Agenda
All times are Eastern Daylight Time (EDT)
|
Day 1
Tuesday, May 3, 2022
|
|
|---|---|
| Time (EDT) | Agenda |
| 11:00 am - 11:10 am |
Welcome and Course Overview
Greetings from NCI Senior Leadership
|
Session 111:10 am - 12:35 pm |
Introduction to Cluster Randomized TrialsThis session introduces the foundational elements to be considered in designing a cluster randomized trial (CRT) versus a traditional patient randomized trial. The session will introduce the audience to the key concepts and considerations for a cluster randomized trial, including the range of different types of cluster randomized designs. |
|
S1.a Presenter: |
An Introduction to the Parallel Cluster Randomized Trial This session introduces the CRT design including key features; different types of interventions (cluster-level, professional-level, individual-level); broad outcome types (differentiating implementation outcomes from clinical outcomes); and participant recruitment. (20 minutes) |
|
S1.b Presenter: |
An Overview of the Statistical Implications of Cluster Randomization This session reviews the statistical considerations associated with cluster randomization (including intra-cluster correlation and degrees of freedom corrections for small samples) and summarizes how and when CRTs are being utilized in the current scientific literature. (21 minutes) |
|
S1.c Presenter: |
An Introduction to Multiple Period Cluster Randomized Trial Designs This session introduces the multiple period CRT design, including the different types of multiple period CRT designs, such as parallel arm with baseline period and cluster cross-over designs. An introduction to cohort and cross-sectional sampling, and an exploration of the trade-offs between statistical efficiency and trial logistics are provided. (15 minutes) |
|
S1.d Presenter: |
An Introduction to the Stepped-Wedge Cluster Randomized Trial This session introduces the stepped-wedge cluster randomized trial (SW-CRT) design and provides an overview of the different types of SW-CRT design (cohort, cross-sectional sampling). It also describes associated methodological complexities (e.g., time confounding, longitudinal correlations, time-varying effects). (28 minutes) |
| 12:35 pm-12:20 pm |
Question and Answer Session Moderator: Dr. Sandra Mitchell |
| 12:50 pm-1:20 pm | BREAK |
Session 21:20 pm-2:55 pm |
Risk of Bias and Design JustificationThis session helps researchers determine and justify a CRT design (e.g., parallel CRT, SW-CRT or other) appropriate for particular settings. Alternatives/extensions within the family of CRT designs, as well as the implications of these decisions for internal validity (risk of bias) and pragmatic considerations, are also addressed. |
|
S2.a Presenter: |
Risk of Bias in Parallel Cluster Randomized Trials This session addresses risk of bias in parallel CRTs, as defined by the Cochrane Risk of Bias tool (RoB2.0), including strategies to prevent or mitigate bias. The prevalence of these risks in contemporary samples of CRTs is reviewed, providing illustrative examples of the most common sources of bias. (24 minutes) |
|
S2.b Presenter: |
Justification for a Cluster Randomized Design This session addresses the justification for choosing a cluster trial design, situating that justification within the context of risk of bias; study aims related to gauging direct and indirect effects; as well as contamination; trial logistics; and unit of delivery. Concepts are illustrated with examples. The implications of post-randomization recruitment on risk of bias and how to mitigate such biases are reviewed. (20 minutes) |
|
S2.c Presenter: |
Risk of Bias in Stepped-Wedge Cluster Randomized Trials This session addresses additional risks of bias in SW-CRTs and other multiple period designs, including consideration of secular trends, contamination, and randomization deviations. (22 minutes) |
|
S2.d Presenter: |
Justification for using a Stepped-Wedge Cluster Randomized Trial This session summarizes and provides illustrative examples of the justifications for using a stepped-wedge trial design, including social acceptability, ethical considerations, and sequential roll-out and other pragmatic/feasibility considerations. (23 minutes) |
| 2:55 pm-3:10 pm |
Question and Answer Session Moderator: Dr. Wynne Norton |
| 3:10 pm-3:25 pm | BREAK |
Session 33:25 pm-4:30 pm |
Trial Implementation and RandomizationThis session will help researchers identify common practical trial implementation issues that might need to be surmounted/avoided when conducting a cluster randomized trial (including parallel designs, stepped-wedge, and cluster cross-over designs). The session will also cover the pros and cons of different randomization methods. |
|
S3.a Presenter: |
Practical Design-Based Considerations in Cluster Randomized Trials Practical issues in the design of CRTs are addressed in this session, including the choice of cluster, transition periods, staggered starts, staircase designs, number of sequences, duration of study and the implications when using a parallel design, stepped-wedge design, or cluster cross-over design. The importance of hybrid and pilot studies are highlighted. (31 minutes) |
|
S3.b Presenter: |
Randomization Methods in Cluster Randomized Trials This session provides an overview of different types of restricted randomization in cluster trials (simple, stratified, paired, covariate constrained, minimization), including their comparative benefits (including variations that allow you to stagger the cluster starts); the importance of design-based adjustment; and implementation of randomization methods. (21 minutes) |
|
4:30 pm-5:00 pm |
Question and Answer Session Moderator: Dr. Sandra Mitchell |
| 5:00 pm |
Wrap Up and Planning for Day 2 Dr. Sandra Mitchell and Dr. Wynne Norton |
|
Day 2
Wednesday, May 4, 2022
|
|
| Time (EDT) | Agenda |
| 11:00 am-11:05 am |
Welcome to Day 2 Dr. Sandra Mitchell, Dr. Wynne Norton |
Session 411:05 am-1:05 pm |
Sample Size and Statistical EfficiencyThis session will allow researchers to design cluster randomized trials to maximize statistical efficiency, given the constraints of the setting. Some of these sessions are designed for methodologists and statisticians and will review methods for sample size estimation in CRTs, including the importance of the consideration of non-exchangeable correlation structures. |
|
S4.a Presenter: |
Statistically Efficient Cluster Randomized Designs This session addresses considerations for statistically efficient cluster trial designs including optimal choice of cluster size, number of clusters, number of crossovers, for parallel, stepped-wedge and cross-over designs. (28 minutes) |
|
S4.b Presenter: |
Sample Size Determination in Parallel Arm Cluster Randomized Trials This session reviews sample size calculations for parallel CRTs (including trials with continuous outcomes, binary outcomes, varying cluster sizes, intra-cluster correlations, small sample corrections). Concepts are illustrated with worked examples and practical resources, including implementation tools and sources for estimating the intra-cluster correlation, including rules of thumb. (32 minutes) |
|
S4.c Presenter: |
Sample Size Determination for Multiple Period Cluster Randomized Trials This session reviews sample size calculations for SW-CRTs and other multiple period CRT designs. The importance of non-exchangeable correlation structures, including cohort or cross-sectional designs, incomplete designs, conditional vs. marginal models, immediate vs. gradual treatment effects, different analytical approximations for non-continuous outcomes, magnitude of time effects, unequal cluster sizes, unequal allocation, cluster-treatment effect heterogeneity are addressed. (56 minutes) |
| 1:05 pm-1:20 pm |
Question and Answer Session Moderator: Dr. Wynne Norton |
| 1:20 pm-1:50 pm | BREAK |
Session 4 (continued)1:50 pm-2:50 pm |
Sample Size and Statistical Efficiency (continued) |
|
S4.d Presenter: |
Practical Trial Implementation: Sample Size Calculation for Multiple Period Cluster Randomized Trial Designs This session focuses on selection of a sample size calculator that supports different design features and illustrates sample size implications of different design features. Sample size calculation implemented using a variety of software platforms is demonstrated. (57 minutes) |
| 2:50 pm-3:05 pm |
Question and Answer Session Moderator: Dr. Wynne Norton |
Session 53:05 pm-4:50 pm |
Analysis of Parallel Cluster Randomized TrialsThis technical session is targeted to methodologists and statisticians, providing an in-depth treatment of the analysis of parallel CRTs, including model specification. |
|
S5.a Presenter: |
Estimands: Thinking Carefully About What We Are Trying to Estimate This session introduces basic considerations for analyzing CRTs, including specification of primary and secondary outcomes and the estimand of interest, intention to treat, covariate adjustment, and analysis such as odds ratios, relative risks, risk differences, and marginal vs conditional effects. (48 minutes) |
|
S5.b Presenter: |
Analysis of Parallel Arm Cluster Randomized Trials This session presents methods of analysis for parallel arm CRTs (continuous and binary outcomes; cluster-level methods, mixed models, generalized estimating equations), including practical applications. The session highlights the importance of small sample corrections and alternatives for estimating risk differences and relative risks (marginal standardization and modified Poisson). (51 minutes) |
| 4:50 pm-5:00pm |
Question and Answer Session Moderator: Sandra Mitchell |
| 5:00 pm |
Wrap Up and Planning for Day 3 Dr. Sandra Mitchell and Dr. Wynne Norton |
|
Day 3
Thursday, May 5, 2022
|
|
| 11:00 am-11:05 am |
Welcome to Day 2 Dr. Sandra Mitchell, Dr. Wynne Norton |
Session 5 (continued)11:05 am-11:30 am |
Analysis of Parallel Cluster Randomized Trials (continued) |
|
S5.c Presenter: |
Small Sample Corrections in the Analysis of Parallel Cluster Randomized Trials This session addresses small sample corrections in parallel CRTs (between-within, Kenward-Roger, Fay-Graubard, KC, etc.) and their comparative performance; and the use of randomization-based tests. (21 minutes) |
Session 611:30 am-12:20 pm |
Analysis of Cluster Randomized TrialsThis technical session is targeted to methodologists and statisticians, providing an in-depth treatment of the analysis of longitudinal cluster randomized trials, including model specification. |
|
S6.a Presenter: |
Methods of Analysis for Stepped-Wedge Cluster Randomized Trials This session focuses on mixed-effects regression for continuous outcomes accounting for complex correlation structures. Sample syntax for analysis in SAS, R and STATA is provided. The implications of misspecification of the correlation structure in the analysis are addressed. (45 minutes) |
| 12:20 pm-12:35 pm |
Question and Answer Session Moderator: Dr. Sandra Mitchell |
Session 6 (continued)12:35 pm-1:15 pm |
Analysis of Cluster Randomized Trials (continued) |
|
S6.b Presenter: |
Advanced Methods of Analysis for Stepped-Wedge Cluster Randomized Trials This session addresses advanced methods of analysis for SW-CRTs including methods for binary outcomes and various extensions of mixed-effects regression to accommodate deviations from model assumptions (e.g., time-varying treatment effects and cluster-treatment effect heterogeneity). Alternatives to mixed effects regression are introduced. (37 minutes) |
| 1:15 pm-1:30 pm |
Question and Answer Session Moderator: Dr. Wynne Norton |
| 1:30 pm-2:00 pm | BREAK |
Session 6 (continued)2:00 pm-2:35 pm |
Analysis of Cluster Randomized Trials (continued) |
|
S6.c Presenter: |
Estimation of Intra-Cluster Correlations to Inform the Design of Cluster Randomized Trials This session addresses the estimation of intra-cluster correlations to inform the design of parallel and multiple-period cluster trials, illustrating the concepts using Stata software. (18 minutes) |
|
S6.d Presenter: |
Estimation of Intra-Cluster Correlations for Multiple Period Cluster Randomized Trials Using SAS This session addresses the estimation of intra-cluster correlations to inform the design of parallel and multiple-period cluster trials, illustrating the concepts using SAS software. (8 minutes) |
| 2:35 pm-2:50 pm |
Question and Answer Session Moderator: Dr. Wynne Norton |
Session 72:50 pm-4:30 pm |
Data Monitoring and ReportingThis session introduces researchers to the unique considerations in monitoring and reporting CRTs and summarizes the concepts of importance when critically appraising studies using these designs. This material will help researchers appropriately consider the ethical implications of their chosen study design, how and when consent should be obtained, and the importance of trial registration. |
|
S7.a Presenter: |
Reporting of Cluster Randomized Trials This session discusses the reporting guideline extensions for CRTs and SW-CRTs. (25 minutes) |
|
S7.b Presenter: |
Ethical Issues in Cluster Randomized Trials This session provides an overview of the Ottawa Statement on the Ethical Design and Conduct of Cluster Randomized Trials, including the importance of justifying the use of cluster randomization, identifying research participants, and addressing informed consent and human subjects protections. (33 minutes) |
|
S7.c Presenter: |
Data Monitoring in Cluster Randomized Trials This session addresses considerations relevant to data monitoring in CRTs including interim analysis, data quality checks, tests for selection bias and sample size re-estimation. (38 minutes) |
| 4:30 pm-4:55 pm |
Question and Answer Session Moderator: Dr. Sandra Mitchell |
| 5:00 pm |
Course Wrap Up Dr. Sandra Mitchell and Dr. Wynne Norton |