survival analysis.
Bayesian Joint Models for Longitudinal Cluster Randomised Trials with Informative Dropout
We recently introduced a frequentist joint modelling approach to account for informative dropout in longitudinal stepped wedge cluster-randomized trials (Gasparini et al., Statistics in Medicine, 2025). This approach combines a linear mixed-effects model for the longitudinal outcome of interest with a survival model for the dropout process, jointly estimating both components. We now extend this methodology to longitudinal cluster-randomized trials (a special case of the stepped wedge design) and adopt a Bayesian framework to enhance the interpretability of trial results and increase statistical power by incorporating prior knowledge or beliefs, when available. We illustrate the Bayesian joint modelling approach in practice using data from the CARING trial, a cluster-randomized controlled trial evaluating an integrated community strategy to promote maternal nutrition and paediatric outcomes in rural eastern India, where undernutrition is highly prevalent.