Real World Health Care Data Analysis: Causal Methods and Implementation Using SAS®: Causal Methods and Implementation Using SAS®

Contents
About the Book
	What Does This Book Cover?
	Is This Book for You?
	What Should You Know about the Examples?
		Software Used to Develop the Book’s Content
		Example Code and Data
	Acknowledgments
	We Want to Hear from You
About the Authors
Chapter 1: Introduction to Observational and Real World Evidence Research
	1.1 Why This Book?
	1.2 Definition and Types of Real World Data (RWD)
	1.3 Experimental Versus Observational Research
	1.4 Types of Real World Studies
		1.4.1 Cross-sectional Studies
		1.4.2 Retrospective or Case-control Studies
		1.4.3 Prospective or Cohort Studies
	1.5 Questions Addressed by Real World Studies
	1.6 The Issues: Bias and Confounding
		1.6.1 Selection Bias
		1.6.2 Information Bias
		1.6.3 Confounding
	1.7 Guidance for Real World Research
	1.8 Best Practices for Real World Research
	1.9 Contents of This Book
	References
Chapter 2: Causal Inference and Comparative Effectiveness: A Foundation
	2.1 Introduction
	2.2 Causation
	2.3 From R.A. Fisher to Modern Causal Inference Analyses
		2.3.1 Fisher’s Randomized Experiment
		2.3.2 Neyman’s Potential Outcome Notation
		2.3.3 Rubin’s Causal Model
		2.3.4 Pearl’s Causal Model
	2.4 Estimands
	2.5 Totality of Evidence: Replication, Exploratory, and Sensitivity Analyses
	2.6 Summary
	References
Chapter 3: Data Examples and Simulations
	3.1 Introduction
	3.2 The REFLECTIONS Study
	3.3 The Lindner Study
	3.4 Simulations
	3.5 Analysis Data Set Examples
		3.5.1 Simulated REFLECTIONS Data
		3.5.2 Simulated PCI Data
	3.6 Summary
	References
Chapter 4: The Propensity Score
	4.1 Introduction
	4.2 Estimate Propensity Score
		4.2.1 Selection of Covariates
		4.2.2 Address Missing Covariates Values in Estimating Propensity Score
		4.2.3 Selection of Propensity Score Estimation Model
		4.2.4 The Criteria of “Good” Propensity Score Estimate
	4.3 Example: Estimate Propensity Scores Using the Simulated REFLECTIONS Data
		4.3.1 A Priori Logistic Model
		4.3.2 Automatic Logistic Model Selection
		4.3.3 Boosted CART Model
	4.4 Summary
	References
Chapter 5: Before You Analyze – Feasibility Assessment
	5.1 Introduction
	5.2 Best Practices for Assessing Feasibility: Common Support
		5.2.1 Walker’s Preference Score and Clinical Equipoise
		5.2.2 Standardized Differences in Means and Variance Ratios
		5.2.3 Tipton’s Index
		5.2.4 Proportion of Near Matches
		5.2.4 Proportion of Near Matches
		5.2.5 Trimming the Population
	5.3 Best Practices for Assessing Feasibility: Assessing Balance
		5.3.1 The Standardized Difference for Assessing Balance at the Individual Covariate Level
		5.3.2 The Prognostic Score for Assessing Balance
	5.4 Example: REFLECTIONS Data
		5.4.1 Feasibility Assessment Using the Reflections Data
		5.4.2 Balance Assessment Using the Reflections Data
	5.5 Summary
	References
Chapter 6: Matching Methods for Estimating Causal Treatment Effects
	6.1 Introduction
	6.2 Distance Metrics
		6.2.1 Exact Distance Measure
		6.2.2 Mahalanobis Distance Measure
		6.2.3 Propensity Score Distance Measure
		6.2.4 Linear Propensity Score Distance Measure
		6.2.5 Some Considerations in Choosing Distance Measures
	6.3 Matching Constraints
		6.3.1 Calipers
		6.3.2 Matching With and Without Replacement
		6.3.3 Fixed Ratio Versus Variable Ratio Matching
	6.4 Matching Algorithms
		6.4.1 Nearest Neighbor Matching
		6.4.2 Optimal Matching
		6.4.3 Variable Ratio Matching
		6.4.4 Full Matching
		6.4.5 Discussion: Selecting the Matching Constraints and Algorithm
	6.5 Example: Matching Methods Applied to the Simulated REFLECTIONS Data
		6.5.1 Data Description
		6.5.2 Computation of Different Matching Methods
		6.5.3 1:1 Nearest Neighbor Matching
		6.5.4 1:1 Optimal Matching with Additional Exact Matching
		6.5.5 1:1 Mahalanobis Distance Matching with Caliper
		6.5.6 Variable Ratio Matching
		6.5.7 Full Matching
	6.6 Discussion Topics: Analysis on Matched Samples, Variance Estimation of the Causal Treatment Effect, and Incomplete Matching
	6.7 Summary
	References
Chapter 7: Stratification for Estimating Causal Treatment Effects
	7.1 Introduction
	7.2 Propensity Score Stratification
		7.2.1 Forming Propensity Score Strata
		7.2.2 Estimation of Treatment Effects
	7.3 Local Control
		7.3.1 Choice of Clustering Method and Optimal Number of Clusters
		7.3.2 Confirming that the Estimated Local Effect-Size Distribution Is Not Ignorable
	7.4 Stratified Analysis of the PCI15K Data
		7.4.1 Propensity Score Stratified Analysis
		7.4.2 Local Control Analysis
	7.5 Summary
	References
Chapter 8: Inverse Weighting and Balancing Algorithms for Estimating Causal Treatment Effects
	8.1 Introduction
	8.2 Inverse Probability of Treatment Weighting
	8.3 Overlap Weighting
	8.4 Balancing Algorithms
	8.5 Example of Weighting Analyses Using the REFLECTIONS Data
		8.5.1 IPTW Analysis Using PROC CAUSALTRT
		8.4.2 Overlap Weighted Analysis using PROC GENMOD
		8.4.3 Entropy Balancing Analysis
	8.5 Summary
	References
Chapter 9: Putting It All Together: Model Averaging
	9.1 Introduction
	9.2 Model Averaging for Comparative Effectiveness
		9.2.1 Selection of Individual Methods
		9.2.2 Computing Model Averaging Weights
		9.2.3 The Model Averaging Estimator and Inferences
	9.3 Frequentist Model Averaging Example Using the Simulated REFLECTIONS Data
		9.3.1 Setup: Selection of Analytical Methods
		9.3.2 SAS Code
		9.3.3 Analysis Results
	9.4 Summary
	References
Chapter 10: Generalized Propensity Score Analyses (> 2 Treatments)
	10.1 Introduction
	10.2 The Generalized Propensity Score
		10.2.1 Definition, Notation, and Assumptions
		10.2.2 Estimating the Generalized Propensity Score
	10.3 Feasibility and Balance Assessment Using the Generalized Propensity Score
		10.3.1 Extensions of Feasibility and Trimming
		10.3.2 Balance Assessment
	10.4 Estimating Treatment Effects Using the Generalized Propensity Score
		10.4.1 GPS Matching
		10.4.2 Inverse Probability Weighting
		10.4.3 Vector Matching
	10.5 SAS Programs for Multi-Cohort Analyses
	10.6 Three Treatment Group Analyses Using the Simulated REFLECTIONS Data
		10.6.1 Data Overview and Trimming
		10.6.2 The Generalized Propensity Score and Population Trimming
		10.6.3 Balance Assessment
		10.6.4 Generalized Propensity Score Matching Analysis
		10.6.5 Inverse Probability Weighting Analysis
		10.6.6 Vector Matching Analysis
	10.7 Summary
	References
Chapter 11: Marginal Structural Models with Inverse Probability Weighting
	11.1 Introduction
	11.2 Marginal Structural Models with Inverse Probability of Treatment Weighting
	11.3 Example: MSM Analysis of the Simulated REFLECTIONS Data
		11.3.1 Study Description
		11.3.2 Data Overview
		11.3.3 Causal Graph
		11.3.4 Computation of Weights
		11.3.5 Analysis of Causal Treatment Effects Using a Marginal Structural Model
	11.4 Summary
	References
Chapter 12: A Target Trial Approach with Dynamic Treatment Regimes and Replicates Analyses
	12.1 Introduction
	12.2 Dynamic Treatment Regimes and Target Trial Emulation
		12.2.1 Dynamic Treatment Regimes
		12.2.2 Target Trial Emulation
	12.3 Example: Target Trial Approach Applied to the Simulated REFLECTIONS Data
		12.3.1 Study Question
		12.3.2 Study Description and Data Overview
		12.3.3 Target Trial Study Protocol
		12.3.4 Generating New Data
		12.3.5 Creating Weights
		12.3.6 Base-Case Analysis
		12.3.7 Selecting the Optimal Strategy
		12.3.8 Sensitivity Analyses
	12.4 Summary
	References
Chapter 13: Evaluating the Impact of Unmeasured Confounding in Observational Research
	13.1 Introduction
	13.2 The Toolbox: A Summary of Available Analytical Methods
	13.3 The Best Practice Recommendation
	13.4 Example Data Analysis Using the REFLECTIONS Study
		13.4.1 Array Approach
		13.4.2 Propensity Score Calibration
		13.4.3 Rosenbaum-Rubin Sensitivity Analysis
		13.4.4 Negative Control
		13.4.5 Bayesian Twin Regression Modeling
	13.5 Summary
	References
Chapter 14: Using Real World Data to Examine the Generalizability of Randomized Trials
	14.1 External Validity, Generalizability and Transportability
	14.2 Methods to Increase Generalizability
	14.3 Generalizability Re-weighting Methods for Generalizability
		14.3.1 Inverse Probability Weighting
		14.3.2 Entropy Balancing
		14.3.3 Assumptions, Best Practices, and Limitations
	14.4 Programs Used in Generalizability Analyses
	14.5 Analysis of Generalizability Using the PCI15K Data
		14.5.1 RCT and Target Populations
		14.5.2 Inverse Probability Generalizability
		14.5.3 Entropy Balancing Generalizability
	14.6 Summary
	References
Chapter 15: Personalized Medicine, Machine Learning, and Real World Data
	15.1 Introduction
	15.2 Individualized Treatment Recommendation
		15.2.1 The Individualized Treatment Recommendation Framework
		15.2.2 Estimating the Optimal Individualized Treatment Rule
		15.2.3 Multi-Category ITR
	15.3 Programs for ITR
	15.4 Example Using the Simulated REFLECTIONS Data
	15.5 “Most Like Me” Displays: A Graphical Approach
		15.5.1 Most Like Me Computations
		15.5.2 Background Information: LTD Distributions from the PCI15K Local Control Analysis
		15.5.3 Most Like Me Example Using the PCI15K Data Set
		15.5.4 Extensions and Interpretations of Most Like Me Displays
	15.6 Summary
	References
Index
	A
		B
		C
		D
		E
		F
		G
		H
		I
		K
		L
		M
		N
		O
		P
		Q
		R
		S
		T
		U
		V
		W
		X
		Y
		Z