NIH: Office of Disease Prevention

Menu


David M. Murray, Ph.D.
Associate Director for Prevention
Director, Office of Disease Prevention

Resources

A free, 7-part, online course presented by Dr. David M. Murray that provides a detailed guide to designing and analyzing group-randomized trials (GRTs). The course includes video presentations, slide sets, suggested reading materials, and guided activities.

Course Videos


Part 1: Introduction and Overview

Recorded on September 15, 2016

Part 1 provides an introduction and overview of the three kinds of randomized trials and their distinguishing characteristics. The differences between group-randomized trials (GRTs) and individually randomized group-treatment trials (IRGTs) are discussed.

Read Video Transcript for Part 1


Part 2: Designing the Trial

Recorded on September 15, 2016

Part 2 considers the design of group-randomized trials (GRTs), with a focus on internal and statistical validity, the factors that influence precision, and the major features that define variations in group-randomized trial designs, including cohort vs. cross-sectional designs; matching, stratification, and constrained randomization; and time as a factor.

Read Video Transcript for Part 2


Part 3: Analysis Approaches

Recorded on September 15, 2016

Part 3 provides analysis approaches to group-randomized trials (GRTs), including a review of model-based approaches, randomization tests, marginal methods like generalized estimating equation (GEE), methods for complex survey samples, two-stage methods, methods for unbalanced designs, methods used with constrained randomization, and methods for individually randomized group-treatment trials (IRGTs).

Read Video Transcript for Part 3


Part 4: Power and Sample Size

Recorded on September 15, 2016

Part 4 explores power and sample size for group-randomized trials (GRTs). The presentation focuses on Cornfield's two penalties of extra variation and limited degrees of freedom (df); strategies to reduce extra variation and increase df; the seven steps involved in any power analysis; estimating the intraclass correlation coefficient (ICC); and adaptations required for unbalanced designs. The course also provides an example of a sample size calculation for a simple GRT.

Read Video Transcript for Part 4


Part 5: Examples

Recorded on September 15, 2016

Part 5 provides examples of group-randomized trials (GRTs) from the Health Care Systems Collaboratory, a project funded by the NIH involving nine independent projects, eight of which are GRTs. All rely on electronic health records as their primary source of data, but they vary in their interventions, outcomes, and design and analytic methods.

Read Video Transcript for Part 5


Part 6: Review of Recent Practices

Recorded on September 15, 2016

Part 6 reviews recent practices in group-randomized trials (GRTs) based on a paper published by Murray, et al., in 2008 in the Journal of the National Cancer Institute. The presentation also explores recent practice[s] for individually randomized group-treatment trials (IRGTs) based on a paper published by Pals et al. in 2008 in the American Journal of Public Health. Both papers review published studies and critique the methods used for design, sample size calculation, and analysis.

Read Video Transcript for Part 6


Part 7: Alternative Designs

Recorded on September 15, 2016

Many alternative designs have been proposed to evaluate interventions delivered to or through groups, including multiple baseline designs, time series designs, quasi-experimental designs, stepped wedge designs, and regression discontinuity designs. Part 7 examines these designs in comparison to group-randomized trials (GRTs) and individually randomized group-treatment trials (IRGTs) with a focus on internal validity and cost in terms of time and participants.

Read Video Transcript for Part 7