All Times EDT
Virtual
Pragmatic Cluster-Randomized Trials for the Fight Against TB: Design and Analytical Challenges (309562)
*Bareng Aletta Sanny Nonyane, Johns Hopkins Bloomberg School of Public HealthKeywords: cluster randomized trials, Tuberculosis, trial design
Background: Cluster randomized trials (CRT) have been widely used to evaluate strategies for increasing tuberculosis (TB) screening and diagnosis, and the feasibility of integrating those strategies into national healthcare systems. This study reviews associated statistical design and analytical challenges. Methods: A review of CRTs for TB research published over the last 15 years. Results: Parallel arm CRT designs were statistically most efficient, stepped-wedge designs were chosen for ethical reasons, and matched and factorial designs were chosen due to resource constraints. Interventions included improving screening protocols; contact tracing strategies; and improving preventative treatment uptake. Challenges included limited sample sizes, withdrawal of clusters and changes in clinical guidelines. Covariate-constrained, stratified and simple randomization were used. Cluster-level analyses were most commonly used. Discussion: CRTs have contributed towards WHO’s sustainable development goal to end the TB epidemic by 2030. Good stakeholder collaboration and informative baseline data facilitate efficient design, mitigation of interruptions and minimize statistical challenges.