A Bayesian hierarchical push-forward method for cluster Randomized Controlled Trial power calculations in complex humanitarian settings
Abstract
The design and evaluation of public health interventions in humanitarian crises demand rigorous yet flexible methodological approaches. Cluster-randomized controlled trials (cRCTs) are increasingly used to generate evidence in such settings, but trial conduct is frequently complicated by displacement, resource constraints, and cluster dropout. Although Bayesian methods provide a natural framework for hierarchical modeling, probabilistic uncertainty quantification, and the integration of prior knowledge, their use in cRCT design remains limited, particularly with respect to anticipating missingness. This paper introduces a Bayesian design strategy tailored for cRCTs in complex settings, motivated by a forthcoming trial in Cox's Bazar, Bangladesh, that will assess a community health worker adherence intervention for cardiovascular disease and hepatitis C. Our approach incorporates prior information on cluster and individual-level dropout into power and sample size calculations via a hierarchical model and push-forward distribution framework. Through simulation studies, we evaluate the performance of the proposed method under varying levels of missingness and correlation structures. The approach is then applied to inform the design of the Bangladesh trial. We disseminate our methods via a publicly-available R Shiny app to promote accessibility and usability to investigators. By formalizing Bayesian strategies that anticipate data loss, this work provides a foundation for more robust and efficient cRCTs in humanitarian contexts; therefore, strengthening the evidence base for health interventions in resource-limited and crisis-affected populations.
Related
- Bayesian Hierarchical Models for Longitudinal Meta-Analysis in Sports Science: A Practical Tutorial Using brms
- Power and sample size calculations for causal mediation analysis with a binary mediator in randomized trials
- A Bayesian hierarchical model for meta-analysis
- Behavioral Mapping of Hierarchical Dynamical Models Accelerates Bayesian Parameter Inference
- A Bayesian hierarchical model for preferentially sampled spatial extremes data: an application to air quality
BibTeX
@misc{elliott2026bayesian,
title = {{A Bayesian hierarchical push-forward method for cluster Randomized Controlled Trial power calculations in complex humanitarian settings}},
author = {Margaret Elliott and Sonia Jain and Adam Richards and Brendan Fries and Ravi Goyal},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.21022},
eprint = {2610.21022},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.21022},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - A Bayesian hierarchical push-forward method for cluster Randomized Controlled Trial power calculations in complex humanitarian settings
AU - Margaret Elliott
AU - Sonia Jain
AU - Adam Richards
AU - Brendan Fries
AU - Ravi Goyal
PY - 2026
DA - 2026/10/14
PB - arXiv
DO - 10.48550/arxiv.2610.21022
UR - https://arxiv.org/abs/2610.21022
AB - The design and evaluation of public health interventions in humanitarian crises demand rigorous yet flexible methodological approaches. Cluster-randomized controlled trials (cRCTs) are increasingly used to generate evidence in such settings, but trial conduct is frequently complicated by displacement, resource constraints, and cluster dropout. Although Bayesian methods provide a natural framework for hierarchical modeling, probabilistic uncertainty quantification, and the integration of prior knowledge, their use in cRCT design remains limited, particularly with respect to anticipating missingness. This paper introduces a Bayesian design strategy tailored for cRCTs in complex settings, motivated by a forthcoming trial in Cox's Bazar, Bangladesh, that will assess a community health worker adherence intervention for cardiovascular disease and hepatitis C. Our approach incorporates prior information on cluster and individual-level dropout into power and sample size calculations via a hierarchical model and push-forward distribution framework. Through simulation studies, we evaluate the performance of the proposed method under varying levels of missingness and correlation structures. The approach is then applied to inform the design of the Bangladesh trial. We disseminate our methods via a publicly-available R Shiny app to promote accessibility and usability to investigators. By formalizing Bayesian strategies that anticipate data loss, this work provides a foundation for more robust and efficient cRCTs in humanitarian contexts; therefore, strengthening the evidence base for health interventions in resource-limited and crisis-affected populations.
ER -