Design and Analysis of Pragmatic Trials

Cover
Half Title
Series Page
Title Page
Copyright Page
Table of Contents
Author Biographies
Chapter 1 Pragmatic Randomized Trials
	1.1 Introduction
		1.1.1 Statistical Issues in Pragmatic Randomization Trials
	1.2 Cluster Randomized Designs
		1.2.1 Completely Randomized Cluster Trial Design
		1.2.2 Restricted Randomized Designs: Strategies to Improve Efficiency for CRTs
			1.2.2.1 Matched-Pair Cluster Randomized Design
			1.2.2.2 Stratified Cluster Randomized Design
			1.2.2.3 Covariate-Constrained Randomized Design
		1.2.3 Multiple-Period Cluster Randomized Designs
			1.2.3.1 Longitudinal Cluster Randomized Design
			1.2.3.2 Crossover Cluster Randomized Design
			1.2.3.3 Stepped-Wedge Cluster Randomized Design
	References
Chapter 2 Cluster Randomized Trials
	2.1 Introduction
	2.2 Continuous Outcomes
		2.2.1 Standard Two-Sample t-Test
		2.2.2 Adjusted Two-Sample t-Test
		2.2.3 Generalized Estimating Equation Method
		2.2.4 Mixed-Effects Linear Regression Models
	2.3 Binary Outcomes
		2.3.1 Standard Pearson Chi-Square Test
		2.3.2 Adjusted Chi-Square Test
		2.3.3 Ratio Estimator Chi-Square Test
		2.3.4 Generalized Estimating Equation Approach
		2.3.5 Generalized Linear Mixed Model Approach
	2.4 Count Outcomes
		2.4.1 Adjusted Normality Test
		2.4.2 Ratio Estimator Method
		2.4.3 Generalized Estimating Equation
	2.5 Cluster Size Determination for a Fixed Number of Clusters
	References
Chapter 3 Matched-Pair Cluster Randomized Design for Pragmatic Studies
	3.1 Introduction of Matched-Pair Cluster Randomized Design
	3.2 Considerations for Pragmatic Matched-Pair CRTs
		3.2.1 Impact of Correlation
		3.2.2 Impact of Missing Data
	3.3 Matched-Pair Cluster Randomized Design with Missing Continuous Outcomes
		3.3.1 Sample Size Estimation Based on GEE Approach
		3.3.2 Relative Efficiency of GEE Approach vs Crude Adjustment
		3.3.3 Adjustment of Inflated Type I Error
		3.3.4 Sensitivity Analysis
		3.3.5 Example
	3.4 Matched-Pair Cluster Randomized Design with Missing Binary Outcomes
		3.4.1 Sample Size Estimation Based on GEE Approach
		3.4.2 Relative Efficiency of GEE Approach vs Crude Adjustment
		3.4.3 Adjustment of Inflated Type I Error and Sensitivity Analysis
		3.4.4 Example
	3.5 Further Readings
	Appendix
		A.1 Eigenvalues of the Correlation Matrix
		A.2 Proof of Theorem 1
	References
Chapter 4 Stratified Cluster Randomized Design for Pragmatic Studies
	4.1 Introduction of Stratified Cluster Randomized Design
	4.2 Considerations for Pragmatic Stratified CRTs
	4.3 Stratified Cluster Randomized Design with Continuous Outcomes
		4.3.1 Sample Size Estimation Based on GEE Approach
		4.3.2 Relative Sample Size Change Due to Varying Cluster Size
		4.3.3 Example
	4.4 Stratified Cluster Randomized Design with Binary Outcomes
		4.4.1 Sample Size Estimation Based on CMH Statistic
		4.4.2 Relative Sample Size Change Due to Varying Cluster Size
		4.4.3 Estimation of Clustering Parameter
		4.4.4 Example
	4.5 Further Readings
	References
Chapter 5 The GEE Approach for Stepped-Wedge Trial Design
	5.1 Introduction
	5.2 A Brief Review of GEE
	5.3 Design SW trials with a Continuous Outcome
		5.3.1 Accounting for Missing Data
		5.3.2 Simulation Research
		5.3.3 Adjusting for Underestimated Variances for Small Sample Sizes
		5.3.4 Consideration of Efficiency and Robustness
	5.4 Design SW Trials with a Binary Outcome
		5.4.1 Extension to Outcomes from the Exponential Family
	5.5 Longitudinal and Crossover Cluster Randomized Trials
		5.5.1 Longitudinal cluster randomized trials
		5.5.2 Crossover Cluster Randomized Trials
		5.5.3 Comparison of  and
		5.5.4 Adjusting for Small Numbers of Clusters by the -Distribution
		5.5.5 Accounting for Randomly Varying Cluster Sizes
	Appendix A: Derivation of Equation (5.8)
	Appendix B: Derivation of Equations (5.9) and (5.10)
	Appendix C: Sample Size for Cross-Sectional SW Trials
	Appendix D: Derivation of Equation (5.15)
	Appendix E: Proof of Theorem 1
	Appendix F: Proof of Theorem 2
	References
Chapter 6 The Mixed-Effect Model Approach and Adaptive Strategies for Stepped-Wedge Trial Design
	6.1 A Brief Review of Mixed-Effect Models
	6.2 Sample Size Calculation Based on Cluster-Step Means
		6.2.1 Commonly Used Correlation Structures
			6.2.1.1 Exchangeable Correlation Structure
			6.2.1.2 Nested Exchangeable Correlation Structure
			6.2.1.3 Block Exchangeable Correlation Structure
			6.2.1.4 Exponential Decay Correlation Structure
			6.2.1.5 Proportional Decay Correlation Structure
	6.3 Adaptive Strategies for SW Trials
		6.3.1 Group Sequential Design for SW Trials
			6.3.1.1 Bayesian Adaptive Design for SW Trials
	6.4 Further Readings
	References
Index