Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

A simulation study of odds ratio estimation for binary outcomes from cluster randomized trials.

Obioha C Ukoumunne1, John B Carlin, Martin C Gulliford

  • 1Clinical Epidemiology and Biostatistics Unit, Murdoch Children's Research Institute and Department of Paediatrics, University of Melbourne, Australia. obioha.ukoumunne@mcri.edu.au

Statistics in Medicine
|December 13, 2006
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

The Ideal Trial: Defining Causal Estimands that Balance Relevance and Feasibility in Target Trial Emulations and Actual Randomized Trials.

Epidemiology (Cambridge, Mass.)·2026
Same author

"Identifying variables that independently predict…" is not a well-defined research task.

Journal of clinical epidemiology·2025
Same author

Causal Machine Learning Methods and Use of Cross-Fitting in Settings With High-Dimensional Confounding.

Statistics in medicine·2025
Same author

On the Uses and Abuses of Regression Models: A Call for Reform of Statistical Practice and Teaching.

Statistics in medicine·2025
Same author

Rejoinder to Commentaries on: On the Uses and Abuses of Regression Models: A Call for Reform of Statistical Practice and Teaching.

Statistics in medicine·2025
Same author

Primary Care and Linked Secondary Care Encounters for Foot and Ankle Problems in Children and Young People: A Population-Based Cohort Study in England.

Journal of foot and ankle research·2025

For cluster randomized trials, the generalized estimating equations (GEE) method offers reliable estimation and confidence interval coverage, especially with adjustments for fewer clusters. Other methods perform well under specific conditions like large clusters and low correlation.

Area of Science:

  • Biostatistics
  • Clinical Trials Methodology
  • Epidemiology

Background:

  • Cluster randomized trials (CRTs) are essential for evaluating interventions in group-level settings.
  • Accurate statistical analysis of binary outcomes in CRTs is crucial for reliable intervention effect estimation.
  • Various analytical methods exist, but their performance under different design scenarios requires thorough evaluation.

Purpose of the Study:

  • To compare the accuracy of estimation and confidence interval coverage of different statistical methods for binary outcomes in CRTs.
  • To identify the most robust methods across a range of simulated trial design parameters.

Main Methods:

  • Simulation study comparing generalized estimating equations (GEE), unweighted cluster-level mean difference (CL/U), weighted cluster-level mean difference (CL/W), and cluster-level random effects linear regression (CL/RE).

Related Experiment Videos

  • Methods evaluated for population-averaged intervention effect on the log-odds scale.
  • Simulations varied the number of clusters, subjects per cluster, intraclass correlation coefficient (rho), and arm proportions.

    • Main Results:

    • The GEE method demonstrated generally acceptable properties, with near-nominal confidence interval coverage, particularly when adjusted for a small number of clusters.
    • CL/U and CL/W performed well when the number of subjects per cluster was large and rho was small.
    • CL/RE showed good performance in similar conditions, provided a t-distribution multiplier was used for confidence intervals with few clusters.

    Conclusions:

    • The GEE method is a reliable choice for analyzing binary outcomes in CRTs, especially with adjustments for limited clusters.
    • Cluster-level methods (CL/U, CL/W, CL/RE) are suitable under specific conditions of large cluster size and low intraclass correlation.
    • All cluster-level methods may exhibit poor performance when cluster size is small and intraclass correlation is high.